红外和可见光图像高效融合的人工智能技术研究  

Research on swarm intelligence optimization method for efficient fusion of infrared and visible images

作　　者：廖宁 陈怡然 LIAO Ning;CHEN Yiran(College of Big Data and Artificial Intelligence,Chongqing Institute of Engineering,Chongqing 40050,China)

机构地区：[1]重庆工程学院大数据与人工智能学院,重庆400050

出　　处：《激光杂志》2022年第3期109-113,共5页Laser Journal

基　　金：重庆市自然基金项目(No.cstc2020jcyj-msxmX0666);重庆市教委科学技术研究项目(No.KJZD-K202001901);重庆市教委科学技术研究项目(No.KJZD-K201901902);重庆市教委科学技术研究项目(No.KJQN201801905)。

摘　　要：为提升红外和可见光图像融合后图像的清晰度和细节信息的丰富程度,提出了基于人工智能技术的红外和可见光图像融合方法。通过NSCT变换分别将红外图像和可见光图像分解成低频低通子带和高频带通子带两个部分,采用低频系数加权平均求均值的选择方法、系数值选大法与局部区域融合规则相结合的方法,完成低频低通子带的图像和高频带通子带图像的融合;采用基于群智能技术的布谷鸟算法优化NSCT方法的融合过程,实现红外和可见光图像的高效融合。实验结果显示:采用该方法进行图像融合的PSNR值和MSSIM值、信息熵、空间频率、平均梯度等评价指标的测试结果均得到了优化,且收敛率均为100%,可有效提升融合后图像的清晰度以及细节信息丰富程度。In order to improve clarity and richness of detail information of infrared and visible images after fusion, a method of infrared and visible images fusion based on swarm intelligence technology is proposed. Through NSCT transform, infrared image and visible light image are decomposed into low frequency low-pass subband and high frequency subband respectively. Combining with the low frequency coefficient of average weighted for average values method, the coefficient values chosen method and the method of combining local fusion rules method, the fusion of the low frequency of low-pass subband images and the high frequency subband images are completed;The Cuckoo algorithm based on swarm intelligence technology is used to optimize fusion process of NSCT to realize efficient fusion of infrared and visible images. Experimental results show that the PSNR value and MSSIM value, information entropy, spatial frequency, average gradient and other evaluation indexes of image fusion with this method are optimized, and the convergence rate is 100%, which can effectively improve the clarity and richness of detail information of fused images.

关 键 词：群智能技术 红外 可见光 图像融合 低频低通子带 高频带通子带 

分 类 号：TN209[电子电信—物理电子学]