【摘要】 基于傅里叶光学中的相关器原理,制作匹配滤波器并做ln-θ坐标变换,从而有效地降低了光学图像特征识别的难度.识别过程中先对原始图像进行ln-θ坐标变换操作,以消除图像间相对缩放、旋转的影响;然后应用相关器原理,通过比较检测图像和参考图像的相关输出以及参考图像的自相关输出,判断检测图像和参考图像的相似程度.实验结果表明:该识别方法实现了较好的识别效果.更多还原

【Abstract】 Based on the correlator principle in Fourier optics,a matched filter was made and ln-θcoordinate transformation was performed,which effectively reduced the difficulty of optical feature recognition.In the process of recognition,ln-θcoordinate transformation was performed to eliminate the effects of relative rotation and scaling between images.Then by comparing the correlation output of inspection image and reference image and the self-correlation output of reference image,the degree of similarity between inspection image and reference image could be judged based on correlator principle.This image feature recognition method achieved satisfactory results in the experiments.更多还原