红外与可见光图像匹配共性特征增强的新方法

New method for common factor enhancement in IR/visible image registration

分析了红外与可见光梯度图像的特点,将梯度图像中的可用信息分为低频分量、弱边缘和强边缘三类,通过梯度图像直方图自适应地求取图像增强的非线性变换参数,同时利用区域主方向和梯度方向抑制噪声的增强,采用梯度方向加权的方法增强弱边缘特征。实验结果表明,算法能够提高红外与可见光梯度图像的协方差相关值,增强了图像间对应边缘的一致性,且对噪声有较好的抑制作用。

The gradient features of IR/visible image are analyzedt,he available information of gradient image is divided into low frequency part,weak edges and strong edges.Transformation parameters of nonlinear enhancement function are calculated automatically from the histogram of gradient imagea,nd the main direction of small area and gradient direction of pixels are used to restrain noise enhancement,weak edges are enhanced by weight of the gradient directions.Experiment results show that the cross-correlation value of gradient images is increased,and the algorithm is also robust to noise.