基于手绘草图骨架特征的三维目标检索方法-《计算机技术与发展》

文章信息/Info

Title:: 3D Object Retrieval Method Based on Skeleton Features of Hand Drawn Sketch

作者:: 赵雪青¹; 2; 3; 师昕¹; 2; 3; 陈惠娟¹; 2; 3; 1. 西安工程大学服装设计智能化陕西省重点实验室,陕西西安 710048;2. 新型网络智能信息服务国家地方联合工程研究中心,陕西西安 710048;3. 西安工程大学计算机科学学院,陕西西安 710048

Author(s):: ZHAO Xue-qing¹; 2; 3; SHI Xin¹; 2; 3; CHEN Hui-juan¹; 2; 3; 1. Shaanxi Key Laboratory of Clothing Intelligence,Xi’an Polytechnic University,Xi’an 710048,China;2. National and Local Joint Engineering Research Center for Advanced Networking & Intelligent Information Service,Xi’an 710048,China;3. School of Computer

Keywords:: skeleton feature; 3D object retrieval; histogram of oriented gradient; support vector machine

摘要:: 随着人们对三维数据使用需求的日益增加,网络上三维目标数据亦迅速增长,传统的基于关键词的检索方法难以满足三维目标的检索需求。因此,针对三维目标检索问题,提出了一种新的基于手绘草图骨架特征的三维目标检索方法。首先,采用有向梯度直方图(HOG)作为特征描述三维目标的骨架信息,使用支持向量机实现三维目标分类检索;其次,手绘三维目标草图并获取其HOG特征在训练好的三维目标数据库中进行检索;最后,采用查全率进行客观评价获得三维目标检索结果。仿真实验结果表明,提出的基于手绘草图骨架特征的三维目标检索方法具有较好的检索性能。

Abstract:: With the increasing demand for 3D data usage,3D object data on the network is also growing rapidly,so that traditional retrieval method based on key words is difficult to meet the demand of 3D object. Therefore,we propose a new 3D object retrieval method based on skeleton features of hand-drawn sketch. Firstly,the 3D skeleton information is described by using the histogram of oriented gradient (HOG) and object classification retrieval is achieved by support vector machine (SVM). Secondly,hand-painted 3D object sketch and its HOG feature are retrieved in the trained 3D object database. Lastly,the recall ratio is used to objectively evaluate the 3D retrieval result. The simulation indicates that the proposed method provides persuasive data towards 3D skeleton based object retrieval.