Realization Of Remote Sensing Image Classification System Using SVM Algorithm In MABLAB

Remote sensing image classification is one of the important steps for the extraction of remote sensing information, the classification accuracy and completeness will directly decide the success or failure of information extraction task。The traditional methods of the remote sensing image classification, not very good use images of the space of semantic information, led to the slow operation speed, operational accuracy at the end.The algorithm of support vector machine (Support Vector Machine, SVM) is first proposed by V. Vapnik in1995, based on statistical learning theory as the basis of a pattern classification algorithm. To solve the problem with limited sample, nonlinear and high dimension pattern recognition problem shows many unique properties, and has strong generalization ability, in recent years has been widely applied to many fields of pattern classification and recognition.However, the SVM algorithm need to have strong mathematical skills and computer programming technology, the majority of scientists is often difficult to apply the algorithm to remote sensing image classification for non-computer professionals.Convenient, fast, high-precision remote sensing image classification as the guiding ideology, the design of an effective use of remote sensing image classification system of the SVM algorithm, the system in MATLAB to achieve development. Experimental results show that this classification system not only has a small sample learning, remote sensing image classification results as well as relatively fast, automatic, stable classification performance, and system interface is very user-friendly, especially non-computer professional research workers to provide a simple, convenient, fast and efficient application of the SVM algorithm to remote sensing information extraction of remote sensing image classification and target objects.