JPH02266683A - Picture processing unit - Google Patents

Abstract

PURPOSE:To attain highly efficient transmission with simple control by setting a single scanning area in matching with the distribution of a significant conversion coefficient for each block. CONSTITUTION:The unit consists of an input terminal 101, an orthogonal conversion circuit 102, an area split selection circuit 103, a coefficient scanning circuit 104, a coder 105, a scanning end detection circuit 106 and a code output terminal 107. Then a single scanning area in matching with the distribution of the significance conversion coefficient is set for each block. Thus, any complicated control to the transmission of the conversion coefficient caused due to the split scanning area is prevented and highly efficient transmission with simple control is attained.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to an image processing device that orthogonally transforms image information in units of blocks each consisting of n rows and 1 columns of pixels, compresses the information, and transmits the compressed information. .

[Prior Art] Generally, in an image signal, low frequency components account for most of the power, and the power decreases as the power increases to high frequency components.

Therefore, when transmitting image information, the image signal is converted into frequency components, and among the conversion coefficients obtained by this conversion, only significant conversion coefficients corresponding to low frequency components with large power are transmitted. can be compressed and transmitted.

A conventional example of a device that compresses and transmits image information using such a method is described in a publicly known document (Collection of Lectures at the General National Conference Commemorating the 70th Anniversary of the Institute of Electronics, Information and Communication Engineers, 1988, Volume 5, Page 12). Equipment can be mentioned.

In the case of this device, an input image is first divided into blocks each having n pixels by 1 pixel, and orthogonal transformation is performed on each block using discrete cosine transform (DCT).

This orthogonal transformation refers to the transformation into the frequency components mentioned above, and the distribution of significant transformation coefficients within the transformation plane (post-transformation block) obtained by this transformation generally depends on the nature of the input image. It has a certain pattern of appearance.

For example, it is a certain appearance pattern, such as a band-like distribution in the horizontal or vertical direction, or a band-like distribution in an oblique direction from low frequency components to high frequency components.

Therefore, in conventional devices, the transmission efficiency is improved by eliminating wasteful scanning of the conversion surface for information transmission.
A scanning area is set for each block obtained by the transformation, taking into consideration the appearance pattern of the significant transformation coefficients.

FIG. 2 shows how to set a scanning area in a conventional device.

That is, in the conventional apparatus, a number of scanning areas (in this example, six areas ``■'' to ``■'') are set on the conversion plane X of the block.

Then, by selecting only the scanning areas where significant conversion coefficients exist and scanning only those scanning areas, unnecessary scanning is avoided.

In the case of this illustrated example, the black part is a part where the DC component of the transform coefficient and the low frequency AC component in the vicinity are distributed, and are always transmitted in a significant block. Furthermore, the portion indicated by the dotted line is a portion where significant conversion coefficients are distributed, and is horizontally band-shaped.

For such a distribution of significant conversion coefficients, scanning only needs to be performed on regions (2), (2), and (2).

[Problems to be Solved by the Invention] However, in the conventional device described above, scanning on the conversion surface Since the data is divided and transmitted, there is a problem in that the control over the transmission becomes complicated, and this problem was considered to be a problem to be solved in the future.

The present invention has been made in view of the above-mentioned circumstances, and includes an image transform encoding device that performs orthogonal transform on image information in blocks of n rows by n columns of pixels, compresses the information, and transmits the information. A single scanning area that matches the distribution of coefficients can be set for each block, which prevents the complexity of controlling the transmission of transform coefficients caused by dividing the scanning area, resulting in high efficiency due to simple control. The purpose is to enable the transmission of

[Means for Solving the Problems] An image processing device according to the present invention includes an orthogonal transform circuit that orthogonally transforms image information in units of blocks consisting of n rows and X n columns of pixels, and a transform coefficient obtained by the orthogonal transform circuit. The image information is compressed and transmitted by comprising an area division selection circuit that determines a scanning area for each block so that the scanning matches the significant transform coefficient distribution of each block.

Further, the area division selection circuit classifies the transform coefficients of the orthogonally transformed block according to the magnitude relationship of frequencies, and subdivides the transform coefficients into a plurality of transform coefficient groups extending in a diagonal direction, and each of the subdivided transform coefficients The three transform coefficients located at both ends and the center of each transform coefficient are extracted for each group, and the shape and size of a single scanning area in each block is determined based on the comparison result between each extracted transform coefficient and the threshold value. It is said that the configuration determines the

Further, the shape of the scanning region in the region division selection circuit is determined in advance by associating three transform coefficients extracted from the transform coefficient group with a threshold value as the shape of a single scan region. This is done by preparing three types of standard formats and selecting one of the applicable standard formats based on the results of actual comparison between the three extracted conversion coefficients and the threshold value.

Further, the size of the scanning area in the region division selection circuit is determined based on the positions of the transform coefficients at both ends of the transform coefficient group where all of the three extracted transform coefficients are larger than a threshold value. This is done by expanding or contracting the size of the selected standard form as appropriate.

[Operation] The image processing device according to the present invention scans each block by the region division selection circuit in order to make the scanning of the transform coefficients obtained by the orthogonal transform circuit match the significant transform coefficient distribution of each block. Determine the area.

In that case, the scanning area is determined by determining the shape and size, but the scanning area shape is determined in advance by dividing the area into a single scanning area that matches the distribution of significant conversion coefficients. This is achieved by selecting one of the three standard forms prepared for the selection circuit, and
The size of the scanning area is determined by appropriately expanding or contracting the size of the standard form selected by the form determination.

Therefore, in the image processing device according to the present invention,
A single scanning area that matches the distribution of significant transformation coefficients can be set for each block, and therefore, it is possible to prevent complicated control over transmission of transformation coefficients due to division of the scanning area, and thus, Highly efficient transmission can be achieved through simple control.

[Embodiment] FIG. 1 is a block diagram showing a schematic configuration of an embodiment of the present invention.

The image processing device of this embodiment includes an input terminal 101, an orthogonal transform circuit 102, a region division selection circuit 103, a coefficient scanning circuit 104, an encoder 105, a scanning completion detection circuit 106, a code output terminal 107, etc. It is something.

First, the processing by each component will be explained in detail, and then the structure and operation of the area division selection circuit 103, coefficient scanning circuit 104, etc., which are the main parts of this embodiment, will be explained in detail.

In one embodiment, the image information signal is inputted to the input terminal 101 for each block consisting of n rows and Xn columns of pixels, and is applied to the orthogonal transformation circuit 102. This orthogonal transformation circuit 102
Then, orthogonal transform such as discrete cosine transform (DCT) is performed on these input signals.

The transform coefficients obtained by this orthogonal transform are output to the region division selection circuit 103 and the coefficient scanning circuit 104.

The area division selection circuit 103 selects the scan area configurations shown in FIGS. 3(a), (b), and (C) that match the distribution configuration of significant conversion coefficients among the conversion coefficients obtained by orthogonal transformation.
Three standard forms I, If, and II shown in Figure 3 (a), (b), and (c) are prepared in advance.
), the shaded area is the area where the significant conversion coefficients are distributed and should be scanned). These standard forms r, n, and m are all set based on appearance patterns that depend on the properties of the input image.

Then, when the transform coefficients are sent from the orthogonal transform circuit 102, the region division selection circuit 103 divides the scanning region in block units so that scanning for transmitting each transform coefficient can be performed efficiently. The form is determined to be one of the three standard forms mentioned above, and the size (range) of the area in that case is determined.
Determine.

The shape and size of the determined scanning area are then sent to the coefficient scanning circuit 104 and the encoder 105 as scanning area setting information.
is output to.

The coefficient scanning circuit 104 selects a region division selection port J from among the transform coefficients of each block output from the orthogonal transform circuit 102.
The conversion coefficients falling within the scan area indicated by the scan area setting information from 3103 are sequentially scanned, and the conversion coefficients obtained by scanning are output to the encoder 105 and the scan completion detection circuit 106.

In the scanning completion detection path 106, the coefficient scanning circuit 104
The output of the coefficient scanning circuit 104 is compared with a threshold value set appropriately in advance, and if the output of the coefficient scanning circuit 104 continues to be lower than the threshold value, a scanning completion signal is output to the encoder 105.

The encoder 10'5 first encodes the scanning area setting information, and then encodes the output of the coefficient scanning circuit 104 until the scanning completion signal is input. When the scan completion signal is input, the coding of the output of the coefficient scanning circuit is stopped, and a block end signal representing the block end code is sent to the output terminal 107, thereby completing the processing of the block.

Next, the area division selection circuit 10 which is the main part of the above embodiment
3 and the coefficient scanning circuit 104 will be explained in detail.

FIG. 4(a) specifically shows the image information input to the area division selection circuit 103 and the coefficient scanning circuit 104. 102 shows an array of transform coefficients obtained by orthogonal transform.

In this figure, a black circle indicates that the conversion coefficient at that position is a significant coefficient (
(a significant conversion coefficient), and a white circle represents an insignificant coefficient.

As shown in FIG. 4(a), the region division selection circuit 103 classifies each transform coefficient on the transform surface X according to the magnitude relationship of frequencies, and divides the transform coefficients into a plurality of transform coefficient groups 3 extending diagonally.
Subdivide into 10,320,330,340.350,360. Then, for each further subdivided conversion coefficient group,
Extract three conversion coefficients A, B, and C. Here, the conversion coefficient A.

C is a transform coefficient located at both ends of each transform coefficient group, and transform coefficient B is a transform coefficient located approximately at the center of each transform coefficient group.

More specifically, as shown in FIG. 4(b), for example, in the transform coefficient group 360, the transform coefficient 361 is extracted as the transform coefficient A, the transform coefficient 368 is extracted as the transform coefficient C1, and the transform coefficient 364 is extracted as the transform coefficient B. Ru.

Note that the position at which coefficient B is extracted does not have to be strictly at the center, and therefore, the conversion coefficient 365 may be extracted as coefficient B.

In this way, the values of coefficients A, B, and C extracted from each conversion coefficient group are compared with the threshold value prepared in the first selection circuit 103 for each of them, and the conversion coefficient is Binarization is performed by setting the value to 1 when the value is larger than the threshold value, and to 0 otherwise.

Such a binarization operation is performed using a conversion coefficient group of 360°350,
. . . 310, the conversion coefficients are sequentially processed starting from the image frequency side.

In the process of repeating the binarization operation, when the binarization result of two or more coefficients A, B, and C becomes 1 for the first time in a certain coefficient group, any of A, B, and C becomes 1. The form of the scanning area is determined depending on whether the Here, the scanning area is determined by selecting one of the standard forms (1), (1) and (2) shown in FIG.

In this area division selection circuit 103, as shown in Table 1 below, each of the three standard formats 1. I
l, I[r, the three extracted conversion coefficients A,
The comparison results (binarized values) for the threshold values of B and C are associated with each other, so that once the comparison results for the threshold values are obtained, the standard form can be immediately selected.

In this example, first three coefficients A, B.

Two or more of C are 1 in the case of the conversion coefficient group 320 in FIG. 4(a). In this case, the coefficients A, B,
All of C becomes I.

Therefore, from Table 1, standard form (2) is selected.

Table 1 In the binarization operation described above, all three coefficients A, B, and C are 1.
Repeat until . Then, first three coefficients A, B
, C are all 1, and the size of the scanning area is set so that the reference position is included in the scanning area, based on the coordinate position on the conversion plane X of the coefficients A and C at both ends of the conversion coefficient group where all of It is determined.

That is, the size of the scanning area is determined by the coefficient A, which uses the boundary line in the previously selected standard form as the reference position.

It is set at a position that includes C, and can be thought of as an expanded or contracted version of the standard form.

In the case of the embodiment, the first conversion coefficient group in which all three coefficients A, B, and C are 1 is the conversion coefficient group 32 in FIG.
0, and the positions of the coefficients at both ends of this coefficient group 320 serve as a reference to determine the size of the scanning area. This size can be expressed using the coordinates (0, l, . . . 6, 7) indicating the position of each coefficient shown on the conversion plane X in FIG. 4 as parameters.

In the case of this embodiment, the coefficient group 32 serving as the standard for the size
Since the coordinates of coefficients A and C at both ends of 0 are 3, the size is 3.

Note that even if the binarization scan is repeated, the three coefficients A, B,
The size of the area in the case where there is no conversion coefficient group in which all of C are 1 is expressed using the coordinate values of the conversion surface as follows: ■
becomes.

Next, the coefficient scanning circuit 104 will be explained.

The coefficient scanning circuit 104 receives the area setting information (i.e., standard form (I[[
) and the area size (3)), set the scanning area Y to be actually scanned (the shaded area in the figure) as shown in Figure 5. , for the transformation coefficients (including null coefficients) included in this scanning area Y,
As shown by arrows in the figure, scanning is performed sequentially from the low frequency component side to the high frequency component side.

As is clear from the above explanation, in the image processing apparatus of this embodiment', a single scanning area Y that matches the distribution of significant conversion coefficients can be set for each block, and therefore, the scanning area Y can be set for each block. It is possible to prevent complicated control over transmission of transform coefficients due to division, thereby enabling highly efficient transmission with simple control.

In the above embodiment, it has been shown that the discrete cosine transform (DCT) is used as the orthogonal transform means, but the orthogonal transform means itself may be, for example, a sine transform (DS).
It is possible to use various known methods such as T) and Hadamard transformation.

[Effects of the Invention] The image processing device according to the present invention uses the region division selection circuit to scan the transform coefficients obtained by the orthogonal transform circuit for each block in order to match the significant transform coefficient distribution of each block. Determine the scanning area.

In that case, the scanning area is determined by determining the shape and size, but the scanning area shape is determined in advance by dividing the area into a single scanning area that matches the distribution of significant conversion coefficients. This is achieved by selecting one of the three standard forms prepared for the selection circuit, and
The size of the scanning area is determined by appropriately expanding or contracting the size of the standard form selected by the form determination.

Therefore, in the image processing device according to the present invention,
A single scanning area that matches the distribution of significant transformation coefficients can be set for each block, and therefore, it is possible to prevent complicated control over transmission of transformation coefficients due to division of the scanning area, and thus, Highly efficient transmission can be achieved through simple control.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing a schematic configuration of an embodiment of the present invention, FIG. 2 is an explanatory diagram of a scanning area set by a conventional device, and FIGS. 3(a), (b), and (C) are respectively Explanatory diagram of the standard form of the scanning area in one embodiment, FIG. 4(a), (
b) is an explanatory diagram of the scanning region setting operation of the region division selection circuit in one embodiment, and FIG. 5 is an explanatory diagram of the scanning region set in the aforementioned one embodiment. lot...Input terminal, 102...Orthogonal transformation circuit, 1
03... Area division selection circuit, 104... Coefficient scanning circuit, 105... Encoder, 106... Scan completion detection circuit, 107... Code output terminal, X...・
Conversion surface, 310°320.330,340,350,3
60... Conversion coefficient group, r, n, m...
Standard form, Y...Scanning area. Diagram explaining the scanning area set by the conventional device

Claims (1)

[Claims] An orthogonal transform circuit that orthogonally transforms image information in units of blocks each consisting of n rows and n columns of pixels, and scans the transform coefficients obtained by the orthogonal transform circuit into a significant transform coefficient distribution for each block. An image processing device that compresses and transmits image information, comprising an area division selection circuit that determines a scanning area for each block in order to match the image information, and the area division selection circuit is configured to perform orthogonal transformation. The transform coefficients of the block are classified according to the magnitude relationship of frequencies and subdivided into a plurality of transform coefficient groups extending in a diagonal direction. The three located transform coefficients are extracted, and the shape and size of a single scanning region in each block is determined based on the comparison result between each extracted transform coefficient and a threshold value, and the To determine the form of the scanning area, three standard forms are prepared in advance as the form of a single scanning area, which correspond to the results of comparison between the three transform coefficients extracted from the group of transform coefficients and the threshold value. The size of the scanning area is determined by selecting one of the applicable standard formats based on the actual comparison results between the three extracted conversion coefficients and the threshold value. I did 3
An image characterized by expanding or contracting the size of the selected standard form as appropriate based on the positions of the transform coefficients at both ends of the transform coefficient group where all of the transform coefficients are larger than a threshold value. Processing equipment.