JPS6174455A - Coding control system of binary picture signal - Google Patents

Abstract

PURPOSE:To code a binary dither picture by using a pattern of a dither matrix period so as to attain forecasting and switching a document picture and an intermediate tone picture so as to increase the compression rate even to a picture where the document and intermediate tone picture exist in mixture. CONSTITUTION:A dither picture signal (a) mixed with the document and intermediate picture from an input terminal 3 of a dither picture signal of a coding control system of a binary picture signal is fed to a ROM 6 via shift registers 4, 5. Further, a line number from an input terminal 12 of a line number and a history signal (h) from a register 7 are fed to the ROM 6 as an address. The ROM 6 and a register 7 detect whether a specific pattern appearing easily in the pattern comprising plural picture elements appears in the signal (a) or not. Further, the output of the register 4 is fed to a variable length coder 8 and an MH coder 9, and forecast coding is applied by the gradation given to the pattern by the coder 8. Outputs of the register 7 and the coders 8, 9 are fed to a multiplexer 10 to switch the document and intermediate tone pictures thereby applying coding to the binary dither picture.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to a binary image signal encoding control method for encoding and band-compressing an image signal containing a mixture of a halftone image and a document, etc. be.

[Conventional technology]

A systematic dither method is known as a method for expressing a halftone image in black and white binary values. In this dither method, as shown in FIG. 8, a multilevel image signal rxy is compared with a dither signal Dxy whose darkness value changes for each pixel in a comparator circuit 1 to generate a binary dither image signal Pxy. This binary dither image signal Pxy is data compressed in the encoding circuit 2.

In the systematic dither method, the dither signal Dxy is set at pixel position x,
y, and in the case of 16 gradations, it is given by the matrix shown in FIG. In the figure, (a) is the Bayer type, (b) is the halftone type, and (C1 is called the spiral type. This dither matrix is the same one that is used repeatedly over the entire screen, A binary black and white dithered image signal is formed from the multivalued image signal by the comparator circuit 1.

The digital facsimile method is normally used to transmit black and white binary document images, but when transmitting halftone images, it is sufficient to convert them into the dithered images described above and transmit them. Furthermore, in the digital facsimile system, data compression processing is performed to shorten transmission time. The international standard MH method (Modif) is used as this data compression method.
ied Huff-man method) and MR method (Mod
IFID READ method) has been proposed.

However, these methods perform data compression using the correlation between adjacent pixels, and because they are targeted at documents, they are not efficient for dithered images where the correlation between adjacent pixels is small. It was impossible to perform good data compression.

For this reason, conventionally, as a method that is compatible with the MH method,
For example, (1) JP-A-57-181270, or (2) JP-A-58-194465 (or Institute of Electronics and Communication Engineers technical report IE82-45° "MH of dithered images").
A method of "encoding") has been proposed.

In the method (1) mentioned above, in a Heyer type dithered image,
This study focused on the fact that the following patterns (black 3. white l), (white 3. black 1), and (white 1° black 1) tend to occur, and for these patterns, as shown in Figure 10. , and the number of consecutive patterns is encoded. That is, (black 3. white 1) or (white 3. black 1)
Assign "01" to the pattern, (black 1. white 3)
Or, assign "00" to the pattern of (white 1. black 3), and assign "00" to the pattern of (black 1. white 1) or (white 1. black 1),
"10" is assigned to "10", and the end of this encoding is "110".
”, and “111” is assigned to run-length encoding.

FIG. 11 shows an example of the above-mentioned method, where (a) is an image signal, (b) is a coded signal, RL is a run-length MH encoding period, and PL is a pattern-length encoding period. The identification code rD indicates switching of the code system. White 4 in MH encoding period RL
The pixel is shown as PL4, and the pixel of black 2 is shown as PL2. Furthermore, when a predetermined number of patterns shown in FIG. 10 are consecutive, the pattern length encoding is switched to by the identification code ID, and white 3. PLI the pattern of black 1, two consecutive white 1. Black 1 pattern is PL3, white 3. The black l pattern is shown in PLI. When a predetermined number of patterns shown in FIG. 10 are no longer consecutive, the end code shown in FIG. 1O is used as the identification code ■D, and the code is switched to the MH code.

In addition, the method (2) described above uses predictive transformation as preprocessing for the MH method, and similarly to the method (1) described above,
If we decide to use image information only on the encoded scan line,
A pixel 4 pixels before, which has the same darkness value as the encoded pixel and has a strong correlation, is set as a reference pixel, and this is used as a prediction signal.

FIG. 12 is an explanatory diagram of the operation of this method, in which a prediction error signal e is formed by performing an exclusive OR between the dither image signal X and a prediction signal X' shifted by 4 pixels from the dither image signal Furthermore, a smoothed prediction error signal e'' is formed from this prediction error signal e by inverting black and white each time a prediction error occurs.

This smoothed predicted erroneous arrival signal e' is run-length encoded as shown by the arrow, and PL2. PL2.
PL12. PL4. PL3 encoding will be performed. Since this smoothed prediction error signal e'' has a relatively long run length, it can be efficiently compressed using the MH method.

[Problem that the invention seeks to solve]

The above-mentioned method (1)" encodes a mixed image including a document image and a halftone image by switching the coding system, and therefore enables relatively efficient data compression. Since the number of consecutive specific patterns in a dithered image is encoded, there is a drawback that the amount of code increases for image areas where the pattern changes rapidly due to large density changes. The Heyer type, which has a high resolution, is suitable for type recording devices, and the halftone type or spiral type, which can compensate for the deterioration of gradation characteristics due to dot collapse, is used for thermal recording devices and laser printers. ) is a Bayer type method, and has the drawback that it is not possible to select the dither type most suitable for the recording apparatus.

Also, in the method (2) mentioned above, since the pixel 4 pixels before is used as the prediction signal, (al~+c+
It is possible to compress the dithered image by creating it with all the 4×4 dither matrices shown in Figure 2, and it is possible to select the type of dithering that matches the recording device, but the reference pixel is far from the encoded pixel. Therefore, for document images in which there is a large correlation between adjacent pixels, there is a drawback that the compression ratio decreases.

The present invention aims to improve the above-mentioned conventional drawbacks.

[Means for solving problems]

The encoding control method for a binary image signal of the present invention includes means for detecting whether a specific pattern that is likely to appear among patterns consisting of a plurality of pixels with a dither matrix period has appeared in a human image signal; means for predictively encoding a pixel pattern to be encoded using a gradation given to the gradation; and means for detecting continuous appearance of a combination of specific patterns and switching from run-length encoding to the predictive encoding;
It encodes a binary dither image.

[Operation] The appearance of a dither matrix periodic pattern is detected to identify whether the mixed image is a document image or a halftone image, and based on this, switching between predictive coding and run-length coding is performed. For tone images, pseudo tone values are assigned to each pattern in the dithered image and predictive encoding is performed, and for document images, run length encoding is performed.

〔Example〕

Embodiments of the present invention will be described in detail below with reference to the drawings.

FIG. 1 is a block diagram of main parts of an embodiment of the present invention, in which 3 is an input terminal for a dither image signal, 4 and 5 are shift registers,
6 is read-only memory (ROM), 7 is register,
8 is a variable length encoder, 9 is an MH encoder, 10 is a multiplexer (MPX), 11 is an output terminal, and 12 is an input terminal for a line number signal. When 4 pixels of the dithered image signal a of a mixed image of a document and a halftone image are added to the input terminal 3 and sequentially shifted to the shift register 4, the 4 pixels that were previously shifted to the shift register 4 are Data b is shifted to shift register 5. The parallel output signals c and d of these shift registers 4.5 are pixel signals of the encoded pattern and the reference pattern, and are input to the read-only memory 6 as part of the address signal. Also, the line number signal e and the history signal ri from the register 7 are input to the read-only memory 6 as part of the address signal.

The line number signal e specifies the line number among the four lines using the lower two bits of the count output signal of the scanning line of the image. The history signal is a signal indicating the number of consecutive gradation patterns other than all white or all black.

The read-only memory 6 uses the above-mentioned human input signal as an address signal to record the appearance order of the encoded patterns.

A signal f indicating either a non-grayscale pattern or an end, a signal g specifying whether a predictive coding system or a run-length coding system is used in the present invention, and a history signal are read out and set in the register 7.

The variable length encoder 8 receives the image signal and the signal f set in the register 7, and performs predictive encoding of the image signal.The output signal i is sent to the multiplexer 1.
Added to 0. The MH encoder 9 receives the image signal j and performs MH encoding, and the output signal j is applied to the multiplexer 10. This multiplexer 10 is switched and controlled by the signal g set in the register 7.
A compressed code signal k is output to the output terminal 11.

When the encoded pattern is a gradation pattern other than all white or all black, a history signal that is +1 to the human history signal is read from the read-only memory 6, and if the number of consecutive specific patterns is 3, the human history signal is read out from the read-only memory 6. When the signal number becomes 3, the same history signal number of 3 is read out. Also, when the encoded pattern is all white, all black, or other than a gradation pattern, a history signal minus 1 from the human history signal is read out,
If the input history signal h7><o, the same 0 is read out. In this way, the history signal indicates the history of the encoding pattern. Furthermore, the MH encoder 9 and the multiplexer IO have three gradation patterns other than all white or all black.
When the history signal 1 indicates 3 or more consecutively, the switching control (2) from the code system of the M)(system to the predictive code system is performed by the signal g.

When the MH encoder 9 is instructed to switch to the predictive coding system by the signal g, it assumes that there is a change point in the image signal and outputs the MH code, and also outputs a signal indicating that it will enter the non-compression mode. Output j.

When the history signal becomes O, the variable length encoder 8 outputs the termination code "00000001T" shown in FIG. 5 by the signal f.
At the same time, the MH encoder 9 is controlled to perform encoding, and switching control between the MH system encoding system and the predictive encoding encoding system is performed.

Next, a case will be described in which a Bayer type dither image signal is used.

FIG. 2 shows a gradation pattern of a Bayer type dithered image, which can express 17 gradations from 0 to 16 with a 4×4 pattern. Moreover, when looking at each line L1 to L4, it can be seen that the pattern shown in FIG. 3 appears. For example, for line L1, gradation 0 is 0, gradation 1.2
is 1, gradations 3-8 are 2, gradations 9 and 10 are 3, gradations 11-
16 becomes 4. Also, for line L4, gradation levels O to 5
is 0, gradations 6 and 7 are 1, gradations 8 to 13 are 3, gradation 14.
15 becomes 3, and gradation 16 becomes 4. Therefore, gradations can also be assigned to a 1×4 pattern. This one
Using the x4 pattern of gradation, the pattern immediately before the encoding pattern is used as a reference pattern, and predictive encoding is performed using this reference pattern as a predicted value.

In a dithered image, patterns other than gradation patterns also appear, but for these non-gradation patterns, for example, if the pixels of each pattern are arranged in descending order of darkness value, as shown in FIG. are five gradation patterns (shown in the gradation 0 to 4 columns) related to lines L1 to L4.

On the other hand, there are 11 non-gradation patterns, and after approximating them to 5 gradation patterns and encoding these non-gradation patterns, the approximated gradation patterns are used as the next reference pattern. It is used.

When the reference pattern is a tone pattern, the same tone is most likely to appear in the encoded pattern. Encoding is performed using such rankings of likelihood of appearance (predicted rankings). Table 1 below shows the ranking of the gradations of the reference pattern and the gradations of the encoded pattern in terms of their likelihood of appearing.

Table 1 In this Table 1, R indicates the gradation of the reference pattern,
A indicates the order of appearance of the encoding pattern.

For example, when the reference pattern's gradation R is 1, the appearance order A of the gradation l of the encoding pattern is 1, the appearance order A of the gradation 2 is 2, the appearance order A of the gradation O is 3, and the appearance order A of the gradation 2 is 2. The appearance order A of tone 3 is 4, the appearance order A of tone 4 is 5, and if the tone R of the reference pattern is 4, the appearance order A of the encoded pattern is
1 to 5 are the ranks of the encoding pattern R) 514.3, 2,
1. The order is O.

The encoding shown in FIG. 5 is performed for the appearance order A of 1 to 5. In the same figure, ××××
indicates the original data of the non-gradation pattern, and T indicates the color of the next run, where "1" indicates black and "0" indicates white. As shown in Fig. 5, the shortest code "1" is assigned to the coding pattern with the highest appearance order, the long code "000011" is assigned to the coding pattern with the lowest appearance order, and the non-grayscale pattern is assigned the shortest code "1". is the code “000010x” in which the original data ×××× of the non-gradation pattern is inserted as is.
xxxl''.The start code is the code that enters the non-compressed mode, and the end code is the code that exits the non-compressed mode.The above-mentioned encoding is performed for the non-gradation pattern, and the next code is As the reference pattern used for the calculation, the approximated gradation pattern shown in FIG. 4 is used.

FIG. 6 is a detailed block diagram of the encoding circuit according to the embodiment of the present invention, in which 21 is an input terminal for a dither image signal;
5 is a 4-bit shift register, 26 is a line counter, and 27 is a glue-only memory for image judgment (
ROM), 29 is a glue-only memory for predictive conversion.
(ROM), 28° 30 is a register, 31 is an MH encoder, 32 is a variable length encoder, 33 is a multiplexer (MP
X), 34 is an output terminal.

Dithered image signal 21a of mixed image of document and halftone image
are applied to input terminal 21 and sequentially shifted to 4-bit shift registers 22-25, respectively. The line counter 26 counts the number of scanning lines of the mixed image and outputs a line number signal 26a indicating which of lines L1 to L4 the lower two bits are. , are added to the read-only memory 27, 29 as part of the address signal.

The read-only memory 27 also contains a 2-bit line number signal 26a from the line counter 26, a 4-bit image signal 22a from the shift register 22, and a 4-bit image signal 22a from the register 22.
A 3-bit history signal 28b set to 8 is added as an address signal, and a signal 28a is set in the register 28.

28b is what is read out. A history signal 28b determines whether the gradation pattern has been input three times in a row, for example.
The history signal 28b is determined by the transition using 3
When , the signal 28a is applied to the multiplexer 33 as an image judgment switching control signal to switch to predictive coding.

Further, the above-mentioned line number signal 26a, the 4-bit image signal 24a from the shift register 24, and the 4-bit image signal 25a from the shift register 25 are added to the read-only memory 29 as address signals, and the register Signals 30a, 30b, which are centered at 30, are read out. 4-bit image signal 24a from shift register 24
indicates the encoding pattern, and 4 from shift register 25
The bit image signal 25a indicates a reference pattern. A signal 30a is read from the read-only memory 29 and sent to the register 30.

30b is input to a variable length encoder 32, and a pattern prediction code 32a is output from the variable length encoder 32. Further, the signal 30a is applied to the multiplexer 33 as a non-gradation pattern switching control signal.

The image signal 25a from the shift register 25 is applied to the MH encoder 31, and an MH code signal 31a is output. Therefore, the MH code signal 31a from the MH encoder 31, the pattern prediction code signal 32a from the variable length encoder 32, and the original data 24a of the image signal from the shift register 24 are applied to the multiplexer 33, and the signal 28a
, 30a, and the compressed code signal is output to the output terminal 34. In the case of a civilian image, since the history signal 28b is 0, the MH code signal 31a is selectively output, and the halftone image is output. In this case, the gradation pattern is continuous and the history signal 28b is 3, so the predicted code signal 32
a or the original data 24a is selectively output, and in the case of a non-gradation pattern, as shown in the code of the non-gradation pattern in FIG. 33, and in the case of a gradation pattern, the pattern prediction code signal 32a is selectively output.

FIG. 7 is a block diagram of a decoding circuit according to an embodiment of the present invention, in which 41 is an input terminal of a compression code, 42 is a demultiplexer (DMPX), 43 is a fixed length decoder, 44 is an MH decoder, and 45 is a line 46 is a memory (ROM) for predictive inverse conversion; 47 is a shift register; 4
8 is a multiplexer (MPX), and 49 is an output terminal for the decoded dither image signal. The compressed code signal is passed through a demultiplexer 42 to a fixed length decoder 43. MH decoder 4
4 and multiplexer 48.

The fixed length decoder 43 receives the pattern prediction code, and
It identifies whether it is a gradation pattern or a non-gradation pattern, and in the case of a non-gradation pattern, outputs a control signal to selectively output the original data and add it to the multiplexer 48, and outputs a control signal to control the demultiplexer 48. 42 and multiplexer 4
8, a prediction error signal is also output and added to the read-only memory 46.

The read-only memory 46 stores the above-mentioned prediction error signal,
The line number signal of the lower 2 bits of the count obtained by counting the scanning lines of the image by the line counter 45 and the 4-bit reference pattern data from the shift register 47 are input as address signals, and the predicted inverse conversion data is read out. and added to the shift register 47.

The MH decoder 44 also receives an MH code signal and decodes it, and also receives a detection signal indicating whether it is in an uncompressed mode or a compressed mode.
In addition to the demultiplexer 42 and the multiplexer 48, the demultiplexer 42 selectively outputs the input compressed code signal to the MH decoder 44 according to the detection signal in the case of compressed mode, and performs fixed length decoding according to the detection signal in the case of uncompressed mode. The input compressed code signal is selectively outputted to the unit 43.

Further, the multiplexer 48 selectively outputs the decoded output signal of the MH decoder 44 according to the detection signal of the compressed mode, and in the case of the non-compressed mode and the gradation pattern, the multiplexer 48 selects and outputs the decoded output signal of the MH decoder 44 according to the detection signal of the compressed mode.
7 output signals are selectively output, and in the case of non-compression mode and non-gradation pattern, the original data is selectively output. Therefore, the dither image signal is decoded and outputted to the output terminal 49, and the compressed code encoded by the encoding circuit shown in FIG. 6 can be decoded.

In the above embodiment, five gradation patterns are used as patterns in which a 1×4 pattern is likely to appear, but in terms of the circuit configuration, it is preferable to use 2 to the power of 2. Since there are many patterns, eight patterns that are likely to appear can be used. Further, although the case of Heyer type dither is described, it can be similarly applied to cases of other types of dither. Furthermore, although the case where a variable length code is used in predictive encoding is explained, since the prediction rank signal (prediction error signal) including a non-grayscale pattern has the property that the first rank appears concentratedly, Conventional means can also be applied to the encoding to improve the compression ratio. Also, if the prediction rankings are grouped into m units, and all m predictions are in the first ranking, the code "0" is given.
”, and if there is a rank other than the first rank among the m, it is encoded using a 1-bit identification code “1” and the m variable length codes mentioned above. It is also possible to do so.

〔Effect of the invention〕

As explained above, according to the present invention, prediction is performed using a dither matrix cycle pattern to switch between a document image and a halftone image. This has the advantage that a large compression ratio can be obtained even for images containing a mixture of grayscale images and halftone images. Again-
If the MH method is used as dimensional run-length encoding, it can be made compatible with international standard data compression methods.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of the main part of an embodiment of the present invention, FIG. 2 is an explanatory diagram of the gradation pattern of the Heyer dither, FIG. 3 is an explanatory diagram of the gradation pattern of each line of the Bayer dither, and FIG. The figure is an explanatory diagram of a gradation pattern and a non-gradation pattern that approximates the gradation, FIG. 5 is an explanatory diagram of code assignment for predictive coding, FIG. 6 is a detailed block diagram of an embodiment of the present invention, and FIG. is a block diagram of a decoding circuit according to an embodiment of the present invention, FIG. 8 is a block diagram for explaining the formation of a dithered image signal, FIGS. diagram,
FIGS. 11 and 12 are explanatory diagrams of an example of a conventional encoding operation, respectively. 3 is an input terminal for dithered image signals, 4 and 5 are shift registers, 6 is a read-only memory (ROM), 7 is a register, 8 is a variable length encoder, 9 is an MH encoder, 10 is a multiplexer (MPX), 11 is the output terminal of the compression code, 1
2 is an input terminal for a line number signal, 21 is an input terminal for a dither image signal, 22 to 25 are 4-bit shift registers,
26 is a line counter, 27 is a read-only memory (ROM) for image judgment, 29 is a read-only memory (ROM) for predictive conversion, 28.30 is a register, 3
1 is an MH encoder, 32 is a variable length encoder, 33 is a multiplexer (MPX), and 34 is an output terminal.

Claims (1)

[Claims] Means for detecting whether a specific pattern that is likely to appear among patterns consisting of a plurality of pixels with a dither matrix period has appeared in an input image signal, and predicting a pixel pattern to be encoded based on the gradation given to the pattern. and means for detecting continuous appearance of the combination of the specific patterns and switching from run-length coding to the predictive coding, when the specific pattern is included in the binary dither image signal. An encoding control method for a binary image signal, characterized in that encoding is performed using the predictive encoding described above.