改进NSDTCT与粒子群优化红外和可见光图像融合算法  

Improved fusion algorithm of infrared and visible images based on NSDTCT and particle swarm optimization

作　　者：杨友良 王梓任 马翠红 YANG Youliang;WANG Ziren;MA Cuihong(College of Electrical Engineering,North China University of Science and Technology,Tangshan Hebei 063210,China)

机构地区：[1]华北理工大学电气工程学院,河北唐山063210

出　　处：《激光杂志》2020年第6期67-71,共5页Laser Journal

基　　金：国家自然科学基金资助项目(No.61171058)。

摘　　要：红外图像在光照不佳情况时因为存在模糊不清的问题,导致图像识别容易出现问题。因此,提出了一种基于非下采样双树复轮廓波与粒子群结合图像融合算法。该算法由于采用了NSDTCT方法,会把图像分成高频与低频两个部分,之后对低频部分采取粒子群算法得到图像块,为了能减少误差,同时采用标签图。将高频部分的邻域系数差与标签图融合,可以得到融合产生的图像。通过实验对比发现本文算法能够减少时间消耗,并一定程度上优化了图像。Infrared images are prone to problems in image recognition because of the ambiguity which leads to poor illumination.Therefore,an image fusion algorithm based on non-sub-sampled double-tree complex contour wave and particle swarm optimization is proposed.Due to the NSDTCT method,the algorithm divides the image into two parts:high frequency and low frequency.Then,the particle swarm algorithm is used to obtain the image block for the low frequency part.In order to reduce the errors,the label map is used.By blending the neighborhood coefficient difference of high frequency portion with the label map,an image produced by the fusion can be obtained.Through experimental comparison,the algorithm can reduce the time consumption and optimize the image to some extent.

关 键 词：红外图像 粒子群算法 非下采样双树复轮廓波 图像融合 

分 类 号：TN219[电子电信—物理电子学]