期刊
INFRARED PHYSICS & TECHNOLOGY
卷 89, 期 -, 页码 8-19出版社
ELSEVIER SCIENCE BV
DOI: 10.1016/j.infrared.2017.12.003
关键词
Image fusion; Gradient domain guided image filtering; Multiscale decomposition; Infrared and visible imaging; Visual saliency
资金
- National Natural Science Foundation of China [61231014]
- Foundation of Army Armaments Department of China [6140414050327]
- Foundation of Science and Technology on Low-Light-Level Night Vision Laboratory [BJ2017001]
For better surveillance with infrared and visible imaging, a novel hybrid multiscale decomposition fusion method using gradient domain guided image filtering (HMSD-GDGF) is proposed in this study. In this method, hybrid multiscale decomposition with guided image filtering and gradient domain guided image filtering of source images are first applied before the weight maps of each scale are obtained using a saliency detection technology and filtering means with three different fusion rules at different scales. The three types of fusion rules are for small-scale detail level, large-scale detail level, and base level. Finally, the target becomes more salient and can be more easily detected in the fusion result, with the detail information of the scene being fully displayed. After analyzing the experimental comparisons with state-of-the-art fusion methods, the HMSD-GDGF method has obvious advantages in fidelity of salient information (including structural similarity, brightness, and contrast), preservation of edge features, and human visual perception. Therefore, visual effects can be improved by using the proposed HMSD-GDGF method. (C) 2017 Elsevier B.V. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据