Concealment of damaged block coded images using intelligent two-step best neighborhood matching algorithm

Imperfect transmission of block-coded images often results in lost blocks. Best neighborhood matching (BNM) is a very good error concealment algorithm to achieve high quality restored image. However, the computation complexity of BNM is high due to the full search method. A fast error concealment method named intelligent two-step BNM (ITSBNM) is proposed, where an intelligent search method is developed for most time consuming stage. We alternatively use browse search and fine search in block matching process. The fine search used herein employs strategy of diamond search to find matching blocks. Simulation experiments have been performed on blocks lost images, and simulation results show that ITSBNM can achieve a good tradeoff between the computation and the visual quality of image restoration, meanwhile it is about only 8-10% of BNM in term of processing speed. Lots of simulation results demonstrate the feasibility of the proposed algorithm.