华南理工大学学报(自然科学版) ›› 2014, Vol. 42 ›› Issue (5): 122-127.doi: 10.3969/j.issn.1000-565X.2014.05.019

• 计算机科学与技术 • 上一篇    下一篇

融合熵和势能的手绘草图特征描述方法

赵鹏 朱伟伟 吴献文 刘慧婷   

  1. 安徽大学 计算智能与信号处理教育部重点实验室,安徽 合肥 230039∥安徽大学 计算机科学与技术学院,安徽 合肥 230601
  • 收稿日期:2013-09-10 修回日期:2014-03-27 出版日期:2014-05-25 发布日期:2014-04-01
  • 通信作者: 赵鹏(1976-),女,博士,副教授,主要从事智能信息处理、多媒体信息检索研究. E-mail:zhaopeng_ad@ahu.edu.cn
  • 作者简介:赵鹏(1976-),女,博士,副教授,主要从事智能信息处理、多媒体信息检索研究.
  • 基金资助:

    国家自然科学基金资助项目(61201413, 61202227);安徽省自然科学基金资助项目(1408085MF122)

Feature Description Method for Freehand Sketch by Combining Entropy and Potential Energy

Zhao Peng Zhu Wei- wei Wu Xian- wen Liu Hui- ting   

  1. Key Laboratory of Intelligent Computing and Signal Processing of the Ministry of Education,Anhui University,Hefei 230039,Anhui,China∥ School of Computer Science and Technology,Anhui University,Hefei 230601,Anhui,China
  • Received:2013-09-10 Revised:2014-03-27 Online:2014-05-25 Published:2014-04-01
  • Contact: 赵鹏(1976-),女,博士,副教授,主要从事智能信息处理、多媒体信息检索研究. E-mail:zhaopeng_ad@ahu.edu.cn
  • About author:赵鹏(1976-),女,博士,副教授,主要从事智能信息处理、多媒体信息检索研究.
  • Supported by:

    国家自然科学基金资助项目(61201413, 61202227);安徽省自然科学基金资助项目(1408085MF122)

摘要: 为更加深刻地刻画手绘草图中的形状主题,并使特征描述方法具有一定的平移、缩放和旋转不变性,提出了一种基于熵和势能的手绘草图特征描述方法.该方法首先提取手绘草图的关键点,并对每个关键点的周围区域进行分块,计算其信息熵;然后根据关键点像素在手绘草图对象中的位置,计算其像素势能,从而获得融合了熵和势能的关键点特征描述向量;最后将训练集中所有图像的全部关键点特征描述向量聚类形成视觉码书,每一幅手绘草图均表示成基于码书的特征描述向量.实验结果表明,文中方法能较好地描述手绘草图的形状特征.

关键词: 手绘草图, 关键点描述, 信息熵, 势能, 聚类

Abstract:

In order to describe the shape of freehand sketch more accurately,a novel feature description method forfreehand sketch,which is of certain translation invariance,scale invariance and rotation invariance,is proposed bycombining entropy and potential energy.First,in this method,the key contour points in freehand sketch are extrac-ted,and the information entropy of each key point is computed by dividing the region around the point into severalsub- regions.Then,according to the spatial position of key point in the object of freehand sketch,the potentialenergy of key contour point is computed,from which the descriptor vector for the key contour point is obtained incombination with entropy and potential energy.Finally,all the descriptor vectors of key contour points in all trai-ning freehand sketches are clustered to construct a visual code book,on the basis of which a freehand sketch is de-scribed as a feature vector.Experimental results verify the effectiveness of the proposed method.

Key words: freehand sketch, key point description, information entropy, potential energy, clustering

中图分类号: