几种压缩算法（Several compression algorithms）.docVIP

几种压缩算法（Several compression algorithms） What is entropy? Data compression is not only the origin of the first in 40s by Claude Shannon of the information theory, and the basic principle of the information can be compressed to exactly how small, still follow a theorem in information theory, this theorem borrowed noun thermodynamic entropy (Entropy) to indicate the amount of information in a real information the need for encoding: Consider the binary code consisting of 0 and 1 for a message containing N encoding symbols, symbols of Fn probability hypothesis is repeated in the whole message in Pn, the symbol entropy is that bits required for the symbol: En = - log2 (Pn) The entropy of the whole message, that is, the number of bits required to represent the whole information is: E = Sigma En For example, for the following string that only appears a, B, C, three characters: Aabbaccbaa The string length is 10, and the characters a, B, and C appear 532 times, and the probability of a, B, C appearing in the information is 0.5, 0.3, 0.2, respectively. Their entropy is respectively: Ea = -log2 (0.5) = 1 Eb = -log2 (0.3) = 1.737 Ec = -log2 (0.2) = 2.322 The entropy of the entire message is the number of bits required to represent the entire string: E = Ea * 5 + Eb * 3 + Ec * 2 = 14.855 bit Recall that if we use the ASCII code commonly used in computers to represent the strings above, we need 80 full bits! Now I know why information can be compressed without losing the original information content. Simply speaking, using fewer bits to represent more frequent symbols is the basic principle of data compression. The attentive reader will soon think how we should use 01 binary digits to represent zeros Its really difficult, but its not impossible. Once we have found a method for accurately representing zeros, we have the right to challenge the limits of lossless compression. Dont worry, youll see it in the fourth chapter. Model From the above description, we understand that to compr