Edge directed filter based error concealment for wavelet-based images

Abstract

Edges in a natural image have important effects on the subjective visual quality. During the transmissions of wavelet-compressed images such as JPEG2000, errors in high frequency subbands will result in the effects like ring or ripple artifacts around edges. In this paper, we propose an error concealment algorithm to remove these annoying artifacts. This algorithm requires an edge directed filter. Although some known filters can be employed, we tailor-make a new edge directed filter which fits well in our algorithm. The proposed scheme firstly enhances the received damaged image using the edge directed filter. Then the recovered wavelet coefficients are rectified using two constraint functions, which are based on the statistical characteristics in the wavelet domain and the observation that correctly received data must remain unchanged. Simulation results show that the image quality has been significantly improved in terms of both objective and subjective evaluation.