Automatic Generation of High Range Resolution Profiles Models for Radar Recognition
-
摘要: 模板的完备性直接决定了基于高分辨距离像的雷达目标识别系统的分类性能；在外场试验中限于目标姿态、环境等因素难以获得准确标定的目标立体角范围内全姿态模板数据。针对一维距离像识别的工程实用化需求，本文基于数据驱动思想，提出了新的一维距离像聚类模板自动生成算法。与传统方法相比，本文方法在提高工程可行性的同时提高了识别性能。为满足实验需要，本文提出了新的基于MSTAR图像的高分辨距离像反演算法，得到更精确的反演数据。基于该数据的实验结果表明算法解决了模板生成姿态角依赖性问题，提高了识别性能。Abstract: The completeness of template directly determines the classification performance of automatic radar target recognition system based on high resolution range profiles (HRRP). It’s difficult to get HRRP training data labeled accurately covering the entire targetaspect angle because of a lot of practical factors in the field experiments， such as target attitude， environment and so on. According to the demand of engineering practical development， the dissertation proposes an algorithm of automatic generation of HRRP template based on data driving means. The proposed approach can be realized much easier with better recognition performance， comparing with the traditional approach. The dissertation puts forward a new HRRP inversion method based on MSTAR image to get more precise HRRP using in cyberemulation. At last，the result of the experiments proves the algorithm.
-
Keywords:
- High Range Resolution Profiles /
- template /
- clustering /
- HRRP inversion
-
计量
- 文章访问数: 895
- HTML全文浏览量: 6
- PDF下载量: 1345
下载:
