J4 ›› 2013, Vol. 51 ›› Issue (01): 107-110.

• 计算机科学 • 上一篇    下一篇

支持向量机在视频运动目标分析中的应用

邢吉生1, 杨礼2, 尚祖飞3, 浦铁成1, 牛国成1, 于哲舟4   

  1. 1. 北华大学 电气信息工程学院, 吉林 吉林 132021;2. 中国科学院 长春光学精密机械与物理研究所光学系统先进制造技术重点实验室, 长春 130033|3.  黑龙江大学 电子工程学院| 哈尔滨 150080|4. 吉林大学 计算机科学与技术学院, 长春 130012
  • 收稿日期:2011-12-16 出版日期:2013-01-26 发布日期:2013-01-31
  • 通讯作者: 于哲舟 E-mail:yuzz@jlu.edu.cn

Application of Support Vector Machine inVideo Moving Target Analysis

XING Jisheng1, YANG Li2, SHANG Zu\|fei3, PU Tiecheng1, NIU Guocheng1, YU Zhezhou4   

  1. 1. College of Electrical &|Information Engineering, Beihua University, Jilin 132021, Jilin Province, China;2. Key Laboratory of Optical System Adva
    nced Manufacturing Technology, Changchun Institute of Optics,Fine Mechanics and Physics, Chinese Academy of Sciences, Changchun, 130033, China|3. College of Electronic Engineering, Heilongjiang University, Harbin 150080, China;4. College of Computer Science and Technology, Jilin University, Changchun 130012, China
  • Received:2011-12-16 Online:2013-01-26 Published:2013-01-31
  • Contact: YU Zhezhou E-mail:yuzz@jlu.edu.cn

摘要:

提出一种基于支持向量机的运动目标分类方法. 先将支持向量机引入分析视频运动目标中, 再在视频中筛选出简单有效的组合特征对目标进行分类. 该方法先使用混合Gauss背景模型提取前景运动目标, 获取目标的形状特征和运动特征, 再利用支持向量机对样本数据进行训练, 得到最优决策函数. 实验结果表明, 利用支持向量机和运动目标特征组合的方法进行运动目标分析实用、 有效.

关键词: 支持向量机, 视频运动目标, 特征提取, 分类器

Abstract:

Support vector machine was introduced into the analysis of  moving objects,  simple and effective combination of extracted target features was performed for the classification of moving objects. We used Gaussian mixture background model to detect moving objects, from which shape features and movement features were extracted. Support vector machine was adopted to obtain the optimal decision function via training marked sample data. The experiment shows that the analysis of moving targets via support vector machine and moving object feature combination is effective.

Key words: support vector machine, moving object detection, feature extraction, classifier

中图分类号: 

  • TP391.4