一种用于遥感图像景象匹配的快速光电混合系统

Fast hybrid opto-electronic system for scene matching of remote sensing images

景象匹配是遥感图像处理的一项关键技术,通过对体全息相关器的原理及特点进行分析,提出了一种快速光电景象匹配方法。利用高密度体全息光学存储技术记录若干幅库图像,识别时在光电系统中输入实时图,可以一次并行地完成输入图像与所有库图像的相关运算,快速返回实时图与各库图像的相似度大小。遥感图像的空间灰度分布在统计上可以看作是一个二维宽平稳各态历经的随机过程,由此采用相应的图像预处理及相关点检测后处理方法可以有效提高相关识别的并行度并抑制光电系统的噪声,完成高精度快速景象匹配任务。

Scene matching of remote sensing images is a key technology of precision control and guidance system. A new fast hybrid opto-electronic scene matching method was proposed based on the analysis of the autocorrelation characteristics of the remote sensing images. Multiple images were stored as a database by using the high density volume holographic storage technology. When the real-time image was uploaded into the system, the correlation calculation between the real-time image and all of the images in the database could be implemented in parallel. The output is values of all the correlation. Since the spatial grayscale distribution of remote sensing images are of a stationary random process approximately, the related image preprocessing and correlation spots postprocessing methods are adopted to enlarge the parallelism of the system and to suppress the noise of the hybrid system, which can help the fast and precise scene matching.