CASIA OpenIR  > 毕业生  > 硕士学位论文
一种基于图像的光照模型研究
其他题名A Study of An Image-Based lllumination Model
沈沉
学位类型工学硕士
导师马颂德
2000-07-01
学位授予单位中国科学院自动化研究所
学位授予地点中国科学院自动化研究所
学位专业模式识别与智能系统
摘要在传统图形学中,基于几何模型的绘制方法用几何基元来描述场景,并通过 传统的光照模型,描述了物体表面反射与光源的分布及物体表面材质的关系,最 后利用绘制方程计算拟真的合成图像。近年来,一种全新的绘制方法正在得到越 来越广泛的重视和深入的研究,这就是所谓的基于图像的绘制技术(Image—Based Rendering,简称IBR)。不同于传统图形学中基于几何模型的绘制方法,在基于 图像的绘制技术中没有物体的几何模型,没有描述物体表面材质的双向反射系数 (BRDF),也没有光源分布,有的只是真实场景的基本组成成分:光线。拍摄图 像作为采集光线的手段,而绘制就成为对这些光线的重组过程,因此基于图像的 绘制技术彻底地摆脱了传统图形学的绘制方式。 随着基于图像的绘制技术的进一步成熟和普及,由此扩展出了一个更具挑战 性的问题:即,基于图像的光照研究。在已有的基于图像的绘制技术中,有一个 共同的假设就是场景本身是静态的,即场景的光照条件固定不变,绘制过程只是 生成不同视点位置或视线方向的场景图像。基于图像的光照研究的重要意义在于 它设法突破静态场景的限制,使用户不仅可以改变视点,还可以改变场景的光照 条件,从而产生出更加丰富多变的场景图像。 全光函数从场景由光线组成这个基本点出发,给出了基于图像的绘制技术的 理论框架,同时它也是基于图像的光照研究的理论基础。我们以全光函数为核 心,提出了一种系统的分析框架,将基于几何模型的绘制方法与基于图像的绘制 方法统一在一起,并进一步的扩展出基于图像的光照问题。在此基础上,本文概 括性地指出基于图像的光照研究的基本任务实际上是对全光函数的采样、重建、 合成和重采样的过程。从而将全光函数对基于图像方法的指导意义上升到了一个 更高的层次。 从这个分析框架出发,本文综述了近期内有关基于图像的光照问题的部分研 究工作,从中总结出了三种可能的解决方案。这三种方案之间的差别在于所使用 的光照模型不同,按这样的方法可以将已有的研究工作分成三种类型,即:利用 传统光照模型的方法、利用基于图像的光照模型的方法以及无需光照模型的方 法。通过进一步的分析可以看出,利用基于图像的光照模型的方法最具生命力, 将会成为未来的研究重点。 为此,我们从传统光照模型出发,结合基于图像方法的特点,提出了一种新 的基于图像的光照模型。该模型最大的特点是实时性,它将尽可能多的光照运算 提前到预处理阶段
其他摘要In traditional computer graphics, the geometry-based rendering method regards the scene as a set of basic geometric elements. By the conventional illumination model which defines the property of light sources and the material of objects, we can create synthetic images using the rendering equation. Recently, Image-based Rendering (IBR) technique has been intensively investigated. Different from the traditional rendering techniques, there is no geometry models, no BRDF, and no light sources in IBR system. It is the RAY that composes the whole scene. Rays are captured by taking photos. Rendering is a process to re-bin those captured rays. Therefore, IBR thoroughly abandons the conventional rendering equation. As IBR is more and more popular and well studied, its extension problem, which is much more challenging, Image-based re-Lighting receives attention. In IBR, there is a significant limitation that the lighting condition of the environment is static and nothing can be altered except for the viewpoint. From this aspect, Image-based re-Lighting relax the constraint to make the scene itself be able to be changed, and then much more colorful lighting effects can be obtained. The plenoptic function is the theory basis of Image-based Rendering. Based on this function, we introduce a framework to unify the GBR and IBR together. Furthermore, the image-based re-lighting problem is extended from this framework. From the aspect of plenoptic function, we indicate that the aim of Image-based re-Lighting is basically to sample, reconstruct, synthesize and re-sample the plenoptic function. From this analysis framework, we summarize the recent work related to the problem of Image-based re-Lighting. Moreover, these techniques are classified into three categories according to which kind of illumination model is used: rendering with the traditional illumination model, rendering with the image-based illumination model, and rendering without illumination model. From further analysis, we indicate that the future work will focus on how to render with image-based illumination model and forecasts some possible breakthroughs. To realize the Image-based re-Lighting problem, we introduce our own image-based illumination model. The most important characteristic of such a model is real-time. It pre-computes most parts of the illumination computation in advance. Therefore, it leads more efficient rendering. The first step to build an image-based illumination system is to create an image-based scene. We introduce the Concentric Sphere technique to describe the 4D ray space. From the Concentric Mosaic, we get the sampling and rendering algorithm for the Concentric Sphere. To realize the image-based illumination model, we introduce the concept of the ex tended ray space. The basic idea is to pre-process the original ray space by re-binning rays and pre-filtering environment maps. Then the Diffuse space and the Specular space are created. Beca
馆藏号XWLW569
其他标识符569
语种中文
文献类型学位论文
条目标识符http://ir.ia.ac.cn/handle/173211/7302
专题毕业生_硕士学位论文
推荐引用方式
GB/T 7714
沈沉. 一种基于图像的光照模型研究[D]. 中国科学院自动化研究所. 中国科学院自动化研究所,2000.
条目包含的文件
条目无相关文件。
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[沈沉]的文章
百度学术
百度学术中相似的文章
[沈沉]的文章
必应学术
必应学术中相似的文章
[沈沉]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。