基于统计特征的水下目标一维距离像识别方法研究

Research on underwater target recognition with statistical features of high resolution range profiles

目标探测与识别是水下预警监视、信息对抗的重要组成部分。针对水下目标一维距离像识别问题,通过提取目标的长度、重心、高阶中心矩等特征,分析了所提取特征的统计分布特性,利用假设检验构建了目标识别特征的统计模型。结合Bayes统计分类器开展了5类水下目标的识别实验,并与基于距离像回波匹配相关的识别方法进行对比分析,对比结果显示所提出的方法在识别率和运算量方面均有明显改善。

Target detection and recognition is the main issue of the underwater surveillance and information warfare. In this paper, five features, （target length, mass, the second center moment, third center moment and fourth center moment）, are extracted for the target recognition application from the original high resolution range profiles. The statistical characteristics of the extracted features are analyzed and the statistical model is built with the method of hypothesis testing. FinaUy, the recognition experiments of the five difference targets are made with the Bayes statistical classifier. The results compared with the traditional template correlation method for the high resolution range profiles show that the proposed method has the obvious advantage in recognition rate and computational complexity.