高分辨雷达信号的平移不变KPCA特征提取算法

Translation Invariant Feature Extraction Algorithm of KPCA Based on High Resolution Radar Signal

研究高分辨雷达信号特征提取,针对传统提取平移不变特征存在信息损失量大、识别的准确率低的问题,提出了一种平移不变KPCA特征提取算法。首先计算高分辨雷达信号的原点矩,并在低信息损失的前提下利用高分辨雷达信号相对原点矩的位置来描述原信号,从而消除高分辨雷达信号的平移敏感性。然后结合KPCA特征提取算法得到平移不变的特征信号。最后应用SVM分类器对特征信号进行分类识别。实验证明,改进算法识别率高于雷达目标识别系统中的传统特征提取算法,略低于KPCA特征提取算法,且在常用雷达探测距离内都能够保持较高的识别率。

Research high resolution radar signal feature extraction algorithm. The problems of information loss and low accuracy exist in traditional translation invariant features extraction algorithm, The paper provid.ed a translation invariant KPCA feature extraction algorithm. Firstly, this method calculated origin moment of high resolution radar signal, and then on the premise of low information loss, the position of high resolution radar signal relative to the origin moment was used to describe original signal, thereby eliminated the high resolution radar signal translation sensitivity. Then, the method combined KPCA feature extraction algorithm to get a translation invariant KPCA feature extraction algorithm. Finally, it applied SVM classifier to classify the high resolution radar signal. Experiments show that the recognition rate of the new algorithm is higher than the traditional one and slightly lower than KPCA feature extraction algorithm, which keeps a high recognition rate in the commonly radar detection range.