高动态范围图像分层阶调映射改进算法  

Improved Layered Tone Mapping Algorithm of High Dynamic Range Image

作　　者：王琪[1] 位春傲[1] 谢德红[1] 李蕊[1] 

机构地区：[1]南京林业大学,南京210037

出　　处：《包装工程》2015年第7期108-112,共5页Packaging Engineering

基　　金：国家级大学生创新创业训练计划(201310298043);江苏省高等学校大学生实践创新训练计划省级重点项目(201310298043Z)

摘　　要：目的针对高动态范围图像分层阶调映射算法中显著边缘处产生的光晕和阶调逆转问题,提出了一种分层阶调映射的改进算法。方法首先通过对图像分层后的基础层图像和细节层图像进行分析,确认导致光晕和阶调逆转的异常信号特征;然后依据其异常信号特征,从细节层图像梯度的角度予以识别、修正,再在修正梯度的基础上,建立对应图像的正规化方程;最后利用共轭梯度算法求解正规化方程。结果抑制了细节层图像的异常信号,去除了映射结果图像中的光晕和阶调逆转。结论该算法不仅可以有效地抑制原分层映射算法中的光晕和阶调逆转现象,还可保有与原分层映射算法同等的细节再现能力。Targeting at the halo or gradient reversal artifacts appearing on salient edges in the layered tone mapping algorithm of high dynamic range images, an improved layered tone mapping algorithm was proposed. In this algorithm,firstly the abnormal signal characteristics causing halo and gradient reversal artifacts were determined by analyzing the base layer image and detailed layer image obtained via image decomposition. Then the artifacts were recognized and corrected from the gradient of detailed layer image according to the abnormal signal characteristics, and the regularization equation was subsequently established for the corresponding image based on the corrected gradient. At last, conjugate gradient method was adopted to solve the regularization equation. The abnormal signals in detailed layer image were inhibited, and halo or gradient reversal was removed from the resulting mapping image. The proposed algorithm could not only effectively inhibit halo and gradient reversal artifacts in the original layered tone mapping algorithm, but also preserve the detail reproduction capacity as the original algorithm did.

关 键 词：高动态范围图像 分层映射 光晕 阶调逆转 梯度修正 

分 类 号：TS801.3[轻工技术与工程]