针对光电联合变换相关器目标识别的实际应用，对待测红外目标图片进行多小波变换，并利用模极大值法提取其边缘.通过获取更多的轮廓信息，从而提高对目标的识别能力.计算机模拟了常用于红外目标处理的多小波GHM和SA4,实验结果表明:基于GHM多小波提取的边缘能获取大量的图像轮廓信息，其识别结果明显优于SA4多小波.将目标原图的光学相关探测结果与基于GHM多小波提取的边缘图像光学相关探测结果进行比较发现，经多小波预处理后的边缘图像能有效增强相关峰强度.

Considering the practical application of target recognition with optoelectronic joint transform correlator, the infrared image detected was processed by multiwavelet transform, and its edge points were extracted by the method of local modulus maximum. Capturing much contour can improve the capacity of target recognition. The multiwavelet functions usually used to infrared target processing are GHM and SA4. Simulation experiments show that the correlation output of edges extracted by GHM is obviously better than that one of edges extracted by SA4, because the former edges carry much contour of image. Compared to the optical correlation detection result of original image with that of edge image extracted by GHM, it is found that the latter effectively enhances the intensity of correlation peaks.