基于人眼视觉特性的红外/可见光图像突显融合方法  

Algorithm of Infrared/Visible Image Fusion with Targets Highlighted Based on Human Visual Characteristics

作　　者：王雅芬 黄猛 马天舒[2] 吕亚飞 苗晶 WANG Yafen;HUANG Meng;MA Tianshu;LYU Yafei;MIAO Jing(No.91977 Troops of PLA,Beijing 102249;Space Star Technology Co.,Ltd.,Beijing 100095)

机构地区：[1]中国人民解放军91977部队,北京102249 [2]航天恒星科技有限公司,北京100095

出　　处：《舰船电子工程》2023年第8期84-88,122,共6页Ship Electronic Engineering

摘　　要：在红外/可见光图像融合中,提出了一种基于视觉感知特性的热红外目标突显的方法。首先对红外图像进行增强并分割出目标区域。然后提取待突显图像的亮度分量并对目标区域对比度进行基于视觉光感响应特性的非线性增强处理,同时选择一种突显颜色对其进行着色,最后以红外图像为参考对待突显图像和着色后的图像进行色彩融合得到输出图像。同时,还提出了在利用简化的Itti视觉注意模型获取融合图像的显著性图基础上,用目标区域的平均显著性值对视觉突显效果进行定量描述和评价的方法。实验表明,论文的融合方法可以在充分反映目标与背景温差信息的同时,有效增强区域内部的细节,大大提高了对目标的目视搜索效率,而所提出的评价方法可以有效对目标的突显程度进行量化,并取得了与主观评价相一致的结论。A thermal targets highlighted method is proposed,based on human visual characteristics,in the domain of infrared/visible image fusion.Firstly,the infrared image is enhanced and the target region is segmented.Secondly,the luminance contrast of source image in this region is non-linear enhanced baced on the brightness-feeling curve.At the same time,a highlighting color is choosed to colour source image luminance component.Finally,the result image is obtained by fusing source and coloured image with the reference of infrared image.This algorithm represents temperature differences between target and background,increases some details in target regions.So search efficiency of intersted targets is largely inproved.Besides,an objective evaluation based on visual attention model is introduced.After simplifing Itti's model to acquire saliency map of fused image,the average saliency value of target region is calculated.This method quantizes the highlighting level,and justifies the effects of our algorithm objectively.

关 键 词：视觉突显 图像融合 视觉注意 显著性评价 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]