• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

计算机工程与科学

• 论文 • 上一篇    下一篇

实数型曲线描述子二值化研究

王志衡,陈璐璐,刘红敏,王静   

  1. (河南理工大学计算机科学与技术学院,河南 焦作 454000)
  • 收稿日期:2015-12-30 修回日期:2016-06-22 出版日期:2017-11-25 发布日期:2017-11-25
  • 基金资助:

    国家自然科学基金(61572173,61472119,61272394,61401150);河南省高校创新人才计划(13HASTIT039);河南省青年骨干教师(2012GGJS-057,2013GGJS-052);河南理工大学创新型科研团队计划(T2014-3)

Binaryzation of real-valued curve descriptor

WANG Zhi-heng,CHEN Lu-lu,LIU Hong-min,WANG Jing   

  1. (School of Computer Science and Technique,Henan Polytechnic University,Jiaozuo 454000,China)
  • Received:2015-12-30 Revised:2016-06-22 Online:2017-11-25 Published:2017-11-25

摘要:

曲线匹配技术在模式识别、计算机视觉和图像理解中具有重要作用。随着移动设备的广泛使用,有必要研究存储空间小、匹配速度快的二值型曲线描述子。针对常见实数型曲线描述子(MSCD、IOMSD、IOCD和TCHP),利用阈值化方法,获得由0、1表示的二值曲线描述子。实验结果表明,在旋转、视角变化和光照变化条件下,提出的曲线二值描述子能够保持实数型描述子的匹配准确性,而占用的内存空间仅为原描述子的1/32或1/16。
 

关键词: 曲线匹配, 实数型描述子, 二值描述子

Abstract:

Curve matching plays an important role in pattern recognition, computer vision and image understanding. With the widespread use of mobile devices, it is necessary to study the binary curve descriptors with small storage space and fast matching speed. We quantize the curve descriptors, such as MSCD, IOMSD, IOCD and TCHP, into binary strings represented by 0 and 1 by thresholding the real-valued vectors of the descriptors. Experimental results show that under the conditions of image rotation, viewpoint change and illumination change, the proposed binary descriptors yield a little higher matching accuracy while only requiring 1/32 or 1/16 storage space of the original curve descriptors.
 
 

Key words: curve matching, real-valued descriptor, binary descriptor