计算机应用 ›› 2013, Vol. 33 ›› Issue (04): 1122-1124.DOI: 10.3724/SP.J.1087.2013.01122

• 多媒体处理技术 • 上一篇    下一篇

基于欧氏距离的人脸检测窗口合并方法

黑建业,熊淑华,马娅丽   

  1. 四川大学 电子信息学院,成都 610065
  • 收稿日期:2012-11-05 修回日期:2012-12-06 出版日期:2013-04-01 发布日期:2013-04-23
  • 通讯作者: 黑建业
  • 作者简介:黑建业(1987-),男,山东胶州人,硕士研究生,主要研究方向:多媒体通信;熊淑华(1969-),女,四川成都人,副教授,主要研究方向:多媒体通信;马娅丽(1989-),女,陕西榆林人,硕士研究生,主要研究方向:多媒体通信。
  • 基金资助:

    欧盟FP7-PEOPLIE-IRSES-S2EuNet(247083)

Method of merging face detection windows based on Euclidean distance

HEI Jianye,XIONG Shuhua,MA Yali   

  1. College of Electronics and Information Engineering, Sichuan University, Chengdu Sichuan 610065, China
  • Received:2012-11-05 Revised:2012-12-06 Online:2013-04-01 Published:2013-04-23
  • Contact: HEI Jianye

摘要: 为了解决人脸检测过程中同一人脸在不同尺度下检测结果的位置和大小不能完全重合的问题,对基于统计训练的人脸检测窗口的合并方法进行了研究,提出了一种基于欧氏距离的人脸检测窗口合并方法,在人脸检测结果不存在误检和漏检的情况下,根据人脸检测窗口中心坐标的分布特征,利用判定圆以及欧氏距离实现了人脸重复检测窗口的合并。分别针对不同情况的图片对方法进行了验证实验,实验结果表明该方法简单有效

关键词: 人脸检测, 尺度, 欧氏距离, 判定圆, 合并

Abstract: To address the problem of the coincidence of the position and size for a same face cannot be guaranteed in different results due to the scale transformation in face detection, research has been done by the method of merging windows in face-detection employing statistical training. And a method based on Euclidean distance was proposed too, without considering the case of error-detected and undetected faces. Judging circle and Euclidean distance were employed to merge face-detection windows according to the distribution of center coordinates. Verifying experiments of the method were conducted according to different pictures, and the experimental results proved the method simple and effective.

Key words: face detection, scale, Euclidean distance, judging circle, merging

中图分类号: