非真实感绘制中的线绘制技术及其水墨仿真研究
文献类型：学位论文
| 作者 | 卢健 |
| 学位类别 | 硕士 |
| 答辩日期 | 2010-05-28 |
| 授予单位 | 中国科学院研究生院 |
| 授予地点 | 北京 |
| 导师 | 刘学慧 |
| 关键词 | 非真实感绘制 线绘制 水墨仿真 LB方法 GPU应用 |
| 学位专业 | 计算机应用技术 |
| 中文摘要 | 绘画是人类捕捉、记录及表现不同创意目的的行为，是一种在二维的平面上以手工方式描绘自然的艺术。从远古的石洞壁画到现代的各种绘画艺术作品和插图，随着人类历史的进步，绘画一直扮演着传递信息和艺术表达的角色：作为表达场景信息的手段，在照相技术出现之前，手工绘画一直是最好的方法；而另一方面，作为艺术行为，绘画可以表达作者的精神世界和自我风格。 随着计算机技术的出现和发展，计算机绘图逐渐成熟，并形成一门独立学科：计算机图形学。非真实感绘制(Non-Photorealistic Rendering, NPR)作为计算机图形学的热点研究课题最为接近手工绘画，有别于传统计算机图形学，它是涉及计算机艺术、数字图形处理、认知心理学等多领域的交叉研究领域。非真实感绘制的一个重要作用就是将计算机技术和绘画艺术结合起来，自动或交互地生成具有手绘艺术风格的图像。其目的不在于绘制结果的真实感，而在于表现绘画艺术特质、模拟艺术作品等。非真实感绘制往往依据艺术家的视觉感官或为了突出物体某方面性质，会着重强调某一方面细节而忽略一些其他细节信息。 本文围绕非真实感绘制技术，从模拟线绘画和中国水墨绘画两个方向进行研究。以简单的三维网格模型为输入，力求完全自动化地进行实时绘制。本文工作主要分为两部分：实时线绘制技术和二维水墨扩散仿真。 线条绘画，是表达物理场景既简单又实用的绘画手段，线绘制技术也是非真实感绘制领域最重要的研究课题之一。无论是插图还是艺术绘画，线绘制图像都可以完全表达出三维场景的形状信息而不用借助于其他绘画手段。本文首先总结了前人的线绘制算法，并针对视点相关特征线和视点无关特征线的优缺点，提出了一种融合两类特征线优点的特征线新定义。基于该定义，本文算法以模型曲面的微分几何信息为依据，运用基于图像的绘制技术，利用GPU硬件，实时地绘制出风格可控的线绘制图像。 对于种目繁多的艺术流派，目前NPR技术已经实现了多种绘画艺术风格的模拟，如水彩画、铅笔画、钢笔画、油画等等。水墨画是作为中国传统绘画艺术形式，画风朴素抽象，笔划简单但富表现力，是我国优秀的文化传统。对水墨画的仿真模拟，不仅是传承我国优秀文化的一种新方式，同时也是对NPR技术的拓展。由于本文线绘制算法有较好的线宽控制以及内部着色（灰度渐变），在实现上可以很连贯地由线绘制图像扩展到水墨仿真。在水墨扩散模拟阶段，将线绘制结果图像作为输入，通过在二维水墨扩散仿真实现水墨风格的绘制。本文运用流体力学中常用的LB方法，通过研究水墨扩散的各种现象，修改LB方法的D2Q9模型二维网格间的流体迁移运算以及网格的宏观量更新，模拟水墨在宣纸内的流动。由于LB方法具有良好的并行性，故本文仿真算法可以利用GPU的并行计算能力实时地完成绘制。 |
| 英文摘要 | Drawing is the human behavior to acquire and record natural scenes with different intentions. It is an art which depicts 3D world in 2D images. From cave painting to modern drawing art, the drawing art has been playing the role of transferring information and expressing arts. Before the photographic technique, the manual drawing had been the best method to express objective scenes. On the other hand, as the art activity, the drawing can express the artists' inner emotion. With the development of computer technique, there appear some computer drawing techniques and finally develop into a special research area: Computer Graphics. NPR(Non-Photorealistic Rendering) which is most closed to drawing behavior becomes one of hot subjects of Computer Graphics. NPR is different from traditional CG techniques which involves knowledge of computer art, digital image process, Cognitive Psychology. It is an important role of NPR to combine computer techniques and arts together to render images with artistic styles. The purpose of NPR is not to render scenes realistically but to show the world in artistic way. Therefore, NPR technique always highlights some details of the scenes but ignores others. This paper focuses on two research fields: line-drawing algorithm and ink dispersion simulation which are both popular topics of NPR. The method strives to render stylized image totally automatically in real time with most simple input, such as 3D triangle mesh. According to the two research fields, this paper is divided into two part: Real-time Line-drawing Algorithm and Ink Dispersion Simulation. It is very effective to drawing with lines and the line-drawing algorithm is one of the most important research areas of NPR. Either in art works or technical illustration, line-drawing is able to give a clear description of objective scenes independently. In this paper, we firstly give a survey about current line-drawing algorithms, and then present a novel definition of feature line which incorporates the both advantages of view-dependent and view-independent feature lines. According to the definition, our approach utilizes differential geometry information and image-based rendering technique to render stylized images in real time with the acceleration of GPU. Currently, NPR technique is able to simulation many drawing arts which include watercolor, pencil drawing, pen-and-ink illustration, oil painting and so on. Chinese ink painting as a traditional artistic form, is simulated by many researchers. The ink painting simulation is not only the development of Chinese traditional culture, but also the expansion of NPR research. Thanks to the well controllability on line width and inner diffuse of our line-drawing approach, we can expand line-drawing results to ink painting style images easily. In ink dispersion simulation, we take the line-drawing image as the input and to simulate the ink dispersion by aid of LB method. In practice, we make some modification due to the special properties of ink painting. Because of the good parallelism of LB method, our approach is able to render ink-diffused painting in real time with GPU acceleration. |
| 学科主题 | 计算机应用 ; 计算机图形学 |
| 语种 | 中文 |
| 公开日期 | 2010-06-07 |
| 源URL | [http://124.16.136.157/handle/311060/2313]  |
| 专题 | 软件研究所_计算机科学国家重点实验室 _学位论文 |
| 推荐引用方式 GB/T 7714 | 卢健. 非真实感绘制中的线绘制技术及其水墨仿真研究[D]. 北京. 中国科学院研究生院. 2010. |
入库方式： OAI收割
来源：软件研究所
浏览0
下载0
收藏0
其他版本
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。