False contour reduction using directional dilation and edge-preserving filtering

Abstract

As display devices have become larger, false contour artifacts invisible in small displays have appeared visible. False contours are artifacts that appear as step edges in relatively smooth regions of an image, mostly caused by image quantization. In this paper, a new false contour reduction algorithm of still images is proposed using directional dilation and edge-preserving filtering. The proposed algorithm consists of three steps: false contour detection, false contour map expansion, and edge-preserving false contour reduction. In the first step, pixels to be processed are detected and in the second step the processing area is expanded using the proposed directional dilation. Finally, only the pixels selected in the first two steps are filtered, in which edge-preserving filtering using multi-dimensional weight functions is employed. The performance of the proposed algorithm is compared with those of two existing methods. Experimental results in various aspects show that the proposed algorithm efficiently reduces false contours