计算机工程与应用 ›› 2021, Vol. 57 ›› Issue (10): 163-172.DOI: 10.3778/j.issn.1002-8331.2008-0245

• 模式识别与人工智能 • 上一篇    下一篇

基于计算机视觉的轮椅跟随系统设计

李艳,唐达明,周莹亮,戴庆瑜   

  1. 1.陕西科技大学 电气与控制工程学院,西安 710021
    2.陕西农产品加工技术研究院,西安 710021
  • 出版日期:2021-05-15 发布日期:2021-05-10

Design of Wheelchair Following System Based on Computer Vision

LI Yan, TANG Daming, ZHOU Yingliang, DAI Qingyu   

  1. 1.School of Electrical and Control Engineering, Shaanxi University of Science and Technology, Xi’an 710021, China
    2.Shaanxi Agricultural Products Processing Technology Research Institute, Xi’an 710021, China
  • Online:2021-05-15 Published:2021-05-10

摘要:

为解决老年人口日益增多而看护人员缺乏的问题,设计了一种具有自动跟随功能的智能轮椅系统。该系统通过计算机视觉传感器获取图像信息后,首先利用自适应高斯混合建模的背景减除法完成对目标的检测,再利用基于多特征融合的粒子滤波的TLD(Tracking-Learning-Detection)跟踪算法完成对目标的跟踪,最后结合双目定位算法得到目标在相机坐标系下的三维位置。根据目标的位置控制轮椅的速度和方向,使轮椅与目标人物保持设定的距离和相对方位,完成对目标的跟随。对轮椅跟随系统各项功能进行测试,结果表明,该系统不仅能够完成对使用者的实时跟随,同时也能实现对障碍物安全的躲避。

关键词: 计算机视觉, 智能轮椅, 自动跟随, 目标跟踪

Abstract:

In order to solve the problem of increasing number of elderly people and lack of nursing staff, an intelligent wheelchair system with automatic follow-up function is designed. After the system obtains image information through the computer vision sensor, it firstly uses the background subtraction method of adaptive Gaussian mixture modeling to complete target detection, then adopts particle filter-based TLD(Tracking-Learning-Detection) tracking algorithm based on multi-feature fusion to complete target tracking, and finally combines the binocular positioning algorithm to get the three-dimensional position of target in camera coordinate system. After that, the speed and direction of wheelchair are controlled according to target position, so that the wheelchair and the target person maintain a set distance and relative orientation to complete the follow-up of target. Experimental results show that the proposed system can not only complete the real-time following of user, but also can realize the safe avoidance of obstacles.

Key words: computer vision, intelligent wheelchair, automatic following, target tracking