基于色貌效应预测的HDR图像显示研究  

Research on HDR Image Rendering Based on Image Appearance Effect Predicting

作　　者：刘真[1] 朱明[2] 陈广学[3] 杨丹丹[2] 

机构地区：[1]上海理工大学出版印刷与艺术设计学院,上海市杨浦区军工路516号200093 [2]南京林业大学轻工科学与工程学院,南京市龙蟠路159号210037 [3]华南理工大学轻工与食品学院,广州市天河区五山路381号510000

出　　处：《武汉大学学报（信息科学版）》2011年第7期871-876,共6页Geomatics and Information Science of Wuhan University

基　　金：国家自然科学基金资助项目(60972134);江苏省自然科学基金资助项目(BK2006210);上海市教委"085"工程建设资助项目;江苏省高校研究生科研创新计划资助项目(CX09B_180Z)

摘　　要：在基于iCAM的HDR图像显示过程中加入亨特效应和史蒂文斯效应的预测模型。针对不同的评价目的(图像偏好性和场景再现准确性评价),通过实验确定了亨特效应和史蒂文斯效应预测模型中调制参数c和m的最佳取值,并将最终得到的两个完整的预测模型用于算法流程中。通过图像偏好性实验和场景再现准确性实验发现,改进后算法在图像偏好性上优于商业软件PhotoShop CS4中的4种算法和改进前的iCAM算法;在场景再现准确性上优于改进前的iCAM算法,并与PhotoShop的算法相当;最后通过离散快速傅立叶变换和降采样技术,提高了算法的计算速度。A new high-dynamic-range image rendering algorithm based on image appearance effect predicting was developed in this paper.The predicting models for ＂Hunt effect＂ and ＂Stevens effect＂ were integrated into HDR image rendering algorithm based on image color appearance model.Then,the best values for the modulating parameter ＂c＂ in the predicting model of ＂Hunt effect＂ and the modulating parameter ＂m＂ in the predicting model of ＂Stevens effect＂ were obtained respectively by the corresponding experiments aimed at the two kinds of purposes of evaluation（rendering preference evaluation and the evaluation for perceptual accuracy of reproducing the real-world scenes）.The final predicting models for ＂Stevens effect＂ and ＂Stevens effect＂ were integrated into the flow of algorithm.The performance of the algorithm improved was evaluated by experiments of rendering preference and perceptual accuracy.The result of rendering preference experiment shows the algorithm improved has significant better performance than four algorithms in PhotoShopCS4 and the iCAM algorithm unimproved.The result of perceptual accuracy experiment shows the algorithm improved has significant better performance than the iCAM algorithm unimproved,and has the approximately equivalent performance to PhotoShopCS4 algorithms.Finally,the computing speed of algorithm was improved by discrete FFT and down-sampling technology.

关 键 词：色貌模型 HDR图像 ICAM 亨特效应 史蒂文斯效应 

分 类 号：P237.3[天文地球—摄影测量与遥感] TS8[天文地球—测绘科学与技术]