利用同场景多幅图像进行图像降噪

摘要

提出一种利用同场景多幅图像作为输入来降低数码照片在高感光度下数码照片噪点的方法.该方法本质上基于按区块匹配加权平均值的方法.该方法在匹配区块的搜索过程中利用了计算机视觉中的追踪、变换等技术来减小搜索范围,从而大幅度降低了整个处理过程的时间复杂度,并且增加了匹配的准确度,满足了高分辨率图像在处理速度上的需求.此外,该方法中的权值设计能够有效避免传统按区块匹配方法带来的残影问题.%A method to reduce digital photo noises under high light sensitivity conditions by inputting multiple images of a same scene is proposed. The proposed method is basically based on block matching weighted average method. During the matching block searching procedure the method uses such computer vision technologies as tracking and transformation to reduce the searching area, so that it greatly reduces the time complexity of the entire processing while increasing its matching accuracy, thus satisfies the speed requirement for high resolution image processing. Meanwhile, the weight design of the method can effectively avoid the Ghost Shadow problem caused by conventional block matching methods.