基于RGF和改进自适应Unit-Linking PCNN的红外与可见光图像融合

为了进一步提高融合图像的对比度和清晰度,提 出一种基于RGF(Rolling Guidance Filter)和改进自适应Unit-Linking PCNN(Unit-Lin king Pulse Coupled Neural Net work)的红外与可见光图像融合方法。首先,源图像利用RGF和高斯滤波器进行多尺度分解。 然后,针对基础层通过计算其最大区域能量,提出了将最大区域能量与源图像相结合的融合 规则；针对细节层利用改进的自适应Unit-Linking PCNN进行处理,得到相应的神经元点火 频率图,计算每个像素点火频率图的边缘特征,并选择两者边缘特征较大的系数作为融合图 像系数。最后,利用多尺度重构融合图像。实验结果表明,本文融合算法能较好地突出图像 的目标信息,提供丰富的背景细节,在图像的清晰度和人眼视觉效果方面取得较好的融合效 果。

Infrared and visible image fusion based on RGF and improved adaptive Unit-Linki ng PCNN

In order to further improve the contrast and sharpness of the fuse d images,a method based on Rolling Guidance Filter and improved adaptive Unit- Linking Pulse Coupled Neural Network for infrared and visible images fusion is p r oposed in this paper.Firstly,the multi-scale decomposition of the source imag e s are carried out by RGF and Gaussian filter.Then,according to the maximum reg ional energy of the basic layer,a fusion rule combining the maximum regional en ergy with the source images is proposed.For detail layers,the corresponding ne uron ignition frequency map is obtained by using improved adaptive Unit-linking Pulse Coupled Neural Network,the edge features of ignition frequency map for ea ch pixel is calculated,and the larger coefficient of the two edge features is s elected as the fusion image coefficient.Finally,the fusion image is reconstruc ted by multi-scale.Experimental results show that the fusion algorithm in this paper can better highlight the target information of the images,provide rich ba ckground details,and achieve good fusion effects in terms of image clarity and visual effect of human eye.