首页 | 本学科首页   官方微博 | 高级检索  
     

多视点成像的多边形模板法
引用本文:孙庆杰,王文成,吴恩华.多视点成像的多边形模板法[J].计算机学报,2002,25(7):737-746.
作者姓名:孙庆杰  王文成  吴恩华
作者单位:1. 中国科学院软件研究所计算机科学重点研究实验室,北京,100080
2. 中国科学院软件研究所计算机科学重点研究实验室,北京,100080;澳门大学科学技术学院,澳门
基金项目:国家自然科学基金 (6990 3 0 0 9),澳门大学 Research Grant的支持
摘    要:该文提出一种对场景进行多视点成像的方法。该方法首先为场景中的多边形生成多边形模板,一个多边形模板,包括一条轮廓路径和一组纹条,而一个纹条是平行成像画的一个平面与多边形相交的直线段。由于纹条相对于不同视点的透视投影的变化是线性的,因此,绘制多边形时可以基于模板逐个纹条地处理,而不必按照传统的扫描转换方法逐个点地处理,绘制速度可以提高很多。同时,与视点无关的光照和纹理可以预先计算并保存在模板中,以便在成像时利用基于图像绘制的技术来生成高质量的图像。该方法中,视点可以放置在三维空间的任意位置,并且在场景漫游时可以根据视点位置自动地实现多分辨率绘制。

关 键 词:多视点成像  多边形模板法  纹条  透视投影  三维场景  图像处理  计算机
修稿时间:2000年11月16

Polygon Splatting for Multiple Viewpoint Rendering
SUN Qing Jie WANG Wen Cheng WU En Hua ,.Polygon Splatting for Multiple Viewpoint Rendering[J].Chinese Journal of Computers,2002,25(7):737-746.
Authors:SUN Qing Jie WANG Wen Cheng WU En Hua  
Affiliation:SUN Qing Jie 1) WANG Wen Cheng 1) WU En Hua 1),2) 1)
Abstract:This paper presents a new method for multiple viewpoint rendering to employ perspective coherence, which allows the viewpoint to be placed freely in the 3D space. At first, it creates an intermediate construct referred to as polygon splat for every polygon in the scene, which consists of a set of stripes, line segments parallel to the image plane and extracted from the polygon. As some linear relations between a stripe and its perspective projection with the viewpoint movement can be derived, polygons can be rendered stripe by stripe instead of the pixel by pixel manner used in the conventional scan conversion method, which greatly reduce the average cost for the ray cast from a pixel to intersect a polygon. For view independent shading and texture, they can be pre computed and stored in the stripes to be reused to render high quality images by image based rendering techniques to save a large amount of on line shading and texture mapping calculations. In rendering, the farther the polygon is from the viewpoint, the fewer the stripes of the polygon are selected to represent the polygon. Thus, the new method can automatically implement the level of detail rendering without constructing many models of different levels of details. From experimental results, the new method is several times faster than the conventional scan conversion method for multiple viewpoint rendering. As the new method only aims to replace the scan conversion method in the rendering pipeline, it is easy to integrate with other efficient rendering methods such as visibility determination and shading methods to further improve the rendering efficiency. Also, the new method can be easily extended to render curve surfaces.
Keywords:stripe  polygon splat  perspective projection  multiple viewpoint rendering
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号