基于矩与支持向量机的舰船目标识别方法  

Ship targets recognition method based on moments and SVM

作　　者：徐芳[1,2] 韩树奎 XU Fang;HAN Shu-kui(Changchun Institute of Optics,Fine Mechanics and Physics,Chinese Academy of Sciences,Changchun 130033,China;University of Chinese Academy of Sciences,Beijing 100049,China;Northeast Electric Power Design Institute Co Ltd,China Power Engineering Consulting Group,Changchun 130021,China)

机构地区：[1]中国科学院长春光学精密机械与物理研究所,吉林长春130033 [2]中国科学院大学,北京100049 [3]中国电力工程顾问集团东北电力设计院有限公司,吉林长春130021

出　　处：《传感器与微系统》2018年第8期43-45,48,共4页Transducer and Microsystem Technologies

基　　金：国家自然科学基金资助项目(60902067);吉林省重大科技攻关资助项目(11ZDGG001)

摘　　要：为可靠快速地识别出各种姿态下的舰船目标,提出了一种基于矩与支持向量机(SVM)的目标自动识别方法。根据实际航空摄影模型的特点,将三维舰船模型相对其俯仰轴,偏航轴和横滚轴作相应旋转,投影到二维图像空间,建立舰船样本训练库与测试库,提取舰船各种姿态下的矩特征;基于SVM设计多类分类器进行识别,并进一步计算不同训练和测试样本数下的分类精度。实验结果证明:提出的方法在舰船模型图像和真实遥感图像中的识别精度高,且样本训练和目标识别时间短,经数据库中多幅图像测试,识别系统鲁棒性强。A ship target automatic recognition method based on moment and support vector machine （SVM） is presented ,in order to reliably and quickly recognize ship target in various postures. Three-dimensional ship models are projected into a two-dimensional image space by means of revolving around their pitch, yaw and roll axis, according to the characteristics of actual aerial model. Training and testing databases of samples are constructed, moment features of ships at different attitude are extracted. Ships are recognized by using the muhi-class classifier. Based on SVM, and classification precision of different training and testing samples numbers are calcutated. The experimental results show that the recognition presion of this method in ship model image and true remote sensing image is high. The training sample and target recognizing time is short and this recognition system is robust by testing on multiple images in database.

关 键 词：矩不变量 目标识别 舰船样本 多类分类器 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]