摘要
基于形态图表示的三维物体识别的基本思路是:首先建立待识别物体的模型库,找出模型集中所有模型物体的形态图和特征视图,并提取出它们的拓扑结构信息和几何信息;其次对物体真实图像作轮廓提取和边界跟踪,得到二维图像的线架图,同时提取出它的拓扑结构信息和几何信息;最后将物体图像的拓扑结构信息和几何信息与模型库中模型物体的拓扑结构信息和几何信息匹配,从而达到识别的目的。文中提出了在生成线图链码时提取其拓扑结构信息和几何信息的方法,由拓扑结构信息和几何信息构造特征矩阵的方法,以及识别过程中特征矩阵的匹配算法。
The fundament principles of 3D objects recognition based on the aspect graph is as follows: First of all, establish the model base of objects to be recognized to find all the aspect graphs and feature views in the model collection and extract their topology information and geometry information. Second, extract the edge of the object image and boundary tracking to get the 2D line drawing and extract the topology information and geometry information of them. At last, match the information of topology and geometry of the object images with that of the model objects in the model base to recognize. It is presented in this context the way to extract topology information and geometry information when forming the chain codes of line drawing, the way of constructing characteristic matrix by topology information and geometry information and the matching algorithmic of characteristic matrix in the recognizing process.
出处
《计算机应用》
CSCD
北大核心
2003年第2期9-12,共4页
journal of Computer Applications
基金
国家自然科学基金资助项目(项目批准号:60275037)
江西省测试技术与控制研究中心开放基金资助项目
