一种基于立体视觉的多视点建模方法

论述了一种基于立体视觉的建模方法。该方法利用立体视觉系统在不同视点对景物观测所获得的局部三维几何模型，通过空间特征点匹配和坐标变换将局部模型融合，从而建立景物的完整描述。文章重点介绍了一种基于空间向量的坐标变换求解方法。

Because of it's difficalt to match the object model feature points in  Multi  Vision system。In this paper a method which obtains moving  object 3D  information based on single  vision be given。This method based on object  model、camera  model and object  shape model  applys a kind of speocial image  processor which may realize the real time extraction、matching of the object model feature points  readize real  time obtain moving  object 3D information。

辽宁省自然科学基金