功率谱处理技术在图像相关识别中的应用

The Application of the Power Spectrum Processing Methods on Image Correlation Recognition

经典的联合变换相关(JTC)在相关面上存在较强和较宽的自相关峰,而用以判断识别的互相关信号本身强度较弱且不够尖锐,其互相关峰信噪比也相对较低。为了提高JTC相关识别性能,本文从图像的频域处理入手,通过对功率谱的放大处理来提高互相关峰的强度,用高通滤波器来强化图像高频部分,提高互相关峰的锐度。仿真计算结果表明,对功率谱放大后再进行高通滤波处理,两种方法的结合不仅增强了互相关信号的强度,而且使互相关信号的信噪比得到明显提高,从而进一步优化了联合变换相关系统的识别性能。

The auto-correlation peak is stronger and wider than the cross-correlation peak on the correlation plane of the classical joint transform correlation (JTC), and the cross-correlation peak which is the signs for the judgment of image correlation recognition is weak and not sharp, and that the Signal to Noise Ratio (SNR) of it is lower also. In order to improve the property of the joint transform correlation, starting from the image frequency domain in the paper, the power spectrum amplification is used to enhance the intensity of the cross-correlation peak and the high-pass filter is used to enhance high-frequency signals of images to sharpen the cross-correlation peak. The simulation computer results show that the processing method of the power spectrum that is amplified and then filtered by the high-pass filter is feasible. The method not only enhances the intensity of the cross-correlation peak signals, but also improves the SNR of the cross-correlation signals substantially, so the property of the joint transform correlation are optimized further.