基于遥感图像梯度特征的运动模糊检测方法  

Detection Method of Motion Blur Based on Remote Sensing Image Gradient Characteristic

作　　者：毛冰晶[1] 陈小梅[1] 薛博[1] 李倩倩[1] 倪国强[1] 

机构地区：[1]北京理工大学光电成像技术与系统教育部重点实验室,北京100080

出　　处：《北京理工大学学报》2013年第10期1083-1086,1092,共5页Transactions of Beijing Institute of Technology

基　　金：国家重点基础研究发展计划项目(2009CB724006)

摘　　要：为了精确描述由于高分辨率卫星在轨动力学环境不确定干扰造成的光学传感器拍摄遥感图像退化的过程,在自然图像的运动模糊点扩散函数检测方法的基础上,提出一种基于遥感图像梯度特征的运动模糊检测方法.该方法利用图像分割和梯度特征选择处理区域,对遥感图像进行预处理,使图像的梯度特征更加符合检测方法的先验知识,采用基于概率分布的点扩散函数估计方法获得图像的退化模型.结果表明该方法提高了图像模糊的检测精度,获得了更准确的遥感图像运动模糊点扩散函数的检测结果.In order to accurately describe the remote sensing image degradation, which is caused by the uncertainty dynamic interference of the optical sensor on the high-resolution satellite, taking the existed natural image motion blur detection method as reference, this paper approaches a vibration detection method based on remote sensing image gradient characteristic. The proposed method uses image segmentation and gradient feature to select processing area at first, and then applies pre-processing to make image feature more consistent with the prior knowledge, at last uses the point spread function estimation method based on probability distribution to obtain the detecting result. It is indicated that the presented method could improve the detection accuracy and achieve a more accurate motion blur point spread function results of remote sensing image.

关 键 词：梯度特征 运动退化 模糊检测 卫星平台 

分 类 号：TN911.73[电子电信—通信与信息系统]