针对违章停车的检测范围广、时间与位置不确定、车辆大小种类多、车辆身份难以自动识别等问题,设计了一种全方位视觉传感器(ODVS)和高速快球摄像机相融合的违章停车智能监控系统;利用ODVS在动态图像序列中进行大范围多目标车辆对象的检测与跟踪,得到大范围内各跟踪车辆对象目标的位置、大小、运动轨迹等信息;一旦发现车辆对象停留在非泊车位上一段时间后,控制高速快球对车辆对象进行抓拍,接着对抓拍图像进行车牌识别;实验结果表明,通过ODVS和高速快球摄像机的视频图像信息融合,即使在车辆对象非约束的情况下也能很好地获取到可用于车牌识别的图像,同时能满足多目标实时检测与跟踪的要求;该系统不仅能给欲违章停车的司机提供警示信息,同时也能给数字城管、执法部门装备一双智能化的慧眼。
Aiming at the problems of peccancy parking detection in a large range, uncertainty of time and location, vehicles with multiple size and species as well as automatic recognition of vehicle license plate, this paper designed a peccancy parking intelligent monitoring system based on the fusion of Omni--Directional Vision Sensors (ODVS) and PTZ camera. The ODVS is used to detect and track multiple targets of vehicle object in a large range to acquire the vehicle information such as location, size and trajectory. Once a tracked vehicle is stopped but not in the parking spaces for a period of time, the PTZ camera will be controlled to capture the vehicle image which will be used to recognize the vehicle license plate. The experimental results show that image used for vehicle license plate recognition can be captured under unlimited conditions by this device, and the demands of the Real--time Multi--objective detection and tracking can also be met. The system will not only provide warning information to drivers who might be peccancy parking, but also gives digital urban management and law enforcement departments a pair of intelligent eye.