Change Detection Of Multi-temporal Remote Sensing Images

With the rapid development of modern sensor technology and information technology, earth observation remote sensing data of different resolutions emerges in large amounts, which provides significant support for the theoretical study and practical application of remote sensing. Remote sensing images change detection is an important aspect of remote sensing applications, and it has been widely used in the field of national economy and national defense construction. By analyzing the same area at different phases of the remote sensing images, change detection provides information on changes in the surface objects, for resource and environmental monitoring, natural disaster assessment, and battlefield situation analysis.So far, many scholars have presented a variety of change detection methods, which, according to the complexity of information processing and stages of development, can be divided into the traditional change detection methods and new change detection methods. This paper firstly summarizes the properties of traditional methods. However, because of its simplicity in processing information, it is not adapted to process modern remote sensing images with high spatial, temporal and spectral resolution, so it can not meet the requirement of modern remote sensing applications to remote sensing technologies. The new change detection methods emerging in recent years have stronger capability to handle complex information, greatly improving change detection accuracy. This article summarizes the key features of traditional change detection methods, and then sums up four new change detection methods presented in recent years. By analyzing and discussing the advantages and disadvantages, we prospect the direction of change detection research.Besides, this paper proposes a multi-scale parcel-based multi-temporal remote sensing image change detection method. First, multi-temporal remote sensing images are segmented into parcels to obtain regions with consistent characteristics. Then the contextual information is analyzed based on the multi-scale parcels, reflecting the object features in the multi-resolution levels. Finally, Support Vector Machine is used to get the change detection map. Experimental results show that this method improves the pixel-based methods with more accurate change detection map.