运动模糊不变量研究  

Study of motion blur invariant

作　　者：骞森[1] 朱剑英[1] 

机构地区：[1]南京航空航天大学机械电子工程研究所,江苏南京210016

出　　处：《光电工程》2007年第11期88-92,共5页Opto-Electronic Engineering

基　　金：国家自然科学基金资助项目(50275078)

摘　　要：图像的运动模糊是摄像机在曝光期间与场景间相对运动而造成的图像退化,但现有算法对运动模糊的恢复精度有限。本文分析了图像运动模糊的退化模型,提出了运动模糊不变量的概念,证明了图像零阶矩、第一个Hu不变矩和SIFT为运动模糊不变量,并给出了测试和应用。结果表明运动模糊不变量是存在的,而且可以用于不确定性运动模糊图像的匹配和识别领域。Image motion blur is image degradation caused by relative motion between camera and scene during camera exposing time. But the restoration precision of motion blur is unsatisfactory by using existing algorithms. The degradation model of image motion blur was analyzed and motion blur invariant was presented in the paper. The zero-order moment, the first Hu moment and scale invariant feature transform （SIFT） were proved to be motion blur invariants, and the test methods and applications were also given. The results show that motion blur invariants exist, and can be applied to matching and recognizing uncertain motion blur images.

关 键 词：运动模糊 不变量 HU矩 SIFT 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]