Remote Sensing Images Change Detection Methods And Application

Remote sensing images change detection is a process that analyzes a pair of remote sensing images acquired on the same geographical area at different times in order to identify changes that may have occurred between the considered acquisition dates. With the develop of satellite, such as NOAA/AVHRR,EOS/MODIS,MSS/TM/ETM,SPOT,CBERS,IRA,IKONOS,QUICK-BIRD and ERS,RADARSAT, remote sensing technology develop rapidly. In order to solve the question of information detection and database update, change detection has become an important research domain of remote sensing images.(1) The change detection methods based on probability and statistics models, such as algorithms Cumulant-Based, are good at processing the images with difference background pixels and target pixels. Considering the advantages of algorithms Cumulant-Based and dynamic clustering algorithm, describing details features of images and getting better clustering results rapidly, a change detection method based on Cumulant-Based KL Approximation with the dynamic clustering is proposed in this paper. This method can get very good image information and edges for small-scale images. Simulation results show that the proposed algorithm make a good performance.(2) A change detection method based on region segmentation and noise removal is proposed. There are different thresholds in different regions. If a threshold is used in the whole image, there are many errors will be generated. Using region segmentation to detect region's change can improve detection accuracy. The new change detection method proposed segments the difference image two times and adds the results to construct a new difference image. Our method can solve the problem of poor continuity about boundary.(3) The adaptive threshold segmentation method has always been playing an important role. Many adaptive threshold segmentation algorithms have been developed. But most of them can give a result containing many wrong pixels when processing images with noises. In this paper, a new threshold image segmentation method based on the relationship of the pixels in the neighbor region is proposed. In order to reject false detection and improve the detection result, the relationship between adjacent pixels is used to process the segmented image. Simulation results show that the proposed method can greatly improve the results of change detection on the condition of storing the target information.This work was supported in part by the National Natural Science Foundation of China (No.60703109) and The Research Fund for the Doctoral Program of Higher Education (No.20070701016).