Adaptive filtering of block effects based on edge map

Block coding, such as DCT coding and VQ, play an important role in image data compression. However, block coding usually produces block effects in the reconstructed images. Block effects result in the artificial contours and so degrade the quality of the decoded images. To improve the visual quality of decoded images, postprocessing is often used. We present an adaptive space-variant filter based edge map for reducing block effects. First we apply an edge detector on the decoded image to obtain an edge map. Then, an environment of each pixel (x,y) in the decoded image f(x,y) is analyzed based on the edge map. Based on the analysis, all pixels in f(x,y) are classified into two classes: edge pixels and non-edge pixels. Finally, a 1-D directional filter and a 2-D mean filter are applied on those pixels accordingly. Simulation results show that the algorithm improved the reconstructed images both in the subject visual quality and signal-to-noise ratio