基于CWT自相似特性的飞机目标HRRP分类法

为了解决高分辨率距离像 (HRRP)分类的姿态敏感和距离像对齐问题 ,基于连续小波变换 (CWT)的自相似特性 ,构造了能量尺度分形维数。分别在 3个极化测量通道和两个尺度区间上 ,从 HRRP的 CWT能量随尺度的演化过程中提取了原始特征集。采用 Fisher可分性准则作判据 ,进一步确定了 3维最佳特征向量。并从理论和实验上证明了 ,能量尺度分形维数特征较不敏感于噪声和目标方位角的变化 ,同时不受目标径向位移和距离像平移的影响 ,是性能良好的飞机目标分类特征

Aircraft HRRP classification method based on self-similar characteristics of CWT

The problems of aspect sensitivity and profile alignment in high range resolution profile (HRRP) classification are addressed by analyzing the energy scale fractal dimension feature based on the self similar characteristic of the continuous wavelet transform (CWT). The original feature set is extracted from the HRRP's CWT energy profile in two scale intervals and three polarimetric channels. A 3 D optimum classification feature vector is then built according to Fisher's separability criterion. Theoretical and experimental results show that the novel fractal dimension features are insensitive to noise, target azimuth variation, or profile shift; therefore, the method holds promise for airplane HRRP classification.