基于目标区域与增强方法的红外与可见光图像融合

Fusing Infrared and Visible Images Based on Target Area and Enhancement Method

为了使红外图像与可见光图像融合较好凸显目标与挖掘更多细节信息,提出了一种提取目标区域与融入更多细节信息的融合方法。首先,对红外图像进行分割获取目标区域,并对可见光图像进行增强以挖掘更多细节信息;然后对原始红外图像与增强后的可见光图像分别进行非下采样contourlet变换(NSCT),得到不同的低频系数与高频系数,依据分割得到的二值化图像,低频部分的目标区域系数选自原始红外图像目标区域低频系数,其余区域选择增强后的可见光对应区域低频系数,高频部分按照邻域方差取大法选择高频系数;最后,进行NSCT反变换,得到融合图像。实验结果表明,与其他3种融合方法对比,主客观评价表明,该算法有效提高了图像的对比度,具有较好的整体视觉效果。

In order to make the fused image obtained by fusing infrared and visible images better highlight the target and mine more detail information,proposes is a method which combining extracting target regions with mining more detail.First,the infrared image is segmented to get target region and the visible image is intensified for mining more detail;then,the original infrared image and intensified visible light image are decomposed by nonsubsampled contourlet（NSCT）,getting the low frequency coefficients and the high frequency coefficients respectively,according to binarized image obtained from the original infrared image,for the low frequency coefficients,target region＇s low frequency coefficients are chosen from infrared image＇s target region,via other region＇s low frequency coefficients are chosen from the intensified visible light image;for the high frequency part,its coefficients are decided according to the principle of picking the neighborhood max value of variation;Finally,the NSCT inverse transform is carried out for the processed NSCT coefficients,the fusion image is obtained.The experimental results show that both objective and subjective evaluation prove that,compared with other three fusion methods,the proposed algorithm can improve the image contrast and own better visual effects.