基于视觉显著性与对比度增强的红外图像融合!.pdfVIP

第39 卷 第5 期 红 外 技 术 Vol.39 No.5 2017 年5 月 Infrared Technology May 2017 基于视觉显著性与对比度增强的红外图像融合 张承泓，李范鸣，吴滢跃 （1.中国科学院红外探测与成像技术重点实验室，上海技术物理研究所，上海 200083 ；2.中国科学院大学，北京 100049 ） 摘要：目前传统多尺度分析红外图像融合算法存在以下不足：融合图像的对比度改善效果有限，无法 获取图像的某些细节信息；融合规则仅考虑单一特征，故未能突出目标特征。针对以上问题，本文提 出一种基于视觉显著性与对比度增强的红外图像融合算法。首先对待融合的图像进行基于自适应引导 滤波的多尺度Retinex 图像增强，然后利用NSCT 对图像进行多尺度分解，最后利用图像视觉显著性 融合低频系数，采用基于窗口的系数融合带通系数。实验证明，该算法获得的红外融合图像效果明显 优于传统方法。 关键词：图像增强；图像融合；视觉显著性；自适应引导滤波；NSCT ；多尺度Retinex 中图分类号：TP391 文献标识码：A 文章编号：1001-8891(2017)05-0421-06 Infrared Image Fusion Based on Visual Saliency and Contrast Enhancement 1,2 1 1 ZHANG Chenghong ，LI Fanming ，WU Yingyue (1.CAS Key Laboratory of Infrared System Detection and Imaging Technology, Shanghai Institute of Technical Physics, Shanghai 200083, China; 2. University of Chinese Academy of Sciences, Beijing 100049, China) Abstract ：Traditional infrared image fusion by multi-scale analysis has some disadvantage: the fused image cannot improve contrast, some details of the image cannot be caught; Fusion rules only depend on single feature, causing that the key target features in the scene of the fused image cannot be highlighted. To solve the above problems, in this paper, an infrared image fusion algorithm based on visual saliency and contrast enhancement is proposed. Firstly, the image is enhanced by using the algorithm of multi-scale Retinex based on adaptive guided filter; then, the image is decomposed by NSCT; finally, we use visual saliency to guide fusing the low frequency coefficient, using coefficient based on window to fuse