A Vector Quantization Method Based on Normal Distribution

Abstract

The invention discloses a vector quantization method based on the normal distribution law. In the codebook generation, the feature center of a training sample set is first used as an initial codeword, and then the initial codeword is divided. When the codeword is not divided for the first time, the width of each codeword and the number of training samples belonging to each codeword are multiplied. The L codewords with the largest product are selected as the codewords to be divided, and a new codeword mean value is obtained to realize the codeword division. Every time after the codeword division, a data partitioning stage starts. In the data partitioning, the codewords are described as normal distribution, the degree of membership of each training sample to each codeword is calculated, and the cell is divided to update the codeword. Whether the quantitative distortion is convergent is judged, and if not, the data partitioning continues. If the quantitative distortion is convergent, whether the total number of the codewords has reached a certain value is judged, if not, the codeword division continues, and if so, a final codebook is output. The method of the invention can improve the accuracy of codeword division and reduce the error of vector quantization.

Description

technical field [0001] The invention relates to a vector quantization method, in particular to a vector quantization method based on normal distribution law. Background technique [0002] With the rapid development of information and communication fields, a large amount of audio, image and other multimedia information needs to be stored, processed and transmitted, requiring a large storage space and channel bandwidth. In order to improve storage efficiency and reduce storage space, redundant information in media information should be eliminated as much as possible under allowable distortion conditions. Quantization is a common technique for data compression, and there are two quantization methods: scalar quantization and vector quantization. Scalar quantization refers to quantizing each sample value of the signal waveform or each parameter value of the signal independently. Vector quantization refers to dividing the sampling value of the signal waveform or the parameter va...

AI Technical Summary

Problems solved by technology

The LBG algorithm uses the current codebook to divide the training vector set into disjoint clusters, and then finds the centroids of these clusters to obtain new codewords. Through the above iterative process, the total average distortion will be gradually improved; however, the codeword The splitting process ignores the distribution characteristics of the sample and is a form of random splitting

In the case of random splitting, because each iteration of vector quantization requires a K-means clustering, the iteration process ends in a local optimal state. The result of this algorithm is closely related to the selection of the initial center. When the initial center is not selected , is easily affected by class edge points and noise points, so the rationality of codeword splitting is relatively low

Images