《中国康复理论与实践》

《中国康复理论与实践》 ›› 2018, Vol. 24 ›› Issue (10): 1210-1214.doi: 10.3969/j.issn.1006-9771.2018.10.016

• 康复工程与辅助技术 • 上一篇    下一篇

基于视觉信息的智能下肢假肢路况识别

喻贝贝1,2,3, 孟青云1,2,3, 喻洪流1,3, 曹武警1,3, 赵伟亮1,3   

  1. 1.上海理工大学康复工程与技术研究所,上海市 200093;
    2.上海健康医学院,上海市 201318;
    3.上海康复器械工程技术研究中心,上海市 200093
  • 收稿日期:2018-05-25 出版日期:2018-10-25 发布日期:2018-10-30
  • 通讯作者: 孟青云。E-mail: mengqy1996@126.com
  • 作者简介:喻贝贝(1993-),男,汉族,湖北孝感市人,硕士研究生,主要研究方向：下肢智能控制。
  • 基金资助:
    1.国家自然科学基金项目(No. 61473193); 2.上海康复器械工程技术研究中心资助项目(No. 15DZ2251700)

Road Condition Identification of Intelligent Lower Limb Prosthesis Based on Visual Information

YU Bei-bei1,2,3, MENG Qing-yun1,2,3, YU Hong-liu1,3, CAO Wu-jing1,3, ZHAO Wei-liang1,3   

  1. 1. Institute of Rehabilitation Engineering and Technology, University of Shanghai for Science and Technology, Shanghai 200093, China;
    2. Shanghai University of Medicine&Health Sciences, Shanghai 201318, China;;
    3. Shanghai Engineering Research Center of Assistive Devices, Shanghai 200093, China
  • Received:2018-05-25 Published:2018-10-25 Online:2018-10-30
  • Contact: MENG Qing-yun. E-mail: mengqy1996@126.com
  • Supported by:
    National Natural Science Foundation of China (No.61473193) and Shanghai Engineering Research Center of Assistive Devices Project (No.15DZ2251700)

PDF (PC)

摘要： 目的 从纯视觉信息角度出发,对智能膝关节进行路况识别的研究。方法 整个实验环境采用卤素灯照明,使得设定的场景的光亮度和色温变化较小。将机器视觉模块嵌入智能膝关节的控制系统,通过摄像头实时采集路况图像数据,并将所采集的图像进行灰度化处理,利用归一化互相关算法(NCC)进行图像模式匹配,从而识别出对应的路况。结果 通过机器视觉模块识别平地、上/下坡、上/下楼梯5种路况的准确率分别为88.6%、85.3%、78.4%、87.5%、77.9%。识别效果较好,识别时间均在1 s以内,实时性较强。结论 基于纯视觉信息进行智能膝关节假肢路况识别具有有效性和可行性。

关键词: 智能膝关节, 路况识别, 机器视觉, 模式匹配

Abstract: Objective To carry out a research of terrain recognition on intelligent knee joint from the perspective of pure visual information.Methods The whole experimental environment was illuminated by halogen lamp, which made the brightness and color temperature of the set scene change less. The machine vision module was embedded in the intelligent knee joint control system. The road image data was collected by the camera in real time, and the captured image was processed in grayscale. The normalized cross correlation algorithm (NCC) was used to match the image pattern to identify the corresponding road conditions.Results The accuracy rates of the road vision, up/down hill, and up/down stairs recognized by the machine vision module were 88.6%, 85.3%, 78.4%, 87.5%, and 77.9%, respectively. The recognition effect was good, and the recognition time was within one second, real-time performance was strong.Conclusion The effectiveness and feasibility of intelligent knee joint prosthetic road condition recognition based on pure visual information is proved.

Key words: intelligent knee joint, terrain recognition, machine vision, pattern matching

中图分类号: