OBJECT RECOGNITION AND POSE ESTIMATION BASED ON BIONIC PATTERN RECOGNITION AND MANIFOLD LEARNING

摘要：

This paper describes a novel method that could recognize object with rotation variance and integrates poses estimation in a whole framework. It takes advantages of bionic pattern recognition theory and the manifold learning algorithms and overcome their drawbacks. Based on the principle of homologycontinuity, we adopt shortest neighborhood graphs to depict the high-dimensional geometrical structure of training image sets. Hence, the recognition task can be converted into the finding of a graph that the tested image lies on.

关键词： Object Recognition Pose Estimation Shortest Neighborhood Graph Local Linear Approximation

作者: Z.H.Hao S.W.Ma

作者单位: School of Mechatronic Engineering and Automation Shanghai Key Laboratory of Power Station Automation Technology, Shanghai University, Shanghai 200072, China

会议类型: 国际会议

会议名称: 2012 International Conference on System Simulation(2012年国际系统仿真学术会议)

会议地点: 上海

会议语种:英文

页码: 494-497

在线出版日期: 2012-04-06（万方平台首次上网日期，不代表论文的发表时间）