rice算法原理阐述以及例子详解.docVIP

The basic idea behind Rice coding is to store as many words as possible with less bits than in the original representation (just as with Huffman coding). In fact, one can think of the Rice code as a fixed Huffman code (i.e. the codes are not determined by the actual statistical content of * the data, but by the assumption that lower values are more common than higher values). * * The coding is very simple: Encode the value X with X 1 bits followed by a 0 bit. However, there are some optimizations in this implementation: 1) The k least significant bits of each word are stored as is, and the N-k most significant bits are encoded with Rice coding. k is chosen as the average number of bits for the previous few samples in the stream. This usually makes the best use of the Rice coding, hides noise from * the Rice coder, and does not result in very long Rice codes for signals with varying dynamic range. 2) If the rice code becomes longer than a fixed threshold, T, an alternate coding is used: output T 1 bits, followed by floor(log2(X-T)) 1 bits, and one 0 bit, followed by X-T (represented by the least significant floor(log2(X-T))-1 bits). This gives pretty efficient coding even for large values, and prevents ridiculously long Rice codes (in the worst case scenario, a single Rice code for a 32-bit word may become as long as 2^32 bits, or 512 MB). X bin Rice Thresholded Rice Difference * ------------------------------------------------------------------ * 0 00000 0 0 * 1 00001 10 10 * 2 00010 110 110 * 3 00011 1110 1110 * 4 00100 11110 11110 * 5 00101 111110 111110 * 6 00110 1111110 +1 * 7 00111 * 8 01000 111111110 1111111000 +1 * 9 01001 1111111110 1111111001 * 10 01010 11111111110