针对高清彩色图像的最佳邻域匹配算法并行化研究

Research on a Parallel Best Neighborhood Matching Algorithm for High Definition Image Restoration

最佳邻域匹配算法是一种优秀的差错掩盖算法,能得到很高的图像恢复质量。但是,该算法计算量大,已很难满足目前图像处理的要求。本文主要针对高清彩色图像,将BNM算法推广到彩色图像,并将该算法并行化。实验表明,在一个4结点的机群系统上,破坏率为15%的条件下,该并行算法的加速比达到7.52,大大提高了原串行BNM算法的效率,并且图像恢复质量没有下降。

The Best Neighborhood Matching （BNM） algorithm is an error concealment algorithm to achieve high quality image restoration. Due to the high computation cost,BNM can not meet the realtime processing requirement at this time. In this paper,we extend the original BNM algorithm which only considers the grey scale images to color images,and use a parallel BNM algorithm for high definition images. Our experiment is done in a fournode Linux cluster system. The results show that the speedup of our parallel algorithm is 7.52 when the damage rate of the image is 15%. It improves the efficiency of BNM,and does not reduce the restoration quality.