通过合理建立针对红外图像的气动光学效应退化模型,根据极大似然估计准则设计复原算法,同时针对随机变化的点扩展函数和噪声对目标图像尤其是图像中细节恢复的影响,对单一规整化方法进行了扩展,采用双重规整化策略,将规整化分为两个各有侧重点的层次来处理。在微机上进行了一些复原实验,并给出对比结果,证实了该算法的有效性。
An aero-optic effect degraded image blind restoration algorithm based on double regularization is proposed. Appropriate aero-optic effect degradation model is established for infrared image and the algorithm is designed in the framework of maximum 4ikelihood principle. To suppress noise and preserve edges more effectively, the proposed algorithm generalizes original nonlinear regularization by double regularization, dividing regularization into two parts with different emphases on noise-suppression and edge-preservation. The algorithm is verified using microcomputer and by some experiment results with contrast to other algorithms.