基于Z变换的局部匀速运动模糊图像恢复算法  

Blurred Image Restoration of Local Uniform Motion Based on Z Transform

作　　者：黎明和[1] 何斌[1] 岳继光[1] 秦健铭 

机构地区：[1]同济大学电子与信息工程学院,上海201804 [2]岑溪市人才交流服务中心,广西岑溪543200

出　　处：《光学学报》2009年第5期1193-1197,共5页Acta Optica Sinica

基　　金：国家自然科学基金(50405045);国家863计划(2007AA04Z253);上海市启明星(05QMX1455)资助课题

摘　　要：针对背景不变的局部匀速运动模糊图像复原问题,提出一种基于Z变换的恢复算法。在前景和背景色差较大的假设前提下,基于背景差方法将前景从图像中分离出来;利用旋转矩阵将前景在像平面内任意方向的运动转换成X轴方向的运动;进而把复杂的局部模糊恢复问题简化为前景模糊恢复及前景和背景融合两个子问题来解决,使求解过程得到最大程度简化;在严格的数学推导基础上建立基于Z变换的退化及恢复模型,将模型中的差分方程转化为简单的代数方程求解;仿真结果表明,提出的算法能正确、有效并且快速地恢复由于局部匀速运动所造成的图像退化;算法对模糊宽度的变化不敏感,较维纳滤波恢复算法有一定的稳健性、优越性。A restoration algorithm of local uniform motion blurred image based on Z transform for invariant background motion image detection was proposed. Suppose the foreground and the background is easy to be separated from blurred image. Image foreground was extracted from blurred image based on background subtraction method. Other direction movement of image foreground in the image space was converted into X axis direction movement by rotation matrix. In order to simplify the complex restoration process of local-blurred image, the whole procedure was divided into two steps： 1） image foreground restoration; 2） integration of foreground and background. Restoration and degradation model was established based on Z transform by strict mathematical deduction. Difference equation was changed into algebraic equation to simplify the solving process. Simulation results shown that the proposed method can be restored blurred image of local uniform motion correctly, effectively and rapidly. The algorithm is not sensitive to fuzzy degree; have a certain stability and superiority compared to Wiener filter restoration algorithm.

关 键 词：图像处理 图像复原 Z变换 局部运动模糊图像 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]