JPH0662259A - Binary image encoding/decoding system - Google Patents

Abstract

PURPOSE:To improve the efficiency of compression by subjecting an input image to the plural kinds of conversion processing, allocating plural encoding systems to the respective processing results and selecting the conversion processing and the encoding system out of the conversion processing results. CONSTITUTION:The input binary image of one scanning line is divided into small blocks by a separation part 1 and sent to a through conversion part 2 and a convolusion conversion part 3. The conversion part 2 performs through conversion to the image and sends it to a one-block delay memory 4 and a through conversion change picture element counter 5 and at the same time, the conversion part 3 performs convolusion conversion and sends it to a one-block delay memory 6 and a convolusion conversion change picture element counter 7. A small block judge part 8 reads the numbers of changed picture elements from the counters 5 and 7 and compares those values. As the result of comparison, when the value of the counter 5 is larger than that of the counter 7, a mode signal addition part 14 is instructed to add a mode signal for selecting and outputting the convolusion conversion, and an encoder selecting selector 13 selects the output of a one-dimensional encoder 10. When the value of the counter 7 is larger than that of the counter 5, the selector 13 selects the output of a two-dimensional encoder 12.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a binary image coding / decoding system capable of high-speed transmission and large-capacity storage.

[0002]

2. Description of the Related Art Conventionally, as a method for encoding a black and white binary image signal obtained by scanning a document such as white and black with a scanner, a modified Huffman (an international standard method of facsimile encoding) Modified Hu
The ffman method (hereinafter referred to as MH method) and the modified read method (hereinafter referred to as MR method) are known. The MH method counts the run lengths of white pixels and black pixels of only the scan lines to be encoded, and encodes the run lengths of the white pixels and black pixels. It is a typical one-dimensional encoding method. is there. Further, the MR system encodes the relative positional relationship of a change pixel from a white pixel to a black pixel or from a black pixel to a white pixel between the scan line to be encoded and the immediately previous scan line that has already been encoded. This is a typical two-dimensional encoding method.

Although these methods are excellent as means for encoding images of black and white such as documents and drawings, they are halftone images often used in newspapers and pseudo halftone images called dither method. Is not suitable for encoding. This is because the halftone image is artificially represented by the white pixels and the black pixels in the halftone dot photograph and the dither, so that the run lengths of the white pixels and the black pixels are shortened and the number of runs is increased. Further, the correlation between the change pixel positions of the already-encoded scan line and the scan line to be encoded from the white pixel to the black pixel or from the black pixel to the white pixel is reduced. However, as a method for encoding the halftone image or the pseudo halftone image subjected to the dither processing, the periodicity of the binarized image is used to convert the binarized image to convert the white pixel and the black pixel. A method is known in which the run length is lengthened and the binary image after conversion is encoded by the MH method. This conversion means is, for example, Japanese Patent Application No. 56-24836 "Image Signal Encoding Device".
Are described in detail in. Hereinafter, this conversion means will be referred to as convolutional conversion.

Further, in general binary images, there are many binary images in which a black and white binary image such as a document or a drawing and a pseudo halftone image represented by a halftone dot photograph or dither are mixed. Conventionally, the encoding method for encoding a binary image in which these black and white binary image and pseudo-halftone image are mixed has been as follows. First, 2 of the scan lines to be encoded that were scanned by the scanner
The value image is divided into a plurality of small blocks. Then, the above-mentioned convolution conversion is applied to each of the divided small blocks.
Here, even when the image scanned by the scanner is not converted, it is referred to as a through conversion as a kind of conversion means.
That is, two types of conversion processing, convolutional conversion and through conversion, are performed on each small block that is scanned by the scanner and divided into small blocks. Then, for each small block, the number of changed pixels of the binary image after being converted by each conversion means is counted, and the number of changed pixels is compared.
The binary image subjected to the processing having the smaller number of changed pixels counted as the value image is selected. After that, the selected plurality of converted small blocks are collectively one-dimensionally encoded as an image of one scanning line, and the selection result for each small block is added as a mode signal on the code string.

[0005]

As described above, according to the conventional encoding method, a binary image in which a black-and-white binary image such as a document or a drawing and a pseudo halftone image represented by a halftone dot photograph or dither are mixed is generated. In the case of encoding, a plurality of selected small blocks after conversion are collectively subjected to one-dimensional encoding as an image of one scanning line.

In general, an image obtained by performing a convolution conversion process on a pseudo-halftone image has almost no correlation between scanning lines. Therefore, when an image is encoded by a two-dimensional encoding system as an encoding system, it is compared with a one-dimensional encoding system. On the contrary, the coding efficiency is reduced. However, with respect to an image obtained by through-converting an image such as a document or a drawing, the coding efficiency is higher when the two-dimensional coding method is used for the coding than when the one-dimensional coding method is used for the coding. Therefore, when encoding a plurality of selected small blocks after conversion, the through-transformed block is a two-dimensional encoding method such as an MR encoding method rather than a one-dimensional encoding such as run-length encoding. It is possible to obtain a higher compression ratio by coding a small block by utilizing the correlation with the already coded block using. However, in the above-described conventional encoding method, one scanning line is composed of a plurality of small blocks after conversion, and one-dimensional encoding is performed in units of one scanning line. All 1
There is a problem that the compression efficiency cannot be improved because it is dimensionally encoded.

[0007]

A binary image encoding / decoding method of the present invention divides a binary image into a plurality of small blocks, and performs a plurality of conversion processes for each small block. The conversion method obtained by the means is preliminarily determined as an encoding method for one-dimensional encoding or two-dimensional encoding, the number of changed pixels of each conversion processing result is obtained, and each of the obtained conversion processing results is calculated. The conversion processing means that minimizes the number of changed pixels is selected in units of the small blocks, the conversion processing result obtained by the selected conversion processing means is encoded by the predetermined encoding method, and the plurality of conversion processing means are selected. A binary image coding apparatus for adding which conversion processing means has been selected among the conversion processing means to the coding result as a mode signal for each small block, and a signal coded by this binary image coding apparatus. Then, a mode signal indicating the conversion processing means selected in each small block is restored from the received signal, and the coding method selected in each small block is one-dimensional coding or two-dimensional according to the restored mode signal. It is determined whether to encode, the decoding result is restored by a decoding method corresponding to the encoding method determined for each small block, and the restored conversion processing result is inversely transformed according to the restored mode signal. It is characterized by comprising an inverse transform processing means and a binary image decoding device for synthesizing a plurality of small blocks inversely transformed by the inverse transform processing means to obtain a binary image.

Further, the binary image coding apparatus divides an input binary image of one scanning line into a plurality of small blocks, and first and second division processing for converting the divided small block images. Conversion processing units, first and second conversion change pixel counters that count the number of change pixels in a small block of the image after the conversion process, and changes read from the first and second conversion change pixel counters. A small block determination unit that compares the number of pixels; first and second 1-block delay memories that store the converted images for one block of the first and second conversion processing units; According to a comparison result of a two-dimensional encoder for two-dimensionally encoding the image of the block delay memory, a one-dimensional encoder for one-dimensional encoding of the image of the second one-block delay memory, and a small block determination unit. The first and second one block A small block selection selector for selecting one of the delay memory outputs, a reference line memory for storing the converted one block image selected by the small block selection selector, and a comparison result of the small block determination unit. An encoder selection selector that selects one of the two-dimensional encoder and the one-dimensional encoder output, and a mode signal addition unit that adds a mode signal to the encoder selection selector output according to the comparison result of the small block determination unit. The first conversion processing unit is a through conversion unit that performs through conversion of the input image of the small block, and the second conversion processing unit is a convolution conversion unit that performs convolution conversion of the input image of the small block. Good or the second
The conversion processing unit may be configured by a filter that removes the periodicity of the pseudo halftone image.

Further, the binary image decoding device restores a mode signal separation unit for separating the mode signal from the received code string, and a designation of the conversion process for the small block selected at the time of encoding from the mode signal. Mode restoring section, a two-dimensional decoder, a one-dimensional decoder for two-dimensionally and one-dimensionally decoding the code sequence into which the mode signal is separated by the designation of the mode restoring section, and the two-dimensional decoder, 1.
The first and second inverse transform units for respectively inverse transforming the results decoded by the dimensional decoder and the plurality of small blocks from the first and second inverse transform units are combined into one scanning line, and 2 A synthesizing unit for outputting a value image, the first inverse transforming unit is a through inverse transforming unit for performing through inverse transform on the output of the two-dimensional decoder, and the second inverse transforming unit is the one-dimensional decoder. It may be a convolutional inverse transformation unit for convolutionally inverse transforming the output, or the first inverse transformation unit may be a notch / isolated point removal filter for improving the image quality of a document drawing.

[0010]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, the present invention will be described with reference to the drawings. 1 and 2 are block diagrams showing a binary image coding device and a binary image decoding device in an embodiment of a binary image coding / decoding system of the present invention.

The binary image coding apparatus according to the present embodiment uses, as conversion processing means, a through conversion unit 2 for performing a through conversion process and a convolution conversion unit 3 for performing a convolution conversion process.

In FIG. 1, first, the through conversion change pixel counter 5 and the convolution conversion change pixel counter 7 are reset. This reset is performed at the start of encoding one block.

Next, the separating unit 1 divides the input binary image of one scanning line into a plurality of small blocks. The divided small blocks are sent to the through conversion unit 2 and the convolution conversion unit 3.

The through conversion unit 2 performs through conversion of the input small block image and sends it to the one-block delay memory 4 and the through conversion change pixel counter 5. The through conversion changed pixel counter 5 counts the number of changed pixels in a small block of the image after through conversion.

At the same time, the convolutional conversion unit 3 performs convolutional conversion on the input small block image and sends it to the one-block delay memory 6 and the convolutional conversion changed pixel counter 7. The convolution conversion changed pixel counter 7 counts the number of changed pixels in a small block of the image after the convolution conversion.

When the conversion of small blocks and the count of changed pixels are completed, the small block determination unit 8 reads the number of changed pixels from the through conversion changed pixel counter 5 and the convolution converted change pixel counter 7. Then, the read changed pixel numbers are compared.

As a result of the comparison, if the value of the through conversion change pixel counter 5 is larger than the value of the convolution conversion change pixel counter 7, the mode signal adding section 14 is instructed to add the mode signal to select and output the convolution conversion. Let next,
The output of the encoder selection selector 13 selects the output of the one-dimensional encoder 10. Then, the small block selection selector 9 selects the output of the 1-block delay memory 6 which is the output of the convolution conversion unit 3.

If the value of the convolution conversion change pixel counter 7 is larger than the value of the through conversion change pixel counter 5 as a result of the comparison, the mode code adding unit 14 is instructed to select and output the through conversion. Is added. Next, the output of the encoder selection selector 13 is the two-dimensional encoder 12
Select the output of. Then, the small block selection selector 9
Then, the output of the 1-block delay memory 4, which is the output of the through conversion unit 2, is selected.

The one-dimensional encoder 10 performs one-dimensional encoding such as run-length encoding on the converted image for one block stored in the one-block delay memory 6. At the same time, the two-dimensional encoder 12 performs two-dimensional encoding such as MR encoding on the converted image for one block stored in the one-block delay memory 4 by referring to the image stored in the reference line memory 11.

The converted image of one block is coded in block units and stored in the reference line memory 11 for reference of coding of the next scanning line, and the contents are updated in one block units. However, the capacity of the reference line memory 11 is set so that the converted image for one scanning line can be stored. In the two-dimensional encoding, the converted image of the small block corresponding to the position immediately above is referred to.

The above operation is continued until one image is completed.

The binary image decoding apparatus in this embodiment is provided on the receiving side, receives the signal encoded by the binary image encoding apparatus on the transmitting side shown in FIG. 1, and decodes and converts the signal. It is for obtaining the previous binary image.

In FIG. 2, first, the mode signal separating section 20 is shown.
Separates a mode signal indicating a conversion process for a small block from the received code string and sends the separated mode signal to the mode restoring unit 21. The code string from which the mode signal has been separated is sent to the one-dimensional decoder 23 or the two-dimensional decoder 25 via the code demultiplexer 22. Since the mode signal is located at the head of each small block, the mode signal is separated and the mode signal is transferred at the head of each small block.

Next, the mode restoring unit 21 is used by the mode separating unit 2
Upon receiving the mode signal separated from 0, the conversion processing designation for the small block selected at the time of encoding is restored, and the output destination of the code demultiplexer 22 is determined from the restored conversion processing designation. Here, if the conversion processing designation of the received small block is through conversion, the received code string is input to the two-dimensional decoder 25 to obtain the decoded converted processing result. If the received conversion processing designation of the small block is the convolutional conversion, the received code string is converted into the one-dimensional decoder 23.
To obtain the processing result after conversion which is decoded by the one-dimensional decoder 23.

In the one-dimensional decoder 23, the mode separation unit 20
The code string received from is restored to the processing result after conversion. The decoding method of the one-dimensional decoder 23 is a method of restoring the encoding used in the one-dimensional encoder 12 on the opposite transmission side. Further, the one-dimensional decoder 23 performs the processing until the restored converted processing result has the number of pixels of a small block. The restored converted processing result is sent to the convolutional inverse conversion unit 27, and the reference line memory 24 is used for reference of decoding of the next scanning line.
, And the contents are updated in block units. However, the capacity of the reference line memory 24 is set so that the processing result after conversion restored for one scanning line can be accumulated.

In the two-dimensional decoder 25, the mode separation unit 20
The stored contents of the reference line memory 24 of the code string received from are referenced to restore the converted processing result. The image to be referred to is the image after the conversion of the small block corresponding to the position immediately above. The decoding method of the two-dimensional decoder 25 is a method of restoring the coding used in the two-dimensional encoder 10 of the opposite transmitter. In addition, the two-dimensional decoder 25 performs the processing until the restored converted processing result has the number of pixels of a small block. The restored converted processing result is sent to the through inverse conversion unit 26, stored in the reference line memory 24 for reference of decoding of the next scanning line, and the content thereof is updated in a block unit.

The converted processing result decoded by the one-dimensional decoder 23 is input to the convolutional inverse conversion unit 27 and inversely converted to obtain an image signal before conversion. Further, the processing result after the conversion decoded by the two-dimensional decoder 25 is inversely converted by the through inverse conversion unit 26 to obtain the image signal before conversion. Here, the inverse conversion processing in the through inverse conversion unit 26 is the reverse processing of the through conversion unit 2 provided on the transmission side so that the binary image before conversion can be restored from the restored converted processing result. is there. Further, the inverse transform process in the convolution inverse transform unit 27 is a process reverse to that of the convolution transform unit 3 provided on the transmission side so that the binary image before conversion can be restored from the restored post-conversion process result. .

Next, when the inverse conversion processing is completed for all the small blocks of one scanning line, the synthesizing unit 28 synthesizes a plurality of inversely converted small blocks into one scanning line, and the synthesized one scanning line is processed. Output a binary image.

The above operation is continued until one image is completed.

As the conversion processing for the pseudo halftone image, not only the convolution conversion processing shown in this embodiment but a filter for removing the periodicity of the pseudo halftone image may be used. Further, as the conversion process for the document and the drawing, not only the through conversion process but also a filter for improving the image quality of the document drawing such as a notch and isolated point removal filter may be used.

[0031]

As described above, according to the present invention, the compression efficiency of a binary image in which a black-and-white binary image such as a document or a drawing and a pseudo halftone image represented by a halftone dot photograph or dither are mixed is improved. The effect that it can be raised is obtained.

[Brief description of drawings]

FIG. 1 is a block diagram showing a binary image encoding device in an embodiment of a binary image encoding / decoding system of the present invention.

FIG. 2 is a block diagram showing a binary image decoding device in an embodiment of a binary image encoding / decoding system of the present invention.

[Explanation of symbols]

 1 Separation Unit 2 Through Conversion Unit 3 Convolution Conversion Unit 4, 6 1 Block Delay Memory 5 Through Conversion Change Pixel Counter 7 Convolution Conversion Change Pixel Counter 8 Small Block Judgment Unit 9 Small Block Selector 10 1-Dimensional Encoder 11, 24 Reference Line Memory 12 Two-dimensional encoder 13 Encoder selection selector 14 Mode signal addition unit 20 Mode signal separation unit 21 Mode restoration unit 22 Code demultiplexer 23 One-dimensional decoder 25 Two-dimensional decoder 26 Through inverse transform unit 27 Convolution inverse transform unit 28 Synthesis Department

Claims (7)

[Claims] 1. A binary image is divided into a plurality of small blocks,
An encoding method for one-dimensional encoding or two-dimensional encoding of the conversion processing results obtained by the plurality of conversion processing means for performing the conversion processing on each small block unit is determined in advance, and each conversion processing result Obtain the number of changed pixels,
The conversion processing means that minimizes the number of changed pixels in each of the obtained conversion processing results is selected for each small block, and the conversion processing results obtained by the selected conversion processing means are selected by the predetermined encoding method. A binary image encoding apparatus that encodes and adds which conversion processing means is selected from the plurality of conversion processing means to the encoding result as a mode signal for each of the small blocks, and this binary image encoding A signal encoded by a device is received, a mode signal indicating the conversion processing means selected in each of the small blocks is restored from the received signal, and an encoding method selected in each of the small blocks according to the restored mode signal. Is one-dimensional coding or two-dimensional coding, restores the coding result by a decoding method corresponding to the coding method determined for each small block, and restores the converted processing result. Is inversely transformed in accordance with the restored mode signal, and a binary image decoding device for synthesizing a plurality of small blocks inversely transformed by the inverse transform processing means to obtain a binary image. Characteristic binary image encoding / decoding method. 2. The binary image encoding device divides an input binary image of one scanning line into a plurality of small blocks, and first and second dividing processes of the divided small block images. Conversion processing units, first and second conversion change pixel counters that count the number of change pixels in a small block of the image after the conversion process, and changes read from the first and second conversion change pixel counters. A small block determination unit that compares the number of pixels; first and second 1-block delay memories that store the converted images for one block of the first and second conversion processing units; According to a comparison result of a two-dimensional encoder for two-dimensionally encoding the image of the block delay memory, a one-dimensional encoder for one-dimensional encoding of the image of the second one-block delay memory, and a small block determination unit. The first and second blocks are delayed A small block selection selector for selecting one of the extended memory outputs, a reference line memory for storing the image of one block after conversion selected by the small block selection selector, An encoder selection selector that selects one of a two-dimensional encoder and a one-dimensional encoder output, and a mode signal addition unit that adds a mode signal to the encoder selection selector output according to the comparison result of the small block determination unit. The binary image encoding / decoding method according to claim 1, wherein 3. The first conversion processing unit is a through conversion unit that performs through conversion of the input image of the small block,
3. The binary image encoding / decoding system according to claim 2, wherein the second conversion processing unit is a convolution conversion unit that performs convolution conversion of the input image of the small block. 4. The binary image encoding / decoding system according to claim 2, wherein the second conversion processing unit is composed of a filter for removing the periodicity of the pseudo halftone image. 5. The binary image decoding device includes a mode signal separation unit that separates the mode signal from a received code string,
A mode restoration unit that restores the designation of the conversion process for the small block selected at the time of encoding from the mode signal,
A two-dimensional decoder and a one-dimensional decoder for two-dimensionally and one-dimensionally decoding the code sequence in which the mode signal is separated by the designation of the mode restoration unit, and the two-dimensional decoder and the one-dimensional decoder for decoding The first and second inverse transforming units for respectively inversely transforming the obtained results and the plurality of small blocks from the first and second inverse transforming units are synthesized into one scanning line to output a binary image. The binary image encoding / decoding method according to claim 1, further comprising: 6. The first inverse transform unit is a through inverse transform unit that performs through inverse transform of the two-dimensional decoder output, and the second inverse transform unit performs convolutional inverse transform of the one-dimensional decoder output. 6. A convolutional inverse transformation unit, wherein
The described binary image encoding / decoding method. 7. The binary image encoding / decoding method according to claim 5, wherein the first inverse conversion unit is a notch for improving image quality of a document drawing and an isolated point removal filter.