摘要
针对一类运动模糊置换混叠图像盲分离问题,提出一种基于NSCT域的二次模糊相关性的盲分离算法.首先,对置换混叠图像进行运动模糊,把原图像与相应的运动模糊图像分别在NSCT域上进行稀疏分解,得到各自的低频子带系数和各带通方向子带系数;其次,利用系数的冗余特性求其和图像;然后,对和图像分块,估计各对应子块的相关系数,对其采用阈值化操作分离出置换图像.实验结果表明,采用本算法在置换图像的运动模糊方向、模糊度、大小、位置和个数未知情况下对运动模糊置换混叠图像实现了有效的分离效果,并且对置换图像的正确分离率高于二次模糊相关性置换混叠图像盲分离和小波变换二次模糊相关性盲分离方法.
Focused on the issue that motion blurred permuted alias image blind separation, an algorithm using double blur correlation blind separation based on NSCT domain was proposed. Firstly, the permuted alias image was blurred again, low frequency subband coefficients and varieties of directional bandpass subband coefficients of the original image and the blurred version were obtained by spare decomposition based on NSCT domain. Secondly, summation image was gotten by coefficient redundancy. Then, the summation image was blocked, the correlation coefficients were estimated by each corresponding sub-block, the permuting image could be separa- ted by using threshold method. Experimental results show that the proposed algorithm can separate the permuting image effectively from the permuted alias image in spite of the motion blurred direction, blur degree, size, location and number of permuting image. And correct separation rate of the permuting image is higher than the Blind Separation of Permuted Image Based on Double Blur Correlation and Blind Separation of Permuted Alias Image using double blur correlation by wavelet transform.
出处
《小型微型计算机系统》
CSCD
北大核心
2016年第5期1057-1061,共5页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(U1204606)资助
河南省科技攻关项目(142102210565)资助
关键词
盲分离
置换混叠图像
NSCT
运动模糊
二次模糊相关性
blind separation
permuted alias image
nonsubsampled contourlet transform （ NSCT ）
motion blurred
double blur correlation
