基于Choquet积分和LP的图像融合方法  

Image Fusion Method Based on Choquet Integral and LP

作　　者：拓浩男 史国川 王荣昌 TUO Haonan;SHI Guochuan;WANG Rongchang(PLA Army Artillery Air Defense Academy,Hefei 230031;Key Laboratory of Polarized Light Imaging Detection Technology of Anhui Province,Hefei 230031)

机构地区：[1]中国人民解放军陆军炮兵防空兵学院,合肥230031 [2]偏振光成像探测技术安徽省重点实验室,合肥230031

出　　处：《计算机与数字工程》2022年第7期1499-1501,1515,共4页Computer & Digital Engineering

摘　　要：针对长波红外偏振图像特征,论文提出了一种基于Choquet模糊积分和拉普拉斯金字塔变换的红外偏振图像融合方法。首先利用Choquet积分选择经偏振解析后的最佳偏振参量图像,然后通过拉普拉斯金字塔分解红外光强图像和被选取偏振参量图像,对顶层采用区域平均梯度最大,对I层采用区域能量选取处理方法,再经过拉普拉斯金字塔反变换得到最终融合图像。实验结果表明,论文提出的融合算法在凸显参加融合图像的共有信息同时保留它们的特有信息,且提高了融合图像的细节信息,利于后期做目标检测的研究。A method of infrared polarization image fusion based on Choquet fuzzy integral and Laplace pyramid transform is proposed.Firstly Choquet integral image is used to choose the optimal polarization parameters after polarization parsing,then through the Laplacian pyramid decomposition strong infrared images,the images are selected and polarization parameters adopted for the top largest regional average gradient,the regional energy selection process is used for I layer,through the Laplace inverse transform to get the final fusion image pyramid.Experimental results show that the fusion algorithm proposed in this paper highlights the common information of the fused images while retaining their unique information,and improves the detailed information of the fused images,which is conducive to the research of target detection in the later stage.

关 键 词：CHOQUET模糊积分 拉普拉斯金字塔 红外偏振 区域平均梯度 区域能量选取 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]