基于NSST和NSGA-Ⅲ的红外和可见光图像融合
Infrared and Visible Image Fusion Based on NSST and NSGA-III
袁东方 1张静 2罗聪 1邢笑雪1
作者信息
- 1. 长春大学(吉林 长春 130022)
- 2. 吉林省特种设备检验研究院(吉林省特种设备事故调查处理服务中心)
- 折叠
摘要
针对图像融合方法容易丢失源图像结构信息的问题,该文提出一种基于非下采样剪切波变换(Non-Subsample Shearlet Transform,NSST)和第三代非支配排序遗传算法(Non-dominated Sorting Genetic Algorithm Ⅲ,NSGA-Ⅲ)的红外和可见光图像融合方法.首先,将源图像进行NSST变换得到低频和高频子带图像;其次,针对低频图像,设计了一种新的基于NSGA-Ⅲ的融合规则,先使用区域能量法对图像进行预融合,以预融合图像作为参考图像,再通过NSGA-Ⅲ算法进行优化获得融合低频图像;针对高频图像,用绝对值取大规则.最后,利用NSST逆变换重构融合图像.算法性能在TNO数据集上进行验证,并与其他先进的融合算法进行比较,定性与定量的结果均表明,提出的方法目标突出,并保留了源图像更多的结构信息.
Abstract
To address the challenge of preserving structural information in image fusion techniques,a novel fusion method based on Non-Subsampled Shearlet Transform(NSST)and Non-dominated Sorting Genetic Algorithm Ⅲ(NSGA-Ⅲ)for infrared(IR)and visible(VIS)images is proposed.Firstly,the source images are decomposed into low-frequency and high-frequency sub-band images by using NSST.Secondly,image fusion is performed on these sub-images.Specifically,for the low-frequency images,a unique fusion rule based on NSGA-Ⅲ is introduced.It initially applies the regional energy fusion rule for pre-fusing the image.Using the pre-fusion image as a reference,an optimization process based on NSGA-Ⅲ algorithm is then carried out to obtain the final fused low-frequency image.Thirdly,the maxi-mum absolute value fusion rule is applied to merge high-frequency sub-bands.Finally,the fused image is reconstructed by using the inverse NSST.The performance of this proposed algorithm is rigorously validated on the TNO data set.Qualitative and quantitative results demonstrate that compared with other state-of-the-art fusion methods,the presented method can effectively enhance target visibility while simultaneously preserve the structural information.
关键词
红外和可见光图像融合/预融合/NSGA-Ⅲ算法/NSSTKey words
infrared and visible image fusion/pre-fusion/NSGA-III algorithm/NSST引用本文复制引用
基金项目
吉林省科技厅项目(20220101133JC)
出版年
2024
NETL
NSTL
万方数据