计算机科学 ›› 2010, Vol. 37 ›› Issue (7): 260-263.

• 图形图像 • 上一篇    下一篇

多特征动态融合的三维模型检索方法

郑赢,周明全,耿国华,高原   

  1. (西北大学信息科学与技术学院 西安710127),(北京师范大学信息科学与技术学院 北京100875)
  • 出版日期:2018-12-01 发布日期:2018-12-01
  • 基金资助:
    国家自然科学基金重点项目(60736008) ,国家863重点基金项目(2008AA101Z301) ,国家白然科学基金项目(60873094,西北大学研究生白主创新基金项目(No. 08YZZ33} ,陕西省自然科学基金(SJ08F29)资助。

3D Model Retrieval on Multi-feature Dynamic Integration

ZHENG Ying,ZHOU Ming-quan,GENG Guo-hua,GAO Yuan   

  • Online:2018-12-01 Published:2018-12-01

摘要: 提出一种基于二维正交投影图像的多特征动态融合的三维模型检索方法。首先计算三维模型的二维正交投影图像,然后提取二维正交投影图像的投影直方图和Zcrnikc矩特征,通过加权求和在输出层融合,得到总体上模型间的相似度。每种特征采用动态权值,针对输入查询模型的不同,根据用户反馈自动更新知识库中的权值。实验表明,该方法在提高检索准确率的同时,也能保证检索效率。

关键词: 三维模型检索,多特征,动态权值,反馈

Abstract: Presented a method by itergrating multi-feature base on two-dimensional orthogonal projection to improve retrieval accuracy. Firstly, we obtained two-dimensional orthogonal projections of the 3D models, and then two features which among two-dimensional orthogonal projections were used and each feature has a weight These features were syncretized on output layer. I3y plusing the two results of the features with their weights, the whole result could be got. According to users' feedback, the system changed the weights of each feature dynamically. Experiments show retriving accuracy and efficiency have a greater improvement than that on single feature.

Key words: 3D model retricval,Multi fcaturcs,Dynamic wcights,Fecdback

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!