CORRECTION OF QUANTIZATION ERRORS IN COMPRESSED COLOR IMAGES

摘要

A method for the reduction of quantization errors in compressed multichannel RGB images is presented. The proposed method adopts fuzzy relations among neighboring pixels in order to reconstruct the data corrupted during the coding process. Results of computer simulations show that the method is simple and effective. Unlike other techniques in the literature, no complicated tuning of the parameters is necessary. Blocking and ringing artifacts are effectively reduced and the quality of the image is satisfactorily preserved.