KR20210042439A - Binary data compression method and appratus thereof - Google Patents

Abstract

According to the present invention, the bit gain of 1 bit from the compressed binary cluster can be utilized as it is by making [compressed binary cluster] + [data that starts with 1 (through proper data modulation or data as it is) and knows how many bits come after it through header bit analysis] + [other data bits to make a fixed bit width].

Description

Binary data compression method and apparatus thereof {BINARY DATA COMPRESSION METHOD AND APPRATUS THEREOF}

Detailed information for carrying out the invention

Detailed information for carrying out the invention

Detailed information for carrying out the invention

Detailed information for carrying out the invention

Detailed information for carrying out the invention

For example, when there is arbitrary binary data,

001110111000001111100111000101111....

Whenever it meets 01, the data is separated and it is called a binary cluster.

001/1101/11000001/1111001/110001/01/111

Let's say that the data that did not meet 01 at the end is the surplus data,

If you remove the lowest "1" from this binary cluster,

Binary data that must end in 0 is compressed by 1-bit, and this is called a compressed binary cluster.

00/110/1100000/111100/11000/0 // 111 (surplus data)

Considering the compressed binary cluster separately,

If this compressed binary cluster is lined up vertically, it is as follows.

00

110

1100000

111100

11000

0

In this way, the compressed binary cluster is set vertically, and in each row, starting with 1, through header bits, etc., a binary data string that can know how many bits are present is combined. For example, a binary data string for UTF-8 characters,

Starting with 0, 0xxxxxxx is a total of 8 bits,

Starting with 110, 16 bits as 110xxxxx10xxxxxx

Starting with 1110, 24 bits as 1110xxxx10xxxxxx10xxxxxx,

Starting with 11110, 11110xxx10xxxxxx10xxxxxx10xxxxxx as 32 bits

By analyzing the header bit sequence, you can see how many bytes are the same data chunk.

In this case, if 1 is added to the front of the string starting with 0, and made into 9 bits with 10xxxxxxx, all cases can be distinguished and combined with the compressed binary cluster as follows, starting with 1.

00 <UTF-8 bit string>

110<UTF-8 bit string>

1100000<UTF-8 bit string>

111100<UTF-8 bit string>

11000<UTF-8 bit string>

0 <UTF-8 bit string>

Now, if the other data for making the same number of bits in each row are sequentially fetched and connected as much as the corresponding bit, bit gains of 0 to 1 bit in each solution unit occur while each row has the same bit width. When it becomes a fixed bit string, it can be combined and transmitted in one row, and each can be decompressed by separating it into fixed bit columns at the receiving side.

00 <UTF-8 bit string> <....other data bits>

110<UTF-8 bit string><....other data bits>

1100000<UTF-8 bit string> <other data bits>

111100<UTF-8 bit string> <.other data bits>

11000<UTF-8 bit string> <..other data bits>

0 <UTF-8 bit string> <.....other data bits>

The present invention is thus,

[Compressed Binary Cluster] + [(through appropriate data modulation or data as it is) data that starts with 1 and knows how many bits are coming after header bit analysis] + [other data bits to make a fixed bit width] It is an invention that can utilize the bit profit of 1 bit from the binary cluster as it is.

In particular, in the middle [data that starts with 1 (through appropriate data modulation or data as it is) and lets you know how many bits are coming after header bit analysis], existing compression algorithm data or any data with the same characteristics can be used.

As another example for intermediate data, to take deflate, deflate is a very popular compression algorithm and is an international standard.

Internally, it has a multi-block structure, and it can be classified for each block. If a simple modulation process is performed by referring to RFC-1951, which defines deflate, a header bit is expressed in the first byte of each block and end-of -Block can be transformed into a form at the end. Data is packed, and the least significant bit of the first byte indicates whether it is the last block (=1) or the middle block (=0), and the subsequent 2 bits (10, 11, 01, 00 indicate the compression method and reserved bits)

After some modulation of the data, the least significant bit indicating whether or not the last block is moved to the highest level, and most of the intermediate blocks are set to 1 because this is 0, and the last block is set to 1 as it is and modulated as follows. Of course, it is also possible to modulate in the form of other bits following together or in reverse order. Importantly, the first bit of the middle block header is set from 0 to 1, and the header bit indicating the end block is 1 as it is. It comes to the forefront. The method of classifying the last block is to include the total number of blocks in the compressed data, or look at the amount of data remaining in the decoding process, and if it remains as long as the bit length of a fixed bit string, it can be naturally known that it is the last block.

Next, the binary cluster is made into a compressed binary cluster according to the rules.

Each of these compressed binary clusters is combined with the data block of the deflate. And, as in the above, other data bits for making a fixed bit string are sequentially fetched and combined as needed. The combined data is combined into one line of data in a row and transmitted, and the receiving side separates and decompresses as much as a fixed bit string.

As shown in the figure below, the first time it encounters 0 and then 1 again is the starting point of the header bit, and the previous is the compressed binary cluster area. Separating 8 bits from this header bit, leaving the most significant bit as 1 ==> 0, and moving to the least significant bit again, the header bit string of the original deflate starts, and if the decoding rule of defalte is applied, the end-of-block naturally goes up. The original is restored when it is divided into compressed data blocks and decompressed, and the original is restored by sequentially combining other data bits. The compressed binary cluster is also restored to the original binary cluster by combining 1 after the least significant bit.

As for the method of classifying the last block, as described above, if the total number of blocks is included in the compressed data, or if the amount of data remaining in the decoding process is viewed and remains as long as the bit length of the fixed bit string, it can be naturally known that it is the last block.

Claims (1)

As a pre-application for claiming domestic priority, no separate claim scope has been stated.