基于高斯chirplet时频原子参数自适应时频分布图的不变矩特征提取方法研究  

Target Feature Extraction Based on Moment Invariants of Adaptive Time-Frequency Distribution on Chirplet Time-Frequency Atom

作　　者：田进[1] 李言俊[1] 张科[1] 郭俊锋[1] 

机构地区：[1]西北工业大学航天学院,西安710072

出　　处：《宇航学报》2008年第5期1656-1661,共6页Journal of Astronautics

基　　金：国家自然科学基金(60575103)

摘　　要：根据时频分布性质,针对雷达目标的特征提取与分类识别问题,提出了一种基于高斯chirplet时频原子参数自适应时频分布图的不变矩特征提取方法。首先分析了各种不变矩,总结了它们各自的应用特点;接着把雷达信号和其时频分布图的不变矩特征结合起来分析了目标的各种行为对时频分布图不变矩的影响,然后对提取时频分布图不变矩的算法进行了详细的研究,识别仿真结果表明,利用信号参数自适应时频分布图像的不变矩特征进行雷达目标识别是一个卓有成效的方法,该方法充分利用了雷达目标的散射信息,具有较高的识别率。Aiming at the problems of feature extraction and recognition of radar target, a target recognition method based on moment invariauts of adaptive time-frequency distribution based on chirplet time-frequency atom was proposed. Firstly various moments were analyzed at length, and then change of time-frequency distribution because of various target behavior was analyzed at length by combining radar signal with moment invariant of time-frequency distribution. And then the algorithm was researched at length. At last, the simulation experiment was carried on, the result indicate that the method has higher recognition rate because of fully utilization of target scattering information.

关 键 词：自适应时频分布 不变矩 匹配追踪法 

分 类 号：TN911.72[电子电信—通信与信息系统]