利用Jetson TK1平台的行人运动目标跟踪系统  

作　　者：许喜斌 赵小蕾[2] XU Xibin;ZHAO Xiaolei(Guangdong Engineering Polytechnic College,Guangzhou 510520;Xinhua College of Sun Yat-Sen University,Guangzhou 510520)

机构地区：[1]广东工程职业技术学院,广州510520 [2]中山大学新华学院,广州510520

出　　处：《计算机与数字工程》2021年第1期99-104,共6页Computer & Digital Engineering

基　　金：广州市科技计划项目(编号:201804010265);中山大学新华学院科研基金一般项目(编号:2018YB013)资助。

摘　　要：针对图像检测识别和目标跟踪技术中存在的抗遮挡性弱、无法应对目标丢失、对目标多尺度变化适应不了等问题,设计开发了基于Jetson TK1平台及计算机视觉OpenCV(Open Source Computer Vision Library)的行人运动目标跟踪系统,利用GPU高效图像处理能力,结合改进的KCF(Kernelized Correlation Filters)跟踪算法,实现对监控区域内行人的运动跟踪。实验结果表明,Jetson TK1平台的行人运动目标跟踪系统稳定可靠,具有良好的抗遮挡性和准确性。Aiming at the problems in detection and recognition and target tracking technologies,such as weak anti-blocking ability,inability to cope with target loss,and inability to adapt to multi-scale changes of targets,a pedestrian motion target tracking system is designed and developed based on Jetson TK1 platform and computer Visual OpenCV(Open Source Computer Vision Li⁃brary).Using GPU efficient image processing capabilities combined with improved KCF(Kernelized Correlation Filters)tracking al⁃gorithm,the system can achieve pedestrian tracking in the monitoring area.The experimental results show that the pedestrian mo⁃tion tracking system based on Jetson TK1 platform is stable and reliable with well anti-blocking and higher accuracy.

关 键 词：JetsonTK1平台 KCF算法 抗遮挡性 目标跟踪 

分 类 号：TN973.1[电子电信—信号与信息处理]