CN102811113B - Character-type message compression method - Google Patents

Abstract

The invention discloses a method for compressing character-type messages. The method provides an optimized adaptive update method for updating the frequency table. One is to update character by character during the encoding process, that is, to update each After the characters are arithmetically coded, the frequency table is updated. Updating the frequency table will take up a certain amount of calculations. If the calculation resources are limited, the first method cannot be applied. The present invention can adopt another method, that is, the update of the frequency table is carried out in units of several messages, after performing arithmetic coding on a single message character by character, only the number of occurrences of each character is recorded, and when the set number of messages is reached, After the text encoding is completed, the frequency table is updated according to the recording situation. The present invention can effectively realize the lossless compression of the message, improve the problems of high delay, occupying excess bandwidth, and using large storage resources encountered in the application process of message sharing, storage, distribution, etc., so that the compression ratio is close to or Reaching the maximum value of entropy coding theory.

Description

A Method for Compressing Character Type Messages

technical field

The invention relates to an effective compression method based on character-type messages. According to the characteristics of character-type messages having a limited character set, a static frequency space is introduced and updated adaptively, and combined with related technologies such as arithmetic coding, a good compression method is obtained. compression effect.

The present invention is applicable to any occasions such as sharing, storage and transmission based on limited character set message compression, especially when the real-time requirement of message transmission is relatively high, and it can well meet the application requirements through actual verification.

Background technique

Data compression methods can be divided into two types according to whether there is loss of information before and after compression, namely lossy compression and lossless compression. Lossy compression refers to the use of compressed data for reconstruction (or restoration, decompression), and the reconstructed data is different from the original data; while lossless compression refers to the use of compressed data for reconstruction, The data is exactly the same as the original data, and the method described in the patent of the present invention is a lossless compression method.

Lossless data compression can be divided into three categories: prediction, dictionary, and statistics according to the implementation technology. Predictive coding is mainly based on the characteristics of a certain correlation between discrete signals, using one or more previous signals to predict the next signal, and then coding the difference between the actual value and the predicted value (prediction error), a typical Methods include DPCM, ADPCM, etc., which are more suitable for compression of sound and image data. Dictionary encoding mainly uses the characteristic that the data itself contains many repeated strings. Its basic principle is to continuously extract new strings from the character stream, and then replace the strings with code names to achieve compression. Typical methods are LZW encoding, etc. LZW encoding achieves compression by dynamically generating a string table during the encoding process and replacing longer strings with shorter code names. Statistical coding is also called entropy coding method, which is mainly compressed according to the distribution characteristics of character occurrence probability. Typical methods include run-length coding, Huffman coding, arithmetic coding and so on. The basic principle of run-length coding is to use a symbol value or string to replace consecutive symbols with the same value, so that the length of the symbol is less than the length of the original data, which is suitable for occasions where the same symbol appears repeatedly; the basic principle of Huffman coding is to Short codewords are coded for information symbols with a high probability of occurrence, and long codewords are coded for information symbols with a low probability of occurrence; the concept of arithmetic coding was proposed by Peter Elias in 1960, but although it is mathematically established, it cannot be realized by a computer. It was not practically applied at the time. In 1976, R.Pasco and J.Rissanen implemented finite-precision arithmetic coding with fixed-length registers, so that it can be realized on a computer. The basic principle is to represent the encoded message as a real number between 0 and 1. Interval, the longer the message, the smaller the interval between the codes and the more binary bits required to represent this interval, the symbols with a higher probability of occurrence make the interval change slower during encoding, and the encoding results produce slower intervals. The whole encoding process adopts the idea of replacing a string of input symbols with a single floating-point number, avoiding the use of a specific codeword to replace an input symbol, that is, the number of bits in Huffman encoding must be rounded The problem. In contrast, arithmetic coding has higher efficiency and superiority, especially when the source contains fewer symbols, such as only two symbols, arithmetic coding is obviously more advantageous, while Huffman coding can hardly achieve any compression effects.

The use of communication messages (hereinafter referred to as messages) is very common, such as radar target information, location information, time information, etc., which are mainly composed of characters. Characters refer to letters, numbers and symbols used in computers, and their storage requires one byte. For details, see the ASCII code table. With the advent of the information age, the storage of various messages presents a massive feature, which brings great pressure to sharing, storage, and distribution. For example, the vehicle (bus, taxi) monitoring and dispatching system covering the entire city, each vehicle transmits its own attributes (such as location, status) to the center through a specific message format, and its mobile characteristics determine that it must communicate wirelessly. At the same time, the center will establish a historical database for each vehicle, and the huge amount of vehicle information will bring inconvenience to communication and storage. In actual use, in order to facilitate observation and interaction, a large number of message formats with character characteristics are used, such as the widely used NMEA-0183 message format, which is a standard format developed by the National Marine Electronics Association for marine electronic equipment , has become a unified standard protocol for GPS navigation equipment.

At present, the use of character-type message formats (such as transmission and storage) is basically processed directly without compression. From the existing literature and published materials, the compression schemes adopted are:

1. Use BCD code to compress the message

BCD code, also known as binary code decimal number or binary-decimal code, is a binary digital encoding form, suitable for processing ten numbers from 0 to 9, and fixedly uses 4 binary numbers to represent ten numbers.

This solution has a limited scope of application and is only suitable for compressing numeric characters, not for letters and the like.

2. Use the extended BCD code to compress the message

Binarize all characters in the character set, and use the binarized data to represent characters to achieve compression. For example, if there is a set of 100 characters, if it is binarized, each character will be assigned 7 binary bits.

This scheme is a typical equal-probability Huffman coding method, which considers that each character is equal-probability, and does not consider the character probability characteristics, which leads to bit waste and limited compression ratio.

3. Use Huffman coding to compress the message

Huffman coding uses a variable-length coding table to encode the source symbols. The variable-length coding table is obtained by a method of evaluating the frequency of source symbols. The symbols with high frequency of occurrence use shorter codes, and vice versa use longer codes. coding. The traditional Huffman coding is a static coding method, which mainly creates a Huffman tree by counting the frequency of each character in the original data to encode the original data. This method is used in practical application systems There are great limitations, especially in real-time transmission and processing systems such as communications. Therefore, it has not been widely used in message compression. Adaptive Huffman coding is a dynamic coding method for the above method, which has been applied in message compression. Its basis for data coding is a dynamically changing Huffman tree, that is, for the N+1th character Encoding is performed based on the Huffman tree obtained from the first N characters in the original data. Every time a character is read, the character count must be adjusted and the Huffman tree updated to ensure the highest encoding efficiency.

This scheme does not consider the joint probability, and because the number of bits must be rounded during the encoding process, the compression efficiency is discounted, resulting in a waste of the output code stream.

Contents of the invention

Purpose of the invention: the present invention is based on the above-mentioned problems encountered in processing character-type message formats, and for character-type message formats, a general lossless message compression method is proposed, which is based on arithmetic coding and introduces static frequency The establishment of the table and the adaptive frequency table can effectively realize the lossless compression of the message, and improve the high delay encountered in the application process of message sharing, storage, distribution, etc., occupying excess bandwidth, using large storage resources, etc. The problem is to make the compression ratio close to or reach the maximum value of entropy coding theory.

Technical solution: a method for compressing character-type messages, comprising the following steps:

Assuming that the character set of the character message is A, the number of characters is n, and the character probability is P _i , then

a _i ∈ A

$\underset{i=1}{\overset{\mathrm{no}}{\mathrm{\Σ}}}{P}_i=1,$ where 1≤i≤n

(1) Pretreatment

When using the character message format for encoding for the first time, the frequency table _Adapt_Table needs to be initialized. There are two specific methods: one is to assign a specific value to P _i according to the characteristics of the character set of the message, combined with the specific use environment, so as to create a character The static frequency table _Exper_Table of the experience value of the collection, and assign its specific value to the frequency table _Adapt_Table; the second is to create an equal probability static frequency table _EqualPro_Table, namely

${\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; no</span>}}$

And assign it to the frequency table _Adapt_Table. In actual use, the initialization method can be selected according to specific needs;

(2) Receive a message

Assume that the received message is Message, the character sequence is B, and the number of sequence elements is m, that is

b _j ∈ A, where 1≤j≤m

(3) Read in characters

Read each character of the received message Message one by one, assuming that the read character is b _j , 1≤j≤m, and its probability is P _bj ;

(4) Arithmetic coding

Perform arithmetic coding on the character according to the current frequency table _Adapt_Table, combined with the current character frequency P _bj ;

(5) Determine whether to update the frequency table

According to actual needs, in the encoding process, the update of the frequency table _Adapt_Table can be performed character by character, that is, after performing arithmetic coding on each character in the message Message, the frequency table is updated; it can also be performed in units of several messages, That is, after performing arithmetic coding on a single message character by character, only record the number of occurrences of each character, and then update the frequency table according to the recording situation after reaching the number of set message encoding; if you need to update the frequency table _Adapt_Table, Then execute the next step (6), otherwise jump to step (7);

(6) Update frequency table

Update the frequency table _Adapt_Table by updating the frequency P _bj of the character b _j ;

(7) Whether the coding of this message is over

If the message encoding of this message is not over, then jump to step (3) and continue to encode the next character, otherwise go to the next step (8);

(8) Determine whether there is a next message

If yes, execute step (9), otherwise execute step (11), that is, end this encoding;

(9) Determine whether to update the frequency table

For the case of using the method of updating the frequency table _Adapt_Table in units of several messages, after the encoding of this message Message is completed, if the frequency table is to be updated, then perform the next step, otherwise skip to step (2) , read the next message and continue encoding;

(10) Update frequency table

Use the recorded number of character occurrences to update the frequency table _Adapt_Table;

(11) end

End this encoding.

The arithmetic coding in the step (4) includes the following steps:

Assume that the initial coding interval used in arithmetic coding is [0,Max], Max is the maximum value of the interval, generally set to 0xFFFF, the interval during the encoding process is [Low,High], and the interval range is Range, where Low is the lower edge of the interval, The initial value is 0, High is the upper edge of the interval, the initial value is Max, the read character is b _j , its frequency is P _bj , and the cumulative frequency is CumP _bj , that is, the sum of the frequencies whose symbol value is less than the symbol.

(41) initialization

Initialize the coding interval [0,Max], and establish a frequency table;

(42) Read in character b _j

Read each character of the message Message one by one, assuming that the read character is b _j , 1≤j≤m, and its probability is P _bj ;

(43) Update interval

According to the current frequency table and P _bj and CumP _bj , update the interval [Low,High], the specific calculation formula is as follows:

Range=High-Low+1

High＝Low+Range*(CumP _bj +P _bj )-1

Low＝Low+Range*CumP _bj

(44) Normalization

Check whether the interval [Low, High] satisfies the conditions for continuing encoding, if so, continue encoding, otherwise normalize the interval [Low, High];

(45) Determine whether to update the frequency table

If so, execute the next step (46), otherwise jump to step (47);

(46) Update frequency table

Updating the frequency P _bj of the coded characters and the corresponding cumulative frequency CumP _bj , that is, updating the frequency table;

(47) Determine whether to end

If so, then end this encoding, otherwise jump to step (42) and continue to encode the next character.

In the step 44, the normalization operation is performed on the interval [Low, High], specifically divided into the following three situations:

Situation 1: The highest bit of the upper edge of the interval is 1, the second highest bit is 0, the highest bit of the lower edge is 0, and the second highest bit is 1, and the second highest bit is removed, that is, the second highest bit is ignored, and the ignored second highest bit is recorded. Times Case1Num;

Case 2: The highest bit of the upper and lower edges of the interval is 0, then perform the operation of shifting the upper and lower edges to the left by 1 bit, and add 1 to the upper edge, and add the shifted bit to the output stream, and then check whether Case1Num is 0, if not 0, the highest bit inversion is called Case1Bit, and Case1Num Case1Bit is output to the output code stream;

Case 3: The highest bit of the upper and lower edges of the interval is 1, then perform the operation of shifting the upper and lower edges to the left by 1 bit, and add 1 to the upper edge, and add the shifted bit to the output stream, and then check whether Case1Num is 0, if not 0, the inversion of the highest bit is called Case1Bit, and Case1Num Case1Bit is output to the output code stream.

The purpose of normalization is to prevent the interval from becoming narrower and narrower as the encoding progresses, resulting in errors in encoding and decoding.

According to the technical solution in this section, decoding is the inverse process of encoding, and will not be repeated here.

Beneficial effects: the present invention has the following beneficial effects through practical application and demonstration:

(1) Based on the characteristics of limited character set messages, arithmetic coding is applied to its lossless compression, which fully utilizes the advantages of arithmetic coding. Compared with BCD code, Huffman coding and other compression methods, it has a higher compression ratio and efficiency .

(2) By using the establishment of the empirical value frequency table, the message can achieve a better compression effect at the initial stage of the compression process;

(3) Two methods of dynamically updating the frequency table are introduced. Among them, the method of updating character by character during the encoding process fully considers the problem of character probability and increases the message compression ratio as much as possible; the unit is several messages The method of updating meets the environment with limited computing resources;

Description of drawings

Fig. 1 is the flowchart of the embodiment of the present invention;

Fig. 2 is a flow chart of arithmetic coding in the embodiment of the present invention.

Detailed ways

Below in conjunction with accompanying drawing and specific embodiment, further illustrate the present invention, should be understood that these embodiments are only for illustrating the present invention and are not intended to limit the scope of the present invention, after having read the present invention, those skilled in the art will understand various aspects of the present invention Modifications in equivalent forms all fall within the scope defined by the appended claims of this application.

As shown in Figure 1, the program completes the initialization of the frequency table in the preprocessing step, which is completed by using an equal-probability static frequency table or an empirical value static probability table. For the update of the frequency table, an optimized adaptive update method is given, which is embodied in steps (6) and (10), which adopts two methods, one is to update character by character during the encoding process, that is, to update the After each character in the text is arithmetically coded, the frequency table is updated. Updating the frequency table will take up a certain amount of calculations. If the calculation resources are limited, the first method cannot be applied. The present invention can adopt another method, that is, the update of the frequency table is carried out in units of several messages, after performing arithmetic coding on a single message character by character, only the number of occurrences of each character is recorded, and when the set number of messages is reached, After the text encoding is completed, the frequency table is updated according to the recording situation.

The variables involved in this scheme are described as follows:

①_Exper_Table: Static frequency table established according to experience value;

②_EqualPro_Table: The probability of each character in the character set is equal, that is, the static frequency table of equal probability;

③_Adapt_Table: Adaptive frequency table in the encoding process.

Assuming that the set of all possible characters in a character message is A, and the number of elements in the set is n, where the probability of occurrence of character a _i is P _i , then:

a _i ∈ A

$\underset{i=1}{\overset{\mathrm{no}}{\mathrm{\&Sigma;}}}{P}_i=1,$ where 1≤i≤n

The technical scheme step that the present invention adopts is as follows, and concrete flow chart sees accompanying drawing 1:

(1) Pretreatment

When using this message format for encoding for the first time, the frequency table _Adapt_Table needs to be initialized. There are two specific methods. One is to assign a specific value to P _i according to the characteristics of the character set of the message, combined with the specific use environment, so as to create a static frequency table _Exper_Table of the experience value of the character set, and assign its specific value to _ Adapt_Table; the second is to create an equal probability static frequency table _EqualPro_Table, namely

${\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; no</span>}}$

And assign it to _Adapt_Table. In actual use, the initialization method can be selected according to specific needs;

(2) Receive a message

Suppose the message is Message, the character sequence is B, and the number of sequence elements is m, that is

b _j ∈ B, where 1≤j≤m

(3) Read in characters

Read each character of the Message one by one, assuming that the read character is b _j , 1≤j≤m, and its probability is P _bj ;

(4) Arithmetic coding

Arithmetic encoding of the character according to the current frequency table combined with the current character frequency P _bj ;

(5) Whether to update the frequency table

According to actual needs, during the encoding process, the update of the frequency table_Adapt_Table can be performed character by character, that is, after the arithmetic encoding of each character in the message Message, the frequency table is updated; it can also be performed in units of several messages, that is, to After a single message is arithmetically coded character by character, only the number of occurrences of each character is recorded, and the frequency table is updated according to the recording situation after reaching the number of set message codes.

The specific steps are if the frequency table _Adapt_Table needs to be updated, then execute the next step (6), otherwise skip to step (7);

(6) Update frequency table

Update the frequency table _Adapt_Table by updating the frequency P _bj of the character b _j ;

(7) Whether the coding of this message is over

If the message encoding of this message is not over, then jump to step (3) and continue to encode the next character, otherwise go to the next step (8);

(8) Whether there is a next message

If yes, execute step (9), otherwise execute step (11), that is, end this encoding;

(9) Whether to update the frequency table

For the above-mentioned second method of updating the frequency table _Adapt_Table, after the encoding of this message Message is completed, if the frequency table needs to be updated, then execute the next step; otherwise, jump to step (2) and read the next message text, continue coding;

(10) Update frequency table

Specifically, update the frequency table _Adapt_Table using the recorded number of occurrences of characters;

(11) end

End this encoding.

For the step (4) arithmetic coding in the technical solution adopted by the present invention, the detailed process is as follows, and the specific flow chart is shown in Figure 2:

Assume that the initial coding interval used in arithmetic coding is [0,Max], Max is the maximum value of the interval, generally set to 0xFFFF, the interval during the encoding process is [Low,High], and the interval range is Range, where Low is the lower edge of the interval, The initial value is 0, High is the upper edge of the interval, the initial value is Max, the read character is b _j , its frequency is P _bj , and the cumulative frequency is CumP _bj , that is, the sum of the frequencies whose symbol value is less than the symbol.

(41) initialization

Initialize the coding interval [0,Max], establish the frequency table, etc.;

(42) Read in character b _j

Read each character of the Message one by one, assuming that the read character is b _j , 1≤j≤m, and its probability is P _bj ;

(43) Update interval

According to the current frequency table and P _bj and CumP _bj , update the interval [Low,High], the specific calculation formula is as follows:

Range=High-Low+1

High＝Low+Range*(CumP _bj +P _bj )-1

Low＝Low+Range*CumP _bj

(44) Normalization

Check whether the interval [Low, High] satisfies the conditions for continuing coding, if so, continue coding, otherwise normalize the interval [Low, High], specifically divided into the following three situations:

Situation 1: The highest bit of the upper edge of the interval is 1, the second highest bit is 0, the highest bit of the lower edge is 0, and the second highest bit is 1, and the second highest bit is removed, that is, the second highest bit is ignored, and the ignored second highest bit is recorded. Times Case1Num;

Case 2: The highest bit of the upper and lower edges of the interval is 0, then perform the operation of shifting the upper and lower edges to the left by 1 bit, and add 1 to the upper edge, and add the shifted bit to the output stream, and then check whether Case1Num is 0, if not 0, the highest bit inversion is called Case1Bit, and Case1Num Case1Bit is output to the output code stream;

Case 3: The highest bit of the upper and lower edges of the interval is 1, then perform the operation of shifting the upper and lower edges to the left by 1 bit, and add 1 to the upper edge, and add the shifted bit to the output stream, and then check whether Case1Num is 0, if not 0, the inversion of the highest bit is called Case1Bit, and Case1Num Case1Bit is output to the output code stream.

The purpose of normalization is to prevent the interval from becoming narrower and narrower as the encoding progresses, resulting in errors in encoding and decoding;

(45) Whether to update the frequency table

If so, execute the next step (46), otherwise jump to step (47);

(46) Update frequency table

Update the frequency P _bj of the encoded character and the corresponding cumulative frequency CumP _bj , that is, update the frequency table;

(47) Is it over?

If so, then end this encoding, otherwise jump to step (42) and continue to encode the next character.

According to the technical solution in this section, decoding is the inverse process of encoding, and will not be repeated here.

The technical solution of the present invention will be described in detail below by taking the format of NMEA-0183 widely used in positioning information as an example, but the protection scope of the present invention is not limited to the embodiments.

Specifically, take the message format representing geographic positioning information in NMEA-0183 as an example, and assume that the Message of this message is "$GPGLL,4250.5589,S,14718.5084,E,092204.999,A*2D", and each field in the message is separated by a comma The specific information represented by each field is as follows:

Field 0: $GPGLL, statement ID, indicating that the statement is Geographic Position (GLL) geographic positioning information;

Field 1: latitude ddmm.mmmm, in degree-minute format (if the leading digits are insufficient, add 0);

Field 2: Latitude N (northern latitude) or S (southern latitude);

Field 3: Longitude dddmm.mmmm, in degree-minute format (if the leading digits are insufficient, add 0);

Field 4: Longitude E (East) or W (West);

Field 5: UTC time, hhmmss.sss format;

Field 6: Status, A=locate, V=not locate;

Field 7: check value.

The specific encoding steps of this message are as follows:

(1) Pretreatment

In view of the characteristics of frequent numbers and commas in the message character set, combined with the specific use environment, create a static frequency table _Exper_Table of the experience value of the character set, and initialize _Adapt_Table to _Exper_Table;

(2) Read in characters

Read each character of the message Message one by one;

(3) Arithmetic coding

Arithmetic encoding of the character according to the current frequency table combined with the current character frequency;

(4) Whether to update the frequency table

According to actual needs, during the encoding process, the update of the frequency table _Adapt_Table can be performed character by character, that is, after performing arithmetic encoding on each character in the message Message, the frequency table is updated; it can also be performed in units of several messages, namely After performing arithmetic coding on a single message character by character, only the number of occurrences of each character is recorded, and the frequency table is updated according to the recording situation after the encoding of the set number of messages is completed.

The specific steps are as follows: if the frequency table _Adapt_Table needs to be updated, execute the next step (5), otherwise skip to step (6);

(5) Update frequency table

Update the frequency table _Adapt_Table by updating the frequency of this coded character;

(6) Is the coding of this message ended?

If the message encoding of this message is not over, then jump to step (2) and continue to encode the next character, otherwise go to the next step (7);

(7) end

End this encoding.

Claims (3)

1. a character type message compression method, is characterized in that: comprise the steps: Assuming that the character set of the character message is A, the number of characters is n, and the character probability is P _i , then a _i ∈ A where 1≤i≤n (1) Pretreatment When using the character message format for encoding for the first time, it is necessary to initialize the frequency table _Adapt_Table and assign it to the frequency table _Adapt_Table; (2) Receive a message Assume that a received message is Message, the character sequence is B, and the number of sequence elements is m, that is b _j ∈ A, where 1≤j≤m (3) Read characters Read each character of the received message Message one by one, assuming that the read character is b _j , 1≤j≤m, and its probability is P _bj ; (4) Arithmetic coding Perform arithmetic coding on the character according to the current frequency table _Adapt_Table, combined with the current character frequency P _bj ; The specific steps of arithmetic coding are: Assume that the initial coding interval used in arithmetic coding is [0,Max], Max is the maximum value of the interval, set to 0xFFFF, the interval during the encoding process is [Low,High], and the interval range is Range, where Low is the lower edge of the interval, and the initial is 0, High is the upper edge of the interval, the initial is Max, the read character is b _j , its frequency is P _bj , and the cumulative frequency is CumP _bj , that is, the sum of the frequencies whose symbol value is less than the symbol; (41) Initialization Initialize the coding interval [0,Max], and establish a frequency table; (42) Read in character b _j Read each character of the message Message one by one, assuming that the read character is b _j , 1≤j≤m, and its probability is P _bj ; (43) Update interval According to the current frequency table and P _bj and CumP _bj , update the interval [Low,High], the specific calculation formula is as follows: Range＝High-Low+1 High＝Low+Range*(CumP _bj +P _bj )-1 Low＝Low+Range*CumP _bj (44) Normalization Check whether the interval [Low, High] satisfies the conditions for continuing encoding, if so, continue encoding, otherwise normalize the interval [Low, High]; Perform normalization operations on the interval [Low, High], specifically divided into the following three situations: Situation 1: The highest bit of the upper edge of the interval is 1, the second highest bit is 0, the highest bit of the lower edge is 0, and the second highest bit is 1, and the second highest bit is removed, that is, the second highest bit is ignored, and the ignored second highest bit is recorded. Times Case1Num; Case 2: The highest bit of the upper and lower edges of the interval is 0, then perform the operation of shifting the upper and lower edges to the left by 1 bit, and add 1 to the upper edge, and add the shifted bit to the output stream, and then check whether Case1Num is 0, if not 0, the highest bit inversion is called Case1Bit, and Case1Num Case1Bit is output to the output code stream; Case 3: The highest bit of the upper and lower edges of the interval is 1, then perform the operation of shifting the upper and lower edges to the left by 1 bit, and add 1 to the upper edge, and add the shifted bit to the output stream, and then check whether Case1Num is 0, if not 0, the highest bit inversion is called Case1Bit, and Case1Num Case1Bit is output to the output code stream; (45) Determine whether to update the frequency table If so, then execute the next step (46), otherwise jump to step (47); (46) Update frequency table Updating the frequency P _bj of the coded characters and the corresponding cumulative frequency CumP _bj , that is, updating the frequency table; (47) Judging whether it is over If so, then end this encoding, otherwise jump to step (42) and continue to encode the next character; (5) Determine whether to update the frequency table According to actual needs, in the encoding process, the frequency table _Adapt_Table is updated or performed character by character, that is, after performing arithmetic encoding on each character in the message Message, the frequency table is updated; or performed in units of several messages, That is, after performing arithmetic coding on a single message character by character, only record the number of occurrences of each character, and then update the frequency table according to the recording situation after reaching the number of set message encoding; if you need to update the frequency table _Adapt_Table, Then execute the next step (6), otherwise jump to step (7); (6) Update frequency table Update the frequency table _Adapt_Table by updating the frequency P _bj of the character b _j ; (7) Is the coding of this article over? If the message encoding of this message is not over, then jump to step (3) and continue to encode the next character, otherwise perform the next step (8); (8) Determine whether there is a next message If so, execute step (9), otherwise execute step (11), that is, end this encoding; (9) Determine whether to update the frequency table For the case of using the method of updating the frequency table _Adapt_Table in units of several messages, after the encoding of this message Message is completed, if the frequency table is to be updated, the next step is performed, otherwise jump to step (2) , read the next message and continue encoding; (10) Update frequency table Use the recorded number of character occurrences to update the frequency table_Adapt_Table; (11) end End this encoding. the 2. the character type message compression method as claimed in claim 1, it is characterized in that: the initialization of frequency table is proposed by empirical value and two kinds of modes of equal probability, wherein, according to the character set characteristic of message in the empirical value mode, Combined with the specific use environment, assign specific values to P _i , thereby creating a static frequency table _Exper_Table of character sets, and assigning it to the frequency table _Adapt_Table; and the equal probability method is to create an equal probability static frequency table _EqualPro_Table, Right now And assign it to the frequency table _Adapt_Table. 3. The character type message compression method as claimed in claim 1, characterized in that: decoding is the inverse process of encoding. the