基于细节特征的高动态范围图像获取算法

High dynamic range image acquisition based on detail features

为了解决真实场景的动态范围与数字图像的动态范围不匹配的问题,提出一种基于细节特征合成的高动态范围图像获取算法。该方法首先提取多幅不同曝光的低动态范围图像的细节特征,细节特征值越大表明所含细节信息越多,然后将归一化的特征值作为权重,合成多幅低动态范围图像,得到一幅高质量的合成图。实验表明,该算法能得到一幅含有较丰富细节特征、动态范围较广的图像,合成图像不需要色调映射就可以较好地在低动态范围显示器上直接显示。

In order to eliminate the gap between the dynamic range of the real scene and that of digital images,an acquisition algorithm based on detail features is proposed.Firstly,the detail features of low dynamic range images in different exposures in terms of detail feature are extracted.It is obvious that large detail feature indicates more details.Secondly,the normalization of local feature are served as weights in image fusion.Finally,a high quality image is generated.Experiments show that the proposed method can provide fusion results with richer details and wider dynamic range.Moreover,the high dynamic range image acquired can be clearly shown on low dynamic range displaying device without tone-mapping work.