SAR图像变化检测综述

A summary of SAR image change detection

合成孔径雷达(synthetic aperture radar,SAR)图像变化检测任务是检测同一地区、不同时间获得的两幅SAR图像的变化情况,在农业、军事等方面都有广泛的应用。首先介绍了实现SAR图像变化检测过程中的三个难点和SAR图像变化检测的一般流程。然后,介绍了图像预处理的方法和产生差异图的步骤。图像预处理是SAR图像进行变化检测前必不可缺的步骤,而差异图的质量影响最终的变化检测结果。随后,分别对传统的变化检测方法和基于深度学习的变化检测方法进行了总结,并分析了每种方法的优缺点。最后,给出了SAR图像变化检测的评价指标。

The task of synthetic aperture radar(SAR)image change detection is to detect the changes of two SAR images in the same area at different times.It is widely used in agriculture,military and other fields.This paper first introduces three difficulties in SAR image change detection and the general process of SAR image change detection.Then,the method of image preprocessing and the steps of generating difference image are introduced.Image preprocessing is an essential step before SAR image change detection,and the quality of difference image affects the final change detection result.Next,the traditional change detection methods and deep learning-based change detection methods are summarized,and the advantages and disadvantages of each method are analyzed.Finally,the evaluation index of SAR image change detection is given.