CN106407285A - RLE and LZW-based optimized bit file compression and decompression method - Google Patents

Abstract

The invention discloses an RLE and LZW-based optimized bit file compression and decompression method. By performing data format analysis on an FPGA configuration bit file, a head control word of the bit file is extracted; preliminary compression is performed by adopting an RLE code with the run length of 4 from real configuration data; and then LZW compression is performed, so that the compression rate is further increased. The decompression is an inverse process of compression; and LZW decompression is performed to recover intermediate data first, and then a data part which does not contain the head control word is subjected to RLE decompression to recover the original FPGA configuration bit file. The method comprehensively considers compression/decompression time and compression rate; compared with a compression tool of Xilinx, single use of an RLE algorithm and single use of an LZW algorithm, the method has the advantages that the win-win purpose of the compression rate and the compression speed is achieved; and the problem of an excessively large FPGA configuration bit file of advanced type of Xilinx is solved, the overhead of a storage chip is reduced, and a key technical support is provided for an FPGA in-orbit reconstruction technology.

Description

An Optimized Bit File Compression and Decompression Method Based on RLE and LZW

technical field

The invention relates to an optimized bit file compression and decompression method based on RLE and LZW, and belongs to the technical field of bit file compression and decompression in communication.

Background technique

With the increase in the complexity of on-board applications, the requirements for on-board data processing capabilities are getting higher and higher, and the FPGA models used have to use higher-level devices with the increase in application complexity. Due to the more powerful functions of these advanced FPGA models, the corresponding bit files have also become larger. The bit files of an FPGA configured with a V5130T have increased by about 3 times compared with the V4SX55, which means that the application of a V5 FPGA requires the past 3 Times the storage resources to store the configuration bits, which is undoubtedly a problem for the limited resources on the star, and certain compression methods must be adopted to reduce storage resources. And the following requirements must be met: 1) the bit file must be restored without loss, and the bit stream must be decompressed to accurately restore the original data; 2) the compression rate should be high, and the compression and decompression time should be short. Aiming at this problem and the above two requirements, an optimized bitfile compression and decompression method based on RLE and LZW is proposed to achieve high-speed and high compression rate bitfile compression and decompression.

Data compression usually includes lossy compression and lossless compression. Lossy compression is compression with distortion, irreversible, and information will be lost. Lossless compression is lossless compression. The redundancy in the data is removed during the compression process, and no information is lost after decompression, which is exactly the same as before compression. Lossless compression is usually used for the compression of text, programs, and important data, and it can guarantee the complete restoration of the original data. Compressing the bit stream requires that the original data can be accurately restored, so it is necessary to use lossless compression technology to compress and decompress the bit file to solve the transmission and storage problems of the configuration bit data.

The classic lossless compression algorithms mainly include: algorithms based on statistical modes, such as Shannon coding, Huffman coding, arithmetic coding, etc.; algorithms based on dictionary modes, such as LZ series algorithms; and some other algorithms, such as run-length coding, JPEG -LS algorithm, DEM algorithm of JPEG2000, BWT algorithm, etc. Considering the design complexity and compression efficiency, the dictionary mode algorithm is more suitable for bit file compression than the statistical mode algorithm. At present, the main compression methods for bit file compression at home and abroad are dictionary compression or run-length coding compression. Xilinx's own compression tool uses the dictionary-style LZ77 method for bitfile compression. It can be seen from the following experimental data that the compression rate of Xilinx compressed bit files is not high.

Contents of the invention

The technical problem solved by the present invention is: to overcome the deficiencies in the prior art, to provide a method for compressing and decompressing optimized bit files based on RLE and LZW, and under the requirements of fast compression and high compression rate, the present invention adopts the bit file data structure Analysis, adaptive optimization improved the RLE algorithm, combined with the LZW method to further improve the compression rate, and designed an optimized bit file compression/decompression method based on RLE and LZW. This method analyzes the bit file data structure, Creatively optimize the RLE algorithm according to the characteristics of the bit file; on the basis of applying the optimized RLE algorithm, in order to further improve the compression rate, creatively combine the optimized RLE algorithm with the LZW algorithm and apply it to the bit file compression and decompression. This method saves storage resources and improves computing efficiency. The design and implementation of compression and decompression verify the correctness, effectiveness, efficiency and feasibility of the method.

The technical scheme that the present invention solves is: a kind of optimization bit file compression method based on RLE and LZW, comprises steps as follows:

Step 1: The FPGA configuration bit file is composed of multiple parts of data, including the header control word, data and command word. The format of the data and command word is in units of four bytes, and the header of the FPGA configuration bit file is removed. control word;

The run length counter in the initialization RLE; The run length of RLE is set to 4 according to the bit file data characteristics described in step 1;

Initialize the compressed dictionary of LZW so that the dictionary contains all possible roots, and the root is a single entry;

Set the prefix pre_char, so that the current prefix pre_char is empty;

Step 2: the string pre_string is set; the first character string of the FPGA configuration bit file that removes the header control word is assigned to the string pre_string in step 1; the FPGA configuration bit file that removes the header control word is set Compressed file;

Step 3: Compress the compressed file, the steps are as follows:

(1) Judging whether there is a character string to be compressed in the compressed file, if there is no character string to be compressed, then first output the character string pre_string, then output the value of the tour length counter, and enter step 4;

(2) If there is a character string that needs to be compressed, set the current character string current_string so that the current character string current_string is equal to the next character string in the character stream;

(3) Determine whether pre_string is consistent with current_string: if pre_string is consistent with current_string, add 1 to the tour length counter, and judge whether the tour length counter reaches 255; if the tour length counter reaches 255, first output pre_string, and then output the tour length counter value, so that the swim length counter value is 1,

Read the uncompressed string of the file to be compressed to pre_string, return to step 3

If the swim length counter value does not count to 255, return to step 3;

If pre_string is inconsistent with current_string, output pre_string, output the value of the tour length counter, assign current_string to pre_string at this time, and return to step 3;

Step 4: Set the current character current_char, according to the output sequence of step 3 (1), arrange the output into a new character stream in turn, read the first character in the character stream and assign it to current_char;

Step 5: Determine whether the entry consisting of pre_char and current_char is in the dictionary: if it is in the dictionary, assign the entry consisting of pre_char and current_char to pre_char; if it is not in the dictionary, output the code of the current prefix pre_char word, add the entries composed of pre_char and current_char in sequence to the dictionary, and assign the value of current_char to pre_char;

Step 6: Determine whether there are characters in the new character stream that need to be compressed. If there are characters that need to be compressed, assign the first character that needs to be compressed to current_char, and return to step 5; if there are no characters that need to be compressed, output the code of the current prefix pre_char word, complete the compression.

LZW's compression dictionary is used to store the entries generated during the compression process.

The swim length counter starts counting from 1, that is, counts 1 when reading the first character string, and can add 1 each time when encountering consecutive repeated strings, until a different character string appears, it can output the swim length counter value, restore the initial value 1; the value of the swim length counter is 1-255, and restore 1 when it reaches 255.

Initialize LZW's compression dictionary with index values 1-4096.

A kind of optimized bit file decompression method based on RLE and LZW, comprises steps as follows:

Step 1: Initialize the decompression dictionary, so that the dictionary contains all possible roots, and the root is a single entry;

Step 2: Set the variable pw, read the first code stream in the data stream to be decompressed, assign it to pw, query the decompression dictionary with pw as the index, read the entry corresponding to the index, set the string variable pre_char, and set The entry corresponding to the index represented by pw is assigned to pre_char, and pre_char is output;

Step 3: Set the variable cw, continue to read the next code stream in the data to be decompressed, assign it to cw, and judge whether there is an entry corresponding to the index represented by cw in the dictionary, if there is an entry corresponding to the index represented by cw , output the entry corresponding to the index represented by cw, that is, the string data corresponding to the index represented by cw, set the variable current_char, assign the string data corresponding to the index represented by cw to current_char, set the character variable pc, and assign pre_char to pc , set the character variable cc, assign the first character of the entry corresponding to current_char to cc, and add the string composed of pc and cc in sequence to the decompression dictionary; if there is no entry corresponding to the index represented by cw, set Assign the value of pre_char to pc, assign the first character of current_char to cc, output the string data composed of pc and cc arranged in sequence, and add the string data composed of pc and cc arranged in sequence to the dictionary;

Step 4: Determine whether there is still data to be decompressed in the data stream to be decompressed, if there is still data to be decompressed, return to step 3; if there is no data to be decompressed, arrange the output of step 3 into new characters in the output order String data, identify the head control word in the new string data; set the string head, and assign the head control word to head;

Step 5: Set the string variable current_string, read the string from the new string data without the header control word and assign it to current_string according to the length of the tour, and then read a character after the string, which is the value of the length counter Value, restore the original compressed data according to the read string and the value of the swim length counter;

Step 6: Determine whether there is data to be decompressed in the new string data with the header control word removed, and if so, replace the new string data with the header control word removed with the data to be decompressed, and return to step 5 ;, if not, arrange the head and the data output in step 5 in order to obtain the original data, and complete the decompression.

The advantage of the present invention compared with prior art is:

(1) Under the requirements of fast compression and high compression rate, the present invention improves the RLE algorithm by analyzing the data structure of the bit file, and improves the RLE algorithm through adaptive optimization, and further improves the compression rate in conjunction with the LZW method, and designs a method based on RLE and Optimized bitfile compression/decompression methods for LZW. This method creatively optimizes the RLE algorithm according to the characteristics of the bit file by analyzing the data structure of the bit file; on the basis of applying the optimized RLE algorithm, in order to further improve the compression rate, the optimized RLE algorithm is creatively combined with the LZW algorithm , applied to bitfile compression and decompression. This method saves storage resources and improves computing efficiency. The design and implementation of compression and decompression verify the correctness, effectiveness, efficiency and feasibility of the method.

(2) the present invention designs a kind of optimized bit file compression and decompression method based on RLE and LZW at the bit file by analyzing the bit file structure and RLE and LZW algorithm, creatively optimizes the RLE algorithm according to the bit file characteristics, In order to further improve the compression rate, the optimized RLE algorithm is creatively combined with the LZW algorithm and applied to bit file compression and decompression, which saves compression time and a large amount of storage resources on the basis of ensuring a certain compression rate, and through corresponding decompression The program design verifies the correctness, efficiency and feasibility of the compression method.

Description of drawings

Fig. 1 is basic digital structure of the present invention;

Fig. 2 is a kind of optimization bit file compression algorithm flowchart based on RLE and LZW of the present invention;

Fig. 3 is a kind of optimized bit file decompression algorithm flowchart based on RLE and LZW of the present invention;

Fig. 4 is a schematic diagram showing the comparison of compression ratios of different compression algorithms.

detailed description

The basic idea of the present invention is: a kind of optimized bit file compression and decompression method based on RLE and LZW, by analyzing the data format of the FPGA configuration bit file, extracting the header control word of the bit file, starting from the real configuration data, RLE coding with a run length of 4 is used for initial compression, and then LZW compression is used to further improve the compression rate. Decompression is the inverse process of compression. First, LZW decompression is performed to restore the intermediate data, and then the data part that does not contain the header control word is decompressed by RLE to restore the original FPGA configuration bit file. This method comprehensively considers the compression/decompression time and compression rate. Compared with Xilinx's built-in compression tool, and the simple application of RLE algorithm and LZW algorithm, it achieves a win-win situation of compression rate and compression speed. It solves the problem of excessively large configuration bit files of Xilinx advanced FPGA models, saves the cost of memory chips, and provides key technical support for FPGA on-orbit reconfiguration technology.

The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

Run-length encoding (RLE), also known as run-length encoding, is a relatively simple data compression technique. The run length refers to the length of a character string formed by the continuous repetition of each character in the data stream composed of source symbols or signal samples, also known as the run length or run length. Run-length encoding is to map this sequence of characters into the length of the string and the position of the string. When the characters of the run-length string, the position and length of the string are given, the original data stream can be recovered. The basic digital structure of the data stream is shown in Figure 1, where the identification code depends on the specific use and can also be omitted.

According to the basic structure of the data stream of the run-length coding, it can be seen that the run-length coding has the function of data compression only when the length of the run length is greater than 3. The efficiency of run-length coding depends on the repetition rate of the source symbols. The higher the repetition rate and the longer the run length, the more obvious the compression effect. Runlength compression logic and hardware implementation are simple and fast. In actual engineering, run-length coding is sometimes mixed with other coding methods to obtain better compression effects.

It can be known from the principle of run-length coding that the compression effect of run-length coding will be better only when the repetition rate of the compressed sample is relatively high. Analyzing the structure of the bitfile, firstly, each bitfile has its own header information, and the repetition rate of this part of information samples is very low. Compared with the whole bitfile, the header information occupies very few bits, so When compressing bit files, the header information bits are cut out, and only the data behind the header information is compressed, which improves the compression speed and compression rate to a certain extent. After extracting the header information, the remaining bit files still have certain rules, that is, both the command word and the data are based on 4 bytes as a unit, so the length selection of bit file compression by using run-length encoding 4 bytes is the most suitable. If you choose less than 4 bytes, more length information will be added in the compressed file, which will reduce the compression rate; if you choose more than 4 bytes, the repetition rate of the sample will decrease. The identification code information will be added to the compressed file, reducing the compression ratio. After implementing the optimized RLE compression algorithm, the analysis results are also confirmed by experiments.

Improved run-length coding Since the compression process does not require storage and calculation, the compression speed is very fast, but limited by the complexity of the bit file and the compression algorithm itself, the compression rate is not particularly high. Therefore, on the basis of optimizing the RLE algorithm, the LZW algorithm is further combined to further process the compressed data to improve the compression rate.

The LZW algorithm was proposed by Terry Welch in 1984. It is an improved algorithm of LZ78. Compared with the LZ78 algorithm, the second field of the identification is removed. The LZW algorithm is a compression algorithm based on the dictionary mode, which does not depend on the probability distribution of the information source. It is a lossless data compression algorithm for general data and easy to implement. LZW algorithm does not depend on any data format, has a wide range of applications, and has fast encoding speed, simple logic and self-adaptive function, which is especially beneficial to hardware implementation and has high real-time performance.

The LZW algorithm first initializes all characters in the alphabet into the dictionary, usually using 8-bit characters, so the first 256 items of the dictionary, ie 0-255, have been occupied before the data is input. Since the dictionary has already been initialized, the next character typed will always be found in the dictionary. In practice, the dictionary encoding is an extension of the 8-bit character set, which is used to represent the strings appearing in the data. The newly added dictionary encoding can be represented by 9 bits, 10 bits, 11 bits, 12 bits or even more bits. For example, if each character string is represented by 9 bits, there can be 512 different 9-bit codes. That is to say, the conversion table has 512 character strings, 256 of which are used to store defined characters, and the remaining 256 are used to store (prefix, string).

The encoding principle of the LZW algorithm is as follows: firstly, the dictionary needs to be initialized, and the first 256 items are allocated to 0~255, and then each time a character pre_char is read, it must first be searched in the dictionary. If the character already exists in the dictionary, update the current character to pre_char, and use the current character pre_char as a prefix, continue to read the next character current_char as the end character, form a string "pre_char, current_char", and look it up in the dictionary again "pre_char, current_char". If not in the dictionary, output the character pre_char position code, and add "pre_char, current_char" to the dictionary, and then use current_char as a prefix. Repeat the above steps until all encodings are completed.

The decoding algorithm of LZW is the inverse process of encoding, which is also an adaptive algorithm based on the dictionary, which generates the dictionary and decodes the output at the same time.

In summary, the present invention proposes a kind of optimized bit file compression method based on RLE and LZW, comprising steps as follows:

Step 1: The FPGA configuration bit file is composed of multiple parts of data, mainly including the header control word, data and command word. The format of the data and command word is four bytes as a unit, and the header of the FPGA configuration bit file is removed. part control word;

The run length counter in the initialization RLE is 1; The run length of RLE is set to 4 according to the bit file data characteristics described in step 1;

Initialize the compression dictionary of LZW to be 4096 rows*32 columns, so that the dictionary contains all possible roots, and the root is a single entry, that is, the first column data of rows 0 to 255 of the initialized dictionary is 0-255;

Set the prefix pre_char, so that the current prefix pre_char is empty;

Step 2: the string pre_string is set; the first character string of the FPGA configuration bit file that removes the header control word is assigned to the string pre_string in step 1; the FPGA configuration bit file that removes the header control word is set Compressed file;

Step three:

(1) Judging whether there is a character string to be compressed in the compressed file, if there is no character string to be compressed, then first output the character string pre_string, then output the value of the tour length counter, and enter step 4;

(2) If there is a character string that needs to be compressed, set the current character string current_string so that the current character string current_string is equal to the next character string in the character stream;

(3) Determine whether pre_string is consistent with current_string: if pre_string is consistent with current_string, add 1 to the tour length counter, and judge whether the tour length counter reaches 255; if the tour length counter reaches 255, first output pre_string, and then output the tour length counter value, so that the swim length counter value is 1,

Read the uncompressed string of the file to be compressed to pre_string, return to step 3

If the swim length counter value does not count to 255, return to step 3;

If pre_string is inconsistent with current_string, output pre_string, output the value of the tour length counter, assign current_string to pre_string at this time, and return to step 3;

Step 4: (1) Set the current character current_char,

(2) According to the output sequence of step 3 (1), arrange the output into a new character stream in turn, read the first character in the character stream and assign it to current_char; in order to improve the compression speed, this step can have data in step 3 Start at the same time when outputting, read the first character output in step 3 to current_char, and then proceed to the subsequent steps;

Step 5: Determine whether the entry composed of pre_char and current_char is in the dictionary: if it is in the dictionary, pay the entry composed of pre_char and current_char to pre_char; if it is not in the dictionary, output the current prefix pre_char Codeword, add the entries composed of pre_char and current_char in sequence to the dictionary, and assign the value of current_char to pre_char;

Step 6: Determine whether there are characters in the new character stream that need to be compressed. If there are characters that need to be compressed, pay the first character that needs to be compressed to current_char, and return to step 5; if there are no characters that need to be compressed, output the code of the current prefix pre_char word, complete the compression.

LZW's compression dictionary is used to store the entries generated during the compression process.

The swim length counter starts counting from 1, that is, counts 1 when reading the first character string, and can add 1 each time when encountering consecutive repeated strings, until a different character string appears, it can output the swim length counter The value of is restored to the initial value 1; the value of the swim length counter is 1-255.

Initialize LZW's compression dictionary with index values 1-4096.

Combining the above-mentioned optimized bit file compression algorithm and program design ideas, a process flow of an optimized bit file compression algorithm based on RLE and LZW is shown in Figure 2.

After the compression is completed, corresponding to the data processed by the compression technology, decompression is the inverse process of compression, which can restore the original data, or use the LZW algorithm to restore the optimized RLE compressed data, and then optimize the RLE decompression Restore the original original data. Decompression is the opposite of compression.

Provide a kind of preferred specific decompression method flow process as shown in Figure 3 below, a kind of optimal bit file decompression method based on RLE and LZW of the present invention preferably comprises steps as follows:

Step 1: Initialize the decompression dictionary as 4096 rows*32 columns, so that the dictionary contains all possible roots, and the root is a single entry, that is, the data in the first column of rows 0 to 255 of the initialized dictionary is 0-255;

Step 2: Set the variable pw, read the first code stream in the data stream to be decompressed, assign it to pw, use pw as the index to query the dictionary, read the entry corresponding to the index, set the string variable pre_char, and represent pw The entry corresponding to the index of the index is assigned to pre_char, and pre_char is output; if pw is 10, and the string data in the dictionary whose index is 10 is searched, the value is assigned to pre_char;

Step 3: Set the variable cw, continue to read the next code stream in the data to be decompressed, assign it to cw, and judge whether there is an entry corresponding to the index represented by cw in the dictionary, if there is an entry corresponding to the index represented by cw , output the string data corresponding to the index represented by cw (output the entry corresponding to the index represented by cw, that is, the string data corresponding to the index represented by cw), set the variable current_char, and assign the string data corresponding to the index represented by cw to current_char, set the character variable pc, assign pre_char to pc, set the character variable, assign the first character of the entry corresponding to current_char to cc, and add the string composed of pc and cc to the decompression dictionary; if If there is no entry corresponding to the index indicated by cw, assign the value of pre_char to pc, assign the first character of current_char to cc, output the string data composed of pc and cc arranged in sequence, and add the sequence composed of pc and cc String data into the dictionary; if the data to be decompressed is “9 18 300…..4095”, read the first pw value 9, and output the string “9256 256….256” represented by the ninth entry in the dictionary to pre_char, and output "9"; continue to read the cw value 18, which is in the initialized dictionary, output the string "18 256 256....256" represented by item 18 of the dictionary to current_char, and output "18", add Add "9 18256....256" to the dictionary, and the entry index is 257; continue to read the cw value 300, which is not in the initialized dictionary, output "18 18", add "18 18 256....256" to the dictionary , and so on, until all data processing is completed.

Step 4: Determine whether there is still data to be decompressed in the data stream to be decompressed, if there is still data to be decompressed, return to step 3; if there is no data to be decompressed, arrange the output of step 3 into new characters in the output order string data, identify the head control word in the new string data; set the string head, assign the head control word to head (such as the new string data is "head control word AA AA AA 02 BB BB BB BB 01 CC CC CC CC09...DD DD DD DD 03", assign the head control word to the head, follow up on "AA AA AA AA 02 BB BB BB BB 01CC CC CC CC 09...DD DD DD DD 03" unzip);

Step 5: Set the string variable current_string, read the string from the new string data without the header control word and assign it to current_string according to the length of the tour, and then read a character after the string, which is the value of the length counter Value, restore the original compressed data according to the read string and the value of the length counter; such as the string "AABB CC DD 02", read the current_string as "AA BB CC DD", the length of the length is 2, and the restored The data is "AA BBCC DD AA BB CC DD";

Step 6: judge whether there is data to be decompressed in the new character string data that removes the header control word, if so, then replace the new character string data that removes the header control word with compressed data, and return to step 5; If not, arrange the head and the data output in step 5 in order to obtain the original data, and complete the decompression. If there is "AAAA AA AA 02 BB BB BB BB 01 CC CC CC CC 09...DD DD DD DD 03" with decompressed data, read "AA AA AAAA" first, and restore the data to "AA" according to "02" AA AA AA AA AA AA"; read "BB BB BB BB" again, according to "01", restore the data to "BB BB BB BB", read all the data in turn until "DD DD DD DD" is read, According to "03", the restored data is "DD DD DD DD DD DD DD DD DD DD DD DD" (spliced with the header control word to complete the complete original data decompression as "header control word AA AA AA AA AA AA AA AA BB BB BB BB...DD DD DD DD DDDD DD DD DD DD DD").

6. The variable current_char is used as the index number to decompress the corresponding string data in the dictionary.

Corresponding to the data processed by the compression technology, the decompression technology first needs to use the LZW algorithm to restore the optimized RLE compressed data, and then perform the optimized RLE decompression to restore the original original data. The algorithm design idea is similar to the compression algorithm, and the specific decompression algorithm flow is shown in Figure 3.

After the compression and decompression program design is completed, five kinds of bit files with different resource utilization rates are used to test the compression rate and decompression restoration test. The restored data is consistent with the original data, which verifies the correctness of the algorithm implementation. The compression rate is shown in the table below:

Table 1 Compression test data

The graphical display of the experimental results is shown in Figure 4. From the experimental results, it can be seen that:

As resources increase, the compression ratios of all compression methods decrease; for data with very little resource usage (1%), the compression ratios of the three methods we developed are all above 95%, and the compression ratio of Xilinx is only 85%;

For data that uses a lot of resources (99%), the compression rate of Xilinx's compression method is only slightly higher than that of optimized RLE, but still far lower than that of LZW and optimized RLE_LZW compression methods;

For data with slightly more resources (20%, 56%, 81%), the compression rate of the LZW compression method is the highest (70% to 80%) compared with the other three methods, and the compression rate of Xilinx is the lowest (only can reach 20-30%);

The RLE_LZW optimized compression method has a slightly lower compression rate than LZW for bits with high resource usage, and the difference is not very large, but the compression time is much improved compared with the LZW method. This is because the data to be compressed is first compressed by the optimized RLE algorithm. When it is sent to LZW for compression, the amount of data has been reduced a lot compared to uncompressed data. Optimizing the RLE algorithm itself is not time-consuming, but the LZW algorithm itself is very time-consuming because the data needs to be compared one by one with the dictionary, and the dictionary itself is dynamically updated. Compared with the simple LZW algorithm, the RLE_LZW optimization algorithm loses compression rate because the complexity of the compressed data processed by the optimized RLE algorithm is higher than that of the original data, resulting in a slight impact on the compression rate. However, considering the compression time and compression rate comprehensively, an optimized bit file compression and decompression technology based on RLE and LZW is optimal.

Claims (5)

1. a kind of optimization bit file compression method based on RLE and LZW it is characterised in that：As follows including step：

Step one：FPGA configuration bit file is made up of many partial datas, comprises head control word, data and command word, wherein, Data and order word format are all with nybble for a unit, scratch the head control word except FPGA configuration bit file；

Brigade commander's enumerator in initialization RLE；The bit file data characteristicses setting according to step one by the run length of RLE For 4；

The compression dictionary of initialization LZW, makes dictionary comprise all possible, and described is single entry；

Setting prefix pre_char, makes current prefix pre_char be sky；

Step 2：Setup string pre_string；The FPGA configuration bit file removing head control word will be scratched in step one First character string is assigned to character string pre_string；Described scratching except the FPGA configuration bit file of head control word is to be compressed File；

Step 3：To be compressed by compressed file, step is as follows：

(1) judge whether have character string to need to compress in compressed file, need to compress without character string, then first output word Symbol string pre_string, then export the value of brigade commander's enumerator, enter step 4；

(2) need to compress if there are character string, current string current_string is set, makes current string current_ String is equal to the character late string in character stream；

(3) judge whether pre_string is consistent with current_string：If pre_string and current_string Unanimously, brigade commander's enumerator adds 1, judges whether brigade commander's Counter Value counts 255；If brigade commander's Counter Value meter, to 255, first exports Pre_string, then export brigade commander's Counter Value, make brigade commander's Counter Value be 1,

The unpressed character string reading file to be compressed is to pre_string, return to step three

If brigade commander's Counter Value does not count 255, return to step three；

If pre_string is inconsistent with current_string, export pre_string, export brigade commander's Counter Value, by this When current_string be assigned to pre_string, return to step three；

Step 4：Set current character current_char, according to step 3 (1) output order, output is arranged in order into new Character stream, the first character reading in this character stream is assigned to current_char；

Step 5：Judge that pre_char and current_char is arranged in order the entry of composition whether in dictionary：If in word In allusion quotation, the entry that pre_char and current_char is arranged in order composition is assigned to pre_char；If not in dictionary, Export the code word of current prefix pre_char, add pre_char and current_char and be arranged in order the entry of composition to dictionary In, the value of current_char is assigned to pre_char；

Step 6：Judge in new character stream, whether also have character to need to compress, need compression if there are character it would be desirable to compress First character be assigned to current_char, return to step five；Need to compress without character, export current prefix pre_ The code word of char, completes to compress.

2. a kind of optimization bit file compression method based on RLE and LZW according to claim 1 it is characterised in that： The compression dictionary of LZW is used for storing the entry producing in compression process.

3. a kind of optimization bit file compression method based on RLE and LZW according to claim 1 it is characterised in that：Trip Long counter, starts counting up from 1, that is, count 1 when reading first character string, can be when running into the character string continuously repeating appearance Jia 1 every time, when different character strings occur, the value of brigade commander's enumerator can be exported, reduce initial value 1；Brigade commander's enumerator It is worth for 1-255, reach and when 255, reduce 1.

4. a kind of optimization bit file compression method based on RLE and LZW according to claim 1 it is characterised in that：Just The index value of the compression dictionary of beginningization LZW is 1-4096.

5. a kind of optimization bit file decompression method based on RLE and LZW it is characterised in that：As follows including step：

Step one：Initial decompression dictionary, makes dictionary comprise all possible, and described is single entry；

Step 2：Setting variable pw, reads first code stream in data flow to be decompressed, is assigned to pw, with pw as search index Decompression dictionary, reads the corresponding entry of this index, setup string variable pre_char, the corresponding word of index that pw is represented Bar is assigned to pre_char, and exports pre_char；

Step 3：Setting variable cw, continues to read the next code stream in data to be decompressed, is assigned to cw, judges in dictionary In whether have the corresponding entry of the index that cw represents, the corresponding entry of index representing if there are cw, the output index that represents of cw Corresponding entry, the corresponding string data of index that is, cw represents, variable current_char, the index that cw is represented are set Corresponding string data is assigned to current_char, arranges character variable pc, and pre_char is assigned to pc, arranges character Variable cc, the first character of corresponding for current_char entry is assigned to cc, pc and cc is arranged in order the character of composition String is added in decompression dictionary；The corresponding entry of index representing without cw, the value of pre_char is assigned to pc, will The first character of current_char is assigned to cc, and output pc and cc is arranged in order the string data of composition, and add pc and Cc is arranged in order the string data of composition in dictionary；

Step 4：Judge in data flow to be decompressed, whether also have data to need to decompress, if also data needs to decompress, Return to step three；Need to decompress without data, the output of step 3 is arranged in new character string number by output order According to the head control word in the new string data of identification；Setup string head, and head control word is assigned to head；

Step 5：Setup string variable current_string, from the new string data removing head control word by Read character string according to brigade commander and be assigned to current_string, then read a character after this character string again, i.e. brigade commander's meter The value of number device, restores original compression data according to the value of the character string reading out and brigade commander's enumerator；

Step 6：Judge the data whether needing in the new string data remove head control word to be decompressed, if having, Data to be decompressed is replaced the new string data removing head control word, return to step five；If no, by head with The data of step 5 output is arranged in order and obtains initial data, completes to decompress.