JP2003174366A - Data decoding apparatus and method - Google Patents

Abstract

(57) Abstract: Provided is a data decoding device and method capable of decoding a JPEG-encoded data stream at a high speed with a simple configuration. A first decoding unit (20) includes a data stream (D).
All the Huffman codes included at the beginning of “in” can be decoded, and the second decoding units 41 to 44 have decoding tables that can decode Huffman codes up to 5 bits in length. According to the data length L1 of the first encoded data, one of a plurality of second decoding units whose data position to be decoded is shifted by one bit is selected, and the decoded data RS
In accordance with 2, the second encoded data length L2 is calculated. The data extraction position is shifted according to the sum of the data length L2 and the data length L1. When the first encoded data length is other than 3 to 6 bits, or when the second Huffman code is 5
If the number of bits is greater than or equal to the number of bits, a signal Siv indicating that the second decoded data RS2 is invalid is generated.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data decoding apparatus and method for decoding a data stream coded by a variable length code such as Huffman code, for example, by a JPEG (joint photographic experts group) method. The present invention relates to a data decoding device and a method for decoding an encoded data stream.

[0002]

2. Description of the Related Art FIG. 17 is a block diagram showing a configuration example of a conventional data decoding apparatus for decoding a data stream coded by Huffman code. The data decoding device shown in FIG. 17 includes a data shift unit 1, a decoding unit 2a ...
It has a decoding unit 2r and a selector 3.

The data shift unit 1 extracts a 16-bit data string S1 from the first bit of the Huffman-encoded data stream Din, and decodes the decoding units 2a to 2r.
To enter. In addition, the extraction position of the data string is shifted from the beginning to the end of the data stream Din according to the shift signal S3.

The decoding unit 2a has a data table in which the Huffman code, its data length, and the decoded word are associated with each other. With reference to this data table, 16 bits from the first bit of the extracted data sequence S1 The Huffman code included in the data string is searched, and the data length L1 of the Huffman code and the decoded word Dout1 are output. However, in this example, the Huffman code has a maximum length of 16 bits.

The decoding units 2b to 2r also have the same data table as the decoding unit 2a, and the extracted data string S1.
The Huffman code included in the 16-bit data string is searched from the bit positions deviated from the first bit by 1 to 16 bits. Then, the decoded word and the data length of the retrieved Huffman code are output to the selector 3. However, the data lengths output from the decoding units 2b to 2r have a value obtained by adding the data length of the searched Huffman code and the data length L1 when each of the decoding units is selected by the selector 3. ing. For example, in the decoding unit 2g, the data length p
When the Huffman code is searched for, the data length output from the decoding unit 2g is (p + 6).

The selector 3 decodes the decoding unit 2b according to the data length L1 of the Huffman code retrieved by the decoding unit 2a.
~ Select any one of the decoding units 2r and output the decoded word Dout2. The data length is output to the data shift unit 1 as the shift signal S3.

According to the data decoding apparatus of FIG. 17 having the above-described configuration, first, the data shift unit 1 extracts the data sequence S1 of 32 bits from the first bit of the data stream Din, and further extracts 16 bits from the first bit.
The Huffman code included in the data string of bits is the decoding unit 2
Searched by a. As a result, the extracted data string S1
The first Huffman code is found from the beginning, and the decoded word Dout1 and the data length L1 thereof are output from the decoding unit 2a.

Since the position of the head bit of the second Huffman code can be known from the data length L1, the selector 3 selects the decoding unit which searches for the Huffman code with this bit position as the head. For example, data length L
When 1 is 7 bits, the decoding unit 2h that searches for a Huffman code by starting the 8th bit of the extracted data sequence S1
Is selected. In this way, the second Huffman code is searched, and the decoded word Dout2 is output from the selector 3.

The shift signal S3 output from the selector 3
Is a signal indicating the sum of the data lengths of the first Huffman code and the second Huffman code, and the first bit of the next Huffman code following the Huffman code decoded second by this shift signal S3. The position is specified. Since the data extraction position of the data shift unit 1 is shifted by the shift signal S3 so as to coincide with the specified head bit position, two Huffman codes are decoded from the head bit in the same manner as described above. By repeating such operations, two Huffman codes are simultaneously decoded for one shift of the data extraction position.

[0010]

However, as shown in FIG.
In the data decoding device shown in FIG. 1, the decoding unit 2a to the decoding unit 2r are included.
In, since a data table for decoding the Huffman code is individually provided, a total of 17 data tables are required. Therefore, there is a problem that the storage capacity for this data table becomes very large and the circuit scale increases. In addition to the Huffman code, a DCT (discrete cosine transfor) called additional bit is added to the data stream compressed and encoded by the JPEG method.
m) Since the data for designating the concrete value of the coefficient is included, the data decoding apparatus of FIG. 17 which considers only the decoding of the Huffman code cannot be applied to the JPEG decoding apparatus.

The present invention has been made in view of the above circumstances, and an object thereof is to provide a data decoding apparatus and method capable of decoding a variable-length coded data stream at high speed with a simple structure. It is in.

[0012]

In order to achieve the above object, a data decoding device according to a first aspect of the present invention is a data decoding device for decoding a supplied encoded data stream. , A data extracting means for extracting a data string of a predetermined data length from the head of the encoded data stream, and a variable length code included in a predetermined variable length code table from the head of the data string extracted by the data extracting means. First decoding means for searching and outputting decoded data corresponding to the searched variable length code, decoded data of the first decoding means, and data of the variable length code corresponding to the decoded data A first data length calculating means for calculating the data length of the encoded data corresponding to the decoded data based on the length and outputting it as the first data length; and the data extracting means. A part of the variable-length code in the variable-length code table is searched from the data positions shifted toward the end by a plurality of different data numbers with respect to the head of the extracted data string, and the searched variable-length code is searched. Of the plurality of second decoding means for outputting the decoded data corresponding to the plurality of second decoding means for outputting the search error signal when the corresponding variable length code is not found in the search. From the second selected according to the first data length
The selecting means outputs the decoded data of the decoding means or the search error signal, and outputs the selection error signal when there is no corresponding second decoding means, and the second decoding means selected by the selecting means. A data length of the encoded data corresponding to the decoded data is calculated based on the decoded data and a data length of the variable length code corresponding to the decoded data, and is output as a second data length. 2 data length calculating means and the second data selected by the selecting means.
When the search error signal is output from the decoding means of
Alternatively, when the selection error signal is output from the selection means, an invalid signal generation means for generating an invalid signal for notifying that the decoded data of the second decoding means is invalid, and the first data length. And the second data length, the data string extraction position of the data extraction means is shifted to the end side of the encoded data stream by the number of data, and the invalid signal is generated, First above
Data extraction position shift means for shifting the extraction position to the end side by the number of data corresponding to the data length of.

According to the data decoding apparatus of the first aspect of the present invention, the data extracting means extracts a data string of a predetermined data length from the beginning of the encoded data stream, and the extracted data string is extracted. The variable length code included in the predetermined variable length code table is searched by the first decoding means from the beginning. Then, the decoded data corresponding to the retrieved variable length code is output. Based on the output decoded data of the first decoding means and the data length of the variable length code corresponding to the decoded data, the data length of the encoded data corresponding to the decoded data is the first Is calculated by the data length calculation means and is output as the first data length. Further, a part of the variable length code in the variable length code table is converted into a second decoding means from the data position shifted toward the end side by a plurality of different data numbers with respect to the head of the data string extracted by the data extracting means. And the decoded data corresponding to the searched variable length code is output. When the corresponding variable length code is not found in the search, a search error signal is output. From the plurality of second decoding means, the decoded data of the second decoding means selected according to the first data length or the search error signal is output from the selection means. If there is no corresponding second decoding means, a selection error signal is output. The decrypted data of the second decrypting means selected by the selecting means,
The data length of the encoded data corresponding to the decoded data is calculated by the second data length calculating means based on the data length of the variable length code corresponding to the decoded data, and is calculated as the second data length. Is output. When the search error signal is output from the second decoding means selected by the selecting means, or when the selection error signal is output from the selecting means, the decoded data of the second decoding means Is generated by the invalid signal generating means for notifying that the is invalid. The data string extraction position of the data extraction means is moved to the end side of the encoded data stream by the data extraction position shift means by the number of data corresponding to the sum of the first data length and the second data length. Be shifted. When the invalid signal is generated, the extraction position is shifted to the end side by the number of data corresponding to the first data length.

Further, the encoded data stream includes encoded data composed of the variable length code and additional data, and the data length of the additional data included in the decoded data of the first decoding means. First additional data extracting means for further extracting the additional data from the extracted data string of the data extracting means based on the information of 1. and data of the additional data included in the decoded data output from the selecting means. It may have a second additional data extracting means for further extracting the additional data from the extracted data string of the data extracting means based on the length information.

Further, according to the information of the continuous number contained in the decoded data of the first decoding means and the decoded data of the second decoding means selected by the selecting means, within the unit block. The number of the decoded data is counted, and when the invalid signal is generated, the information of the continuous number included in the decoded data of the selected second decoding means is invalidated and the counting is performed. If the count value reaches the predetermined number of data, a counting means for initializing the count value may be provided. The counting means outputs a counting error signal when the count value is larger than the predetermined number of data, and the invalid signal generating means adds the end of the unit block to the decoded data of the first decoding means. The invalid signal may be generated when the predetermined information shown is included, or when the counting error signal is output from the counting means.

Further, from at least a part of the plurality of second decoding means, the first data length to the nth data length (the code n is 1≤n≤N with respect to a predetermined natural number N). The decoded data of the second decoding means selected according to the sum of (which represents an arbitrary natural number to be satisfied) and the search error signal are output, and if there is no corresponding second decoding means, the nth
A plurality of selecting means for outputting the selection error signals, the decoded data of the second decoding section selected by the selecting means according to the sum of the first data length to the nth data length, and the decoding data. A plurality of second data that calculates the data length of the encoded data corresponding to the decoded data based on the data length of the variable length code corresponding to the encoded data and outputs the data length as the (n + 1) th data length. When the search error signal is output from the length calculating means and the second decoding means selected by the selecting means according to the sum of the first data length to the n-th data length, or from the selecting means. When the n-th selection error signal is output, the first data length to the m-th data length (the symbol m represents an arbitrary natural number satisfying n ≦ m ≦ N with respect to a predetermined natural number N). ) To the sum of A plurality of invalid signal generating means for generating an n-th invalid signal informing that all the decoded data of the second decoding means selected by the second decoding means are invalid. The data string extraction position of the data extraction means is shifted to the end side of the encoded data stream by the number of data corresponding to the sum of the first data length to the Nth data length, and the nth invalid signal is generated. In this case, the extraction position may be shifted to the end side by the number of data corresponding to the sum of the first data length to the nth data length. In this case, further, in accordance with the information of the continuous number included in the decoded data of the first decoding means and the decoded data of the second decoding means selected by the plurality of selecting means, the unit block If the number n of the decoded data is counted and the nth invalid signal is generated, the second decoding selected according to the sum of the first data length to the mth data length. It has a counting means for invalidating all the information of the continuous number contained in the decrypted data of the means to perform the counting, and when the counted value reaches the predetermined number of data, initialize the counted value. May be. In addition, the counting means selects the second data length selected according to the sum of the first data length to the m-th data length when the count value is larger than the predetermined data number.
In the minimum natural number n in which the count value obtained by invalidating all the information of the continuous number included in the decoded data of the decoding means is larger than the predetermined data number, the nth counting error signal is output and the invalid signal is generated. The means generates a first invalid signal when the decoded data of the first decoding means includes predetermined information indicating the end of the unit block, and outputs a first data length to an nth data length. When the decoded data of the second decoding means selected according to the total includes the predetermined information, the (n + 1) th invalid signal is generated, and the counting means outputs the nth counting error signal. In this case, the nth invalid signal may be generated.

A data decoding method according to a second aspect of the present invention is a data decoding method for decoding a coded data stream supplied, wherein the data having a predetermined data length from the beginning of the coded data stream. A step of extracting a column, and a variable length code included in a predetermined variable length code table is searched from the beginning of the extracted data string, and first decoded data corresponding to the searched variable length code is generated. A step, and the first decrypted data generated above,
Calculating the data length of the encoded data corresponding to the first decoded data as the first data length based on the data length of the variable length code corresponding to the first decoded data, Retrieving a part of the variable-length codes of the first variable-length code table from the data positions shifted toward the end by a plurality of different numbers of data with respect to the beginning of the data string extracted in the data extraction step, A second decoding step of generating second decoded data corresponding to the searched variable length code, and generating a search error signal when the corresponding variable length code is not found in the search; A second recovery code generated by searching the variable length code at a data position corresponding to the first data length from a plurality of second decoded data or search error signals. The selected decoded data or the search error signal, and generating the selected error signal when there is no corresponding second decoded data or the search error signal, the selected second decoded data, and the selected second decoded data. Calculating the data length of the encoded data corresponding to the second decoded data as the second data length based on the data length of the variable length code corresponding to the second decoded data; Generating an invalid signal for notifying that the second decoded data is invalid when the search error signal is selected in the step or when the selection error signal is generated; The data string extraction position of the data extraction step is shifted to the end side of the encoded data stream by the number of data corresponding to the sum of the data length and the second data length. Is bets, when the disable signal is generated, and a step of shifting only the extraction position number data corresponding to the first data length to said trailing side.

A variable length code at a data position corresponding to the sum of the first data length to the nth data length from at least a part of the plurality of generated second decoded data or search error signals. Select the second decoded data or search error signal generated by the search of
If there is no corresponding second decoded data or search error signal, a step of generating an nth selection error signal and a sum of the first data length to the nth data length in the selecting step are performed. When the search error signal is selected, the sum of the first data length to the m-th data length (the code m represents an arbitrary natural number satisfying n ≦ m ≦ N with respect to a predetermined natural number N) is added. A step of generating an n-th invalid signal for notifying that all of the second decoded data selected accordingly are invalid, and selected according to the sum of the first data length to the n-th data length. Second
Of the decoded data and the data length of the variable length code corresponding to the second decoded data, the data length of the coded data corresponding to the second decoded data is set to (n
+1) as a data length, and the step of shifting the extraction position of the data is repeated for a predetermined natural number N within a range satisfying 1 ≦ n ≦ N. When the n-th invalid signal is generated by shifting the extraction position to the end side of the encoded data stream by the number of data corresponding to the sum of the N data lengths, the first data length or The extraction position may be shifted to the end side by the number of data corresponding to the sum of the nth data lengths.

[0019]

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, two embodiments of the present invention will be described by taking as an example the case where the present invention is applied to a JPEG decoding device for decoding a JPEG encoded data stream. In order to facilitate understanding of the above description, an outline of the JPEG encoding method will be briefly described first.

In a natural image, adjacent pixel values often have a correlation, and there is a property that the spatial variation (spatial frequency) of the pixel value when viewed in a certain minute area becomes small. When such a natural image is transformed into data in the spatial frequency domain by orthogonal transformation, the transformed data tends to have a low frequency component larger than a high frequency component. Therefore, when encoding data in the spatial frequency domain, if the code length assigned to the data on the high frequency side is made shorter than the data on the low frequency side, the average code length of the entire data becomes shorter, so the amount of information is compressed. be able to.

In the JPEG DCT-based coding system,
The input image data is divided into blocks of 8 × 8 pixels, and DC which is one of orthogonal transforms is applied to each block.
T (Discrete Cosine Transform)
Is executed. As a result, one block of 64 image data is converted into 64 DCT coefficients. This DCT
The coefficient is quantized by a quantization step determined for each coefficient, and then one DC component and the other 63 AC components are encoded into Huffman codes by different methods.

When the DCT coefficient is the DC component, a Huffman code determined according to the magnitude of the DCT coefficient difference between consecutive blocks is generated.
This is the D between adjacent blocks in a natural image.
This is because the C component is often correlated and the variance of the difference value is small. The DC component Huffman code corresponds to 4-bit data called a category SSSS, and additional bits having a data length designated by the category SSSS are added subsequent to the Huffman code to form one encoding. The data is composed. A specific magnitude of the difference value is designated by the additional bit.

When the DCT coefficient is an AC component, a data string of DCT coefficients rearranged in an order corresponding to the magnitude of the spatial frequency is created, and a zero-value DCT is generated in this data string.
The number of consecutive coefficients (zero run length RRRR),
One Huffman code is generated according to the combination with the category SSSS determined according to the magnitude of the non-zero DCT coefficient following the zero-value DCT coefficient. This is to improve the compression efficiency by utilizing the property that the quantized DCT transform coefficient tends to be zero on the high frequency side.

It should be noted that in the above data string, a zero value A
When 16 or more C components continue, one code ZRL (ze
ro run length) is generated. When a zero-value AC component continues to the end of one block, one code EOB (end of block) is generated regardless of the number of AC components. In either case, compression efficiency is improved by collectively giving one code to a plurality of zero-valued AC components.

<First Embodiment> FIG. 1 is a block diagram showing a schematic configuration example of a JPEG decoding apparatus according to the present invention. The JPEG decoding device shown in FIG.
001, Huffman decoder 1002, inverse quantizer 100
3 and an inverse DCT device 1004.

The buffer 1001 is a JPEG input.
The encoded data is temporarily stored and sequentially output to the Huffman decoding device 1002 in response to a request. The Huffman decoder 1002 sequentially reads the JPEG encoded data accumulated in the buffer 1001, decodes the Huffman code, and reproduces the DCT coefficient. The inverse quantizer 1003 performs an inverse quantization process of multiplying the DCT coefficient reproduced by the Huffman decoder 1002 by a predetermined inverse quantization coefficient. The inverse DCT unit 1004 performs inverse DCT on the DCT coefficient inversely quantized by the inverse quantizer 1003 for each 8 × 8 pixel block, and reproduces each pixel data of the original image.

A more detailed configuration example of the Huffman decoder 1002 described above will be described. FIG. 2 shows the first of the present invention.
2 is a schematic block diagram showing a configuration example of a Huffman decoder 1002 of FIG. 1 according to the embodiment of FIG. The Huffman decoder shown in FIG. 2 includes a data extraction unit 10 and a first decoding unit 2.
0, the first data length calculation unit 30, the second decoding unit 41 to the second decoding unit 44, the selection unit 50, the second data length calculation unit 6
0, invalid signal generator 70, data extraction position shifter 8
0, counting unit 90, DCT coefficient output unit 101 and DCT
It has a coefficient output unit 102.

The data extraction unit 10 extracts a 27-bit data string from the beginning of the JPEG encoded data stream Din accumulated in the buffer 1001. The maximum length of Huffman code is 16 bits, and the maximum number of additional bits is 11
Since the number of bits is one, the maximum data length of one piece of JPEG encoded data composed of Huffman code and additional bits is 16 + 11 = 27 bits. Therefore, the data extraction unit 10 surely extracts one JPEG encoded data. The data extraction unit 10 also shifts the data extraction position from the beginning to the end of the JPEG encoded data stream Din according to the shift signal Sft.

The first decoding unit 20 inputs a 16-bit data string corresponding to the longest data of the Huffman code from the first bit of the data string Ds extracted by the data extraction unit 10, and the input data string. Search the Huffman code included in the given Huffman code table from
The corresponding decoded data RS1 is output. If the input encoded data is the DC component of the DCT coefficient, this D
Category S of additional bits that specify a specific value of the C component
The SSS is output as the decrypted data RS1. Also,
When the input encoded data is the AC component of the DCT coefficient, a zero run length RRRR that specifies the number of consecutive AC components with a value of zero, and the non-zero AC component that follows the zero-value AC component The category SSSS of the additional bit that specifies a specific value is output as the decoded data RS1. In this way, the first encoding unit 20 determines the DCT coefficient D
Both the C and AC components can be decoded.

The first decoding unit 20 retrieves a predetermined Huffman code from an input 16-bit data string and generates an address of a decoding table corresponding to the Huffman code, as shown in FIG. 2, for example. The section 201 and the category SS of the DC component stored at this generated address
It can be configured by the decoding table 202 that outputs the SS or the zero run length RRRR of the AC component and the category SSSS as the decoded data RS1.

The first data length calculation section 30 includes the decoded data RS1 of the first decoding section 20 and the decoded data RS.
Based on the data length of the Huffman code corresponding to 1, the data length of the encoded data (Huffman code + additional bits) corresponding to the decoded data RS1 is calculated and output as the data length L1.

The second decoding section 41 to the second decoding section 44 are
A data string of 5 bits is input from a data position shifted to the end side of the data stream by 3 to 6 bits with respect to the first bit of the data string Ds extracted by the data extraction unit 10, and the input data string is input. Each of the predetermined Huffman codes included in is searched. Then, the decoded data RS21-corresponding to the retrieved Huffman code
The decoded data RS24 is output. As described above, the Huffman code has a maximum data length of 16 bits, but the Huffman code that can be decoded by the second decoding unit has a maximum data length of 5 bits. The data table of 1 need only be a part of the data table of the first decoding unit 20 capable of decoding all Huffman codes.

Further, the second decoding unit 41 to the second decoding unit 4
4 is a search error signal Esr1 to a search error signal Esr4 when the data length of the Huffman code included in the input data string is longer than 5 bits and the corresponding Huffman code cannot be found from the data table.
Is output.

The selecting unit 50 selects one of the second decoding unit 41 to the second decoding unit 44 according to the data length L1.
, And outputs the decoded data Rs2 or the search error signal Esr. In the example of FIG. 2, the second decoding unit 41 is used when the data length L1 is 3 bits, the second decoding unit 42 is used when the data length L1 is 4 bits, and the second decoding unit 43 is used when the data length L1 is 5 bits. In the case of bits, the second decoding unit 44 is selected. If there is no corresponding second decoding unit, that is, if the data length L1 is not included in the range of 3 bits to 6 bits, the selection error signal Esel is output.

The second data length calculation unit 60 includes a selection unit 50.
The decoded data RS2 of the second decoding unit selected in
And the data length of the Huffman code corresponding to the decoded data RS2, the data length of the coded data corresponding to the decoded data RS2 is calculated and output as the data length L2.

The invalid signal generating section 70 decodes when the second decoding section selected by the selecting section 50 outputs the search error signal Esr or when the selecting section 50 outputs the selection error signal Esel. An invalidation signal Siv is generated to notify that the encoded data RS2 is invalid. Further, when the decoded data RS1 of the first decoding unit 20 includes the code EOB indicating the end of the unit block for encoding, or when the counting error signal E in the counting unit 90 described later is included.
Even when cnt is output, the invalid signal Siv is generated.

The data extraction position shift unit 80 shifts the shift signal Sft for shifting the data string extraction position of the data extraction unit 10 to the end side of the data stream by the number of data corresponding to the sum of the data length L1 and the data length L2. To generate. When the invalid signal Siv is generated, the data length L
The shift signal Sft for shifting the extraction position of the data extraction means 10 to the end side by the number of data corresponding to 1 is generated.

The counting section 90 includes a zero run length RRRR included in the decoded data RS1 and the decoded data RS2.
According to, the DCT coefficient of the decoded AC component in the unit block consisting of 64 DCT coefficients is counted. When the invalid signal Siv is generated in the invalid signal generator 70, the zero run length RRRR included in the decoded data RS2 is invalidated and the DCT coefficient of the AC component is counted. That is, the zero run length RRRR of the decoded data RS2 is not included in the count value. The count value of the AC component is 63 (64 DCT coefficients of the whole unit block-D
When the DC component of the C component reaches 1), the count value Scnt is initialized to zero and the AC component is counted again.

When the AC component count value exceeds 63, the count error signal Ecnt is output and the decoded data RS2 is invalidated. AC component count value is 6
When the number is larger than 3, the end of the unit block is reached by the DCT coefficient decoded by the first decoding unit 20, so the DCT coefficient decoded by the decoded data RS2 is the same as the head data of the next unit block. Become. However, since the head data of the unit block is the DC component, the second
The decoding units 41 to 44 cannot perform decoding. Therefore, in this case, the counting error signal Ecnt is output so that the decoded data RS2 becomes invalid.

The DCT coefficient output unit 101 extracts additional data from the extracted data string Ds based on the decoded data RS1, and extracts the DC component or AC from the extracted additional data.
The DCT coefficient D1 of the component is calculated and output. The DCT coefficient output unit 102 extracts additional data from the extracted data string Ds based on the decoded data RS2, calculates the DCT coefficient D2 of the AC component from the extracted additional data, and outputs it.

Here, in the Huffman decoder shown in FIG. 2, a concrete example of a DCT coefficient output section for outputting two DCT coefficients based on two pieces of decoded data (decoded data RS1 and decoded data RS2). A configuration example will be described with reference to FIG. In FIG. 3, reference numerals 1011 and 1021 denote additional data extraction units, reference numeral 1012.
Reference numeral 1022 denotes a DCT coefficient calculation unit, respectively.

The additional data extraction unit 1011 inputs all 27-bit data strings from the extracted data string Ds of the data extraction unit 10, and adds from this input data string based on the category SSSS included in the decoded data RS1. The data AD1 is extracted.

The additional data extraction unit 1021 inputs a data sequence of 13 bits from a bit position shifted by 3 bits with respect to the leading bit of the extraction data sequence Ds of the data extraction unit 10, and from this input data sequence, The additional data AD2 is extracted based on the category SSSS included in the decrypted data RS2. The longest data of the Huffman code decoded by the second decoding unit is 5 bits, and the longest data of the additional bits in this case is 5 bits. Therefore, the extraction range of the additional bits is the data input to the second decoding unit. The range is from the first bit of the column to 10 bits. Therefore,
The extraction range of the additional data AD2 is from the first bit of the data string input to the second decoding unit 41 to the second decoding unit 44.
The range of 13 bits to the end of the 10-bit data string input to is.

The DCT coefficient calculation unit 1012 calculates the DC component or AC based on the additional data AD1 and the decoded data RS1 extracted by the additional data extraction unit 1011.
Calculate the DCT coefficient of the component. The DC component is the additional data AD added to the DC component calculated in the previous unit block.
It is calculated by adding a positive or negative difference value obtained from 1 and the category SSSS. Also, the AC component of zero value is calculated according to the zero run length RRRR,
AC component of non-zero value is additional data AD2 and category SS
Calculate according to SS. DCT coefficient calculation unit 1022
Is AC based on the additional data AD2 and the decoded data RS2 extracted by the additional data extraction unit 1021.
Calculate the DCT coefficient of the component.

Next, the JPEG shown in FIGS. 1 to 3 described above.
Regarding the operation of the decoding device, the Huffman decoder 1002
I will explain mainly.

In FIG. 4, 2 extracted by the data extraction unit 10
It is a figure which shows the range of the coded data which can be decoded by the decoding part 20 and the decoding parts 41-44 in a 7-bit data string. As shown in FIG. 4, the first decoding unit 20
Can decode the encoded data from the first bit "0" to the last bit "26" of the extracted data string Ds. On the other hand, the second decoding unit 41 uses the extracted data string Ds
From bit '3' to bit '12', the second decoding unit 42 from bit '4' to bit '13', and the second decoding unit 43 from bit '5' to bit '14'.
The decoding unit 44 can decode the encoded data in the range from bit '6' to bit '15'. Which of the second decoding units the output is selected is determined according to the data length L1 of the first encoded data calculated by the first data length calculation unit 30.

FIG. 5 is a diagram showing an example of two pieces of encoded data included in the extracted data string Ds. In FIG.
From the first bit "0" to the bit "4" of the extracted data string Ds
Up to 5 bits are the first coded data 1, and 6 bits from bit "5" to bit "10" are the second coded data 2. Further, the first two bits of the encoded data 1 are Huffman codes and the latter three bits are additional bits, and the first three bits of the encoded data 2 are Huffman codes and the latter three bits are additional bits.

In this case, the 27-bit extraction data Ds extracted from the buffer 1001 by the data extraction unit 10
, A data string for 16 bits from the leading bit “0” is input to the first decoding unit 20, and the Huffman code is searched from the leading 2 bits of the input data string. And
Zero run length R corresponding to the retrieved Huffman code
The RRR and the category SSSS are output as the decoded data RS1.

Decoded data RS1 of the first decoding section 20
Is input to the DCT coefficient output unit 101, and the DCT coefficient D1 of the encoded data 1 is calculated based on the decoded data RS1 and the extracted data string Ds.

In the first data length calculation unit 30, the data length of the additional bit designated by the category SSSS of the decoded data RS1 and the data length of the Huffman code corresponding to the decoded data RS1 are summed up. , The data length L1 of the encoded data 1 is calculated. In the example of FIG. 5, 5 bits are calculated as the data length L1 from the sum of 2 bits of the Huffman code and 3 bits of the additional bits.

The calculated data length L1 is input to the selection unit 50, and the second decoding unit is selected according to the data length L1. In the example of FIG. 5, the second decoding unit 43 that searches for the Huffman code from the sixth bit “5” for the first bit “0” of the extracted data string Ds is selected. Then, the decoded data RS23 and the search error signal E
sr3 is output from the selection unit 50 as the decoded data RS2 and the search error signal Esr, respectively.

However, in the example of FIG. 5, since the data length L1 of the encoded data 1 is 5 bits and is included in the range of 3 bits to 6 bits, in this case, the selection error signal E
sel is not output. Further, since the Huffman code of the encoded data 2 has 3 bits and does not exceed 5 bits, the search error signal Esr3 is not output from the second decoding unit 43. Therefore, the search error signal Esr is not output from the selection unit 50.

The decoded data RS2 selected by the selection unit 50 is input to the DCT coefficient output unit 102, and based on this decoded data RS2 and the extracted data string Ds,
The DCT coefficient D2 of the encoded data 2 is calculated.

In the second data length calculation unit 60, according to the sum of the data length of the additional bit designated by the category SSSS of the decoded data RS2 and the data length of the Huffman code corresponding to the decoded data RS2. , The data length L2 of the encoded data 2 is calculated. In the example of FIG. 5, 6 bits are calculated as the data length L2 from the sum of 3 bits of the Huffman code and 3 bits of the additional bits.

Calculated data length L1 and data length L
2 are both input to the data extraction position shift unit 80,
When the invalid signal Siv is not generated, the shift signal Sft corresponding to the sum of these data lengths is generated. As a result, in the example of FIG. 5, bit “1” of the extracted data string Ds
The data extraction position is shifted to 1 ', and the next two pieces of encoded data are decoded from this position. When the invalid signal Siv is generated, since the decoded data RS2 of the encoded data 2 is invalid, the shift amount due to the data length L2 is ignored and the shift signal Sf corresponding to only the data length L1.
t is generated. As a result, in the example of FIG. 5, the data extraction position is shifted to bit “5” of the extraction data string Ds.

The decoded data RS1 output from the decoding unit 20 and the decoded data RS2 output from the selecting unit 50 are both input to the counting unit 90, and the zero run length RRRR included in these decoded data is included. From, the number of decoded AC components in the unit block is counted. For example, when the zero run length RRRR is 7, since seven zero value and one non-zero value AC components are decoded, the decoding number 8 is added to the count value Ecnt. Such a decoding number is calculated for each zero run length RRRR of the decoded data RS1 and the decoded data RS2, and added to the count value Ecnt. Count value Scn
When t exceeds 63, the encoded data 2 is the head data of the unit block, and therefore the counting error signal Ecnt for invalidating the decoded data RS2 is output.

In the example of FIG. 5, neither the selection error signal Esel nor the search error signal Esr is output, so that the invalid signal Siv is not generated. When the decoded data RS1 includes the code EOB indicating the end of the unit block, or when the counting unit 90 causes the counting error signal E
When cnt is output, the invalid signal Siv is generated. When the invalid signal Siv is generated, the DCT coefficient D of the AC component decoded based on the decoded data RS2
2, the data length L2, the count value Scn in the counting unit 90
t and the like are processed as invalid data.

In FIG. 6, since the Huffman code length included in the encoded data 2 exceeds 5 bits, the search error signal Es
An example in which r is output is shown. In FIG. 6, five bits from the first bit “0” to the bit “4” of the extracted data string Ds is the first encoded data 1, and 14 bits from the bit “5” to the bit “18” are two. It is the second encoded data 2. The first half 2 bits of the encoded data 1 are Huffman codes and the latter half 3 bits are additional bits. The first half 7 bits of the encoded data 2 are Huffman codes and the latter half 7 bits are additional bits.

Also in the example of FIG. 6, since the data length L1 of the encoded data 1 is 5 bits, the second decoding unit 43 is selected by the selection unit 50, but the Huffman code length of the encoded data 2 is 7 bits. Therefore, the second decoding unit 43 cannot search for this Huffman code, and the second decoding unit 43
The search error signal Esr3 is output from. Since this search error signal is selected, the search error signal Esr is output from the selection unit 50 to the invalid signal generation unit 70, and the invalid signal Siv is generated.

Therefore, in the example of FIG. 6, the data extraction position shift unit 80 invalidates the data length L2, and the data extraction position is shifted according to only the data length L1. That is, the leading bit for decoding the next encoded data is set to bit '5' of the extracted data string Ds. Further, since the invalid signal Siv is generated, the DCT coefficient D2 output from the DCT coefficient output unit 102 is processed as invalid data in the dequantizer 1003 and the inverse DCT device 1004. Further, the decoding number corresponding to the zero run length RRRR of the decoded data RS2 is the counting unit 9
It is invalidated at 0, and the number of decodings for this is the count value Sc.
Not added to nt.

FIG. 7 shows an example in which the selection error signal Esel is output because the data length of the encoded data 1 is not included in the range of 3 bits to 6 bits. In FIG. 7, 7 bits from the first bit “0” to the bit “6” of the extracted data string Ds is the first encoded data 1, bit “7”.
5 bits from to 11 are the second encoded data 2. The first 4 bits of the encoded data 1 are Huffman codes and the latter 3 bits are additional bits. The first 2 bits of the encoded data 2 are Huffman codes and the latter 3 bits are additional bits.

In the example of FIG. 7, the data length L1 of the encoded data 1 is 7 bits, which exceeds the range of 3 bits to 6 bits. Therefore, the second decoding unit corresponding to this is selected by the selection unit 5.
No selection can be made at 0, and the selection error signal Esel is output from the selection unit 50. As a result, the invalid signal Siv is generated in the invalid signal generator 70.

Therefore, also in the example of FIG. 7, the data extraction position shift unit 80 invalidates the data length L2, and the data extraction position is shifted according to only the data length L1. That is, the leading bit for decoding the next encoded data is set to bit '7' of the extracted data string Ds. The DC output from the DCT coefficient output unit 102
The T coefficient D2 is processed as invalid data in the inverse quantizer 1003 and the inverse DCT device. Furthermore, the decoding number corresponding to the zero run length RRRR of the decoding data RS2 is invalidated in the counting unit 90, and the decoding number for this is not added to the count value Scnt.

FIG. 8 shows a typical 20 images in JPEG.
It is a figure which shows an example of the result of having investigated the number of appearances for every data length of Huffman code, when encoding. Image sa
The number of appearances and the appearance rate (%) of the Huffman code from 2 bits to 6 bits are shown for each of the mpl1 to the image sample20. The rightmost column shows the total number of Huffman codes in each image. In addition, FIG.
FIG. 9 is a diagram showing an example of the result of checking the number of appearances in the same manner as in FIG. 8 for the data length of the sum of Huffman code and additional bits when JPEG-encoding 20 general images.

As can be seen from FIG.
Regarding the data length of the Huffman code in the case of PEG coding, about 75% or more of the whole is included in the range of 2 bits to 4 bits, and about 85% or more is included in the range of 2 bits to 5 bits. As can be seen from FIG. 9, about 55% of the total data length of the Huffman code and the additional bits is included in the range of 3 bits to 5 bits, and about 70% is included in the range of 3 bits to 6 bits. included.

The Huffman decoder shown in FIG. 2 reduces the required storage capacity of the Huffman code data table by utilizing such a bias in the appearance rate of the Huffman code length. That is, the first decoding unit 20 has a data table capable of decoding all Huffman codes of the DC component and the AC component, while the second decoding unit 41 to the second decoding unit 44 When the data length (Huffman code length + additional bit length) of the first encoded data 1 is from 3 bits to 6 bits (appearance rate is about 70%), and the second encoded data 2 is 2 When a Huffman code of 5 bits to 5 bits is included (appearance rate is about 85%), a minimum data table capable of decoding this coded data is provided.

Therefore, as compared with the conventional example of FIG. 17, the storage capacity required for the data table can be greatly reduced and the circuit structure can be simplified. In addition, FIG.
Although it is not possible to decode two pieces of encoded data 100% as in the conventional example, the frequency is relatively high (70% × 8).
5% = about 60%), two pieces of encoded data can be simultaneously decoded, and high-speed decoding of a JPEG encoded data stream can be realized.

<Second Embodiment> Next, a second embodiment of the present invention will be described. In the first embodiment, up to two pieces of encoded data can be simultaneously decoded, but in the second embodiment, more encoded data can be simultaneously decoded.

FIG. 10 is a schematic block diagram showing a configuration example of the Huffman decoder 1002 of FIG. 1 according to the second embodiment of the present invention. The same reference numerals in FIGS. 2 and 10 denote the same components. In addition, the Huffman decoder shown in FIG. 10 includes a second decoding unit 45 to a second decoding unit 410, a selection unit 51 and a selection unit 52, and a second data length calculation unit 61.
And the second data length calculation unit 62, the invalid signal generation unit 71
And the invalid signal generation unit 72, the data extraction position shift unit 8
It has 0 ', a counting unit 90', and a DCT coefficient output unit 103.

Second decoding section 45 to second decoding section 410
Is the data string Ds extracted by the data extraction unit 10.
5 bits from the data position shifted toward the end of the data stream by 7 to 12 bits with respect to the first bit of
A data string for bits is input, and a predetermined Huffman code included in the input data string is searched for. Then, the decoded data RS corresponding to the retrieved Huffman code
25 to decrypted data RS210 are output. Further, when the data length of the Huffman code included in the input data string is longer than 5 bits and the corresponding Huffman code cannot be found from the data table, the search error signal Esr5 to the search error signal Esr10, respectively.
Is output.

The selection unit 51 is a block equivalent to the selection unit 50 of FIG. 2, and has a second decoding unit 41 to a second decoding unit 44.
One of the second decoding units is selected according to the data length L1, the decoded data Rs2 or the search error signal Esr_1 is output, and if there is no corresponding second decoding unit, The selection error signal Esel_1 of 1 is output.

The selecting section 52 selects one second decoding section from the second decoding section 44 to the second decoding section 410 in accordance with the sum of the data length L1 and the data length L2. The decoded data Rs2_2 or the search error signal Esr_2 is output. In the example of FIG. 12, the data length L1 and the data length L2
And the second decoding unit 44 when the sum is 6 bits, the second decoding unit 45 when the sum is 7 bits, and the second decoding unit 45 when the sum is 8 bits.
Of the second decoding unit 47 for 9 bits.
The second decoding unit 48 is selected for 10 bits, the second decoding unit 49 is selected for 11 bits, and the second decoding unit 410 is selected for 12 bits. If there is no second decoding unit corresponding to this, that is, if the sum of the data length L1 and the data length L2 is not included in the range of 6 bits to 12 bits, the second selection error signal Esel_2 is set. Output.

The second data length calculation unit 61 uses the second data length calculation unit 61 shown in FIG.
The data length calculation unit 60 calculates the data length L2 of the encoded data corresponding to the decoded data RS2_1 based on the decoded data Rs2_1 and the corresponding data length of the Huffman code.

The second data length calculation section 62 includes a selection section 52.
The decrypted data Rs2 of the second decryption unit selected in
_2 and the data length of the Huffman code corresponding to the decoded data RS2, the data length of the encoded data corresponding to the decoded data RS2_2 is calculated, and the data length L
Output as 3.

The invalid signal generator 71 is a block equivalent to the invalid signal generator 70 of FIG.
When the r_1 or the first selection error signal Esel_1 is output, the first invalidation signal Siv_1 that informs that the decoded data RS2_1 and the decoded data RS2_2 are invalid is generated. Also, the decrypted data RS1
Also includes the code EOB, or when the counting unit 90 ′ outputs the first counting error signal Ecnt_1, the first invalid signal Siv_1 is generated.

The invalid signal generator 72 receives the search error signal Esr_ from the second decoder selected by the selector 52.
2 is output, or the selection unit 52 outputs the second selection error signal Esel_2, the invalid signal S indicating that the decoded data RS2_2 is invalid.
iv_2 is generated. The second invalid signal Siv_2 is also generated when the decoded data RS2_1 includes the code EOB or when the counting unit 90 ′ outputs the second counting error signal Ecnt_2.

The data extraction position shift unit 80 'shifts the shift signal Sft for shifting the data string extraction position of the data extraction unit 10 to the end side of the data stream by the number of data corresponding to the sum of the data lengths L1 to L3. To generate.
Further, when the first invalid signal Siv_1 is generated, the shift signal Sf that shifts the extraction position of the data extraction unit 10 toward the tail side by the number of data corresponding to the data length L1.
generate t. When the second invalid signal Siv_2 is generated, the shift signal Sft that shifts the extraction position of the data extraction unit 10 toward the tail side by the number of data corresponding to the sum of the data length L1 and the data length L2 is generated.

The counter 90 'counts the DCT coefficient of the AC component decoded in the unit block according to the decoded data RS1, the decoded data RS2_1 and the zero run length RRRR contained in the decoded data RS2_2. .

When the invalid signal generator 71 generates the first invalid signal Siv_1, both the zero run length RRRR contained in the decoded data RS2_1 and the decoded data RS2_2 are invalidated and the AC component of the AC component is deleted. Count the DCT coefficients. That is, the decrypted data R
The zero run length RRRR of S2_1 and the decoded data RS2_2 is not included in the count value. On the other hand, when the invalid signal generator 72 generates the first invalid signal Siv_2, only the zero run length RRRR included in the decoded data RS2_2 is invalidated, and the AC component D
Count the CT coefficients. In this case, the decrypted data RS2_
The zero run length RRRR included in 1 is the count value Scn.
included in t.

When the count value Scnt of the AC component reaches 63 (64 DCT coefficients of the entire unit block-1 DCT coefficient of the DC component), the count value Scnt is initialized to zero and the AC value is returned to the AC value again. Count the components.

When the AC component count value Scnt exceeds 63, the decoded data RS2_1
Even if the zero run length RRRR of the decoded RS2_2 is both disabled, if the count value Scnt is larger than 63, the first count error signal Ecnt_1 is generated. In this case, the second encoded data of the extracted data string Ds is the head data of the unit block. Further, the count value Scnt becomes larger than 63 when the zero run length RRRR of the decoded RS2_2 is invalidated, and the count value Scnt becomes 63 when the zero run length RRRR of the decoded data RS2_1 and the decoded RS2_2 is invalidated. If the number is equal to or less than the second count error signal Ecn
Generate t_2. In this case, the third coded data is the head data of the unit block.

The DCT coefficient output unit 103 extracts additional data from the extracted data string Ds based on the decoded data RS3, calculates the DCT coefficient D3 of the AC component from the extracted additional data, and outputs it.

Here, in the Huffman decoder shown in FIG. 10, three pieces of decoded data (decoded data RS1, decoded data RS2_1 and decoded data RS2_2) are included.
A specific configuration example of the DCT coefficient output unit that outputs three DCT coefficients based on the above will be described with reference to FIG. The same reference numerals in FIG. 3 and FIG. 11 indicate the same components.
Further, in FIG. 11, reference numeral 1031 indicates an additional data extraction unit, and reference numeral 1032 indicates a DCT coefficient calculation unit.

The additional data extraction unit 1021 inputs a data string of 13 bits from a bit position shifted by 3 bits with respect to the leading bit of the extracted data string Ds of the data extraction unit 10, and from the input data string, The additional data AD2 is extracted based on the category SSSS included in the decrypted data RS2_1. Additional data extraction unit 1031
Is a 16-bit data string from the bit position shifted by 6 bits with respect to the first bit of the extracted data string Ds of the data extraction unit 10, and the category included in the decoded data RS2_2 from this input data string. SSS
The additional data AD3 is extracted based on S.

The DCT coefficient calculator 1022 calculates the DC of the AC component based on the additional data AD2 and the decoded data RS2_1 extracted by the additional data extractor 1021.
The T coefficient D2 is calculated. The DCT coefficient calculation unit 1032
AC based on the additional data AD3 and the decoded data RS2_2 extracted by the additional data extraction unit 1031.
The DCT coefficient D3 of the component is calculated.

Next, the operation of the Huffman decoder shown in FIGS. 10 and 11 will be described.

FIG. 12 shows that in the 27-bit data string extracted by the data extracting unit 10, the decoding unit 20 of FIG.
FIG. 11 is a diagram showing a range of encoded data that can be decoded by the decoding units 41 to 410. As illustrated in FIG. 12, the first decoding unit 20 can decode the encoded data from the first bit “0” to the last bit “26” of the extracted data string Ds. On the other hand, the second decoding unit 41
Is the bit "3" to the bit "12" of the extracted data string Ds
Up to bit “4” to bit “1”
Up to 3 ', the second decoding unit 43 from bit "5" to bit "14", the second decoding unit 44 from bit "6" to bit "15", and the second decoding unit 45 from bit "7". 'From bit' 16 ', second decoding unit 46 from bit' 8 'to bit' 17 ', second decoding unit 47 from bit' 9 'to bit' 18 ', second decoding unit 48.
Is from bit '10' to bit '19', the second decoding unit 49 is from bit '11' to bit '20', second
The decoding unit 410 can decode the encoded data in the range from bit '12' to bit '21'.

Decoded data RS1 used for decoding the first encoded data of the extracted data sequence Ds is generated by the first decoding section 20. The decoded data RS2_1 used for decoding the second encoded data is the second decoding unit 4 according to the data length L1 of the first encoded data.
It is selected from the decoded data in any of the first to second decoding units 44. The decoded data RS2_2 used for decoding the third coded data is the second data depending on the sum of the data length L1 of the first coded data and the data length L2 of the second coded data. Selected from the decoded data of any of the decoding unit 44 to the second decoding unit 410.

FIG. 13 shows 3 included in the extracted data string Ds.
It is a figure which shows an example of one encoded data. In FIG. 13, 6 bits from the first bit “0” to bit “5” of the extracted data string Ds is the first encoded data 1, and 5 bits from bit “6” to bit “10” is 2. Second encoded data 2, bit '11' to bit '2
The 10 bits up to 0'are the third encoded data 3. The first 3 bits of the encoded data 1 are Huffman codes, and the latter 3 bits are additional bits. The first 2 bits of the encoded data 2 are Huffman codes, and the latter 3 bits are additional bits. First half 5 of encoded data 3
The bits are Huffman codes, and the latter 5 bits are additional bits.

In this case, the 16-bit data string from the first bit "0" of the extracted data Ds is the first decoding unit 20.
The Huffman code is searched from the leading 3 bits of the input data string. Then, the zero run length RRRR and the category SSSS corresponding to the retrieved Huffman code are output as the decoded data RS1.

Decoded data RS1 of the first decoding unit 20
Is input to the DCT coefficient output unit 101, and the DCT coefficient D1 of the encoded data 1 is calculated based on the decoded data RS1 and the extracted data string Ds.

In the first data length calculating section 30, the sum of the data length of the additional bit designated by the category SSSS of the decoded data RS1 and the data length of the Huffman code corresponding to the decoded data RS1. Accordingly, the data length L1 of the encoded data 1 is calculated. In the example of FIG. 13, 6 bits are calculated as the data length L1 from the sum of 3 bits of the Huffman code and 3 bits of the additional bits.

The calculated data length L1 is input to the selection unit 51, and the second decoding unit is selected according to the data length L1. In the example of FIG. 13, the second decoding unit 44 that searches for the Huffman code from the 7th bit “6” for the first bit “0” of the extracted data string Ds is selected. Then, the decoded data RS24 and the search error signal Esr4 are output from the selection unit 51 as the decoded data RS2_1 and the search error signal Esr_1, respectively.

However, in the example of FIG. 13, the encoded data 1
Has a data length of 6 bits and is included in the range of 3 bits to 6 bits. In this case, therefore, the selection error signal Es
el_1 is not output. Further, since the Huffman code of the encoded data 2 has 2 bits and does not exceed 5 bits, the search error signal Esr4 is not output from the second decoding unit 44. Therefore, the search error signal Esr_1 is not output from the selection unit 51.

The decoded data RS2_1 selected by the selection unit 51 is input to the DCT coefficient output unit 102,
The DCT coefficient D2 of the encoded data 2 is calculated based on the decoded data RS2_1 and the extracted data string Ds.

In the second data length calculation unit 61, the data length of the additional bit designated by the category SSSS of this decoded data RS2_1 and the decoded data RS2_1
The data length L2 of the encoded data 2 is calculated according to the sum with the data length of the Huffman code corresponding to. In the example of FIG. 13, 5 bits are calculated as the data length L2 from the sum of 2 bits of the Huffman code and 3 bits of the additional bits.

Calculated data length L1 and data length L2
Are both input to the selection unit 52, and the second decoding unit is selected according to the sum of the data length L1 and the data length L2. In the example of FIG. 13, the second decoding unit 49 that searches for the Huffman code from the bit “11” of the extracted data string Ds is selected. Then, the decoded data RS29 and the search error signal Esr9 are output from the selection unit 52 as the decoded data RS2_2 and the search error signal Esr_2, respectively.

The decoded data RS2_2 selected by the selection unit 52 is input to the DCT coefficient output unit 103,
The DCT coefficient D3 of the encoded data 3 is calculated based on the decoded data RS2_2 and the extracted data string Ds.

However, in the example of FIG. 13, the sum of the data lengths of the coded data 1 and the coded data 2 is 11 bits, which is included in the range of 6 bits to 12 bits.
In this case, the selection error signal Esel_1 is not output. The Huffman code of the encoded data 3 has 5 bits and does not exceed 5 bits. Therefore, the second decoding unit 49
Therefore, the search error signal Esr9 is not output. Therefore, the search error signal Esr_2 is not output from the selection unit 52.

Calculated data length L1 to data length L3
Are both input to the data extraction position shift unit 80 ′,
The first invalid signal Siv_1 and the second invalid signal Siv
When neither _2 is generated, the shift signal Sft is generated according to the total sum of these data lengths. In this case, in the example of FIG. 13, bit “2” of the extracted data string Ds
The data extraction position is shifted to 1 ', and the next three pieces of encoded data are decoded from this position. First invalid signal S
If iv_1 has been generated, the decrypted data RS
Since 2_1 and the decoded data RS2_2 are both invalid, the shift amount due to the data length L2 and the data length L3 is ignored, and the shift signal Sf corresponding to only the data length L1 is ignored.
t is generated. In this case, in the example of FIG. 13, the data extraction position is shifted to bit “6” of the extraction data string Ds. When the second invalid signal Siv_2 is generated, since the decoded data RS2_2 is invalid, the shift amount due to the data length L3 is ignored, and the shift signal according to the sum of the data length L1 and the data length L2. Sft is generated. In this case, in the example of FIG. 13, the data extraction position is shifted to bit “11” of the extraction data string Ds.

Further, the decoded data RS1 output from the decoding section 20 and the decoded data R output from the selection section 51
S2_1 and the decoded data RS2_2 output from the selection unit 52 are both input to the counting unit 90 ′, and from the zero run length RRRR included in these decoded data, the number of AC components decoded in the unit block. Are counted. When the count value Scnt exceeds 63 and the end data of the unit block consisting of 64 DCT counts is included in the encoded data 1, the first counting error signal Ecnt_1 is generated and the encoded data 2 is generated. , The second counting error signal Ecnt_2 is generated. If this end data is included in the encoded data 3, the count value Scnt does not exceed 63, so
No counting error signal is generated.

In the example of FIG. 13, the first selection error signal E
sel_1, second selection error signal Ssel_2, search error signal Esr_1 and search error signal Esr_2
Is not output, no invalid signal is generated. The first invalid signal Siv_1 is generated when the decoded data RS1 includes the code EOB indicating the end of the unit block or when the counting unit 90 ′ outputs the first counting error signal Ecnt_1. The second invalid signal Siv_2 is generated when the decoded data RS2_1 includes the code EOB indicating the end of the unit block or when the counting unit 90 ′ outputs the second counting error signal Ecnt_2. .

FIG. 14 shows an example in which the search error signal Esr is output because the Huffman code length of the encoded data 3 exceeds 5 bits. In FIG. 14, the extracted data string Ds
5 bits from the first bit "0" to bit "4" of the first encoded data 1 and 4 bits from bit "5" to bit "8" of the second encoded data 2 Twelve bits from bit "9" to bit "20" are the third encoded data. The first 2 bits of the encoded data 1 are Huffman codes, and the latter 3 bits are additional bits. The first two bits of the encoded data 2 are Huffman codes and the second two bits are additional bits. The first 6 bits of the encoded data 3 are Huffman codes, and the latter 6 bits are additional bits.

In the example of FIG. 14, since the data length L1 of the encoded data 1 is 5 bits, the selection unit 51 selects the second decoding unit 43. Since the sum of the data lengths of the encoded data 1 and the encoded data 2 is 9 bits, the selection unit 52 selects the second decoding unit 47. This second
Since the Huffman code length of the encoded data 3 input to the decoding unit 47 of 6 is 6 bits as shown in FIG.
The Huffman code cannot be searched for in the decoding unit 47, and the search error signal Esr7 is output from the second decoding unit 47. Since this search error signal is selected by the selection unit 52, the search error signal Es is output from the selection unit 52.
r_2 is output and the second invalid signal Siv_2 is generated.

Therefore, in the example of FIG. 14, the data extraction position shift unit 80 'invalidates the data length L3, and the data extraction position is shifted according to the sum of the data length L1 and the data length L2. That is, the first bit for decoding the next three pieces of encoded data is set to bit '9' of the extracted data string Ds. In addition, the second invalid signal Siv
Since _2 is generated, the DCT coefficient D3 output from the DCT coefficient output unit 103 is processed as invalid data in the inverse quantizer 1003 and the inverse DCT unit 1004. Further, the decoding number corresponding to the zero run length RRRR of the decoding data RS2_2 is invalidated in the counting unit 90 ′, and the decoding number for this is not added to the count value Scnt.

In FIG. 15, since the sum of the data length of the encoded data 1 and the data length of the encoded data 2 is not included in the range of 6 bits to 12 bits, the second selection error signal Es
An example in which el_2 is output is shown. In FIG. 15, 6 bits from the first bit “0” to the bit “5” of the extracted data string Ds is the first encoded data 1, bit “6”.
To bit '15' are the second coded data 2, and 5 bits from bit '16' to bit '20' are the third coded data 3. The first 3 bits of the encoded data 1 are Huffman codes, and the latter 3 bits are additional bits. The first 5 bits of the encoded data 2 are Huffman codes, and the latter 5 bits are additional bits. The first 2 bits of the encoded data 3 are Huffman codes, and the latter 3 bits are additional bits.

In the example of FIG. 15, since the data length L1 of the encoded data 1 is 6 bits, the selection unit 51 selects the second decoding unit 44. On the other hand, since the sum of the data length L1 of the encoded data 1 and the encoded data length L2 is 16 bits, which exceeds the range of 6 bits to 12 bits, the selection unit 52 selects the second decoding unit. However, the second selection error signal Esel_2 is output. As a result, the invalid signal generator 72 generates the second invalid signal Siv_2.

Therefore, even in the example of FIG.
Similarly, the data length L3 is invalidated in the data extraction position shift unit 80 ′, and the data extraction position is set to the data length L1.
And the data length L2. In this example, the first bit of the encoded data extracted next is
It is set to bit '16' of the extracted data string Ds. In addition, since the second invalid signal Siv_2 is generated, DC
The DCT coefficient D3 output from the T coefficient output unit 103 is
Inverse quantizer 1003 and inverse DCT device 1004 process as invalid data. Further, the decoding number corresponding to the zero run length RRRR of the decoding data RS2_2 is invalidated in the counting unit 90 ′, and the decoding number for this is not added to the count value Scnt.

In FIG. 16, the data length of the encoded data 1 is 3
An example in which the first selection error signal Esel_1 is output because it is not included in the range of bits to 6 bits is shown. Figure 1
In FIG. 6, 7 bits from the first bit “0” to bit “6” of the extracted data string Ds is the first encoded data 1, and 5 bits from bit “7” to bit “11” are two. 10th bit of coded data 2 from bit '12' to bit '21' is the third coded data 3
Has become. The first half 4 bits of the encoded data 1 are Huffman codes, and the latter half 3 bits are additional bits.
The first two bits of encoded data 2 are Huffman codes, the second half 3
Bits are additional bits. The first half 5 bits of the encoded data 3 are Huffman codes, and the latter half 5 bits are additional bits.

In the example of FIG. 16, since the data length L1 of the encoded data 1 is 7 bits, which exceeds the range of 3 bits to 6 bits, the selecting unit 51 should select the second decoding unit. The first selection error signal Esel
_1 is output. As a result, the invalid signal generation unit 71 generates the first invalid signal Siv_1.

When the first invalid signal Siv_1 is generated, the decoded data RS2_1 and the decoded data R are generated regardless of whether or not the second invalid signal Siv is output.
Both S2_2 are invalid. Therefore, in the example of FIG. 16, the data length L2 and the data length L3 are both invalidated in the data extraction position shift unit 80 ', and the data extraction position is shifted according to only the data length L1. In this example, the leading bit of the encoded data to be extracted next is set to bit '7' of the extracted data string Ds. In addition, since the first invalid signal Siv_1 is generated, DC
DCT coefficient D2 output from T coefficient output section 102 and DCT coefficient D output from DCT coefficient output section 103
3 is processed as invalid data in the inverse quantizer 1003 and the inverse DCT device 1004. further,
Decoded data RS2_1 and decrypted data RS2_2
Any of the decoding numbers corresponding to the zero run length RRRR of is invalidated in the counting unit 90 ′, and the decoding number of this is not added to the count value Scnt.

By the way, in the Huffman decoder of FIG. 10, three encoded data can be simultaneously decoded by 1
The data length of the third encoded data is 3 to 6 bits, the data length of the second encoded data is 3 to 6 bits, and the Huffman code length of the third encoded data is 5 bits. Within the case. By multiplying the frequencies of appearance of the respective Huffman codes with reference to the examples of FIGS. 8 and 9, the frequency with which three pieces of encoded data can be simultaneously decoded is 70% × 70% × 85% = about 42% Becomes Further, the Huffman decoder of FIG. 10 can only decode one piece of encoded data when the data length of the first encoded data is not included in the range of 3 bits to 6 bits, and its frequency Is about 30%. Therefore, the average number of coded data that can be simultaneously decoded by the Huffman decoder of FIG. 10 is 1 × 30% + 2 × (100% −42% −30%) + 3 ×
42% = about 2.1 pieces. Thus, the Huffman encoder of FIG. 10 can decode the Huffman-encoded data stream at a speed equal to or higher than that of the conventional example shown in FIG. Further, as compared with the conventional example shown in FIG. 17, the required storage capacity of the Huffman decoding data table can be reduced, so that the circuit configuration can be simplified.

The present invention is not limited to the above-described embodiments, and various modifications obvious to those skilled in the art can be made. For example, JPE is used as an example for explaining the embodiment of the present invention.
Although a G decoding device is shown, the invention is not limited to this example. The present invention is also applicable to a data decoding device for a data stream encoded by another similar method.

The data length and bit position of the data to be processed and the number of blocks such as the second decoding unit in the examples for explaining the embodiments of the present invention do not limit the present invention at all. It can be modified arbitrarily.

In the example of FIG. 10, the selecting unit 51 selects one from the four decoding units (second decoding unit 41 to second decoding unit 44), but the number of decoding units to be selected is further increased. You can also For example, all 10 decoding units (second
One of the decoding units 41 to 410) may be selected. This increases the probability that the second coded data can be decoded, and thus the decoding efficiency can be further improved.

The second embodiment described above shows an example of a Huffman decoder capable of simultaneously decoding three pieces of coded data. However, according to the present invention, four or more pieces of coded data are decoded. A data decoding device capable of decoding at the same time can also be realized.

[0117]

According to the present invention, a variable length coded data stream can be decoded at high speed with a simple structure.

[Brief description of drawings]

FIG. 1 is a block diagram showing a schematic configuration example of a JPEG decoding device according to the present invention.

FIG. 2 is a schematic block diagram showing a configuration example of the Huffman decoder of FIG. 1 according to the first embodiment of the present invention.

3 is a schematic block diagram showing a configuration example of a DCT coefficient output unit in the Huffman decoder shown in FIG.

FIG. 4 is a diagram showing a range of encoded data that can be decoded by each decoding unit in FIG. 2 in a 27-bit data string extracted by a data extraction unit.

FIG. 5 is a diagram showing an example of two pieces of encoded data included in a data string extracted by a data extraction unit.

FIG. 6 shows an example in which a search error signal is output because the Huffman code length of the second encoded data exceeds 5 bits.

FIG. 7 is a data length of the first encoded data is 3 bits to
An example in which the selection error signal is output because it is not included in the 6-bit range is shown.

FIG. 8 is a diagram showing an example of a result of examining the number of appearances of each Huffman code for each data length when 20 general images are JPEG encoded.

FIG. 9 is a diagram showing an example of the result of examining the number of appearances in the same manner as in FIG. 8 for the data length of the sum of the Huffman code and additional bits when JPEG encoding 20 general images. .

FIG. 10 is a schematic block diagram showing a configuration example of the Huffman decoder of FIG. 1 according to the second embodiment of the present invention.

11 is a DC in the Huffman decoder shown in FIG.
It is a schematic block diagram which shows the structural example of a T coefficient output part.

12 is a diagram showing a range of coded data that can be decoded by each decoding unit in FIG. 10 in a 27-bit data string extracted by the data extraction unit.

FIG. 13 is a diagram showing an example of three pieces of encoded data included in a data string extracted by a data extraction unit.

FIG. 14 shows an example in which a search error signal is output because the Huffman code length of encoded data 3 exceeds 5 bits.

FIG. 15 shows that the sum of the data length of the first encoded data and the data length of the second encoded data is not included in the range of 6 bits to 12 bits, so that the second selection error signal is An example of output is shown below.

FIG. 16 shows an example in which the first selection error signal is output because the data length of the first encoded data is not included in the range of 3 bits to 6 bits.

FIG. 17 is a block diagram showing a configuration example of a conventional data decoding device that decodes a data stream encoded by a Huffman code.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Data shift part, 2a-2r ... Decoding part, 3 ... Selector, 10 ... Data extraction part, 20 ... 1st decoding part, 30 ...
1st data length calculation part, 41-49,410 ... 2nd decoding part, 50, 51, 52 ... Selection part, 60, 61, 62 ...
Second data length calculation unit, 70, 71, 72 ... Invalid signal generation unit, 80, 80 '... Data extraction position shift unit, 90,
90 '... Counting unit, 101-103 ... DCT coefficient output unit,
1011, 1021, 1031 ... Additional data extraction unit, 1
012, 1022, 1032 ... DCT coefficient calculation unit, 10
01 ... buffer, 1002 ... Huffman decoder, 100
3 ... Inverse quantizer, 1004 ... Inverse DCT device.

   ─────────────────────────────────────────────────── ─── Continued front page    F term (reference) 5C059 KK09 KK11 MA00 MA23 MC11                       MC24 MC33 MC34 MC38 ME02                       ME08 PP01 RC40 TA58 TB08                       TC00 UA05 UA32                 5C078 AA04 BA57 CA31 DA02 DA06                 5J064 AA03 AA04 BA09 BA16 BB05                       BC01 BC16 BC25 BC28

Claims (16)

[Claims] 1. A data decoding device for decoding a supplied encoded data stream, comprising: data extracting means for extracting a data string of a predetermined data length from the beginning of the encoded data stream; and the data extracting means. First decoding means for searching a variable-length code included in a predetermined variable-length code table from the head of the data string extracted in, and outputting decoded data corresponding to the searched variable-length code; Based on the decoded data of the first decoding means and the data length of the variable length code corresponding to the decoded data,
A first data length calculating means for calculating the data length of the encoded data corresponding to the decoded data and outputting it as the first data length, and a plurality of data lengths for the head of the data string extracted by the data extracting means. From the data position deviated to the end side by the different number of data, each of the variable-length codes in the variable-length code table is searched, decoded data corresponding to the searched variable-length code is output, and in the search, When a corresponding variable length code is not found, a plurality of second decoding means for outputting a search error signal and a plurality of second decoding means are selected according to the first data length. Selecting means for outputting the decoded data of the second decoding means or the search error signal, and for outputting the selection error signal when there is no corresponding second decoding means, and the selecting means. Calculating the data length of the encoded data corresponding to the decoded data based on the decoded data of the second decoding means and the data length of the variable length code corresponding to the decoded data, When the search error signal is output from the second data length calculating unit that outputs the second data length and the second decoding unit selected by the selecting unit, or the selection error signal is output from the selecting unit. In this case, an invalid signal generating means for generating an invalid signal notifying that the decoded data of the second decoding means is invalid, and a sum of the first data length and the second data length. When the invalid signal is generated by shifting the data string extraction position of the data extraction means to the end side of the encoded data stream by the corresponding number of data, depending on the first data length. And a data extraction position shift means for shifting the extraction position to the end side by the number of data. 2. The encoded data stream includes encoded data composed of the variable length code and additional data, and the data length of the additional data included in the decoded data of the first decoding means. First additional data extracting means for further extracting the additional data from the extracted data string of the data extracting means based on the information of 1., and the data of the additional data included in the decoded data output from the selecting means. The data decoding device according to claim 1, further comprising a second additional data extraction unit that further extracts the additional data from the extracted data string of the data extraction unit based on the length information. 3. The encoded data stream is a unit block of a predetermined number of data of a desired data stream according to the number of consecutive data of a predetermined value and data of another value following the data of the predetermined value. The data of the continuous number included in the decoded data of the first decoding means and the decoded data of the second decoding means selected by the selecting means. Accordingly, the number of the decoded data in the unit block is counted, and when the invalid signal is generated, the information of the continuous number included in the decoded data of the selected second decoding means. The data decoding apparatus according to claim 1, further comprising: a counting unit that invalidates the number, performs the counting, and initializes the count value when the count value reaches the predetermined number of data. 4. The counting means outputs a counting error signal when the count value is larger than the predetermined number of data, and the invalid signal generating means adds the unit block to the decoded data of the first decoding means. 4. The data decoding device according to claim 3, wherein the invalid signal is generated when predetermined information indicating the end of is included, or when the counting means outputs a counting error signal. 5. The first data length to the nth data length (the code n is 1 ≦ n ≦ N with respect to a predetermined natural number N) from at least a part of the plurality of second decoding means. The decoded data of the second decoding means and the search error signal selected according to the total sum of (which represents an arbitrary natural number to be satisfied) are output, and the nth selection error signal is output when there is no corresponding second decoding means. Corresponding to a plurality of selection means for outputting the decoded data of the second decoding unit selected according to the sum of the first data length to the n-th data length in the selection means, and the decoded data. A plurality of second data length calculating means for calculating the data length of the encoded data corresponding to the decoded data based on the data length of the variable length code, and outputting the data length as the (n + 1) th data length. , In the selecting means When the search error signal is output from the second decoding means selected according to the sum of the first data length to the nth data length, or the nth selection error signal is output from the selection means. In this case, the first data length to the m-th data length (code m
Is an arbitrary natural number satisfying n ≦ m ≦ N with respect to a predetermined natural number N). The n-th signal notifying that all the decoded data of the second decoding means selected according to the total sum are invalid.
A plurality of invalid signal generation means for generating the invalid signal, and the data extraction position shift means is the data extraction means for the number of data corresponding to the sum of the first data length to the Nth data length. When the n-th invalid signal is generated by shifting the data string extraction position of the above to the end side of the encoded data stream, the data corresponding to the sum of the first data length to the n-th data length. The data decoding device according to claim 1, wherein the extraction position is shifted to the end side by a number. 6. The encoded data stream includes encoded data composed of the variable length code and additional data, and the data length of the additional data included in the decoded data of the first decoding means. The first additional data extracting means for further extracting the additional data from the extracted data string of the data extracting means, and the decoded data of the second decoding means selected by the plurality of selecting means based on the information of The data according to claim 5, further comprising a plurality of second additional data extracting means for further extracting the additional data from the extracted data string of the data extracting means based on the information of the data length of the additional data included. Decoding device. 7. The encoded data stream is a unit block of a predetermined number of data of a desired data stream according to the number of consecutive data of a predetermined value and data of another value following the data of the predetermined value. Data stream generated by encoding for each of the above, the number of consecutive numbers included in the decoded data of the first decoding means and the decoded data of the second decoding means selected by the plurality of selecting means. According to the information, the number of decoded data in the unit block is counted, and when the nth invalid signal is generated, the number of the decoded data is determined according to the sum of the first data length to the mth data length. When the count value reaches the predetermined number of data, all the information of the continuous number contained in the decoded data of the selected second decoding means is invalidated, and when the count value reaches the predetermined number of data, the count value is changed. Initialize Having a counting means, a data decoding apparatus according to claim 5. 8. The decoding means of the second decoding means selected according to the sum of the first data length to the m-th data length when the count value is larger than the predetermined data number. When the count value obtained by invalidating all the information of the continuous number included in the encoded data is larger than the predetermined number of data n, the n-th counting error signal is output, and the invalid signal generating means is A first invalid signal is generated when the decoded data of the first decoding means includes predetermined information indicating the end of the unit block, and selected according to the sum of the first data length to the nth data length. If the predetermined data is included in the decoded data of the second decoding means, the (n + 1) th invalid signal is generated, and if the counting means outputs the nth counting error signal, the nth counting error signal is output. Generates an invalid signal for The data decoding apparatus according to claim 7. 9. A data decoding method for decoding a supplied encoded data stream, comprising: extracting a data string of a predetermined data length from the beginning of the encoded data stream; A step of searching a variable-length code included in a predetermined variable-length code table from the beginning and generating first decoded data corresponding to the searched variable-length code; and the generated first decoded data And a step of calculating the data length of the encoded data corresponding to the first decoded data as the first data length based on the data length of the variable length code corresponding to the first decoded data. And the first variable length code table from the data position shifted toward the end side by a plurality of different data numbers with respect to the head of the data string extracted in the data extraction step. Each part of the variable length code is searched, second decoded data corresponding to the searched variable length code is generated, and if a corresponding variable length code is not found in the search, a search error signal is generated. A second decoding step of generating, and a variable length code generated at a data position corresponding to the first data length from the plurality of generated second decoded data or search error signals are generated. Selecting the second decoded data or search error signal, and generating a selection error signal if there is no corresponding second decoded data or search error signal; and the selected second decoding Based on the data and the data length of the variable length code corresponding to the second decoded data, the data length of the encoded data corresponding to the second decoded data is set to the second data. A step of calculating the data length, and an invalid signal for notifying that the second decoded data is invalid when the search error signal is selected in the selection step or when the selection error signal is generated. And the data string extraction position of the data extraction step is shifted to the end side of the encoded data stream by the number of data corresponding to the sum of the first data length and the second data length. If the invalid signal is generated, a step of shifting the extraction position to the end side by the number of data corresponding to the first data length, the data decoding method. 10. The encoded data stream includes encoded data composed of the variable length code and additional data, and is based on data length information of the additional data included in the first decoded data. The step of further extracting the additional data from the extracted data string in the data extracting step, and the data extracting step based on the data length information of the additional data included in the selected second decoded data. 10. The data decoding method according to claim 9, further comprising the step of further extracting the additional data from the extracted data string. 11. The encoded data stream is a unit block of a predetermined number of data of a desired data stream according to the number of consecutive predetermined value data and data of another value following the predetermined value data. Decoding in the unit block according to the information of the continuous number included in the first decoded data and the selected second decoded data. The number of the converted data is counted, and when the invalid signal is generated, the information of the continuous number included in the selected second decoded data is invalidated to perform the counting, and the count value 10. The data decoding method according to claim 9, further comprising the step of initializing the count value when the predetermined number of data has been reached. 12. In the counting step, a counting error signal is generated when the counted value is larger than the predetermined number of data, and in the step of generating the invalid signal, the unit block is added to the first decoded data. 12. The data decoding method according to claim 11, wherein the invalid signal is generated when predetermined information indicating the end of is included or when the counting error signal is generated. 13. A variable length code at a data position corresponding to the sum of the first data length to the nth data length from at least a part of the plurality of generated second decoded data or search error signals. Selecting the second decoded data or search error signal generated by the search, and generating the n-th selection error signal if there is no corresponding second decoded data or search error signal, When the search error signal is selected according to the sum of the first data length to the nth data length in the selection step, the first data length to the mth data length (symbol m is a predetermined natural number). The second selected according to the sum of N), which represents an arbitrary natural number that satisfies n ≦ m ≦ N
Generating an n-th invalid signal notifying that all of the decoded data are invalid, and second decoded data selected according to the sum of the first data length to the n-th data length. , The data length of the encoded data corresponding to the second decoded data is calculated as the (n + 1) th data length based on the data length of the variable length code corresponding to the second decoded data. In a range satisfying 1 ≦ n ≦ N for a predetermined natural number N, and shifting the extraction position of the data,
When the extraction position is shifted toward the end of the encoded data stream by the number of data corresponding to the sum of the first data length to the Nth data length, and the nth invalid signal is generated, 10. The data decoding method according to claim 9, wherein the extraction position is shifted to the end side by the number of data corresponding to the sum of the first data length to the nth data length. 14. The coded data stream includes coded data composed of the variable length code and additional data, and has information of data length of the additional data included in the first decoded data. Based on the step of further extracting the additional data from the extracted data string in the data extracting step, and the data extraction based on the information of the data length of the additional data included in the selected second decoded data. The data decoding method according to claim 13, further comprising: extracting the additional data from the extracted data string of the step. 15. The encoded data stream is a unit block of a predetermined number of data of a desired data stream according to the number of consecutive data of a predetermined value and data of another value following the data of the predetermined value. Each of which includes a data stream generated by encoding for each of the first and second decoded data and the selected plurality of second decoded data.
When the number of decoded data in the unit block is counted and the nth invalid signal is generated, the first
Data length to the sum of the m-th data length, all the information of the continuous number contained in the second decoded data is invalidated and the counting is performed, and the count value becomes the predetermined data number. The data decoding method according to claim 13, further comprising the step of initializing the count value when the count is reached. 16. The second decoded data selected according to the sum of the first data length to the m-th data length when the count value is larger than the predetermined data number in the counting step. In the step of generating the nth counting error signal and generating the invalid signal in the smallest natural number n in which the count value obtained by invalidating all the information of the continuous number included in is larger than the predetermined data number, When the first decoded data includes predetermined information indicating the end of the unit block, the first invalid signal is generated, and the first invalid signal is generated.
(N + 1) th invalid signal is generated when the predetermined information is included in the second decoded data selected in accordance with the sum of the data length of the first to the mth data length, and the nth counting error is generated. The data decoding method according to claim 15, wherein the nth invalid signal is generated when the signal is output.