高动态图像视觉保真实时显示变换技术

【作者】 彭韬；

【导师】 李在铭；

【作者基本信息】 电子科技大学 ， 通信与信息工程， 2005， 硕士

【摘要】 尽管对图像生成技术的研究给人们提供了越来越好的结果和越来越快速的方法,目前的图像产生过程仍然不足以确保感觉上的逼真性。由于常规图像显示设备只适合输出小动态范围亮度图像,使人们不能观察到场景图像的全部信息,而这一过程又恰恰是整个图像产生过程的重要部分。特别是近年来高动态范围图像(high dynamic range image-HDRI)在计算机图形学领域变得越来越普遍而且重要,真实场景的HDR 图像也变得非常容易获取,图像显示硬件的局限性也就越发暴露出来。本文通过在亮度图像梯度域上对大梯度进行衰减,压缩图像亮度的动态范围,可以使高动态范围图像在被显示时,既能够适应常规的显示硬件,同时又充分保留了原始图像的细节信息,使得图像在被观察时能够重现真实场景的亮度效果。通过对小梯度的适当提升,使得低动态范围图像得以增强,从而更易于观察图像的细微细节。当前,GPU的图形管线为我们提供了良好的可编程性支持,计算机图形硬件架构已经越来越强调其通用性。主流PC 机上的GPU 除了可以作为向量以及流式信息处理器,它完全适用于更加强大的通用数值计算。本文采用适合由图形处理器加速的快速算法,借助可编程图形硬件和先进的图形应用程序加速接口,将动态范围变换的整个处理过程通过可编程图形硬件实现,建立快速的图像动态范围压缩技术,建立起适用于高动态范围图像显示的实时应用框架,使之不仅适用于基于图像的动态范围调整的绝大部分情况,应用于数字摄影、电影艺术、科学图像增强,还能够成为目前交互式图形应用的核心技术,成为实现虚拟实景、交互3D 应用的基础之一。更多还原

【Abstract】 While research into ways of creating images provides us with better and faster methods, we may not see the full effect of these techniques due to display limitations. However, displaying an image is also an important part of the overall process, and weaknesses in this area may significantly detract from advances made in image creation. Today, high dynamic range (HDR) radiance maps are becoming increasingly common and important in computer graphics and HDR maps of real scenes are very easy to construct, while the limitations of display devices are more and more exposed. We present a system for compressing high dynamic range images to fit conventional display devices that are only capable of outputting a low dynamic range. In addition to manipulating the gradient field of luminance image by attenuating the large gradients‘magnitudes, it can preserve fine details, resulting in an image which provokes the same responses as someone would have viewing the scene in the real world. Recently, graphics hardware architectures have begun to emphasize versatility, offering rich new ways to programmatically reconfigure the graphics pipeline. As a result, powerful and potentially general-purpose constructs not unlike vector and stream processors are appearing in commodity PC machines, thanks to their graphics chips. Specifically, with the whole process built on programmable graphics hardware, we present an efficient algorithm based on GPU acceleration and provide a fast dynamic range compression technique. Furthermore, we describe a framework for rendering high dynamic range images in real time. It not only can be used in image based tone mapping but also is able to be a core technique in interactive graphics applications.更多还原