The invention relates to road congestion detection, in particular to a congestion detection method based on an embedded terminal, and the method comprises the steps: obtaining real-time video stream data in a road monitoring scene, carrying out the vehicle recognition, and determining a target vehicle in a preset region; utilizing a first neural network model to extract a road feature map used for representing a road free area and a road occupation area from the real-time video stream data; acquiring a driving area map according to the real-time video stream data, and determining an optical stream matrix corresponding to the driving area map by using a second neural network model; obtaining a plurality of tracking images based on the tracked target vehicle, and determining the traffic flow speed in the preset area by using the tracking images; determining a vehicle mask image and a position mask image in the preset area, and determining traffic flow density in the preset area based on the vehicle mask image and the position mask image; according to the technical scheme provided by the invention, the defect of poor road congestion detection accuracy in the prior art can be effectively overcome.
本发明涉及道路拥堵检测,具体涉及一种基于嵌入式终端的拥堵检测方法,获取道路监控场景下的实时视频流数据,并进行车辆识别,确定预设区域内的目标车辆;利用第一神经网络模型从实时视频流数据中提取用于表征道路空闲区域和道路占用区域的道路特征图;根据实时视频流数据获取行车区域图,并利用第二神经网络模型确定行车区域图对应的光流矩阵;基于跟踪目标车辆得到多张跟踪图像,并利用跟踪图像确定预设区域内的车流速度;确定预设区域内的车辆掩膜图像和位置掩膜图像,基于车辆掩膜图像和位置掩膜图像确定预设区域内的车流密度;本发明提供的技术方案能够有效克服现有技术所存在的道路拥堵检测准确性较差的缺陷。
Congestion detection method based on embedded terminal
一种基于嵌入式终端的拥堵检测方法
2022-11-22
Patent
Electronic Resource
Chinese
European Patent Office | 2022
|European Patent Office | 2020
|European Patent Office | 2018
|European Patent Office | 2023
|European Patent Office | 2022
|