JPH07298062A - Data compression decoding system - Google Patents

Abstract

PURPOSE:To compress image data with high correlation in an excellent way without decreasing the processing speed by combining a run pack coding section with a universal coding section independently of statistic property of an image. CONSTITUTION:When a white or black level picture element appears in input image data D, a white/black run counter section 101 counts the length when runs of white or black level picture elements are consecutive, and a comparator section 102 receives the count N and compares the count N with a predetermined threshold level. Then data are coded by either of a universal coding section 103 and a run pack coding section 104 selected by a control code U/R. When the control code is U, the universal coding section 103 is selected and when the control code is R, the run pack coding section 104 is selected, and the data are coded based on each input value. Thus, an image having long consecutive runs such as a text or an image changing in details such as a dither image is compressed in an excellent way with simple configuration.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data compression and decoding system for a print image using a universal code such as a printer or a facsimile.

[0002]

2. Description of the Related Art In recent years, due to the development of computers and image input / output devices, data including various images such as texts, figures, halftone dots and dither images have been handled in the field of OA. Further, as the resolution of images has been increased, the amount of image data has become enormous. Therefore, a data compression method for compressing various and high-resolution image data is currently required. Since the statistical properties of the image data differ locally in the image data as described above, data with universal coding that does not significantly reduce the data compression efficiency even if the statistical properties of the data change as the data compression process progresses. A compression method is desired. As such a data compression method, by constructing a reference dictionary that optimizes the input data from the already processed part of the input data, the optimum compression can be achieved without depending on the statistical property of the data. A possible Lempel-Ziv encoding method has been proposed.

The Lempel-Ziv encoding method is classified into a dynamic dictionary method and a slide dictionary method according to the method of creating a reference dictionary (specifically, the interface 1
992 vol. 8 No. 183 "Data compression algorithm and its implementation"). In the dynamic dictionary method, the input data string is decomposed into partial data strings by a method called incremental decomposition, the partial data strings are registered in the dictionary, and the encoding process of the input data proceeds while referring to the dictionary. It is an algorithm to go. The slide dictionary method is an algorithm in which encoded data strings are sequentially stored in a reference buffer and the encoding process of input data proceeds while referring to the reference buffer as a dictionary. The dynamic dictionary method is superior to the slide dictionary method due to the ease of the algorithm,
On the contrary, the slide dictionary method is said to be superior in compression rate to the dynamic dictionary method, and is widely used in the field of data compression. Furthermore, as an improvement of the dynamic dictionary method, LZW (L
LZSS (Lempel-Ziv-S) as an improvement of the empel-Ziv-Welch) code and the slide dictionary system.
Torer-Szymanski) codes have been proposed.

The LZW encoding / decoding process will be described with reference to FIG. Here, each word unit in the input image data will be referred to as a character, and an arbitrary number of characters will be referred to as a character string.

FIG. 4 shows the flow of LZW encoding processing.
It shows in (a). First, the reference dictionary is initialized (S40).
1). The reference dictionary is initialized by registering all possible characters in the input image data in the dictionary entry. Next, image data is input (S402), and the longest character string that matches the input character string is searched from the dictionary (S4).
03). Then, the index attached to the searched character string is encoded and output (S404). At this time, since the code must be able to completely identify the dictionary entry, the code length must be at least an integer value of log (M) or more, where M is the current number of dictionary entries. However, the base of the logarithm at this time is 2. Next, a new character string will be registered in the dictionary, but generally the upper limit is set on the number of entries registered in the dictionary due to the limitation of the hardware scale. Usually, the number is about 2 to the 16th power. Therefore, it is determined whether the number of entries in the dictionary has reached the predetermined size (S405),
When the number of entries is the maximum, the process returns to S401, the dictionary is reinitialized, and then the encoding process is performed. If the number of entries has not reached the maximum, a new character string is registered in the dictionary (S406). The character string registered here is the character to be input next at the end of the character string encoded immediately before.
It is a combination of letters. Then, when the dictionary is updated, the process returns to S402 again to continue the sequential encoding process,
The processing is continued until there is no image data to be processed.

FIG. 4 shows the flow of LZW decoding processing.
It shows in (b). First, similarly to the case of the encoding process, the reference dictionary is initialized by registering all possible characters in the image data in the dictionary entries (S411). Next, the compressed code is input (S412), and the index of the dictionary is obtained from the code. A character string registered in the dictionary is searched from the obtained index (S413),
The character string is output as image data (S414).
Next, although a new character string will be registered in the dictionary based on the decoded character string, it is determined whether or not the number of entries in the dictionary is the same as in the encoding process (S415). When the number of entries is the maximum, the process returns to S411, the dictionary is reinitialized, and then the decoding process is performed. If the number of entries has not reached the maximum, a new character string is registered in the dictionary (S416). The character string registered here is one character that is the leading character of the currently decoded character string connected to the end of the character string decoded immediately before. Then, when the dictionary is updated, the process returns to S412 again, the decoding process is continued in sequence, and the process is continued until there is no compression code to be processed.

Here, in the LZW system, there are a variable length code and a fixed length code as a code format. The minimum code value of the LZW method is determined according to the number of entries in the dictionary, but the number of entries in the dictionary increases as the encoding process progresses. Therefore, when the code length is minimized, it is necessary to increase the code length as the number of entries in the dictionary increases, which results in a variable length code. In addition, regardless of the number of entries in the dictionary, if the code length is equal to or more than the code length determined by the maximum number of entries in the dictionary, a fixed length code is obtained. The variable-length code has a higher compression rate than the fixed-length code because it eliminates unnecessary bits that have not been used.

When the LZW method explained above is applied to image data, the following problems occur. That is, LZ
In the W method, an optimum dictionary is constructed after a sufficient time has elapsed by adding dictionary entries while performing encoding processing, and high compression is possible. But,
Image data generally has a very high correlation between the data,
In a binary image for printing, very long continuous white pixels (white run) and continuous black pixels (black run) occur.

As a solution to such a problem, Japanese Patent Laid-Open No.
There is a method disclosed in 5-176187. The encoding method is shown in FIG. 5 (a), and the decoding method is shown in FIG. 5 (b).

In the encoding method disclosed herein, image data is input, and after the intermediate data converting means 501 converts the image data into intermediate data from which the correlation of the image data has been removed, the encoding means 502. The intermediate data is encoded by the LZW method and the compressed code is output. In the decoding method disclosed herein, a compression code is input, and the decoding means 511 performs LZW.
After decoding by the method to create the original intermediate data, the intermediate data inverse conversion means 512 restores the intermediate data to image data and outputs it.

[0011]

However, in the prior art, in the encoding process, the intermediate data converting means and the encoding means of the LZW system are arranged in series, and the decoding means performs the decoding of the LZW system. Since the means and the intermediate data inverse conversion means are arranged in series,
In particular, when the configuration of the prior art is realized by software, there is a problem that the compression processing speed is reduced as compared with the case where universal encoding and decoding are performed alone. Particularly in the printing field such as a printer, it is fatal that the expansion speed is delayed with respect to the processing speed of the engine portion, and the high speed of the data decoding unit becomes a problem. In view of such a point, the present invention is to provide a data compression / decompression method capable of favorably compressing highly correlated image data without reducing the processing speed of data compression and decompression.

[0012]

SUMMARY OF THE INVENTION The present invention is to solve these problems, and encodes input image data by referring to a dictionary created based on the encoded image data. A first encoding unit that outputs a compression code of 1;
A second encoding unit that encodes the number of consecutive white pixels or black pixels of the input image data so as to be distinguishable from the first compression code and outputs a second compression code; and a white pixel of the input image data. Alternatively, it comprises a counting unit for counting the number of consecutive black pixels and a comparing unit for comparing a count value counted by the counting unit with a predetermined threshold value, and based on the comparison result by the comparing unit, A data compression method is provided in which a first encoding unit or the second encoding unit is selected and the input image data is compression-encoded by the selected encoding unit.

Further, the compression-coded coding unit analyzes the first coding unit or the second coding unit, and the analysis unit codes the first coding unit by the first coding unit. And a second decoding unit that decodes the data encoded by the second encoding unit, and the selected decoding unit includes a first decoding unit that decodes the encoded data and a second decoding unit that decodes the data encoded by the second encoding unit. Provided is a data decompression method of decompressing and decoding a compression code by a unit.

Further, the first encoding unit generates an LZW (Lempel-Ziv-Wel) for generating a fixed length compression code.
ch) an encoding unit, and the first compression encoding unit and the second compression encoding unit identify which compression encoding unit the compression unit generated by the analysis unit. A data compression method according to claim 1, wherein a code is added to the first compression code and the second compression code.

[0015]

According to the structure of the present invention, by switching the universal code and the run pack code according to the count value of the white run length or the black run length, an image having a long run such as a text can be dot / dithered by a simple structure. Images that change finely, such as images, can be compressed well.

[0016]

EXAMPLES In the following examples, image data is 8 bits
When the pixel value is represented by a hexadecimal number (0x is added before the numerical value), 0x00 is a white pixel and 0xff is a black pixel.

An embodiment of the data compression and decoding system of the present invention will be described with reference to the drawings. FIG. 1 shows the most basic configuration of the data compression and decoding system of the present invention. First, the basic configuration of the data compression method will be described with reference to FIG. When a white or black pixel appears in the input image data D, the white / black run counting unit 101 counts the length of a continuous run of white or black pixels. The discrimination code W / B indicating whether the counted value N and the counted pixel are white or black
Is output. The comparator 102 receives the count value N and compares the count value N with a predetermined threshold value Nth.

When the count value N is less than or equal to the threshold value Nth, the control code U is output, and the count value N is equal to the threshold value Nth.
When it is larger than th, the control code R is output. Data encoding is performed by selecting either the universal encoding unit 103 or the run pack encoding unit 104 according to the control code U / R. When the control code is U, the universal encoding unit 103 is selected, and the input image data D, the discrimination code W / B, and the count value N are the universal encoding unit 1.
03, and encoding is performed based on each input value. When the control code is R, the run pack encoding unit 1
04 is selected, the discrimination code W / B and the count value N are input to the run pack encoding unit 104, and encoding is performed based on each input value. The code generated by each coding unit is output as the output code C. Here, in the universal encoding unit 103 and the run pack encoding unit 104, an identification code for identifying which encoding method is used is added to the head of the code C.

Next, the basic structure of the data decoding method will be described with reference to FIG. The analysis unit 111 analyzes the identification code included in the input code C and identifies whether the input code C is a code to be decoded by the universal decoding unit 112 or a code to be decoded by the runpack decoding unit 113. . Then, the control code U / R is output according to the identification result, and the input code C required for decoding is output. Data decoding is performed by selecting either the universal decoding unit 112 or the run pack decoding unit 113 according to the control code U / R. When the control code is U, the universal decoding unit 112 is selected, the input code C is input to the universal decoding unit 112, and decoding is performed. When the control code is R, the runpack decoding unit 113 is selected, the input code C is input to the runpack decoding unit 113, and decoding is performed. The image data restored by each decoding unit is output as output data D.

Next, with respect to an embodiment in which the present invention is realized by using the LZW system, the flow of the encoding process will be described with reference to FIGS. Here, the maximum number of entries in the reference dictionary is 2 to the 14th power, and the output code is 2 bytes (= 16
(Bit) fixed length. First, the encoding process is initialized (S201). here,
A reference dictionary for LZW encoding is initialized, and a pointer n is set at the head of the input image data. Then, the image data is sequentially input from the nth (S202).
Next, white or black runs are counted (S203). Here, if the image data Dn is a white pixel, the discrimination code W is set, and if it is a black pixel, the discrimination code B is set. Then, the succession of white pixels or black pixels is counted, and the count value is substituted into N. Next, the count value N obtained in S203 is compared with a predetermined threshold value Nth (S204), and the count value N becomes the threshold value Nt.
If smaller than h, LZW encoding is performed (S205),
When the count value N is greater than or equal to the threshold value Nth, run pack coding is performed by packing the runs of white / black pixels counted in S203 (S206). Further, the pointer n of the input image data is increased by the number of encoded data, the process returns to S202 again, and the process is continued. This S20
Image data is compressed by continuing the steps from 2 to S206 until there is no input image data.

The details of the LZW coding process and the run pack coding process will be described with reference to FIG. First, LZ
Details of the W encoding process will be described with reference to FIG. LZ
A count value N of white / black runs and a pointer n of the current input image data are passed from the main routine to the W encoding processing section. First, a temporary variable L for storing the encoded image data length is prepared and its value is initialized to 0 (S601). Then, the image data is sequentially input from the one indicated by the pointer value n (S602).
Next, the longest character string that matches the input character string is searched from the dictionary, and the value of the temporary variable L is increased by LD using the longest matching character string length as LD (S603). next,
The index attached to the longest matching character string searched in the reference dictionary is encoded and output. The code length at this time is 2 bytes regardless of the number of entries in the dictionary. As shown in FIG. 8A, the code has a code in which the first bit and the second bit of the code are masked with 0, and the remaining 14 bits (indicated by X) represent the index of the dictionary. . Further, since the longest matching character string length LD is equal to the image data length to be encoded, the value of the pointer n is increased by LD as a pointer of the image data to be encoded next (S
604).

Next, the reference dictionary is updated. The dictionary is updated by comparing whether the number of entries registered in the dictionary reaches the maximum number that can be registered in the dictionary, as described in the related art. That is, in the present embodiment, the number of entries in the dictionary is compared with the maximum number of entries 2 to the 14th power (S605). If the number of entries is not the maximum, a new character string is registered (S606). Since the character string cannot be registered as described above, the dictionary is initialized (S607). Next, the temporary variable L and the count value N are compared. Temporary variable L is L once
It represents the image data length encoded by the ZW encoding processing routine, and if the temporary variable L is smaller than the count value N, a white / black run smaller than Nth remains. Therefore, it is obvious that the LZW encoding process is performed without returning to the main routine to perform the white / black run count. Therefore, if the temporary variable L is smaller than the count value N, the process returns to S602 again to continue the LZW encoding process, and if not, the LZW encoding process is terminated and the process returns to the main routine (S608).

Details of the run pack encoding process are shown in FIG.
This will be described with reference to (b). The count value N of white / black runs and the discrimination code W / B of the runs are passed from the main routine to the run pack encoding processing section. First, since the length of the image data encoded here is equal to the count value N, the value of the pointer n is incremented by N as a pointer of the image data to be encoded next (S611). Next, the value of the discrimination code is checked (S
612), if the discrimination code is W, 0x8000 is set in the mask M.
NM, which is the maximum value of runs that can be coded by setting
Is set to 2 to the 15th power (S613). If the discrimination code is B, 0 × 4000 is set in the mask M and 2 of 14 is set in the variable NM (S614). As shown in FIG. 8B, when the white run is encoded by the mask at this time, the first bit of the code is 1 and the remaining 15 bits (denoted by X) are the white run length. Will be represented. Further, as shown in FIG. 8C, when the black run is coded, the first bit of the code is 0, and the second bit is 0
The bit becomes 1, and the remaining 14 bits (denoted by X) represent the black run length. Generally, a white run tends to be longer than a black run in a printed image, and thus the white run can be packed longer.

Next, white / black runs are encoded. In this embodiment, the maximum run length that can be packed in one code is NM, so when the count value N is larger than that, the run length is divided into several. And encode. Therefore, the count value N and the variable NM
Are compared (S615). Then, the count value N is the variable N
When it is larger than M, NM-1 is assigned to the temporary variable c representing the run length, the logical sum (represented by the symbol |) of the temporary variable c and the mask M is taken, and the synthesized code C is output.
Then, the coded run length NM is subtracted from the count value, and the process returns to S615 (S616). When the count value N is less than or equal to the variable NM, N-1 is substituted for the temporary variable c representing the run length, the temporary variable c and the mask M are ORed, and the combined code C is output (S61).
7). Here, the value obtained by subtracting 1 from the actual run length is used as a code, from 1 to 2 15 powers or 2 14 power white or black runs from 0x0000 to 0x7fff or 0x3f
This is to correspond to ff. Then, the run pack encoding process is terminated and the process returns to the main routine. The code generated as described above is used as an identification code as shown in FIG.
Since 00 is coded by the ZW method, 1 is coded by the run pack when the white run length is coded, and 01 is coded when the black run length is coded by the run pack, each is uniquely identified. can do.

Next, with respect to an embodiment in which the present invention is realized by using the LZW system, the flow of the decoding process will be described with reference to FIGS. 3 and 7. First, as initialization of the decoding process, initialization of a reference dictionary of the LZW system, setting of a pointer to the head of an input compression code string, and the like are performed (S3).
01). When the nth compression code Cn is input (S30
2) First, the value of the first bit and the second bit is extracted by taking the logical product (represented by the symbol &) of the compression code Cn with 0xc000, and the value is examined (S303). Since the structure of the compression code is as shown in FIG. 7, if the compression code is coded by the LZW system, the value of the above logical product becomes 0, and if it is coded by the run pack, For example, the value of the logical product is not 0. Therefore, when the logical product value is 0, LZW decoding is performed (S30
4) If the logical product value is other than 0, run pack decoding is performed (S305). In this way, when one compression code is restored to the image data, the process returns to S302 to input the next compression code, and the decoding process of the image data is performed by continuing the sequential decoding process until the compression code is exhausted. .

Details of the LZW decoding process and the run pack decoding process will be described with reference to FIG. First, LZ
Details of the W decoding process will be described with reference to FIG. LZ
The compression code Cn is passed from the main routine to the W decoding processing unit. First, the index of the reference dictionary is obtained from the compression code, and the character string registered in the reference dictionary is searched from the index value (S701). Then, the character string is output as image data, and the value of n is increased by 1 in order to move the pointer to the compression code to be processed next (S702). Next, in order to update the dictionary, it is determined whether the number of entries in the dictionary has reached the 14th power of the maximum number of entries 2 (S703). If the number of entries has not reached the maximum, the method described in the prior art is used. A new character string is registered in the dictionary by the same method as in (S704), and when the number of entries is maximum, the dictionary is initialized (S705). After performing the series of processes described above, the process returns to the main routine.

Details of the run pack decoding process are shown in FIG.
This will be described with reference to (b). The run pack decoding processing unit also receives the compression code Cn from the main routine. First, by checking the first bit of the compression code, it is determined whether this compression code packs a white run or a black run (S711). That is, since the structure of the compression code is as shown in FIG. 8, the first bit is 1 when the white run is packed, and the first bit is 0 when the black run is packed. It can be carried out.
Here, when the first bit is 1, the white run is packed, and the configuration of the compression code at that time is represented by the second bit to the 16th bit as shown in FIG. 8B. Therefore, the compression code Cn is ANDed with 0x7fff and masked, and the value NC obtained by adding 1 to the value becomes the run length. Therefore, the image data can be restored by outputting NC white pixels 0x00. Further, the value of n is incremented by 1 to move the pointer to the compression code to be processed next (S712). Similarly, when the first bit is 0 in S711, the compression code Cn is logically ANDed with 0x3fff and masked, and black pixel 0xf is obtained by adding 1 to the value NC.
Image data can be restored by outputting f, and
The value of n is incremented by 1 to move the pointer to the compression code to be processed next (S713). After performing the series of processes described above, the process returns to the main routine.

[0028]

As described above, according to the present invention,
By combining the universal encoding unit that does not depend on the statistical properties of the image with the run pack encoding unit, data having a long run of white / black such as text is highly compressed while preserving the universality of the encoding unit. be able to.

The Rampack encoding / decoding unit can be realized with a simple structure, and the counting unit and the comparing unit for switching between the two encoding units are composed of one counter and a comparator. Since the analysis unit for switching the decoding unit is composed of one comparator, the processing amount is hardly increased and the processing speed is not deteriorated as compared with the case where the universal encoding / decoding process is performed by itself. In particular, the image data can be compressed / decoded without deteriorating the speed of the decoding process required for the printing process of the printer or the like.

[Brief description of drawings]

FIG. 1 is a diagram showing a basic configuration of a data compression and decoding system of the present invention.

FIG. 2 is a diagram showing a flow of an encoding process using the LZW method.

FIG. 3 is a diagram showing a flow of a decoding process using the LZW method.

FIG. 4 is a diagram showing a flow of LZW encoding and decoding processing.

FIG. 5 is a diagram showing a configuration of a conventional data compression and decoding system.

FIG. 6 is a diagram showing a processing flow of each encoding unit when the LZW method is used.

FIG. 7 is a diagram showing a processing flow of each decoding unit when the LZW method is used.

FIG. 8 is a diagram illustrating a code structure when the LZW method is used.

[Explanation of symbols]

 101 White / Black Run Counting Unit 102 Comparing Unit 103 Universal Encoding Unit 104 Run Pack Encoding Unit 111 Code Analysis Unit 112 Universal Decoding Unit 113 Run Pack Decoding Unit

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI technical display area H04N 1/413 D

Claims (3)

[Claims] 1. A first encoding unit for encoding input image data by referring to a dictionary created based on encoded image data and outputting a first compression code, and input image data A second encoding unit that encodes a continuous number of white pixels or black pixels so as to be distinguishable from the first compression code and outputs a second compression code; The first code is composed of a counting unit for counting the number of consecutive times and a comparing unit for comparing the count value counted by the counting unit with a predetermined threshold value, and based on the comparison result by the comparing unit. An encoding unit or the second encoding unit,
A data compression method, wherein input image data is compression-encoded by the selected encoding unit. 2. The compression-encoded encoder is the first encoder.
An encoding unit or an analyzing unit that analyzes the second encoding unit; a first decoding unit that decodes the data encoded by the first encoding unit by the analyzing unit; And a second decoding unit that decodes the data encoded by the second encoding unit, and the compression decoding code is expanded and decoded by the selected decoding unit. 3. The first encoding unit is composed of an LZW (Lempel-Ziv-Welch) encoding unit for generating a fixed-length compression code, and the first compression encoding unit and the second compression encoding unit. The unit adds an identification code for identifying which code is generated by the analysis unit by the analysis unit to the first compression code and the second compression code. The data compression method according to item 1.