基于OpenMV辅助进食机械手人脸检测算法研究  

Research on face detection algorithms of auxiliary feeding manipulator based on openMV

作　　者：刘祚时[1] 许志良 张平 LIU Zuoshi;XU Zhiliang;ZHANG Ping(School of Mechanical and Electrical Engineering,Jiangxi University of Science and Technology,Ganzhou 341000,China)

机构地区：[1]江西理工大学机电工程学院,江西赣州341000

出　　处：《传感器与微系统》2021年第1期39-41,45,共4页Transducer and Microsystem Technologies

基　　金：国家自然科学基金资助项目(71361014);江西省科技计划资助项目(2015BBE50038);江西省研究生创新专项资金资助项目(YC2018-S325)。

摘　　要：针对手脚不便的老人或者上体截肢的特殊人群用餐问题,提出了基于机器视觉的辅助进食机械手设计方案。针对系统的设计需求,分析了基于Haar特征Adaboost人脸检测分类器的组织结构以及检测原理,结合基于视觉的目标追踪比例-积分-微分(PID)控制算法,利用其响应快鲁棒性强等特点,有效实现人脸移动的追踪要求。最后通过辅助进食调试,机械手能够完成对人脸的检测和辅助进食工作,达到了预期的控制效果。Aiming at the the problem of feeding for the elderly or the upper limb amputees,a design scheme of feeding assistant manipulator based on machine vision is proposed.Aiming at the design requirement of the system,the structure and detection principle of Adaboost face detection classifier based on Haar feature are analyzed,and the visual-based object tracking proportional-integral-differential(PID)control algorithm is combined to make use of its features of fast response and strong robustness,effective realize requirement of face tracking.Finally,the manipulator can complete the detection of the human face and feeding-assistant work through feeding-assistant debugging,and achieve the desired control effect.

关 键 词：OpenMV平台 HAAR特征 人脸检测 比例-积分-微分(PID)追踪算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]