欢迎访问《图学学报》 分享到:

图学学报

• 计算机图形学 • 上一篇    下一篇

融合信息熵和CNN 的基于手绘的三维模型检索

  

  1. 1. 中国石油大学计算机与通信工程学院,山东青岛 266580;
    2. 中国科学院计算技术研究所智能信息处理重点实验室,北京 100190;
    3. 中国科学院大学,北京 100190
  • 出版日期:2018-08-31 发布日期:2018-08-21
  • 基金资助:
    国家自然科学基金项目(61379106,61379082,61227802);山东省自然科学基金项目(ZR2013FM036,ZR2015FM011)

Sketch-Based 3D Shape Retrieval with Representative View and Convolutional Neural Network

  1. 1. College of Computer & Communication Engineering, China University of Petroleum, Qingdao Shandong 266580, China;
    2. Key Laboratory of Intelligent Information Processing, Institute of Computing Technology Chinese Academy of Sciences, Beijing 100190, China;
    3. University of Chinese Academy of Sciences, Beijing 100190, China
  • Online:2018-08-31 Published:2018-08-21

摘要: 基于手绘草图的三维模型检索(SBSR)已成为三维模型检索、模式识别与计算机视
觉领域的一个研究热点。与传统方法相比,基于卷积神经网络(CNN)的三维深度表示方法在三
维模型检索任务中性能优势非常明显。本文提出了一种基于手绘图像融合信息熵和CNN 的三
维模型检索方法。首先,通过计算模型投影图的信息熵得到模型的代表性视图,并将代表性视
图经过边缘检测等处理得到三维模型投影图的轮廓图像;然后,将轮廓图像和手绘草图输入到
CNN 中提取特征描述子,并进行特征匹配。本文方法在Shape Retrieval Contest (SHREC) 2012
数据库和SHREC 2013 数据库上进行实验。实验证明,该方法的效果较其他传统方法检索准确
度更高。

关键词: 三维模型检索, 卷积神经网络, 代表性视图, 信息熵

Abstract: Sketch-based shape retrieval (SBSR) has become a hot research spot in the field of model
retrieval, pattern recognition, and computer vision. 3D deep representation based on convolutional
neural network (CNN) enables significant performance improvement over state-of-the-arts in task of
3D shape retrieval. Motivated by this, in this paper a sketch-based 3D model retrieval algorithm by
utilizing entropy representative views and CNN feature matching is proposed. The representative
views are obtained by viewpoint entropy. And the representative views are processed by edge
detection to get the contour image of 3D model projection. The CNN descriptors extracted as features
for representative view of each object. And the method of feature matching is based on CNN
descriptors. Our experiments on Shape Retrieval Contest (SHREC) 2012 database and SHREC 2013
database demonstrate that our method is better than state-of-the-art approaches.

Key words: 3D shape retrieval, convolutional neural network, representative view, entropy