摘要
多尺度曝光图像融合是一种有效的高动态范围(high-dynamic range,HDR)场景图像增强技术。提出了一种新的多尺度曝光图像融合算法,利用加权引导图像滤波器(weighted guided image filtering,WGIF)平滑所有低动态图像(low-dynamic range,LDR)的加权映射高斯金字塔,对不同曝光程度的LDR图像进行融合。该算法在不改变融合图像相对亮度的情况下,较好地保留了HDR图像场景中最亮和最暗区域的细节,避免了光晕伪影现象。此外,对传统的曝光图像融合算法中的曝光质量评价指标由像素级改为图像级,解决传统算法融合之后的图像亮度过于均匀而缺乏对比度的问题,增强图像亮暗区域的信息细节并减少图像失真。融合之后的图像可以兼顾过曝光图像和欠曝光图像中各自保留的细节信息,最大程度的还原真实场景的细节信息。
Multi-scale exposure image fusion is an effective high-dynamic range(HDR)scene image enhancement technique.In this paper,a novel multi-scale exposure image fusion algorithm is proposed which utilizes weighted guided image filtering(WGIF)to smooth all low-dynamic images range(LDR),weighted mapping Gaussian pyramid,fusion of LDR images with different exposure degrees.The algorithm preserves the details of the brightest and darkest regions in the HDR image scene without changing the relative brightness of the fusion image,and avoids the halo artifact phenomenon.In addition,the evaluation index of exposure quality in the traditional exposure image fusion algorithm is changed from pixel level to image level,which solves the problem that the image brightness after fusion is too uniform and lacks contrast,enhances the information details of the light and dark areas of the image and reduces the image distortion.After fusion,the image can take into account the details retained in the over-exposed image and under-exposed image to restore the details of the real scene to the greatest extent.
作者
史艳琼
张文亮
SHI Yan-qiong;ZHANG Wen-liang(School of Mechanical and Electrical Engineering,Anhui Jianzhu University,Hefei,Anhui 230601,China)
出处
《井冈山大学学报（自然科学版）》
2023年第6期93-100,共8页
Journal of Jinggangshan University (Natural Science)
基金
安徽省科技重大专项(202203a05020022)
安徽建筑大学校引进人才及博士启动基金项目(2019QDZ16)。
