基于显著性图的红外与可见光图像融合  

Image Fusion of Infrared and Visible Images Based on Saliency Map

作　　者：林子慧 魏宇星[1] 张建林[1] 徐智勇[1] LIN Zihui;WEI Yuxing;ZHANG Jianlin;XU Zhiyong(Institute of Optics and Electronics of the Chinese Academy of Sciences,Chengdu 601209,China;University of Chinese Academy of Sciences,Beijing 100049,China)

机构地区：[1]中国科学院光电技术研究所,四川成都610209 [2]中国科学院大学,北京100049

出　　处：《红外技术》2019年第7期640-645,共6页Infrared Technology

摘　　要：为使红外图像与可见光图像融合后的图像能获得更多目标信息和细节信息,本文提出了一种基于显著性图的图像融合方法。使用改进的Frequency Tuned(FT)算法提取红外图像的显著性图,并使用对比度受限的自适应直方图均衡化(Contrast Limited Adaptive Histgram Equalization,CLAHE)算法增强可见光图像的对比度。将红外图像与增强后的可见光图像进行非下采样轮廓波变换(Nonsubsampled Contourlet Transform,NSCT)后,根据所设定的融合规则分别对红外与可见光图像的低频部分与高频部分进行融合,最后对融合系数进行NSCT 逆变换操作后获到融合图像。实验表明,该融合方法相较于其他方法而言,保留了更多的目标信息和细节信息,可以取得更好的视觉效果。To obtain more details and target information from fused infrared and visible images,an image fusion method based on saliency map is proposed.First,the frequency-tuned algorithm is improved to extract the saliency map of the infrared image,and the contrast of visible image is enhanced by contrast limited adaptive histogram equalization.Then,the infrared and enhanced visible images are decomposed by nonsubsampled contourlet transform,and the low and high frequency coefficients are fused according to the fusion rules.Finally,the fusion image is obtained by the inverse transformation of the low and high frequency fusion coefficients.Experimental results show that compared to the other methods,this fusion method retains more details and target information with good visual effects.

关 键 词：图像融合 显著性图 NSCT 红外图像 可见光图像 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]