Wavelet Image Compression Based on Potential Fuzzy Clustering Quantization
YANG Xu-dong, WANG Wan-liang (Information Engineering Institute, Zhejiang University of Technology, Hangzhou 310014)
The wavelet-based image compression is a successful technology, and plays more and more important role in the image compression fields. But the edge fuzzy phenomenon, which occurs in the wavelet-based image compression algorithms under low bit rates, remains an open question. In order to reduce the edge fuzzy phenomenon under low bit rates to some extent, a new method of wavelet image compression based on potential fuzzy clustering quantization has been presented in this paper. The potential fuzzy clustering method is applied to quantize the detail sub band images' wavelet coefficients after the image has been decomposed by the wavelet transform. This method has two advantages. One is it considers the statistical characteristics of each detail sub band images' wavelet coefficients and the importance of detail sub band images' wavelet coefficients for saving the edge and texture information of the original image. The second advantage is it makes use of the characteristics of fuzzy set. The experimental results show that this method can get satisfying results, the edge and texture information can be saved well under low bit rates, the edge fuzzy phenomenon is reduced to some extent. And the subjective quality of the reconstructed image is improved. This paper has made some tries on fuzzy clustering quantization method in the wavelet-based image compression fields.