基于红外目标特征提取的图像融合方法

为有效突出红外目标，同时尽可能多地保留可见光图像中的纹理细节信息，使得最终的融合图像更符合人类视觉感知效果，本文提出一种新的基于红外目标特征提取的红外与可见光图像融合方法。首先，利用高斯滤波器将源图像分解为粗略尺度信息和边缘纹理细节信息；对红外图像的边缘纹理细节信息进行去“光晕”分解，在此基础上进一步利用OTSU多阈值分割算法将红外图像分割为目标区域、过渡区域和背景区域；最后，依据分割结果确定各分解子信息的融合权重，以有效地将红外目标信息注入到可见光图像中，同时尽可能多地保留可见光图像中重要的场景细节信息。实验结果表明，本文方法无论从主观视觉还是客观评价指标上，都要优于目前常用的有代表性的图像融合方法。 

Image fusion based on infrared object extraction

In order to effectively highlight the infrared objects, while preserving the texture details in the visible image as much as possible, so the final fused image is more consistent with human visual perception effect. In this paper, a new fusion method for the infrared and visible images based on infrared target feature extraction is proposed. Firstly, the source images are decomposed into the coarse-scale information and edge texture information by the Gaussian filter; then, the “halo” decomposition is conducted in the edge texture information, based on which, the OTSU multi-threshold segmentation algorithm is used to segment the infrared image into the object region, transition region and background region; finally, the fused weight of each decomposition information is determined according to the segmented results, so as to effectively inject the infrared target information into visible image, and retain the scene details of the visible image as much as possible. Experimental results show that the method can effectively enhance the infrared targets, and retain the visual information as much as possible, and is superior to the commonly-used representative fusion methods both in subjective and objective evaluation.