目的 近年来遥感技术发展迅速,遥感影像变化检测作为最重要的关键技术之一,越来越多的应用在国土资源管理、地物变化、农林业的监测等领域。目前依据变化检测的流程的综述相对较少,大多数综述只针对变化信息提取的方法进行论述,为使国内外研究者对遥感影像变化检测理论、流程及其现存问题有一个比较全面的认识,对其进行系统综述。方法 通过对国内外大量的遥感影像变化检测算法进行总结、分类及比较,按照技术流程的各个环节分别论述,重点论述了变化信息提取环节中的图像分割、特征提取和分类的发展现状、基本思想及其趋势。结果 目前,多数变化检测算法主要是针对特定的条件具有较好的效果,还没有通用性算法,且现有算法在效率、精度、智能性等方面存在的问题,大多算法解决的问题及理论相对分散。结合现存问题及目前大数据影响下的技术发展状况,从数据类型、预处理方法、变化信息提取方法、算法效率、算法理论创新5个方面对遥感影像变化检测领域的未来发展趋势进行预测和展望。结论 遥感影像变化检测在多领域具有较高的研究价值,但针对目前变化检测存在的一些局限性还需要进行深入的研究,针对变化检测的研究需要从研究热点中挖掘创新思路、引入深度学习等发展趋势。

Objective In recent years,remote sensing technology has developed rapidly. And remote sensing image technology has been applied in more and morefields, especially inland and resources management, earth's surfacechange,agroforestry monitoring and other fields. At present,there wererelatively few reviews based on change detection process. Most algorithms were only for change information extraction method.In order to make more researchers have a more comprehensive understanding in remote sensing image change detection theory, process and its existing problems, a detailed introductionwas reviewed. Method a large number of remote sensing image change detection algorithms were summarized, classified and compared. A deep description based on the flow of the change detection technology was given in this paper. And the development status and trends of the image segmentation, feature extraction and classification algorithms in the step of change information extraction were mainly discussed. Result Most of the change detection methods have good performance for specified condition. There are no generic algorithm, and the existing algorithms have problems in the perspectives of efficiency, accuracy, intelligent. The majority of algorithms have solved the relatively scattered problems and theories. Combined with the existing problems and the present state of development under the influence of the big data technology, the future development trend of remote sensing images change detection field was forecasted and prospected-from five aspects, which are data types, pre-processing method, the change information extraction method, the algorithm efficiency and theoretical innovation. Conclusion Remote sensing image change detection has high research value in many areas, but some of the limitations of change detection at the present time also need further research. The study of change detection needs creative thinking from the research hot spot and introducing deep learning and other development trends.