JPH02194772A - Method and device for encoding - Google Patents

Abstract

PURPOSE:To obtain the high compression rate of a code by selecting an encoding processing method for every symbol series divided into every prescribed number of sections by cutting off from the inputted symbol series. CONSTITUTION:The dividing means 101 of the symbol series segments the set of the prescribed number of symbols from the input symbol series, and an encoding method selecting means 105 measures a statistic property in the segmented set of the symbol series, and executes the selection of the encoding method suitable for the encoding of this symbol series, i.e., block encoding and non-block encoding, etc. Then, the encoding processing of the symbol in the set of this symbol series is executed based on this selected result. Thus, the encoding of the high code compression rate becomes possible.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to an encoding method capable of efficiently encoding and compressing an output signal from an information source, and particularly to encoding method for encoding and compressing one image signal. The present invention relates to an encoding method and an encoding apparatus suitable for use in the present invention.

[Prior Art] Examples of conventional technologies related to encoding processing methods for efficiently compressing image signals and the like include IE Transactions on Communication and COM 29. No. 6 (1981), pp. 858-867 (IEEE.

Trans,on Communications,C
OM-29, No. 6 (1981) PP858-
867) is known.

This conventional technology repeatedly performs processing for each symbol to make the input symbol sequence to be compressed and encoded correspond to a probability number line divided according to the probability of occurrence of the symbol, and the value obtained as a result of the processing is expressed in binary form. This is what is output by. This conventional technique is characterized in that as the number of symbols to be encoded increases, the compression rate of the code in the encoding process for each symbol approaches the ideal compression rate (entropy rate).

Also, as other conventional techniques regarding code compression.

The technology disclosed in Japanese Patent Application Laid-open No. 62-29372 is equipped with means for switching the dominant symbol estimating means using the transmitted signal for each pixel signal to be encoded. Arithmetic sequence coding using the relative frequency of occurrence of known or estimable information source symbols, a technique disclosed in Japanese Patent Application Laid-open No. 126484/1984, which is equipped with a means for counting the number of dots and determining features within a region. According to
JP-A-53-110338 for efficiently performing arithmetic coding
The techniques described in the above publications are known.

[Problems to be Solved by the Invention] The conventional technology disclosed in the above paper performs encoding processing for each symbol even when the probability of appearance of an input symbol sequence is set with high precision, and as a result, This method has the problem that the number of output code words increases, that is, the compression rate of the code cannot be improved beyond a certain level.

Furthermore, each of the conventional techniques described in the above publications also has the problem that it is difficult to improve the compression rate of codes beyond a certain level.

SUMMARY OF THE INVENTION An object of the present invention is to provide an encoding device that solves the problems of the prior art and makes it possible to obtain an even higher code compression rate.

[Means for Solving the Problems] According to the present invention, the object is to measure the statistical properties of a symbol sequence to be encoded in advance, and, based on the measurement results, to develop an encoding processing method that is optimal for the symbol sequence. By selecting the symbol sequence to be encoded,
By cutting out and dividing the actually input symbol sequence into a certain number of symbol sequences, and performing the selection of the encoding processing method for each divided symbol sequence, information can be generated as the symbol sequence. This is achieved by using a comparison result between a symbol sequence output from the source and a symbol sequence estimated from a symbol sequence that has already been encoded.

That is, the purpose is to provide a symbol sequence dividing means, an encoding method selection means, an encoding means, a signal output means,
This is achieved by comprising a signal estimation means and a symbol sequence comparison means.

[Operation] The symbol sequence dividing means cuts out a set of a fixed number of symbols from the input symbol sequence, thereby making it possible to execute the encoding method using this set as a unit.

The encoding method selection means measures the statistical properties within the set of symbol sequences extracted above, and selects an encoding method suitable for encoding the symbol sequence, such as block encoding or non-block encoding. I do.

The encoding means is configured to be able to execute a plurality of encoding methods, selects one encoding method based on the selection result by the encoding method selection means, and selects one encoding method to encode the symbols in the set of symbol sequences. Execute encoding processing.

The signal output means outputs the signal processing result by the encoding means and the selection result by the encoding method selection means.

The signal estimating means estimates the appearance probability of a symbol sequence to be encoded from the symbol sequence that has already undergone signal processing, and creates a dominant symbol sequence.

The symbol sequence comparison means compares the symbol sequence to be encoded with the predominant symbol estimated above, determines whether they match, and outputs the result.

The number of symbols that constitute a set of symbol sequences for performing the signal processing as described above can be adaptively selected using, for example, a symbol number setting means, for example, by comparing symbol sequences. If the number of matches is large as a result of the procedure, the code compression rate can be increased more efficiently by performing processing such as increasing the number of symbols to be combined into a set.

[Embodiment] Hereinafter, an embodiment of the encoding method and encoding apparatus according to the present invention will be described in detail with reference to the drawings.

FIGS. 1(a) and 1(b) are block diagrams showing the configuration of an embodiment of the present invention, FIG. 2
FIG. 3B is a diagram illustrating a dominant symbol estimating device that obtains an estimated value of a symbol to be encoded.

Figure 3 is a flowchart explaining the signal processing operation;
The figure is a diagram explaining the operation of the encoding method selection device, and FIG.
FIG. 1(a) is a diagram illustrating the number of divisions of an input symbol sequence, and FIG. 6 is a diagram illustrating a symbol sequence by subsampling.

In (b), 101 is symbol sequence dividing means.

102 is a memory, 103 is a dominant symbol estimation means, 10
4 is a comparison means, and 105 is an encoding method selection means.

106 is a signal switching means, 107 is an encoding processing means a
, 108 is an encoding processing means, 109 is a signal output means,
110 is a line memory, and 111 is an n-line memory.

First, the principle and operation of an embodiment of the present invention will be briefly explained with reference to FIGS. 1(a) and 1(b).

An input signal that is a symbol sequence from an information source to be encoded, for example, an input signal obtained by scanning an image, is input to the line memory 110, and then its bits are compared. At the same time, n
The data are sequentially stored in the line memory 111. The dominant symbol estimating means 103 is constituted by, for example, a ROM, predicts the value of the symbol to be encoded given to the comparing means 104, and supplies the prediction result to the comparing means 104. This prediction is performed using a plurality of symbols that exist in the n-line memory 111 and have a strong correlation with the symbol to be encoded that has been input to the n-line memory 111 before the symbol to be encoded. Details of this prediction will be described later.

The comparing means 104 compares the symbol to be encoded given from the line memory 110 and the dominant symbol estimating means 1.
The predicted value of the symbol given by 03 is compared with the predicted value of the symbol, and the comparison result is provided to the symbol sequence dividing means 101 as an input symbol sequence.

The symbol sequence dividing means 101 divides a given input symbol sequence into a set of a fixed number of symbols,
This set of symbols is provided to memory 102. In subsequent encoding processing, this set of symbols is used as a unit of encoding.

The memory 102 supplies the divided set of a certain number of symbols to an encoding method selection means 105 for selecting and determining an encoding processing means, and also inputs each symbol of the set of symbols into a signal for encoding. Switching means 10
6 sequentially. The encoding method selection means 105 measures the statistical properties of the set of symbols given from the memory 102, determines the encoding method, and selects the encoding processing means a107 or the encoding processing means b108. The signal is output to the signal switching device 106 and the signal output device 109. Details of the operation of the encoding method selection means 105 will be described later.

The encoding processing means a107 or encoding processing means b108 selected by the encoding method selection means 105 encodes the set of divided symbols that are sequentially input from the memory 102 via the signal switching means 106, and the result is The obtained code word is output to signal output means 109. The signal output means 109 includes the aforementioned encoding method selection means 10.
The selection result in step 5 and the code word output from the encoding processing means a107 or encoding processing means b108 are combined and outputted to the outside as a signal format suitable for external output.

FIG. 2(a) shows an example of a symbol sequence from an information source, that is, an input image signal when the present invention is applied to encoding processing of a binary image signal.

In this example, the symbol sequence from the information source to be encoded is a signal obtained by scanning an image with a photoelectric conversion device, etc., and the symbol sequence has properties that depend on the pixel configuration of the image. It is equipped with

Therefore, the unit of the set of symbol series mentioned above.

It is assumed to be a line unit of the image, or a unit in which one line is divided into a plurality of parts. In general, image signals are known to have two-dimensional signal correlation, and by determining the unit of a set of symbol sequences as described above, it is possible to use a set of symbol sequences that have already undergone signal processing. Therefore, it is possible to estimate the dominant symbols within the set of symbol sequences to be encoded.

Specifically, the statistical distribution of two-dimensional signal correlation can be measured in advance using training images, and the above-mentioned method for estimating the dominant symbol can be determined in advance.

For example, in the case of the example shown in FIG. 2(a), the value of the dominant symbol of the symbol X of the pixel to be encoded in the symbol sequence to be encoded is, for example, the value of the dominant symbol of the symbol It can be estimated based on the values of symbols A-G.

The dominant symbol estimating device 103 converts the results of the estimation method determined above into RO as shown in FIG. 2(b), for example.
By using this, it is possible to estimate the dominant symbol of the symbol to be encoded based on the signal-processed symbol sequence.

The encoding method selection means 105 measures the statistical properties of the input symbol sequence as a result of comparing the dominant symbol of the encoding target symbol estimated above with the encoding target symbol. It determines whether or not all are composed of "0" and outputs the result.

The fourth @ shows the relationship between the conditions of the input symbol sequence given to the encoding method selection means 105, the value of the 1 selection signal, and the encoding processing result output from the signal output means 109.

In this example, the encoding method selection means 105 determines that all the values of each symbol of the input symbol sequence given are II OI
If N, that is, if the results of comparison between the symbol to be coded and the estimated dominant symbol by the comparison means 104 all match within the set of nine divided input symbols, "0" is set as the selection signal. In other cases, the selection signal J (139 is output.The selection signal It
In the case of the present invention, the encoding processing means selected by O11 operates so as to suppress the given symbol sequence and not output it. As a result, when the selection signal is "0", only this selection signal 11071 is output from the signal output means 109, and when the selection signal is "1", this selection signal II 117 and the symbol sequence are output. The code word that is the result of the encoding process will be output.The aforementioned selection signal “′
If only 0'' is output, this signal is determined to be the same symbol sequence as the previous symbol sequence on the code decoding side.

The encoding processing method by the encoding processing means a107 and the encoding processing means b108 is 1, for example, 110 of the symbol sequence.
It is possible to use a method of allocating a Huffman code or MR code in which the number of consecutive II and 1''' is measured as a run length, or a method of using an arithmetic code or predictive code as described above as the prior art, but there are other methods. You can also use this method.

In addition, in the above, the selection signal is one bit, but the selection signal is made into multiple bits, and one of the encoding processing means most suitable for the nature of the symbol sequence to be encoded is selected from among many encoding processing means. You can choose one.

In one embodiment of the present invention configured as described above, the symbol sequence dividing means 101 of FIG.
It may be located at the location of the line memory 110 in a) or between the line memory 110 and the comparing means 104.

FIG. 3 is a flowchart for explaining the operation of one embodiment of the present invention configured as described above, and the signal processing operation will be explained below in accordance with this flow.

(1) A symbol sequence to be encoded from an information source is divided into a set of symbols having a predetermined number of symbols by the symbol sequence dividing means 101, and each symbol is sequentially provided to the comparing means 104. .

(2) On the other hand, the signal-processed symbol sequence of the symbol sequence from the information source is given to the dominant symbol estimating means, and is given to the comparing means 104 as a dominant symbol sequence in which the value of the symbol to be encoded is estimated.

(3) The comparing means 104 compares each symbol of the symbol sequence to be encoded with each symbol of the dominant symbol sequence from the dominant symbol estimating means 103, and sets it Ots when they match and when they do not match. '1'' is output. The output of this comparison result is input to the encoding processing means a107 or encoding processing means b108, and is used as an input symbol sequence to be subjected to actual encoding processing.

(4) This input symbol sequence is
05 and encoding processing means a107 or encoding processing means b108, and encoding method selection means 105 outputs a selection signal in the same manner as described above based on this input symbol sequence, and also performs encoding One of the processing means is caused to encode the input symbol sequence.

(5) The selection signal from the encoding method selection means 105 and the code word from one of the encoding processing means are transmitted to the signal output means 10
9, and is converted into a signal format suitable for external output and output.

In the embodiment of the present invention described above, the unit of execution of the encoding process is a symbol sequence from an information source, for example, one line of an image input signal, but the input symbol sequence is the unit of execution of this encoding process. By making the number of divisions variable using the statistical properties of the processed signal,
It becomes possible to perform more efficient code compression. This point will be explained below with reference to FIGS. 5(a) and 5(b).

Assume that the input image is a document image as shown in FIG. 5(a), and is to be encoded.

Generally, when encoding such a document image,
White areas tend to appear relatively concentrated.The present invention takes advantage of this property of document images to estimate dominant symbols using signal-processed symbol sequences adjacent to the symbol sequence to be encoded. By changing the number of symbols to be divided depending on whether the matching rate with the symbol to be encoded is high or low, the optimal encoding processing method is adaptively selected and the code compression rate is further improved. It is possible to improve the What is the agreement rate of the above comparison results?

During signal processing in the present invention, this affects the encoding method selection result and the encoding processing result, and the setting of the number of symbols to be divided is performed based on the encoding method selection result and the encoding processing result. be able to.

In this case, the result of symbol division for the input image shown in FIG. 5(a) is, for example, as shown in FIG. 5(b).

In other words, in image parts where white areas are continuous, the matching rate of the above-mentioned symbol comparison results is high, so the set of divided symbols is made larger.

For example, if l lines are taken as one symbol sequence to be encoded, and in certain parts such as characters, the above-mentioned matching rate will be low, so the set of divided symbols will be reduced to 1, for example, 1/
Two lines are treated as one symbol sequence to be encoded. This division is not limited to 1/2 line, but may be 1/n line.

Although the embodiment of the present invention described above encodes a binary image as an example, the present invention can be similarly applied to encoding processing of multi-valued images, color images, etc. It is.

Furthermore, the present invention can be applied not only to scanning input for each line in which the input order of image signals is input, but also to encoding processing of a symbol sequence obtained by subsampling as shown in FIG. .

Further, the present invention can be applied not only to encoding processing of a single image signal, etc., but also to a communication device that performs data communication via a transmission path, a filing device that stores data in a storage device, and the like.

Furthermore, in the configuration of the embodiment of the present invention described in FIGS. , as a direct input symbol sequence,
It may also be input to the symbol sequence dividing means shown in FIG. 1(b), and the present invention can perform encoding with a high code compression rate even in this case.

Note that decoding of the code word encoded according to the present invention and output to the outside can be easily realized by performing the signal processing procedure for encoding described above in the reverse direction.

It goes without saying that the encoding device according to the embodiment of the present invention described above can be realized using a combination circuit of individual parts, and can also be realized using a one-chip device incorporating a data memory, an arithmetic processing section, etc. It can be realized using a semiconductor integrated circuit.

[Effects of the Invention] As explained above, according to the present invention, the amount of information required when encoding an input symbol sequence can be reduced, and encoding with a high compression rate can be easily realized. Can be done.

[Brief explanation of the drawing]

FIGS. 1(a) and 1(b) are block diagrams showing the configuration of an embodiment of the present invention, FIG. 2(a) is a diagram illustrating an image signal using encoding target symbols and signal processed symbols, 2
Figure (b) is a diagram explaining the dominant symbol estimating device that calculates the estimated 5 μm value of the symbol to be encoded, and Figure 3 is a flowchart explaining the signal processing operation. Diagram explaining the operation, 5th
Figures (a) and (b) are diagrams for explaining the number of divisions of an input symbol sequence, and Fig. 6 is a diagram for explaining a symbol sequence by subsampling. 101...Symbol series dividing means, 102...
... Memory, 103 ... Dominant symbol estimation means, 104 ... Comparison means, 105 ...
- Encoding method selection means, 106...signal switching means, 107...encoding processing means a, 108
... Encoding processing means b, 109 ... Signal output means, 110 ... Line memory, 111
・・・・・・N line memory. Fig. 3 Bon Fig. 2 Cσ No AAG: Meeting construction plan Fig. 2 rb Fig. 4

Claims (1)

[Claims] 1. In an encoding method for encoding a symbol sequence from an information source, an input symbol sequence is divided into a set of a certain number of symbols, and a statistical analysis of the symbols in the divided set is performed. An encoding method characterized by selecting a symbol encoding method based on a property, and performing symbol encoding processing using the selected encoding method. 2. The input symbol sequence is an estimated symbol sequence for a symbol sequence from an information source to be encoded, which is estimated using a symbol sequence that has already been encoded;
2. The encoding method according to claim 1, wherein the symbol sequence is obtained by comparing a symbol sequence from an information source to be encoded. 3. The number of symbols in the divided set is set based on the result of encoding method selection and the result of encoding processing for symbol sequences that have already been output as signals. The encoding method according to item 1 or 2. 4. In an encoding device that encodes a symbol sequence from an information source, a symbol sequence dividing means that divides an input symbol sequence into a set of a certain number of symbols, and statistical properties of the symbols in the divided set. a selection means for selecting a symbol encoding method based on the selection means; an encoding processing means for encoding the symbols in the divided set based on the result of the selection means; and a result of the encoding method selection means. and signal output means for outputting the result of the encoding processing means. 5. The input symbol sequence is a means for estimating a symbol sequence from an information source to be encoded using a symbol sequence that has already been encoded, and an estimated symbol sequence from the estimating means and an encoding target. 5. The encoding device according to claim 4, wherein the encoding device is calculated using a comparing means for comparing symbol sequences. 6. The symbol sequence dividing means sets the number of symbols in the divided set based on the result of the encoding method selection means and the result of the encoding processing means, which have already output signals. Characteristic Claim No. 4
6. The encoding device according to item 5.