Image Compression using ANFIS in Wavelet Domain

Image compression is backbone in the field of communications and multimedia. The objective of Image compression is to minimize the size in bytes of a file by reducing the redundancy of the image data without degrading the quality of image, resulting in reduction of file size so that more images can be stored in a given amount of disk or memory space and also reduces the time required to send the images over the network [1]. In this paper we present a novel approach towards image compression based on Adaptive Neural Fuzzy inference system(ANFIS) with Differential Pulse code modulation(DPCM) in Wavelet domain so that better compression ratio can be achieved with minimum error. The overall implementation of the work consists of four steps: first, seven band wavelet. Based upon statistical properties of sub bands, different quantization and coding schemes are used. Second, first sub band is compressed using differential pulse code modulation (DPCM), the coefficient corresponding to other sub bands are compressed using Adaptive Neural Fuzzy Inference system (ANFIS). In the third stage, the result obtained is fed as input to the fuzzy vector quantizer and in fourth stage the output obtained from third stage is fed to the K means quantizer for further compression. Finally the results are compared for MSE, PSNR and the visual appearance after decoding of image.