计算机应用 ›› 2016, Vol. 36 ›› Issue (11): 2974-2978.DOI: 10.11772/j.issn.1001-9081.2016.11.2974

• 第十六届中国粗糙集与软计算联合学术会议(CRSSC 2016)论文 • 上一篇    下一篇

基于加速鲁棒特征和多示例学习的目标跟踪算法

白晓红1, 温静1, 赵雪1, 陈金广2   

  1. 1. 山西大学 计算机与信息技术学院, 太原 030006;
    2. 西安工程大学 计算机科学学院, 西安 710048
  • 收稿日期:2016-04-25 修回日期:2016-05-30 出版日期:2016-11-10 发布日期:2016-11-12
  • 通讯作者: 温静
  • 作者简介:白晓红(1989-),女,山西大同人,硕士研究生,主要研究方向:计算机视觉;温静(1982-),女,山西晋中人,副教授,博士,CCF会员,主要研究方向:计算机视觉、图像处理、模式识别;赵雪(1992-),女,山西晋中人,硕士研究生,主要研究方向:计算机视觉;陈金广(1977-),男,河南镇平人,副教授,博士,主要研究方向:多源信息融合、目标跟踪。
  • 基金资助:
    国家自然科学基金资助项目(61201453,61201118);山西省基础研究计划项目(2014021022-2);山西省高等学校科技创新项目(2015108)。

target tracking algorithm based on the speeded up robust features and multi-instance learning

BAI Xiaohong1, WEN Jing1, ZHAO Xue1, CHEN Jinguang2   

  1. 1. School of Computer and Information Technology, Shanxi University, Taiyuan Shanxi 030006, China;
    2. School of Computer Science, Xi'an Polytechnic University, Xi'an Shaanxi 710048, China
  • Received:2016-04-25 Revised:2016-05-30 Online:2016-11-10 Published:2016-11-12
  • Supported by:
    This work is partially supported by the National Natural Science Foundation of China (61201453, 61201118), the Basic Research Project in Shanxi Province (2014021022-2), the Higher School Science and Technology Innovation Project in Shanxi Province (2015108).

摘要: 针对照明变化、形状变化、外观变化和遮挡对目标跟踪的影响,提出一种基于加速鲁棒特征(SURF)和多示例学习(MIL)的目标跟踪算法。首先,提取目标及其周围图像的SURF特征;然后,将SURF描述子引入到MIL中作为正负包中的示例;其次,将提取到的所有SURF特征采用聚类算法实现聚类,建立视觉词汇表;再次,通过计算视觉字在多示例包的重要程度,建立“词-文档”矩阵,并且求出包的潜在语义特征通过潜在语义分析(LSA);最后,通过包的潜在语义特征训练支持向量机(SVM),使得MIL问题可以依照有监督学习问题进行解决,进而判断是否为感兴趣目标,最终实现视觉跟踪的目的。通过实验,明确了所提算法对于目标的尺度缩放以及短时局部遮挡的情况都有一定的鲁棒性。

关键词: 加速鲁棒特征, 多示例学习, 潜在语义分析, 目标跟踪, 支持向量机

Abstract: Concerning the influence of changing light, shape, appearance, as well as occlusion on target tracking, a target tracking algorithm based on Speeded Up Robust Feature (SURF) and Multi-Instance Learning (MIL) was proposed. Firstly, the SURF features of the target and its surrounding image were extracted. Secondly, SURF descriptor was introduced to the MIL as the examples in positive and negative bags. Thirdly, all the extracted SURF features were clustered, and a visual vocabulary was established. Fourthly, a "word document" matrix was establish by calculating the importance of the visual words in bag, and the latent semantic features of the bag was got by Latent Semantic Analysis (LSA). Finally, Support Vector Machine (SVM) was trained with the latent semantic features of the bag, so that MIL problem could be handled in accordance with the supervised learning problem. The experimental results show that the robustness and efficiency of the proposed algorithm under the variation of scale, gesture and appearance, as well as short-term partial occlusion.

Key words: Speeded Up Robust Feature (SURF), Multi-Instance Learning (MIL), Latent Semantic Analysis (LSA), target tracking, Support Vector Machine (SVM)

中图分类号: