高级检索
张淑芳, 刘孟娅, 韩泽欣, 郭志鹏. 基于Retinex增强的单幅LDR图像生成HDR图像方法[J]. 计算机辅助设计与图形学学报, 2018, 30(6): 1015-1022. DOI: 10.3724/SP.J.1089.2018.16598
引用本文: 张淑芳, 刘孟娅, 韩泽欣, 郭志鹏. 基于Retinex增强的单幅LDR图像生成HDR图像方法[J]. 计算机辅助设计与图形学学报, 2018, 30(6): 1015-1022. DOI: 10.3724/SP.J.1089.2018.16598
Zhang Shufang, Liu Mengya, Han Zexin, Guo Zhipeng. Generation Method of High Dynamic Range Image from a Single Low Dynamic Range Image Based on Retinex Enhancement[J]. Journal of Computer-Aided Design & Computer Graphics, 2018, 30(6): 1015-1022. DOI: 10.3724/SP.J.1089.2018.16598
Citation: Zhang Shufang, Liu Mengya, Han Zexin, Guo Zhipeng. Generation Method of High Dynamic Range Image from a Single Low Dynamic Range Image Based on Retinex Enhancement[J]. Journal of Computer-Aided Design & Computer Graphics, 2018, 30(6): 1015-1022. DOI: 10.3724/SP.J.1089.2018.16598

基于Retinex增强的单幅LDR图像生成HDR图像方法

Generation Method of High Dynamic Range Image from a Single Low Dynamic Range Image Based on Retinex Enhancement

  • 摘要: 针对目前生成高动态范围(HDR)图像的方法对采集设备要求高且不适用于动态场景的问题,提出一种基于Retinex增强的单幅低动态范围(LDR)图像生成HDR图像方法.首先基于Retinex方法将单幅LDR图像映射生成多幅不同曝光的图像,然后结合图像的对比度、饱和度等4个测量因子计算各幅图像的权重,最后利用金字塔分解来融合生成HDR图像.在多幅图像上的实验结果表明,该方法生成的HDR图像纹理更清晰,更符合人眼视觉效果.

     

    Abstract: According to problems that current high dynamic range(HDR) image generation methods require high acquisition equipment and don't apply to the acquisition of dynamic scenes, a generation method of HDR image from a single LDR image based on Retinex enhancement is proposed. Firstly, a single LDR image is mapped to generate multiple different exposure images based on Retinex algorithm. Secondly, the weight of each image is computed by combining four measurement factors, such as contrast and saturation of the image. Finally, HDR image is generated by using the pyramid decomposition. Experimental results on multiple images show that the HDR image generated by the proposed method has a clearer texture and a better human visual effects.

     

/

返回文章
返回