Mapping Features to Semantic Space for 3D Model Retrieval

摘要：

In this paper.a noveI offiine supervised learning method is proposed to map IOW-level features to high-level semantic space for 3D model retrieval.Firstly.users are expected to retrievaI models through relevance feedback Iearning.and the retrievai history will be recorded in a distance matrix Which can be renewed gradually.Then,Laplacian Eigenmaps is used to establish the representations of models in semantic space by means of retrieval history.Experimental evaluation Oil the standard repositories PSB shows that our method can effectively capture the user’s knowledge in mind and embody it in the retrieval results.The algorithm also displays its strong generalization ability in ourexperiments.

作者: Liqun Li Zheng Qin Biao Leng

会议类型: 国际会议

会议名称: The International Conference Information Computing and Automation(2007国际信息计算与自动化会议)

会议地点: 成都

会议语种:英文

页码: 847-850

在线出版日期: 2007-12-19（万方平台首次上网日期，不代表论文的发表时间）