Multi-temporal Remote Sensing Images Change Detection Based On TPCA

At present, the multi-temporal remote sensing image change detection has been widely used in the field of the national economy and national defense construction. By analyzing the remote sensing images which were obtained in the same area but different time, we take the change detection technology to extract the change information of objects which are used in urban resources management, environmental monitoring, disaster prevention and mitigation, and so on. Although remote sensing image change detection research has been made a lot of development and haven a variety of detection methods, it is still at the exploratory stage and automatization degree is not high, often through artificial intervention to achieve it. As a result, the paper is to find some technology of automatic detection changes, so that it can analyze and compare the remote sensing images which obtained in the same area but different time on the computer and obtain the change information of objects. At the same time, it can be able to accurately determine the changes in the region and extract change information of the features as much as possible.This thesis introduces the basic concepts, mathematical models and the processing flow of remote sensing images change detection. The existing change detection methods are also more comprehensive and summarized comments. And, the theory of knowledge and processing flow of image differencing and traditional PCA method in remote sensing images change detection are described in detail. Furthermore, the TPCA method was realized on the basis of the traditional PCA method and combined some view of many foreign papers. It is an improvement of the traditional PCA method. It overcomes the shortcoming that traditional PCA method needs to depend on experience to choose which components can characterize the change information. This method is not to analyze the principal components of per image separately, but to analyze the principal components of two combined image together in order to obtain change detection information directly. Therefore, it can improve automatization degree for change detection.We select two temporal images of some area in ChengDu, deal the images with geometric correction, and implement the method of image differencing, conventional PCA, and TPCA based on IDL. Experimental results show that using of the TPCA method can judge exactly that the ground object is changed or not and where it is changed in the area. It can detect more information than image differencing and conventional PCA. At the same time, it shows that the method is accuracy and reliability.