基于支持向量机的遥感矿化蚀变信息提取方法研究-地图学与地理信息系统专业论文.docxVIP

ABSTRACTThe ABSTRACT The western high-cold mountainous districts are the most important mineral-forming areas，with extremely high mineral exploration potential． Exploring and assessing the mineral resources in the areas to turn resource advantages into economic advantages is one of important approaches in terms of boosting the economy of West China．Compared with traditional geological prospecting，Remote Sensing technique has all technical advantage beyond description for mineral resources prospecting and assessment in similar natural landscape．So，_it is considerably significant to explore new suitable approach of Remote Sensing altered rock’S information extraction for improving geologic prospecting reliability and efficiency． Support Vector Machine is a new machine learning approach based on Statistic Learning Theory and has been successfully applied in different fields owing to its advantages，such as global optimization，strong flexibility and good generalization．But．how tO select the model parameters much suffers from man-made influence．Considering the problem,on a basis of fully discussing SVM and PSO basic theory,the paper proposes a new technique based on PSO to select the model parameters for SVM．And，on a basis of the above technique,the paper has an experiment on altered rock’S information extraction using SVM．Major findings ale as follows： 1．Using the swarm intelligence character of PSO，the paper proposes a technique based on PSO to select the model parameters for SVM．Compared with the common used grid way in LibSVM．it behaves an advantage of less learning times． 2。On a basis of the technique based Oil PSO to select the model parameters for SVM,the paper has an experiment on altered rocks information extraction using SVM and proposes a full frame of the way．It shows the way is an effective for altered rock’S information extraction by comparing the extracted information with the data of the known alteration points． 3．Based on Uniform Design，one strategy placi