Optimization Method of Parallel Processing for Remote Sensing Image Cloud Detection

High computational complexity and long time computation are the main characteristics for cloud detection based on texture feature. It affects the real time of cloud detection based on texture feature. So a parallel architecture for cloud detection is proposed, according to the characteristics of the algorithm on the basis of analyzing the computing property of the algorithm. The new parallel architecture is on the basis of the multi processing element(PE) parallel optimization method and pipeline optimization method base on run time and resource consumption. The parallel structure can improve the real time for the cloud detection algorithm. In order to verify the parallel structure for cloud detection proposed in the thesis, comparative analysis for run time, resource consumption and dynamic power between methods which have been proposed and the proposed method. It shows that the proposed method has the best run time, and decreases the area consumption effectively at the same time. It also decreases the dynamic power effectively.