首页 | 本学科首页   官方微博 | 高级检索  
     

基于立体视觉的影像多视点建模方法探讨
引用本文:赵晓东,张军. 基于立体视觉的影像多视点建模方法探讨[J]. 测绘科学, 2008, 0(Z1)
作者姓名:赵晓东  张军
作者单位:[1]辽宁工程技术大学研究生学院 [2]中国测绘科学研究院
摘    要:随着GIS用户对地理信息可视化性能要求的不断提高,虚拟地理场景的景物实体建模方法成为VRGIS领域的热点问题之一。本文探讨了一种基于立体视觉的影像多视点建模方法,包括景物实体表面影像纹理的采集和重建。影像纹理采集方面,提出了根据景物实体形状和景物实体表面纹理分布特点分配采集点的纹理采集方法。影像纹理重建方面,根据多幅影像对应特征点创建采样点模型,最后与几何模型进行空间配准,创建景物实体模型。以一实例进行实验论证,实验结果表明,该建模方法能够有效的提高虚拟地理环境中模型的真实感。

关 键 词:立体视觉  影像纹理  多视点建模  采样点模型  纹理配准

Discuss the method of image Multi-View modeling based on stereo vision
Abstract:With the continuous improvement of requirement visualization performance of the geographic information by GIS users,the object modeling method of virtual geographic scenes becomes one of the hot issues in the VRGIS.This paper discuss a method of image Multi-View modeling based on stereo vision,including acquisition of the surface images texture and reconstruction of the texture.In the section of image texture acquisition,to propose distribution acquisition points according to object shape and distribution of surface texture. In the section of image texture reconstruction,according to create sampling model many based on images' corresponding feature points, At last,matching sampling model with geometry model,and create a object modeL One example of feasibility studies,the experimental results showed that Multi-View modeling method may improve the reality of model in virtual geographic scenes effectively.
Keywords:stereo vision  image texture  Multi-View modeling  sampling model  texture registration
本文献已被 维普 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