应用像素运动模糊特征分割的空间移变降质复原  

Space-variant blurred image restoration based on pixel motion-blur character segmentation

作　　者：张玉叶[1,2] 周晓东[2] 王春歆[2] 

机构地区：[1]海军航空工程学院青岛分院,山东青岛266041 [2]海军航空工程学院控制工程系,山东烟台264001

出　　处：《光学精密工程》2009年第5期1119-1126,共8页Optics and Precision Engineering

基　　金：国家863高技术研究发展计划资助项目(No.2006AA0619)

摘　　要：为了复原由动机座上的相机拍摄运动目标得到的降质图像,提出了基于像素运动模糊特征分割图像的复原方法。根据运动模糊的卷积作用,分析了图像中目标与背景不同的运动特性及像素平移叠加机理,得出了目标背景的叠加区域中相邻像素灰度值的变化程度相似的结论。基于该灰度变化特征检测运动模糊方向上的目标边缘,并利用Canny算子检测平行于模糊方向的边缘,两部分边缘进行求或运算并进行形态学处理,从而分割得到完整的目标模糊图像。剔除该目标图像叠加区的背景灰度信息,并补零扩充为完全卷积的模糊图像,最后利用反卷积复原算法去除模糊。复原实验结果表明,该方法复原得到的目标图像内部纹理清晰,复原图像的信噪比为9.64 dB,即主观和客观评价都表明该方法能够有效处理该类空间移变降质图像的复原问题。An image restoration method based on pixel motion-blur character image segmentation is presented to restore a kind of space-variant blurred image taken by a moving camera. According to the convolution of motion blur, the motion characteristics of a target and a background and the shift superposition theory are analyzed. It is concluded that the gray changes of neighboring pixels within a superposition region are similar. According to the feature, the blurred edge of the target in the superposition region can be detected, and the target edge parallel to the motion blur direction is detected by a Canny operator. Then the background information of the superposition region is subtracted from the target image,and the target image is restored by adding zeros and a deconvolution algorithm. The experimental results show that the restored target image has clear textures and the improved Signal-to- Noise Ratio （SNR） is 9.64 dB, which shows proposed approach can effectively realize the restoration for this kind of space-variant motion blurred image.

关 键 词：图像复原 空间移变降质 图像分割 运动模糊 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]