Research And Application Of Remote Sensing Image Subblock And Classification

Remote sensing image classification is an important aspect of remote sensing image processing and widely applied to ocean, agriculture, geology, meteorology, environmental, military and other fields. With the development of satellite sensor and data communication, remote sensing has been in a era of dynamic, real-time and nicety, provided growing numbers of data. On the one hand, increasing of data bring on much time in remote sensing image processing. On the other hand, there are many idle computing resource in LAN. How does it quickly finish remote sensing image classification without affecting accuracy and making use of the idle computing resource become a hotspot The popularity of distributed computing technology provided a new direction of remote sensing image classification.From the application aspect, the paper analysis existing problems in current remote sensing image classification, indicated the necessity of distributed computing in remote sensing image classification. For the long time and low efficiency in remote sensing image classification, the paper research on remote sensing image classification using distributed computing, study and use distributed computing library MPI(Message Passing Interface), integrate message passing mechanism and LAN, implement high performance distributed computing without changing hardware environment.Based on analysis of speedup and speedup efficiency, in order to achieve higher speedup, the paper propose two kind of image subblock algorithm, namely Load Balance Priority Subblock Algorithm and Efficiency Priority Subblock Algorithm., detailed state the thought and implement of the two algorithm. Through the comparison of the two algorithms' speedup and speedup efficiency, the paper demonstrates the application environment of the two algorithms, and analyzes the advantages and disadvantages of them.The paper designs a module of remote sensing image subblock and classification, and implements it based on MPI library. Through the analysis of remote sensing image subblock and classification, the paper divided the module into several function modules, introduce the function of every module and implement it using C++. The module of remote sensing image subblock and classification at the cost of increasing image subblock and transmit time, divide the complex computing task on single computer into independent computing node, decrease the computing time.