JPS5921235B2 - encoding device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an encoding device that encodes output symbols of an information source in which the probabilities of appearance of binary symbols are close to each other. On the other hand, as a practical code for effectively encoding this, the one shown in FIG. 1 is known.

However, when considering the case where the target information source is a memoryless information source, the probability of appearance of a dominant symbol with a high occurrence frequency is approximately 0.62.
The problem with the following is that compression cannot be performed. The present invention has been made in view of this point.

When the probability of occurrence of n+1 dominant symbols is P, P>(
1-P) select n that satisfies n, and divide the output symbols from the information source into a pattern including a total of (n+1) dominant symbols and a pattern including a total of n inferior symbols;
When the delimited pattern consists only of dominant symbols, an n-bit code word is added to it, when it consists only of inferior symbols, an (n+1) bit code word is added to it, and in other cases, an n-bit code word is added to it. By providing a code word with the same number of bits as the number of binary symbols, we have developed an encoding device that can always perform encoding compression, no matter how close the probability of appearance of a dominant symbol is to 0.5, as long as the appearance process follows a memoryless process. is intended to provide.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 5 is a schematic configuration diagram of an encoding apparatus according to an embodiment of the present invention, in which 100 is an "o" symbol counter that counts the dominant symbols "0" among the symbols to be encoded, and 101 is an "o" symbol counter that counts the dominant symbols "0" among symbols to be encoded; '', the ``o'' symbol counter 100 counts each time its counted value reaches (n+1), and the ``l'' symbol counter 101 counts each time its counted value reaches n. It outputs a signal every time the signal changes. Here, the above n is selected as follows.

That is, when the target information source outputs symbols according to the memoryless process, n is the probability of appearance of the "0" symbol as P(>1/
2), where the probability of appearance of the "1" symbol is (1-P), pnf1>(1-P)n is satisfied.
102 is an encoder, and the encoder 102 is the above-mentioned ゛o'' symbol counter 100, or 1'' symbol counter 1.
Separate the symbol sequence every time a signal is output from 01,
If the content of the delimited pattern is only the "o" symbol, convert it into a code word with n consecutive 1" bits.

If it is only the r゛ symbol, it is a code word with (n+1) consecutive bits of ``0'', but if there is a mixture of ``0'' and 1'', this is converted into a code word with the same pattern and output. be. Fig. 2 shows an example of a code format when n=2 in the present invention, and Fig. 3 A, b, and c show correspondences between a binary symbol sequence, a coded symbol sequence, and a code string. It is.

Next, the operation will be explained with reference to FIGS. 2 and 3.

Now, since n=2, the binary output symbol sequence in Figure 3a is divided by whether the symbol 60'' appears three times or the symbol ``1'' appears two times, and the symbol in Figure 3b A series is obtained. Assigning the code word in Figure 2 to this, the third
The code sequence shown in figure c is obtained. In the example code word in Figure 2, 3
For a pattern consisting only of (n+1) "0" symbols, 2(n) "F" symbols are used as codewords, and two "F" symbols are used as code words.
For patterns consisting only of 1'' symbols, three ``O'' symbols are used as code words, and for other patterns, the original symbol sequence is used as a code word.

Moreover, such a configuration makes the configuration of the encoder 102 extremely easy. On the other hand, on the receiving side, a break is placed when the total number of "0" symbols reaches 3(o+l) or the total number of "1" symbols reaches 2(n), and three "O" symbols are added. A pattern consisting of only two "F" symbols is converted into two "1" symbols, a pattern consisting of only two "F" symbols is converted into three O" symbols, and the other patterns are converted into decoded signals with their codes as they are.

Therefore, it can be easily decoded on the receiving side as well. Note that FIG. 4 shows the encoding efficiency for a memoryless information source, and the shaded area is the portion that cannot be encoded and compressed. As described above, according to the present invention, when the probability of occurrence of a dominant symbol is P, n is selected that satisfies PO+1>(1-P)1, and the output symbols from the information source are summed (n+
1) Divide into a pattern containing a total of n dominant symbols and a pattern containing a total of n recessive symbols, and when the divided pattern consists of only dominant symbols, add an n-bit code word to it and consist of only recessive symbols. Time is like this (n
+1) Since the bit code word is given as the same number of bits as the number of binary symbols in the pattern at other times, it is possible to extremely efficiently identify information sources with probabilities of appearance of binary symbols that are close to each other. It has the effect of being encoded.

[Brief explanation of drawings]

FIG. 1 is a diagram showing a conventional code format, FIG. 2 is a diagram showing an example of a code according to an embodiment of the present invention, and FIGS. 3A, b, and c are diagrams for explaining the operation of an embodiment of the present invention. FIG. 4 is a diagram showing encoding efficiency when the code according to the present invention is applied to a memoryless information source, and FIG. 5 is a block diagram of an encoding apparatus according to an embodiment of the present invention. 100...゛0'' symbol counter, 101...
・・・・・・“1゛Symbol counter, 102...
・Encoder.

Claims (1)

[Claims] 1. In an encoding device that gives a code word to an output symbol from an information source in which the appearance probability P of a frequently occurring dominant symbol and the appearance probability (1-P) of a less frequently occurring symbol are close to each other, the output symbol The sequence has a total of (n+1) dominant symbols (n:P^n^+^1>(1-P)^
(a positive integer that satisfies n) or each time a total of n inferior symbols appear, and when the divided pattern consists of only dominant symbols, an n-bit code word is given to it, and when it consists of only inferior symbols, it is given an n-bit code word. An encoding device characterized in that it gives a code word of (n+1) bits, and when it consists of both symbols, it gives a code word of the same number of bits as the pattern.