摘要
遥感卫星传感器与观测目标之间的相对运动是导致遥感图像退化的常见因素之一。对运动模糊图像的频谱特性进行了分析,基于Radon变换估算运动模糊角度和长度,进而准确估计退化图像的点扩散函数进行图像复原。针对实际模糊图像频谱中出现的十字亮线严重干扰Radon变换准确性的问题,提出频谱分块与边缘检测相结合的改进算法,提高了检测精度。最后对实际发生运动模糊的对月遥感图像进行了模糊参数估计,并采用FTVd(fast total variation(TV)deconvolution)算法进行图像复原,实验证明参数估计精确,复原效果好。
The relative motion between remote sensing satellite sensor and objects is one of the common reasons for remote sensing image degradation. This paper calculated motion blur orientation and blur extension by using Radon transform based on the analysis of the feature of blurred image spectrum. Then it accurately estimated the PSF of degradation to restore the blurred image. As for some actual remote sensing images, there might be some cross-shaped bright lines in the spectrum which could interrupt Radon transform, this paper developed an improved method which chose the suitable sub-block in the spectral image before edge detection, to effectively increase detection accuracy of Radon transform. In the end, it applied FTVd algorithm to restore the actual remote sensing images of the moon after estimating the blur parameters. The experiments show the effectiveness of the algorithm.
出处
《计算机应用研究》
CSCD
北大核心
2015年第12期3798-3800,3809,共4页
Application Research of Computers
