Two Methods of Data Mining Technique for Target Recognition
Based on clustering analysis by Euclide distance and optimal hyperplane structured by Support Vector Machines (SVM), this paper deals with two target recognition approaches on Data Mining technique. The characters of the two methods are compared with each other. By means of power cepstrum analysis and envelope analysis of ship-radiated noise, the target characteristics are abstracted and optimized. Using pattern recognition algorithm, quadratic programming arithmetic and SVM, it completed the classification experiments on three kinds of underwater targets at the same operation condition. The practical result is satisfactory.
Tian Na Wang Haiyan
College of Marine Engineer Northwestern Polytechnical University Xian 710072 P. R. China
国际会议
北京
英文
352-356
2007-10-01（万方平台首次上网日期，不代表论文的发表时间）