倒谱和快速全变差去卷积的运动模糊图像复原  

Motion blur image restoration used by cepstrum and fast total variation de-convolution

作　　者：章忠宪[1] 黄德根[2] 顾亚丽[1] 

机构地区：[1]漳州职业技术学院计算机工程系,福建漳州363000 [2]大连理工大学软件研究所,辽宁大连116000

出　　处：《计算机工程与应用》2015年第8期174-177,199,共5页Computer Engineering and Applications

基　　金：福建省教育厅A类科技项目(No.JA1310)

摘　　要：物体和成像系统之间的相对运动导致图像产生运动模糊,降低了图像质量,为了获得更加理想的复原图像,提出一种基于倒谱和快速全变差去卷积相结合的运动模糊图像复原算法。利用倒谱法对运动模糊图像的点扩散函数参数(模糊角度和模糊尺度)进行辩识,采用快速全变差去卷积法对模糊图像进行复原,采用多幅图像进行仿真实验测试算法的性能。仿真结果表明,相对于经典图像复原算法,该算法复原图像的主观视觉效果以及客观评价指标均更优,具有一定的实际利用价值。The motion blur generated by the relative motion during the object and the imaging system is one of the most common reasons for image degradation, in order to obtain good restoration results, a novel motion blur image restoration algorithm is proposed based on cepstrum and fast total variation de-convolution in this paper. The motion blur parameters （blur orientation and length）of point spread function are estimated based on the cepstrum method, and then the motion blur image is restored by fast total variation de-convolution method, the simulation experiments are carried out to test the performance of algorithm on many images. The results show that, compared with classical image restoration algorithms, the proposed algorithm can obtain better restoration affects, and has a certain practical reference value.

关 键 词：运动模糊 图像复原 点扩散函数 倒谱法 全变差算法 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]