JPH06112839A - Encoder - Google Patents

Abstract

PURPOSE:To obtain an encoder optimizing a compression rate of binarized data by generating a Huffman code table in response to a characteristic of each of binarized data and executing the coding (and decoding processing) through the use of the table. CONSTITUTION:Approximate equations P(n) expressing a graph of occurrence probability distribution expressed by an exponential function or a sum of exponential functions are shown in equations I, II in an approximate processing section 300 and obtain each coefficient. In the equations, n is a run length and alpha, alpha1,..., alpham, a, a1,..., am, and (b) are coefficients obtained by the least square method. After each coefficient is obtained, a coding processing section 500 executes run length coding. Coefficient data obtained by both the equations are added to data obtained as the result to obtain coding data, which are outputted, The quantity of data of each coefficient is very small in comparison with data quantity by the coding, then the compression rate of the outputted code data is not reduced even when the data of each coefficient are added.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a coding apparatus for coding binary data and decoding coded data.

[0002]

2. Description of the Related Art In a facsimile machine or the like, binarization processing is performed after image data of an original image is read by an image reader. The encoding device has a value of 2 obtained by the binarization process.
The binarized data is run-length encoded and output to the transmission unit and the like, and the received code data is decoded into the binarized data and output to the control unit and the like. Conventionally, when executing the above run length encoding, an appropriate table is selected from a fixed code table or several types of code tables prepared in advance and the encoding is executed.

Here, the run length coding means the following. First, for standard binarized data, a graph of the occurrence probability distribution for the length of white or black pixels following (hereinafter referred to as run length) is obtained. Statistically, the graph of the occurrence probability distribution is represented by an exponential function. That is, in the graph of the occurrence probability distribution, the number of occurrences of run length of 0 is the largest, and the number of occurrences decreases exponentially as the run length becomes longer. Here, give a short run length an identification code with a small number of bits,
An identification code with a larger number of bits is given as the run length becomes longer, and a Huffman code table is created based on this. In the run-length encoding, the Huffman code table created in this way is used to encode the binary data, and the information amount of the binary data is compressed.

[0004]

However, in the conventional facsimile apparatus, run length coding is executed using a Huffman code table prepared in advance. Therefore, the compression rate of the information amount may not increase depending on the input binarized data.

Therefore, it is an object of the present invention to provide an encoding device for optimizing the compression ratio of binary data by performing encoding that matches the characteristics of each binary data.

[0006]

According to a first aspect of the present invention, there is provided a coding device, a memory for storing inputted binary data, and a run-length occurrence probability of the binary data stored in the memory. A run length calculation means for obtaining the distribution, a coefficient calculation means for calculating each coefficient of the approximate expression of the occurrence probability distribution obtained by the run length calculation means, and a Huffman code table is created based on the coefficients calculated by the coefficient calculation means. A table creating means, a counter means for sequentially obtaining and outputting a run length value of the binarized data stored in the memory, a run length value output by the counter means, and a Huffman created by the table creating means. Coding means for executing a coding process using the code table, and data obtained by the coding means are obtained by the coefficient calculating means. Adding data of each coefficient of the approximate expression of the probability distribution, and a additional means for outputting it as coded data.

In the encoding device according to the second aspect, when the code data including the data of each coefficient of the approximate expression of the run probability occurrence distribution of the binarized data is input, the code data is added to the code data. Extracting means for extracting the data of each added coefficient, table creating means for creating an Huffman code table by obtaining an approximate expression of the occurrence probability distribution based on the value of each coefficient extracted by the extracting means, and table creating means The Huffman code table created by the above, and the decoding means for executing the decoding process from the input code data and outputting the binarized data.

[0008]

In the encoding device according to the first aspect, after the binarized data is input and stored in the memory, the run-length calculating means calculates the occurrence probability distribution of the binarized data stored in the memory. Then, the coefficient calculating means calculates the value of each coefficient of the approximate expression of the occurrence probability distribution, and the table creating means creates the Huffman code table based on the calculated coefficients. The encoding means performs a run length encoding process by using each run length value of the binarized data stored in the original memory by the counter means and the Huffman code table created by the table creating means. After execution, the adding means adds the data of each coefficient obtained by the coefficient calculating means to the data output by the encoding means, and outputs this as coded data.

In the encoding device according to the second aspect of the present invention, the data of each coefficient added to the input code data is extracted by the extracting means, and the occurrence probability is extracted from the data of each extracted coefficient by the table creating means. An approximate expression of the distribution is obtained and a Huffman code table is created. The decoding means decodes the input code data by using the created Huffman code table and outputs the binarized data.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An encoding apparatus according to the present invention will be described below with reference to the accompanying drawings. (1) Control block, (2) Encoding process,
(3) The decoding process will be described in detail in this order.

(1) Control Block FIG. 1 is a diagram showing a control block diagram of the coding apparatus of the present invention. The encoding device stores an image reader 10 that reads image data, a binarization processing unit 20 that converts the read image data into binarized data, and binarized data created by the binarization processing unit 20. A binary data memory 100, a CPU 50 for controlling the entire encoding device,
A probability distribution calculation unit 200 that calculates a run-length occurrence probability distribution of binarized data, an approximation processing unit 300 that calculates the value of each coefficient of the approximation formula of the probability distribution graph obtained by the calculation unit 200, and an approximation A Huffman code table creation unit 400 (and 80) that creates a Huffman code table based on the value of each coefficient of the approximate expression obtained by the processing unit 300.
0), an encoding processing unit 500 that performs encoding processing based on the created code table, and a code data memory 700 that stores code data input via a telephone line.
And a decoding processing unit 900 that executes the decoding processing based on the code data stored in the code data memory.

(2) Encoding processing The encoding apparatus of the present invention does not use a Huffman code table prepared in advance as in the conventional case when executing run-length encoding of input binary data, A feature is that a Huffman code table is created for each binarized data, and the coding process is executed using this.

The Huffman code table is created in the following order. First, the image data read by the image reader 10 is binarized by the binarization processing unit 20 and stored in the binarized data memory 100. Next, in the probability distribution calculation unit 200, the 2
After the graph of the run probability occurrence probability distribution of the binarized data is obtained, the approximation processing unit 300 calculates the exponential function represented by the following "Equation 1" or the sum of the exponential functions represented by "Equation 2". The value of each coefficient of the approximate expression of the graph of the occurrence probability distribution represented is calculated.

[Formula 1] P (n) = a · exp (−αn) + b

## EQU00002 ## P (n) = a ₁ .exp (-α ₁ n) + a ₂・ exp (-α ₂ n) + ... + am _m exp (-α _m n) + b where P (n) Shows an approximate expression of the graph of the occurrence probability distribution. n indicates a run length value. Where α,
α ₁ , ..., α _m , a, a ₁ , a ₁ , ..., _Am and b are coefficients obtained by an appropriate calculation means such as the least square method.

After the values of the respective coefficients of the approximate expression of "Equation 1" or "Equation 2" are obtained, the Huffman code table creating unit 400 creates a Huffman code table based on the approximate expression. Next, run length coding is executed in the coding processing unit 500 using the created code table.

In the encoding process of the present invention, the data of each coefficient is added to the data obtained by run-length encoding the binarized data, and this is output as encoded data. Since the data of each coefficient is smaller than the amount of data obtained by encoding, adding the data of each coefficient does not reduce the compression rate of the output code data.

When decoding code data input via a telephone line into binary data, first, data of each coefficient added and added to the code data is extracted, and based on this data. The Huffman code table creating unit 800 creates a code table. After that, the decoding processing unit 900 executes the decoding processing using the Huffman code table.

FIG. 2 is a diagram showing each functional block used in the encoding process of the run length encoding device. The run length encoding process according to the present invention will be described below with reference to this drawing. The image data read by the image reader 10 is converted into binarized data by the binarization processing unit 20 and stored in the binarized data memory 100. The probability distribution calculation unit 200 calculates the occurrence probability distribution of each run length of the binarized data stored in the memory 100.

The run length counter 201 provided in the probability distribution calculation unit 200 counts the number of times each run length existing in the binarized data stored in the memory 100 is generated. The probability distribution memory 202 stores the result counted by the counter 201 for each run length.

Based on the run-length occurrence probability distribution calculated by the probability distribution calculation unit 200, the next approximation processing unit 300 minimizes each coefficient of the exponential function shown in the above "Equation 1" or "Equation 2". Calculate using the square method.

The Huffman code table creating unit 400 creates a Huffman code table based on the value of each coefficient of "Equation 1" or "Equation 2" obtained by the approximation processing unit 300 (Huffman table calculation unit 401). , And stores this (Huffman table memory 402).

The encoding apparatus of the present invention executes the above-mentioned processing, and stores the binary data stored in the binary data memory 100.
Based on the binarized data, the run-length counter 501 provided in the encoding processing unit 500 causes the memory 100
The number of occurrences of each run length existing in the binarized data stored in is counted. Next encoding block 502
Performs the encoding process based on the result of the run length counter 501 and the Huffman code table created by the table creation block 400. Encoding block 502
The data output from the
The data of each coefficient of the approximate expression of “Equation 1” or “Equation 2” obtained by the above is added in the adding unit 503, and then output as coded data.

(3) Decoding process FIG. 2 is a diagram showing each functional block used in the decoding process of the encoding apparatus of the present invention. The decoding process will be described below with reference to this figure. The code data input to the encoding device via the telephone line is temporarily stored in the code data memory 700. The CPU 50 extracts the data of each added coefficient from the code data stored in the code memory 700. Huffman code table creation unit 80
Then, based on the value of each coefficient extracted by the CPU, an approximate expression of the graph of the occurrence probability distribution represented by “Equation 1” or “Equation 2” is obtained, and a Huffman code table is created (Huffman code table calculation unit 801), and stores this (Huffman code table memory).

In the encoding processing unit 900, the encoded data read from the encoded data memory 700 is subjected to the decryption processing using the Huffman code table output from the Huffman code table creating unit 800, and the decrypted binary value is obtained. Output the converted data.

As described above, by using the encoding device of the present invention, it becomes possible to execute the encoding process according to the characteristic of each binary data. Further, by using this for sending and receiving documents and the like, it becomes possible to hold confidential documents.

[0025]

The encoding apparatus of the present invention can create a Huffman code table according to the characteristics of each binary data, and use this to execute encoding and decoding processing. The compression rate of the digitized data can be improved.

[Brief description of drawings]

FIG. 1 is a diagram showing a control block of an encoding device.

[Fig. 2] Fig. 2 is a diagram illustrating each functional block used in an encoding process of an encoding device.

[Fig. 3] Fig. 3 is a diagram illustrating each functional block used for decoding processing of the encoding device.

[Explanation of symbols]

 100 ... Binary data memory 200 ... Probability distribution calculation block 300 ... Least squares calculation unit 400 ... Table creation block 500 ... Run length counter 600 ... Huffman coding block 700 ... Code memory 800 ... Table creation block 900 ... Huffman decoding block

Claims (2)

[Claims] 1. A memory for storing input binarized data, a run length calculation means for obtaining a run length occurrence probability distribution of the binarized data stored in the memory, and a run length calculation means. Coefficient calculating means for calculating each coefficient of the approximate expression of the occurrence probability distribution, table creating means for creating a Huffman code table based on the coefficient calculated by the coefficient calculating means, and binary data stored in the memory. A counter means for sequentially obtaining and outputting the run length value, a run length value output by the counter means, and an encoding means for executing an encoding process using the Huffman code table created by the table creating means, Add the data of each coefficient of the approximate expression of the occurrence probability distribution calculated by the coefficient calculating means to the data calculated by the encoding means. And an adding unit for outputting the data as coded data. 2. Extraction means for extracting the data of each coefficient added to the code data when the code data including the data of each coefficient of the approximation formula of the probability distribution of occurrence of run length of binarized data is input. And a table creating means for creating an Huffman code table by obtaining an approximate expression of the occurrence probability distribution based on the value of each coefficient extracted by the extracting means, and a Huffman code table created by the table creating means, An encoding device comprising decoding means for executing decoding processing from coded data and outputting binarized data.