Image Compression Based on Wavelet Transform with Fuzzy Inference

A quantization method based on fuzzy inference filter is proposed. First, the original image is decomposed to obtain wavelet coefficients. Second, a fuzzy inference filter is used to non-uniformly determine whether a wavelet tree contains higher entropy energy. A general if-then-else structure is utilized to model the fuzzy rule, and an energy tree is then obtained by the entropy energy which is concern of the basic fuzzy rules affected. Finally, we use energy tree for both VQ and SQ quantization. The proposed approach occupies a lot of characteristics, in terms of fuzzy justification, energy tree, scalar quantization, progressive transmission, and vector quantization; thus, an optimal quantization is obtained for lossy data compression. Experiments show that both VQ and SQ can achieve high compression ratio and lower distortion.