利用奇异值特征的激光成像雷达目标识别  

Target recognition using singular value feature for laser imaging radar

作　　者：陈晓清[1] 马君国[1] 付强[1] 庄钊文[1] 

机构地区：[1]国防科学技术大学ATR实验室,湖南长沙410073

出　　处：《红外与激光工程》2011年第9期1801-1805,共5页Infrared and Laser Engineering

基　　金：国家自然科学基金(60972113)

摘　　要：随着激光技术的发展,激光成像雷达在现代战争复杂战场环境中逐渐获得了广泛的应用,目前激光成像雷达目标识别技术已成为国内外研究的热点。基于激光成像雷达的距离像,以地面军事目标为识别对象,提出了针对奇异值特征的激光成像雷达目标识别算法。利用奇异值特征表征图像具有良好的稳定性,通过分析识别率与奇异值特征数目的关系,从激光成像雷达目标的距离像中提取奇异值特征,该特征有效抑制了噪声,最后应用最优参数支持向量机对3种地面军事车辆目标进行识别。仿真实验表明,该算法取得了比较高的识别率,识别效果良好。With the development of laser technology, laser imaging radar gradually obtains wide application in complicated battlefield of modem warfare. The target recognition technology for laser imaging radar has become a hot at home and abroad recently. Target recognition algorithm based on singular value feature for laser imaging radar was put forward in this paper, which regarded military vehicles as recognizing objects. The singular value had good performance that was invariant to image disturbance and it was credible to represent the image. According to the relationship between recognizing rate and the number of singular value feature, singular value feature was extracted from range image of laser imaging radar targets. This feature restrained noise effectively. Support vector machine of optimum parameter was used to recognize three kinds of ground military vehicles. Simulation experiments gain high recognition rate and good recognition effect.

关 键 词：目标识别 激光成像雷达 距离像 奇异值 

分 类 号：TN958.58[电子电信—信号与信息处理]