摘要
本文通过深入研究匀速直线运动模糊图像的点扩展函数,提出一种识别运动模糊图像模糊方向的频域方法。首先对模糊图像的频谱图出现平行暗条纹的原因进行分析。从理论证明和实验说明指出暗条纹的方向与模糊方向垂直的特性,并采用RADON变换方法检测这些暗条纹。最后,给出了运动模糊方向精确的数值估算方法。实验结果表明,该算法可以准确地识别图像运动模糊方向,绝对误差不超过2°。
Through exploration of the point spread function of the blur of uniform linear motion, a novel method based on frequency-domain algorithm for identification of motion-blur direction was presented. Firstly, the reason for the occurrence of the black strips in the spectrum image of motion-blur image was analyzed. Then the characteristic that the black strips are perpendicular to the motion-blur direction was proved in theory and experiment, and the black strips were detected by RADON transformation. Finally the estimation algorithm was proposed for the exact motion-blur direction. Experimental results show that the proposed method can accurately identify the motion-blur direction and the absolute errors does not exceed 2 degree.
出处
《光电工程》
EI
CAS
CSCD
北大核心
2007年第10期98-101,共4页
Opto-Electronic Engineering
基金
国家973项目(2003CB716207)
关键词
图像复原
频谱图
RADON变换
图像识别
image restoration
spectrum image
RADON transformation
image recognition
