基于LDA算法的一维距离像特征提取

Feature Extraction of HRRP Based on LDA Algorithm

在线性分辨分析(LDA)基础上,通过遗传算法寻找Fisher准则下最优的线性映射中心和相应的最优映射,使不同模式在特征空间内具有最大的可分离性,并将该算法用于雷达目标一维距离像特征提取与识别中。实验表明,和原算法相比,新算法在特征提取性能和目标正确识别率上有较大提高,说明了算法的有效性。

Finding the right features is a key part of a pattern recognition system. This paper presents a linear feature extraction algorithm which research the optimal pattern centers and projection in Fisher＇ s criterion by genetic algorithm（GA）. And then the new algorithm is used in the range profile features extraction and recognition. Experiments show that the new algorithms are more effective than IDA in Fisher＇s function value and classifying results.