摘要
针对欧氏距离受模式特征量纲影响的缺点,且影响制导精度和抗干扰的能力,采用一种基于马氏的一维距离像目标识别方法。通过对毫米波雷达目标回波进行IFFT变换得到一维距离像,用主分量分析算法对距离像进行降噪和特征提取,取最小马氏距离判别目标类别。马氏距离考虑了模式特征参数的大小以及特征间的相关性,克服了欧氏距离受量纲影响的缺点。与欧氏距离的分类算法仿真结果比较,马氏距离算法具有较好的识别性能;在不同信噪比下的仿真结果表明,方法适用于毫米波雷达一维距离像目标识别。
For Euclidean distance ignoring data distribution tendency, this paper presents a new method of target recognition based on Mahalanobis distance in millimeter wave radar （MMW）. The Mahalanobis distance takes the magnitude and relativity of pattern character into account, and has a better performance than Euclidean distance. Using IFFT Transform to MMW target echo, one - dimensional range profile was obtained. Then, the primary compo- nent analysis （PCA） was analyzed for noise -reduction and feature extraction. At last, the rain - mahalanobis distance was used for recognition. Experimental results on both Euclidean distance and Mahalanobis distance show that the later method has a better performance. Simulation on different Signal - to - Noise Rate （SNR） shows weighted Mahalanobis distance method is the same with MMW target recognition.
出处
《计算机仿真》
CSCD
北大核心
2010年第3期31-34,84,共5页
Computer Simulation
关键词
马氏距离
一维距离像
主分量分析
Mahalanobis distance
One - dimensional range profile
Primary component analysis （PCA）
