多目标识别的联合变换相关器的研究

提出了一种可用于多目标识别的联合变换相关器。为改善相关信号的性能,对功率谱作了优化处理。为消除相关面上的零级项和目标间的相关项,可用联合功率谱减去纯目标输入的功率谱和参考图像的功率谱;为增强和锐化相关峰,将相减的功率谱作指数函数滤波处理。分析了指数滤波参数对相关结果的影响。计算机模拟结果表明,这种相关器所输出的相关信号比经典联合变换相关器和二元联合变换相关器输出的相关信号更好,互相关得到了抑制,自相关得到了增强,具有很好的抗噪能力。

Multi-object recognition using power spectrum optimized joint transform correlator

A new joint transform correlator for multi-object recognition is proposed.For improving the correlation capability,the joint power spectrum was optimized in this correlator.The joint power spectrum was subtracted by the power spectrum of the object image and the power spectrum of the reference image respectively for eliminating the strong dc component and cross-correlations between each input objects.The subtracted power spectrum was filter with exponential filter for enhancing and sharpening the correlation peak and reducing the cross-correlations between the reference and input objects.The exponential filter parameter is discussed.Simulation results show that the correlation output of the new proposed joint transform correlator is better than the common joint transform correlator and binary joint transform correlator.The anti-noise ability is also well.