提出和实现了一种基于嵌入式设备的停车场视频监控和车位检测方案.在嵌入式Linux操作系统下,通过调用计算机视觉库OpenCV中的图像处理函数,实现多车位的检测,检测结果保存到嵌入式数据库中.通过嵌入式设备将车位信息发送给PC或手持设备,并以图像的形式显示,达到远程车位检测的目的.本系统还移植了Mjpgstreamer实现停车场的远程视频监控.车位状态的检测采用2种判据:一是车位范围内差影的均值;二是车位区域内的边缘点数.测试结果表明,该方案可以很好地运行在嵌入式平台,为停车诱导提供支持.
In this paper,one of the parking lot video monitoring and parking space detection scheme was put forward and implemented.Detection of multiple spaces was realized by the image processing function of computer vision libraries OpenCV under the embedded Linux operating system,,the detection result was stored in the embedded database.In order to achieve the purpose of remote parking detection,the parking information was transmitted to the PC or handheld devices by embedded devices and displayed as an image.The video surveillance of remote parking lot was realized by transplanting Mjpg-streamer to embedded devices.The detection algorithm is based on two criterias which are the mean of the car shadow and the edge pixels of parking area.Experiment result show that the scheme can run well on embedded platform and provide support for parking guidance.