细节保留的多曝光图像融合

针对传统的多曝光图像融合算法存在的细节丟失严重和鬼影现象,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。

Detail preserving multi-exposure image fusion

A detail preserving multi-exposure image fusion algorithm was proposed to address the problem of the loss of visual details and ghost artifacts in traditional multi-exposure images. Firstly, three image features, image details, exposure brightness and color information,were calculated. In which,the image details were obtained by using a guided filter, the each exposure intensity was weighted by a Gaussian function and the color information was measured by color saturation. Then, the difference maps and correlation coefficients were used to detect the motion objects in dynamic scenes and the focused weight map of static and dynamic scenes were calculated respectively by using feature indexes and detection results. In order to remove the noise effect, a recursive filter was used to correct the focused weight image, and the focused image was obtained by a pixel-by-pixel weighted sum of the input images. Ten kinds of multi-exposure image sequences were tested in the experiments and obtained results were compared with that of six kinds of traditional methods. The experimental results demonstrate that the proposed algorithm exhibits good visual appearance and preserves more details. It also effectively removes ghost artifacts in dynamic scenes. It concludes that the proposed algorithm is better than 6 classical methods and it produces desirable images in both static scenes and dynamic scenes.