基于机器视觉的目标识别追踪算法及系统设计

Design of object recognition and tracking algorithm and system based on machine vision

为实现对特定动态目标的准确识别与平稳追踪,提出了一种基于机器视觉的智能算法及系统设计。系统以Jetson TX2为控制核心,采用基于YOLOv3算法的卷积神经网络本地实现动态目标的识别,并结合基于模糊逻辑思想的追踪避障算法识别出目标运动的转向、转动幅度及障碍物位置,实现了对系统的运动进行精确控制且准确避障。系统采用的目标识别追踪算法可以实现对前端摄像头传感器采集的视频流进行实时本地处理及反馈。经测试,系统工作稳定,具有准确性高、实时性强等优点,可以广泛应用于智能跟随行李箱、安防巡逻机器人等智能产品,具有极高的市场应用价值。

In order to achieve accurate identification and stable tracking of specific dynamic targets,an intelligent algorithm and system based on machine vision is proposed. The system uses the Jetson TX2 as the control core,making full use of the YOLOv3 algorithm-based convolutional neural network to realize the dynamic target recognition locally. Combined with the fuzzy logic-based tracking obstacle avoidance algorithm,the target ’s steering,rotation amplitude and obstacle position are identified. It is realized that the precise control of the motion of the system and accurate avoidance for obstacle. The target recognition algorithm used in the system performs real-time local processing and feedback on the front-end video stream. After testing,the system has stable operation,high accuracy and strong real-time performance. It can be widely used in intelligent following luggage,security patrol robots and other intelligent products,with high market application value.