细节保留的多曝光图像融合

发布时间：2018-02-17 01:15

本文关键词： 多曝光图像图像融合图像细节引导滤波鬼影　出处：《光学精密工程》2016年09期 　论文类型：期刊论文

【摘要】：针对传统的多曝光图像融合算法存在的细节N失严重和鬼影现象,提出了一种细节保留的多曝光图像融合算法。该算法首先计算曝光序列的3个特征指标:图像细节、曝光亮度和色彩信息,其中图像细节通过引导滤波计算,曝光亮度的权值由高斯方程分配,而曝光序列的色彩信息用色彩饱和度表示。然后,利用差分图和邻域相关系数检测多曝光序列中运动物体,利用3个特征指标和运动目标检测结果分别计算静态场景和动态场景的融合权值图。为了消除噪声的影响,采用递归滤波器来修正融合权值图。最后,采取加权融合的方式得到融合图像。选取10组不同的曝光序列,分别从主观和客观两方面与6种传统的融合算法进行了比较。实验结果表明,本文算法保留了丰富的细节信息,呈现出了更加生动自然的现实场景,并且有效去除了由运动物体产生的鬼影现象,效果优于其他比较算法,在静态场景和动态场景的曝光融合中都取得了好的效果。
[Abstract]:In view of the serious loss of detail N and ghost phenomenon in the traditional multi-exposure image fusion algorithm, a new multi-exposure image fusion algorithm with preserving details is proposed, which first calculates three characteristic indexes of the exposure sequence: image detail. Exposure brightness and color information, in which the details of the image are calculated by guided filtering, the weight of the exposure brightness is distributed by Gao Si's equation, and the color information of the exposure sequence is represented by color saturation. The moving objects in multi-exposure sequences are detected by differential map and neighborhood correlation coefficient. The fusion weights of static scene and dynamic scene are calculated by using three characteristic indexes and the result of moving target detection, respectively. The fusion weight map is modified by recursive filter. Finally, the fusion image is obtained by weighted fusion. 10 groups of different exposure sequences are selected. Compared with six traditional fusion algorithms in subjective and objective aspects, the experimental results show that the proposed algorithm retains abundant detail information and presents a more vivid and natural reality scene. And the ghost phenomenon produced by moving objects is removed effectively, and the effect is better than other comparison algorithms, and good results are obtained in the exposure fusion of static and dynamic scenes.
【作者单位】：武汉大学电子信息学院;湖北工程学院物理与电子信息工程学院;地球空间信息技术协同创新中心;
【基金】：国家自然科学基金资助项目(No.61471272)
【分类号】：TP391.41

【相似文献】