针对高清彩色图像的最佳邻域匹配算法并行化研究

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。

Research on a Parallel Best Neighborhood MatchingAlgorithm for High Definition Image Restoration

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.