Fast image restoration using the multi-layer best neighborhood matching approach
Ching-Tang Hsieh, Yen‐Liang Chen, Chih-Hsu Hsu · 2007
Abstract:- This paper presents an accurate and fast image restoration for lost block images via wireless image transmission. In the network transmission of block-coded images, fading in wireless channels and congestion in packet switched networks could cause entire blocks to be lost. Instead of using retransmission query protocols, we reconstruct the lost block on the wavelet domain that considers the relation between the lost block and its neighbor coefficients. Thus, this method could use the transmitted image immediately and increase the transmission rate of the network bandwidth. The algorithm considers both the best neighborhood matching (BNM) and the different frequency contents of the multi-layer on wavelet domain, which is referred as the Multi-Layer Best Neighborhood Matching (MLBNM). The MLBNM is proposed to improve the efficiency of search processing and to reconstruct the different frequency composition separately. This method considers the different veins of the directional composition and the independent directional estimation, so that the veins will not influence each other. The experimental results demonstrate the high restoration quality of the proposed system. Key-Words: Wavelet, Image Restoration, Wireless, BNM and MLBNM. 1