基于最小平均相关能量滤波器的畸变目标识别

通过对多种畸变目标识别算法的研究,对用于匹配滤波器的最小平均相关能量(MACE)滤波器进行了改进。利用MACE滤波器算法的基本思想构造滤波器函数,在合成参考图像前对训练图像进行边缘提取,并对联合图像的功率谱进行了拉普拉斯锐化,成功实现了在联合变换相关器中对旋转目标图像的探测和识别,提高了畸变图像相关峰的亮度,拓展了联合变换相关器的探测和识别的范围。作为实例,对旋转战机目标图像进行了计算机模拟实验和光学实验,验证了该算法的可行性。

Research of Distortion Target Recognition Based on Minimum Average Correlation Energy Filters

One of the bottleneck techniques of correlation pattern recognition is the accurate detection of distortion targets such as rotation and scale. Through researching on the variety algorithms of distortion target recognition, the minimum average correlation energy (MACE) filter used in matched filter was modified in this paper. Firstly, based on the basic idea of MACE filter, the filter function was constructed. Then, the edge extraction of training images was performed before composing reference images. Lastly, Laplace sharpening for the joint power spectrum of joint image was used. In this way, the rotated target image was recognized in joint transform correlator(JTC)successfully, and the correlation peak brightness of distortion images is increased, and the range of detection and recognition is improved. And also, as a practical example, the computer simulation experiments and optical experiment of a rotated aircraft target image were carried out, proving the feasibility of this algorithm.