• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>International Conference on Cloud Computing and Security >View-Based 3D Model Retrieval Based on Distance Learning
【24h】

View-Based 3D Model Retrieval Based on Distance Learning

机译：基于远程学习的基于视图的3D模型检索

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

As information technologies develop, 3D model retrieval is paid more and more attentions by researchers. But the limitations of image features poses a great challenge to view-based 3D model retrieval. In this paper, a novel 3D model retrieval method based on distance learning is introduced. The objective function with respective to two latent variables was formulated especially. The variables are the clique information in the original graph and the pairwise clique correspondence constrained by the one-to-one matching. The proposed method has the following benefits: (1) the local and global attributes of a graph with the designed structure can be preserved; (2) redundant and noisy information can be eliminated by strengthening inliers and suppressing outliers; and (3) the difficulty of defining high-order attributes and solving hyper-graph matching can be avoided. By extensive experiments on ETH, NTU60 and MV-RED datasets with Zernike moments, Histograms of Oriented Gradients (HoG) and convolutional neural networks (CNN) features, the effectiveness of the proposed method could be tested.
机译：随着信息技术的发展，研究人员越来越重视3D模型检索。但是图像特征的局限性给基于视图的3D模型检索带来了巨大挑战。本文介绍了一种基于远程学习的新型3D模型检索方法。特别制定了具有两个潜在变量的目标函数。变量是原始图中的集团信息以及受一对一匹配约束的成对集团集团对应关系。所提出的方法具有以下优点：（1）可以保留具有设计结构的图的局部和全局属性; （2）可以通过加强孤立点和抑制孤立点来消除多余和嘈杂的信息; （3）可以避免定义高阶属性和解决超图匹配的难题。通过对具有Zernike矩，定向梯度直方图（HoG）和卷积神经网络（CNN）特征的ETH，NTU60和MV-RED数据集进行广泛的实验，可以验证该方法的有效性。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号