多视角图像对的生成及渲染优化技术  

Multi-views stereo pairs creating and rendering optimizing

作　　者：姜太平[1] 陈丛辉 邰伟鹏[1] 

机构地区：[1]安徽工业大学计算机学院,安徽马鞍山243002

出　　处：《计算机工程与设计》2014年第9期3163-3168,共6页Computer Engineering and Design

基　　金：安徽省教育厅自然科学科研基金项目(2012Z027)

摘　　要：针对立体片源生成方式中离线渲染不能与用户形成交互、在线渲染难以实时生成多视角立体图像对且渲染速度慢等缺点,提出在构造虚拟摄像机模型的基础上,采用DLL拦截策略,在opengl32.dll中构建并生成立体图像对,并对多视角渲染采用并行渲染进行优化的方法。实验结果表明,该方法可以提供实时在线渲染的3D内容,为当前大量基于OpenGL的系统提供了一种无需修改源程序、可自动转换为多视角立体图像的新方法。The 3D contents are mainly produced by two rendering methods including the offline rendering and the online rendering. To avoid the disadvantages thai the offline rendering can not supply a human-computer interaction and the online rendering can not build multi views easily to create stereo pairs in real-time and the rendering speed is slow, a new way was proposed that built the virtual cameras acquiring stereo pairs automatically and applied the parallel rendering optimizing to improve the speed in the opengl32, dll using the DLL interception technique. Experimental results show that the method can provide the real-time online 3D sources which present a novel approach to create the multi-view stereo images automatically for numerous current systems based on OpenGL without modifying the source codes.

关 键 词：立体显示 多视角 立体图像对 并行渲染 动态库拦截 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]