基于Walsh-Hadamard投影的快速Nonlocal-Means图像去噪

摘要

A fast Nonlocal-means image denoising method with projection is proposed. After analyzing the bottleneck of Noncal-means, the fast and auto-recursive computation of Walsh-Hadamard transform is used to project the image blocks into Walsh-Hadamard space. Because of efficient energy packing of Walsh-Hadamard projection, the algorithm can fast discard those impossible blocks by using a rejection scheme in the matching procedure of image blocks in the Nonlocal-means image denoising method. Experimental results show that the proposed method can accelerate the Nonlocal-means algorithm without image quality degradation.%针对Nonlocal-means图像去噪算法计算强度较高的问题,提出了一种投影加速算法.分析了Nonlocal-means算法的计算瓶颈,利用Walsh-Hadamard变换的快速运算和自递归特性,使用一组完备的Walsh-Hadamard 变换将图像块投影到其张成的空间中;利用Walsh-Hadamard投影的能量集中特性和匹配计算过程中的拒绝策略,在Nonlocal-mean,去噪算法的图像块匹配计算中快速丢弃无法匹配的图像块.实验结果表明,该算法获得了较好的加速性能,且去噪效果没有受到影响.