畸变不变联合变换相关器目标识别技术研究

针对传统目标识别方法对目标图像的尺寸、方位敏感问题,采取最大平均相关高度功率谱滤波的方法,对联合变换功率谱面进行滤波,成功地实现并改进了原算法.通过大量计算机模拟和光学实验,证实了在光电混合实时联合变换相关器中应用改进的最大平均相关高度法可以解决存在旋转和大小变化的目标识别问题,使得相关峰的对比度明显增强,从而也提高了目标识别率.作为实例,给出复杂背景下地面目标汽车大小变化的实验结果,验证了该算法的可行性.

Research on distortion invariant target recognition technology in joint transform correlator

Traditional target recognition methods are sensitive to scale and rotation variation of target images, the algorithm of maximum average correlation height (MACH) is adopted and modified successfully by filtering the joint transform power spectrum plane. Many computer simulation and experimental results show that the problems of target recognition with rotation variation and scale variation can be resolved by using the modified MACH algorithm in hybrid optoelectronic real-time joint transform correlator. The contrast of correlation peak is enhanced greatly, and the target recognition ratio is also increased. As an example, the experimental result of the scale variant of a ground target-automobile under cluttered background is presented. It proves that the modified MACH algorithm is feasible.