Image Enhancement for Low Bit-Rate JPEG and MPEG Coding via Postprocessing

JPEG and MPEG compression standards are based on block discrete cosine transform (BDCT) and, when the bit rate becomes low, visually annoying blocking artifacts appear in decompressed images and videos. In this work, we attempt to characterize and quantify the blocking artifact, and then propose an iterative method for its removal by using block classification and space-frequency filtering. The proposed new method is better than the POCS (projection onto convex sets) method in terms of convergence rate, visual appearance and objective mean square error measure.