摘要
为了解决移动机器人快速精准检测和跟踪目标的问题,提出一种基于目标颜色和目标形状的自主跟踪目标的算法,该算法在TLD(tracking-learning-detection)框架下,用KCF(核相关滤波器)算法在TLD(tracking-learning-detection)框架下替换的跟踪器,并采用HOG(Histogram of Gradient)特征和CN(Color Names)颜色特征的目标识别方法,让HOG特征与CN颜色特征相互补充,并在该算法的基础上设计了一款基于视觉目标跟踪的全向轮移动机器人。实验结果表明,该目标跟踪算法具有较好的实时性、精准度和鲁棒性,提升机器人在遮挡处理和目标丢失等情况的鉴别能力。采用该视觉目标跟踪的机器人的目标跟踪精度可以达到91.5%,跟踪成功率达到71.2%。
In order to solve the problem of rapid and accurate detection and tracking of targets by mobile robots,I propose an algorithm for autonomous target tracking based on target color and target shape.Under the framework of TLD(tracking-learning-detection),this algorithm uses the KCF(Kernel Correlation Filter)algorithm to replace the TLD algorithm tracker,and uses HOG(Histogram of Gradient)features and CN(Color Names)color features for target recognition method,make HOG feature and CN color feature complement each other,and design an omnidirectional wheel mobile robot based on visual target tracking based on this algorithm.Furthermore,the experimental results show that the target tracking algorithm has good real-time performance,accuracy and robustness,and improves the robot's ability to discriminate in occlusion processing and target loss.And the target tracking accuracy of the robot using this vision target tracking can reach 91.5%,and the tracking success rate can reach 71.2%.
作者
缪文南
周政
陈雪娇
曹闹昌
MIAO Wen-nan;ZHOU Zheng;CHEN Xue-jiao;CAO Nao-chang(School of Electronic&Information Engineering,Guangzhou College of South China University of Technology,Guangzhou 510800,China)
出处
《组合机床与自动化加工技术》
北大核心
2021年第7期52-56,共5页
Modular Machine Tool & Automatic Manufacturing Technique
基金
2019年华南理工大学广州学院百人计划“优秀骨干教师”科研项目(53-CQ18YG23)。
