中文摘要：

为了抑制气动退化复原图像中的振铃效应,提出了一种气动退化复原图像多尺度局部去振铃算法。首先,对复原后图像进行非下采样轮廓波变换（Nonsubsampled Contourlet Transform,NSCT）分解;然后进行多尺度边缘检测、融合和二次跟踪细化实现图像主轮廓的提取;最后,对图像的主轮廓进行去振铃局部模糊滤波。仿真实验结果表明,在运算效率与其他算法相当的情况下,滤波后图像的边缘振铃去除效果更好,纹理细节恢复更加显著,基本满足了振铃去除、纹理细节保护以及实时性的要求。

英文摘要：

In order to solve the problem of ringing effect in the aero-degraded restoration image,a multi-scale local deringing algorithm is put forward.First,disassembly of nonsubsampled contourlet transform（NSCT）was performed.Then multi-scale edge detection,edge fusion and secondary track-thinning were carried out to get main contour.Finally,local deringing fuzzy filter to main contour was finished.Experimental results indicate that deringing effect is better and the preservation of texture detail is more obvious under the same calculatiom efficiently as other algorithms.The proposed algorithm can satisfy the requirements of ringing removal,preservation of texture detail and real time.