Change Detection Methods Research In Remote Sensor Images Based On Cluster Analysis

With the fast development of aviation and aerospace remote sensing technology, the remote data are tending to three-multi (.i.e. multi-sensor, multi-platform and multi-angle) and three-high (i.e. high spatial resolution, high time phase resolution and high spectral resolution). How to use these remote sensing images to realize change detection of targets and take decision and after-service for battlefield command system will become a popular direction in military affairs region.The change detection based on remote sensing images is the technology to analyze and determine the change characters and process quantificationally for the remote sensing images obtained from different time. This plays important role in civil and military regions such as resource survey, environment surveillance, and military spy and so on and becomes one of the most important research directions in remote sensing technology.A change detection method in remote sensing image based on cluster analysis is presented through the research and analysis on the existing problems of the traditional change detection methods. The main works are as following:a) Have the theory analysis for cluster analysis deeply; discuss some basic concepts such as sample type, sample similarity and cluster definition and so on by mathematics form expression. Discuss some common methods for statistical cluster and the definition of distance between clusters. Analyze the advantage of cluster analysis in data processing and these results can offer theory basis for change detection method in this paper;b) Study on the imaging principle of remote sensing image and study the principle, flow and algorithms of change detection in image systematacially and deeply. Analyze the advantage and disadvantage of some common change detection methods at present, and moreover compare them based on the experimental results. And offer the basic for next method put away by this paper;c) The change detection method in remote sensing image based on cluster analysis is put away. Firstly, the algorithm and flow is presented by mathematics expression, and then use this method to take simulation experiments for SAR image and multi-spectral image respectively. The results given are better than the traditional methods. Finally, the relation between image matching precision and detection are put away from the results analysis.