计算机与现代化

• 人工智能 •    下一篇

基于SVM和HMM二级模型的行为识别方案

  

  1. (1.厦门大学计算机科学系,福建 厦门 361005; 2.厦门大学智能科学系,福建 厦门 361005)
  • 收稿日期:2015-02-02 出版日期:2015-05-18 发布日期:2015-05-18
  • 作者简介:苏竑宇(1988-),男,福建安溪人,厦门大学计算机科学系硕士研究生,研究方向:人工智能,信息安全; 陈启安(1963-),男,福建永安人,教授,硕士,研究方向:多媒体技术及应用,嵌入式系统,软件人机界面设计及网络应用; 吴海涛(1990-),男,厦门大学智能科学系硕士研究生,研究方向:人工智能,信息安全。

Human Activity Recognition Based on Combined SVM&HMM

  1. (1. Department of Computer Science, Xiamen University, Xiamen 361005, China; 2. Department of Cognitive Science, Xiamen University, Xiamen 361005, China)
  • Received:2015-02-02 Online:2015-05-18 Published:2015-05-18

摘要: 人体行为识别对于个人辅助机器人和智能家居等一些智能应用,是非常必要的功能,本文运用SVM&HMM混合分类模型进行日常生活环境的人体行为识别。首先,使用微软的Kinect(一种RGBD感应器)作为输入感应器,提取融合特征集,包括运动特征、身体结构特征、极坐标特征。其次,提出SVM&HMM模型,SVM&HMM二级模型发挥了SVM和HMM各自的优点,既结合了SVM适于反映样本间差异性特点,又发挥了HMM适合处理连续行为的特点。该二级模型克服了单一SVM模型、传统HMM模型和在人体复杂和相似行为建模过程中精度、鲁棒性和计算效率上的不足。通过大量实验,结果表明SVM&HMM二级模型对室内日常行为的识别具有较高的识别率,且具有较好的区分性和鲁棒性。

关键词: Kinect, 行为识别, 融合特征, SVM, HMM

Abstract: Being able to recognize human activities is essential for several intelligent applications, including personal assistive robotics and smart homes. In this paper, we perform the recognition of the human activity based on the combined SVM&HMM in daily living environments. Firstly, we use a RGBD sensor (Microsoft Kinect) as the input sensor, and extract a set of the fusion features, including motion, body structure features and joint polar coordinates features. Secondly, we propose a combined SVM&HMM model which not only combines the SVM characteristics of reflecting the difference among the samples, but also develops the HMM characteristics of dealing with the continuous activities. The SVM&HMM model plays their respective advantages of SVM and HMM comprehensively. Thus, the combined model overcomes the drawbacks of accuracy, robustness and computational efficiency compared with the separate SVM model or the traditional HMM model in the human activity recognition. The experiment results show that the proposed algorithm possesses the better robustness and distinction.

Key words: Kinect, activity recognition, fusion features, SVM, HMM

中图分类号: