THT机制耦合高斯模糊的图像融合方案  

An Image Fusion Algorithm Based on THT Mechanism Coupled with Gaussian Fuzzy Logic

作　　者：张炜[1] 

机构地区：[1]安阳工学院计算机科学与信息工程学院,河南安阳455000

出　　处：《西南大学学报（自然科学版）》2017年第12期143-151,共9页Journal of Southwest University(Natural Science Edition)

基　　金：国家自然科学基金项目(U1204613);河南科技攻关计划重点项目(122102210138)

摘　　要：为了提高当前红外(IR)与可见光(VI)图像的融合质量,保持良好的可视化细节和突出IR目标,有效减少模糊,降低信息冗余,通过改进THT(Top-Hat Transform)机制,设计了一种新的红外-可见光图像融合算法.首先,为了有效地提取源图像的特征和细节,在传统的高帽变换(Top-Hat Transform,THT)中引入了多尺度计算,分别在不同尺度上提取红外和可见光图像的不同亮(暗)特征区域;其次,通过多判别对比融合规则来综合两个图像中的同一尺度的特征区域,将所有尺度上的特征进行累加,获取特征图像;然后,通过最大尺度结构元素的开、闭运算得到源图像的平滑明亮、暗图像,在平滑图像上利用Gaussian模糊逻辑融合规则获取基础图像;最后,将提取的亮与暗特征图像导入基础图像中形成新图像,输出图像融合.实验表明:与当前流行的融合方法相比,本文方法具有更好的视觉质量和定量分析结果.In order to improve the fusion performance of the current infrared（IR）and visible（VI）images,maintain good visual details and highlight IR goals,effectively reduce fuzziness and reduce information redundancy,a new top hat transform coupled Gaussian fuzzy logic infrared-visible fusion scheme was designed.Firstly,multi-scale calculation was introduced into the traditional top hat transform（THT）,and different brightness（dark）feature regions of infrared and visible images were extracted at different scales so as to effectively extract the features and details of the source image.Next,through the combination of multiple judgment and contrast fusion rules,the feature regions of the two images in the same scale were combined,the features were accumulated at all scales and the feature images were obtained.Then,smooth,bright and dark images of the source image were obtained by means of the open and close operations of the largest scale structuring elements and,on the smooth image,the basic image was acquired by the Gaussian fuzzy logic fusion rule.Finally,the extracted bright and dark feature images were imported into the base image to form a new image.Experimental results showed that compared with current popular fusion methods,the method reported herein had better visual and quantitative analysis results.

关 键 词：图像融合 改进高帽变换 特征区域 基础图像 均值加权 Gaussian模糊逻辑 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]