A Compensated Fuzzy c-Means Applied on Image Compression

Abstract

Due to the internet being widespread, the growing rate of data in computer network is increasing each year. It forces scientists to develop efficient algorithms to compress huge data using least code to represent source data in order to avoid overflowing storage capacity, consuming transmission time, and expending broadband. In this paper, a Compensated Fuzzy c-Means (CFCM)for vector quantization in image compression is proposed. Vector quantization is one of techniques in the lossy data compression. The purpose of vector quantization is to create a codebook for which the average distortion generated by approximating a training vector and a codeword in codebook is minimized. Such a method results in massive reduction of the image information in image transmission. The CFCM, modified from penalized fuzzy c-means algorithm (PFCM), is to speed up the convergence rate for clustering procedure. From the experiment results, the CFCM algorithm shows promising clustering results in comparison with FCM and PFCM algorithms. And in the image compression application, the proposed CFCM algorithm to design codebook for vector quantization produces better PSNR than the generalized Lloyd algorithm about 2dBs.