基于SOM和SVM的遥感图像目标识别-中国科技论文在线.PDFVIP

系统工程与电子技术 　 　 第 24 卷 第 7 期 Systems Engineering and Electronics Vol24 ,No7 2002 ( ) 　文章编号 :1001506X 2002 　 　 基于 SOM 和 SVM 的遥感图像目标识别 张艳宁 , 郑江滨 , 候云舒 , 赵荣椿 (西北工业大学 , 陕西 西安 710072) ( ) ( ) 摘 　要 : 提出了一种基于自组织特征映射 SOM 神经网络和支撑矢量机 SVM 相结合的遥感图像目标识别方 法。该方法首先利用 SOM 对目标进行聚类 ,然后应用 SVM 方法对其进行分类识别。最后将该方法应用于二值遥 感图像的目标识别 ,与仅用支撑矢量机方法的识别结果进行了比较 ,结果表明 ,这一方法对二值遥感图像目标具有 很好的分类识别效果 ,且训练时间大幅度缩短。 关键词 : 图像识别; 支撑矢量机 ; 遥感图像 ; 自组织特征映射神经网络 中图分类号:TN9198 　　　文献标识码 :A Remote Sensing Target Recognition Based on SOM and SVM ZHANG Yanning , ZHENGJiangbin , HOU Yunshu , ZHAO Rongchun ( Northwestern Polytechnical University , Xi ’an 7100 2 , China) ( ) Abstract : An efficient method of remote sensing target recognition based on selforganizing map SOM neural network incorpo rating with support vector machines is proposed in this paper. Firstly , SOM is applied to cluster for the remote sensing target , then SVMs are used to classify each cluster. Finally , this method is used to recognize some binary remote sensing targets. Compared with SVM , the experiment results show that the method proposed will have better classification and recognition results for the binary remote images , which only contains object and background , and the tranning time is shorter than that by