Postprocessing Algorithm for Quantization Noise Reduction Using Block Classification and Adaptive Filtering

摘要

In this paper, we proposed a postprocessing algorithm for the quantization noise reduction in the block coded images using the block classification and the adaptive filterings. The proposed algorithm consists of the block classification, the adaptive inter-block filtering, and the intra-block filtering. First, each block is classified into one of seven classes based on the characteristics of 8 x 8 DCT coefficients. And then, according to the information of various patterns and frequency distributions, in which are given by the block classification, the adaptive inter-block filtering is performed at the horizontal and vertical block boundary to reduce the blocking artifacts. Finally, within blocks which are classified into complex class, the intra-block filtering is performed to reduce the ringing noise without blurring edge. Experimental results show that the proposed algorithm gives better results than the conventional algorithms from both a subjective and an objective viewpoint.