基于直方图与图像分块融合的阶调映射算法  

Histogram based tone mapping algorithm using image segmentation and fusion

作　　者：蒋昊[1] 徐海松[1] JIANG Hao;XU Hai-song(State Key Laboratory of Modern Optical Instrumentation,College of Optical Science and Engineering,Zhejiang University,Hangzhou 310027,China)

机构地区：[1]浙江大学光电科学与工程学院,现代光学仪器国家重点实验室,浙江杭州310027

出　　处：《浙江大学学报（工学版）》2022年第11期2224-2231,共8页Journal of Zhejiang University：Engineering Science

摘　　要：阶调映射算法(TMO)旨在将高动态范围(HDR)图像以最符合人眼感知的形式复现于低动态范围显示设备上.为此基于图像分块与融合,提出一种直方图调整的阶调映射算法.高动态范围图像被分为若干非重合的长方形区域,在每一块区域中阶调映射问题被视为基于直方图的K均值聚类问题并且建立相应的求解目标函数.各区域中解得的映射函数根据该区域的均匀性进行调整,以避免对比度增强过度,从而减少伪影.最后,提出一种双边滤波形式的图像融合策略以保证区域边界的平滑性,兼顾位置及亮度差,进而提升映射结果的自然性.在实验中,采用阶调映射图像质量指标来对算法进行客观评估,结果表明所提算法相对于经典的阶调映射方法有更好的阶调映射效果,在细节增强与全局外貌保留间达到平衡,并在局部阶调映射算法中具有较好的运算效率.The tone mapping operators(TMO)aims at reproducing the visual perception of high dynamic range(HDR)images on the low dynamic range media.A histogram based tone mapping algorithm was proposed to use image segmentation and fusion.The HDR images were segmented into non-overlapping rectangle blocks.In each block,the tone mapping problem was regarded as the K-means clustering problem based on the image histogram with a proposed loss function.The derived mapping function was adaptively adjusted considering the uniformity of the block.The detail-enhancing effect of the mapping function was then restrained by the block uniformity for less arising artefacts.An image fusion scheme was presented to deal with the artefacts on the segmentation boundaries.With the form of a bilateral filter,the fusion scheme took into account both the difference of position and luminance,ensuring the naturalness of the tone-mapped results.The proposed algorithm was evaluated to use the tone-mapped image quality index and the results illustrated a better image quality compared with the typical tone mapping methods.A balance between the detail enhancement and the global appearance preserving has also been achieved with an impressed execution efficiency among local TMOs.

关 键 词：图像处理 阶调映射 高动态范围图像 直方图调整 图像融合 K均值聚类 

分 类 号：TN27[电子电信—物理电子学] TP751[自动化与计算机技术—检测技术与自动化装置]