Adjustment of Wavelet Filters for Image Compression Using Artificial Intelligence

Abstract

A method for lossless image compression using wavelet lifting transform with automatic adjustment of wavelet filter coefficients for better compression is presented. The proposal is based on pattern recognition by 1-NN classifier. Using the pattern recognition, the lifting filter coefficients are optimized globally for each image. The proposed technique was aplied to test images and the compression results were compared to the results produced by the standard CDF (2,2) and CDF (4,4) wavelet filters. The results obtained with the optimized wavelet filters are better in terms of the achieved entropy for all images, as well as for the overall performance.