【摘要】 针对红外图像和可见光图像的特性,提出了基于梯度信息和区域互信息的红外与可见光图像配准。为了提高配准的速度,采用分层配准方法:使用小波分解将原始图像分为两层,在小尺度层上使用梯度信息和PSO方法找到粗配准参数,大尺度层上使用区域互信息和Powell搜索。其中,Powell搜索的初始值为粗配准参数,最终实现红外图像和可见光图像的有效配准。仿真实验结果表明:该算法配准精度高、速度快、鲁棒性好。更多还原

【Abstract】 Aiming at the characteristics of infrared and visible images,an images registration method based on gradient information and regional mutual information is proposed in this paper. The hierarchical registration method is used in order to improve the registration speed. The original images are divided into two layers by wavelet decomposition. The coarse registration parameters are got by gradient information and PSO algorithm at the small scale layer. The fine parameters are found by using regional mutual information and Powell algorithm at the large scale layer,and the initial values of Powell algorithm are set as the coarse parameters. In the end the effective registration can be achieved. The experimental results show that the proposed method is of high accuracy,high speed and good robustness.更多还原