摘要
高动态范围(HDR)图像是一类包含丰富的图像细节、具有较高明暗对比度、接近真实场景色彩的图像.HDR图像具有广阔的研究前景和实际应用价值.与传统图像相比,这类图像具有很高的动态范围,因而HDR图像在一般显示设备上难以得到比较满意的显示效果.为了克服这一问题,本文提出一个基于协同滤波的多尺度色调映射模型,该模型先利用滤波方法将HDR图像分解为不同细节尺度的图像层,再对各个图像层的动态范围进行不同程度的调整,从而同时达到压缩动态范围和增强细节的效果.实验表明该模型能较好地在一般显示设备重现原场景,具有较好的视觉效果.
High dynamic range （HDR） images are special kind of images, which have abundant details, high contrast and high fidelity of original scene. HDR images have a lot of applications. HDR image display and capture are hot research topics in image processing. Compared with traditional digital images, HDR images have a wide dynamic range, and they are hard to represent on traditional display devices satisfactorily. In order to overcome this problem, a multi-scale tone mapping model based on collaborative filtering is proposed. Thi~ model uses collaborative filtering technique to decompose original HDR image into different image layers with different detail-scale, then adjust dynamic range on each image layer and reconstruct original scene. Experimental results show this model can satisfactorily represent original scenes on traditional display device.
出处
《汕头大学学报（自然科学版）》
2015年第1期64-70,共7页
Journal of Shantou University：Natural Science Edition
基金
汕头大学科研启动基金资助项目(NTF12009)
汕头大学学术创新团队建设项目(ITC12001)
广东省高校工程技术研究中心(GCZX-A1306)
关键词
协同滤波
高动态范围
块匹配
多尺度色调映射
collaborative filtering
high dynamic range
block matching
multi-scale tonemapping
