基于曝光适度评价的多曝光图像融合方法  

Multi-exposure image fusion based on well-exposedness assessment

作　　者：江燊煜 陈阔[1] 徐之海[1] 冯华君[1] 李奇[1] 陈跃庭[1] 

机构地区：[1]浙江大学现代光学仪器国家重点实验室,浙江杭州310027

出　　处：《浙江大学学报（工学版）》2015年第3期470-475,481,共7页Journal of Zhejiang University：Engineering Science

基　　金：国家自然科学基金资助项目(61107009)

摘　　要：针对数字相机在拍摄宽亮度范围场景时动态范围不足的问题,提出一种高效的多曝光图像融合方法.该方法通过场景的多次采样曝光,分析不同像素的亮暗变化,利用高斯数学模型计算评价数值.将结合图像亮度信息的曝光适度评价方法应用于图像分块区域,并计算每个像素的权重值,消除图像块效应.将不同曝光度下成像的图像序列融合为一幅局部自适应曝光的高动态范围(HDR)图像,实现动态范围扩展.与另外3种方法进行对比分析,结果表明:该方法简便有效,细节纹理与颜色信息具有更强的表现能力,且熵值更高.Aiming at the problem that digital camera is insufficient in dynamic range when taking pictures of wide dynamic range scenes,a spatial-domain fusion method for multi-exposure images was proposed.Scenes were sampled and exposed repeatedly,and brightness changes of different pixels of the images were analyzed.Then,assessment value was calculated using the Gauss model.The well-exposedness assessment method was applied to sub-areas of the images.Weighted value of each pixel was calculated to eliminate the image＇s block effect.Image sequences that imaged under different exposures were fused into high dynamic range（HDR）images with local adaptive exposure,which extended the dynamic range.Compared with other methods,the proposed method is simple and effective,producing results with better performance in scene details and color information,and even with higher entropy.

关 键 词：高动态范围成像 多曝光图像 图像融合 曝光适度 

分 类 号：TN911[电子电信—通信与信息系统]