Abstract: Best Neighborhood Matching (BNM) is a very good error concealment algorithm to achieve high quality image restoration. However, BNM needs intensive computation, which restricts its real application. A fast image restoration algorithm: Jump and Look all around BNM (JLBNM) is proposed, where an intelligent search method is developed for the most time consuming stage. Meanwhile, an adaptive method is employed for optimization. Theoretical analysis has been carried out for the proposed algorithm. Experiment results show that our JLBNM can efficiently reduce computation time for image restoration, while the restored image with high quality can be obtained.