Karhunen Loeve变换用于光学相关识别旋转目标研究

提出利用Karhunen Loeve变换（K-L变换）识别旋转畸变目标的方法.利用K-L变换法对不同旋转目标进行处理得到不同特征值,取前两个最大的特征值所对应的特征向量分别进行滤波器编码,把得到的匹配滤波器分别加载到VLC相关器上进行光路识别.针对两个滤波器产生的相关峰值曲线波动性较大的缺点,采用线性叠加法进行改进.实验结果表明,该方法使峰值的波动性得到降低.

Research on Recognition of Rotating Targets in Optical Correlation Using Karhunen Loeve Transform

The K-L transform is used in optical correlation recognition to deal with the rotating objects and different eigenvalues are obtained. The two eigenvectors, corresponding to the two biggest eigenvalues, are coded and used in VLC correlation experiment. According to the wave property of the correlation peak, the linear superposition method is put forward to improve it. The experiment shows that the wave property falls by this method.