Two-Stage False Contour Detection Using Directional Contrast Features And Its Application To Adaptive False Contour Reduction

In the era of digital television, artifacts such as false contours unnoticed in small television sets appear conspicuously as the size of display devices increases. This paper proposes a two-stage false contour detection algorithm and its application to false contour reduction. In stage 1 of false contour detection, smooth regions are first removed by bit-depth reduction or re-quantization. In stage 2, false contours are separated from edges or texture regions using directional cantrast features, yielding the binary false contour decision map that show's possible candidate regions of false contours. Also a number of non-directional and directional features for possible false contour detection are considered and their performance is compared. For false contour reduction, variable-size directional smoothing is applied only to candidate regions that are specified by the false contour decision map. We use one-dimensional variable-size directional smoothing masks whose directions are orthogonal to the directions determined by the directional contrast,features found in the false contour detection step. Computer simulations with several test images with various types of false contours show the effectiveness of the proposed false Contour reduction algorithm in terms of visual quality and the peak signal-to-noise ratio. The proposed algorithm can also be used as a post-processing method for artifact reduction in various display devices.