结合编码曝光和运动先验信息的局部模糊图像复原

针对局部运动模糊图像复原的病态性和背景被破坏的问题，提出基于编码曝光和运动先验信息的局部模糊图像复原方法. 分析编码曝光成像理论模型，建立最优码字选取的适应度函数准则. 通过物像关系，获得运动目标的点扩散函数（PSF）像移尺度初步估计参数，作为运动先验信息. 采用背景差分法进行目标提取，综合编码曝光运动模糊图像的叠加特性，实现对运动模糊目标区域的精确提取. 结合先验信息，引入基于贝叶斯最大后验概率框架的student-t复原算法进行PSF精确估计和复原重建，快速迭代得到复原结果. 搭建实验仿真系统，并开展针对实际运动目标的复原实验. 实验结果表明，该方法能有效改善传统曝光中运动模糊复原的病态性问题，抑制复原过程中目标图像边缘振铃及背景噪声的放大效应，所复原图像具有更好的主客观评价结果.

Restoration of local blurred images based on coded exposure and motion prior information

A local blurred image restoration method based on the coded exposure and motion prior information was proposed, aiming at the problem of ill-posed and background damage along with the local motion blurred image restoration process. The theoretical model of coded exposure imaging was analyzed, and the fitness function criterion for the optimum code sequence selection was established. Through the object-image relationship, the preliminary estimation scale parameters of point spread function (PSF) for moving target were obtained as the prior information of motion. The background difference subtraction method was used to extract the motion blurred target area accurately by synthesizing the overlapping characteristics of the motion blurred image by coded exposure. Combined with the prior information, the student-t restoration algorithm based on the Bayesian maximum posteriori probability framework was introduced to estimate and reconstruct the PSF accurately. And the restoration results were obtained by several iterations fast. The experimental simulation system was built and the reconstruct experiments aming at the real motion targets were carried out. Experimental results show that the method can improve the pathological problem of motion blur restoration in traditional exposure effectively, as well as restrain the magnification effect of edge ringing and background noise in the restoration process. The images restored by the proposed method have better subjective and objective evaluation results than that of other methods.