基于局部结构和视觉显著特征的红外和可见光图像泊松重构融合算法

研究了一种基于泊松重构的红外和可见光图像融合算法,算法在梯度域内实现图像信息的融合,可有效避免传统空域和变换域方法在融合图像中由于局部亮度不一致而产生伪边缘。另外,提出的算法在源图像梯度融合时,同时考虑了图像的局部结构显著和视觉显著特征,能够在保留源图像更多细节的同时突出输入图像的视觉显著目标信息。通过与其他最新融合算法的对比实验结果显示本文算法获得的融合图像既有突出的红外目标,又有清晰的可见光背景细节,并且不会产生伪影和噪声,同时客观评价指标也有显著的优势。

Local Structural and Visual Saliency Based Infrared and Visible Image Fusion with Poisson Reconstruction

This paper presents an infrared and visible images fusion algorithm based on Poisson reconstruction.The algorithm achieves the fusion of image information in the gradient domain,which can effectively avoid the occurrence of false edges due to the local brightness inconsistency in the traditional methods.The proposed algorithm uses local structure calculated from gradients as the pixel activity measure which retains more details of the source image.Moreover,the visual saliency targets of the input images can be highlighted in the fused image due to the visual saliency features which are considered in the fusion scheme.Compared with the other latest infrared and visible image fusion algorithms,the experimental results show that the fusion images obtained by our algorithm demonstrate both highlighted infrared targets and clear visible light background details,and will not produce artifacts and noises.Moreover,the proposed method also performs better results in terms of higher objective evaluation metrics than other methods.