Image Fusion of Infrared and Visible Images Based on Residual Significance
-
摘要: 为了将可见光图像与红外图像中的细节信息更多的呈现在融合图像中,突出目标特征并获得更好的图像视觉效果,本文提出一种基于改进谱残差显著性图的红外与可见光图像融合方法。首先用改进的谱残差显著性检测算法提取红外图像的显著性图并获得融合图像的显著性系数,然后对源图像进行双树复小波分解,并根据特定的融合规则分别对图像的低频部分以及高频部分进行融合,最后采用双树复小波逆变换重构获得最终的融合图像。实验表明,本文融合方法相较于传统融合方法融合质量更高并且在视觉效果上有显著提升。Abstract: To make the fusion image show more image details and to obtain a better image visual effect, a fusion method based on residual significance is proposed. First, the infrared image is analyzed using residual significance to obtain its significance coefficients. Then, the source images are decomposed using a dual-tree complex wavelet transform, and the low- and high-frequency components are fused according to different fusion rules. Finally, the fusion image is reconstructed using the inverse transformation of a dual-tree complex wavelet. Experimental results showed that the fusion method proposed in this paper produced higher quality images and better visual effects than those of the traditional fusion method.
-
表 1 融合图像客观评价结果
Table 1. The objective evaluation results of fused images
Scene Fusion method IE SD SF Qc MI Scene 1 LP 5.2477 25.0859 6.4781 0.6714 2.6909 DWT 5.9016 20.5764 8.4659 0.7026 3.0138 GFNSCT 5.9431 25.5093 12.0113 0.7223 3.1173 Ours 5.9672 36.6538 12.6746 0.8148 3.1984 Scene 2 LP 7.2489 44.5980 10.5057 0.7089 2.1191 DWT 7.0574 40.5774 12.7919 0.6914 2.2907 GFNSCT 7.1928 42.3394 15.7901 0.7218 2.3410 Ours 7.2985 46.4514 15.9041 0.7901 2.3667 Scene 3 LP 6.5528 26.4718 12.2714 0.7551 3.0173 DWT 6.4503 25.7847 14.1124 0.8081 3.4493 GFNSCT 6.6167 32.1633 16.5527 0.8271 3.5117 Ours 7.0221 37.5441 16.9096 0.8650 3.6869 -
[1] 敬忠良, 肖刚, 李振华.图像融合——理论与应用[M].北京:高等教育出版社, 2007.JING Zhongliang, XIAO Gang, LI Zhenhua. Image Fusion—Theory and Application[M]. Beijing: Higher Education Press, 2007. [2] 刘信乐.热红外图像与可见光图像融合方法研究[D].成都: 电子科技大学, 2013.LIU Xinle. Research on Fusion Method of infrared image and visible image[D]. Chengdu: University of Electronic Science and Technology of China, 2013. [3] 余美晨, 孙玉秋, 王超.基于拉普拉斯金字塔的图像融合算法研究[J].长江大学学报:自然科学版, 2016, 13(34): 770-776. http://www.cnki.com.cn/Article/CJFDTOTAL-CJDL201634005.htmYU Meichen, SUN Yuqiu, WANG Chao. Image fusion algorithm based on Laplacian pyramid[J]. Journal of Yangtze University: Natural Science, 2016, 13(34): 770-776. http://www.cnki.com.cn/Article/CJFDTOTAL-CJDL201634005.htm [4] Ashish V Vanmali, Vikram M Gadre. Visible and NIR image fusion using weight-map-guided Laplacian–Gaussian pyramid for improving scene visibility[J]. Springer Nature, 2017(6): 1063-1082. [5] Gonzalo P, Jesus M A wavelet-based image fusion tutorial[J]. Pattern Recognition, 2004, 37(9): 1855-1872. [6] Seal Ayan, Bhattacharjee Debotosh, NasipuriMita. A trous wavelet transform based hybrid image fusion for face recognition using region classifiers[J]. Expert Systems, 2018(12): 2185-2188. [7] 王少杰, 潘晋孝, 陈平.基于双树复小波变换的图像融合[J].核电子学与探测技术, 2015, 35(7): 726-728.WANG Shaojie, PAN Jinxiao, CHEN Ping. Image fusion based on DT-CWT[J]. Nuclear Electronics & Detection Technology, 2015, 35(7): 726-728. [8] 林子慧, 魏宇星, 张建林, 等.基于显著性图的红外与可见光图像融合[J].红外技术, 2019, 41(7): 640-645. http://hwjs.nvir.cn/oa/DArticle.aspx?type=view&id=201810021LIN Zihui, WEI Yuxing, ZHANG Jianglin, et al. Image fusion of infrared image and visible image based on saliency map[J]. Infrared Technology, 2019, 41(7): 640-645. http://hwjs.nvir.cn/oa/DArticle.aspx?type=view&id=201810021 [9] TIAN Huawei, FANG Yuming. Salient region detection by fusing bottom-up and top-down features extracted from single image[J]. IEEE Transaction on Image Processing, 2014, 23(10): 4389-4397. doi: 10.1109/TIP.2014.2350914 [10] HOU X, ZHANG L. Salient detection: A spectral residual approach[C]//IEEE Conference on Computer Vison and Pattern Recognition, 2007: 18-23. [11] 郭玲, 杨斌.基于视觉显著性的红外与可见光图像融合[J].计算机科学, 2015, 42(6): 211-214. http://www.cnki.com.cn/Article/CJFDTotal-GXYQ201604005.htmGUO Ling, YANG Bin. Image fusion of infrared image and visible image based on visual saliency[J]. Computer Science, 2015, 42(6): 211-214. http://www.cnki.com.cn/Article/CJFDTotal-GXYQ201604005.htm [12] 张承鸿, 李范鸣, 吴滢跃.基于视觉显著性与对比度增强的红外图像融合[J].红外技术, 2017, 39(5): 421-426. http://hwjs.nvir.cn/oa/DArticle.aspx?type=view&id=201612035ZHANG Chenghong, LI Fanming, WU Yingyue. Image fusion of infrared image and visible image based onvisual saliency and contrast enhancement[J]. Infrared Technology, 2017, 39(5): 421-426. http://hwjs.nvir.cn/oa/DArticle.aspx?type=view&id=201612035 [13] ZHENG Y F, Essock E. A new metric based on extended spatial frequency and its application to DWT based fusion algorithms[J]. Information Fusion, 2007, 8(2): 177-192. doi: 10.1016/j.inffus.2005.04.003 [14] QU G, ZHANG D, YAN P. Information measure for performance of image fusion[J]. Electron Lett, 2002, 38(7): 313-315. doi: 10.1049/el:20020212