基于GLSL的多重视频纹理映射与融合  

Multiple video texture mapping and fusion based on OpenGL shading language

作　　者：简洪登[1,2] 范湘涛 

机构地区：[1]中国科学院遥感与数字地球研究所数字地球重点实验室,北京100094 [2]中国科学院大学,北京100049

出　　处：《计算机工程与设计》2014年第11期3873-3878,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(41101436)

摘　　要：为将实时视频与三维场景进行融合来构建大范围增强虚拟环境(augmented virtual environment,AVE),消除融合后出现的反向投影、纹理拉伸、拼缝等现象,实现基于OpenGL着色语言(OpenGL shading language,GLSL)的反向投影剔除算法和深度测试算法,提出基于GLSL的多重视频纹理的横向加权融合算法和对角加权融合算法。实验结果表明,所提算法能够在显存为1GB的计算机上使包含200万三角面和6路实时视频的三维场景的运行帧率保持在40fps以上,高效地实现了实时视频与三维场景的融合。To establish the wide-field augmented ~rtual environment （AVE） by fusing multiple real-time videos with 3D models, and to e- liminate the errors of reverse projections, texture distortions and obvious video seams, the reverse projection eullkng algorithm and the depth test algorithm were applied to the fragment shader base on the OpenGL shading language （GLSL）. Then the horizontal weighted algorithm and the diagonal weighted algorithm based on the GLSL were implemented and applied to fuse the overlap area of two adjacent videos. The results show that the proposed algorithms can fuse multiple videos with 3D models efficiently, which runs a three-dimensional scene contai- ning 2 million triagular facets and 6 real-time videos at more than 40 fps on a computer with 1 GB graphics card memory.

关 键 词：增强虚拟环境 实时视频 投影纹理映射 反向投影 深度测试 对角加权融合 多重视频纹理融合 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]