使用物体交互模型的车辆违停事件检测

随着平安城市与监控摄像头的发展,车辆违章停靠事件的自动检测在视频自动检测方法中具有重要意义.当前的目标检测算法,例如基于特征+分类器、基于视频背景差分、基于深度学习的方法等,大多关注车牌、交通标志的定位与识别,而忽略了对车辆违章等事件的判别.因此,发展了一种基于物体交互模型的车辆违章停靠事件检测方法:首先基于轮廓提取和SVM进行车牌检测;其次基于形状和颜色进行交通标志检测;最后基于物体交互模型来判断是否违停.上述方法在真实视频场景中的实验验证了方法的有效性和可靠性.

Illegal parking detection using object interaction model

With the development of safe city and surveillance cameras,automatic detection of illegal parking events is of great sig nificance in the video automatic detection methods.Lots of target detection algorithms exist including feature + classifier method,video background subtraction method,and the deep-learning method.These methods are mosdy concerned about the position and confirmation of license plates and traffic signs.According to this,a method based on object interaction model to detect illegally parking event is proposed in this paper:Firstly license plate is detected based on contour extraction and SVM;secondly the traffic sign is detected based on the shape and color;the last part judges whether illegal parking exists based on object interaction model.The above-mentioned method is regarded as a validity and reliability method by being tested in the real video scene.