【摘要】 在车型识别系统中,车辆的特征提取是非常重要的。传统的检测方法受摄像机的成像精度与架设方式等因素的影响,给车辆特征的提取带来困难。提出了一种采用虚拟线圈识别车型的方法。该方法将检测线与虚拟线圈相结合,对提取的车辆信息进行预处理,以检测线检测到达预先设置位置的车辆,然后触发虚拟线圈来提取车辆特征,通过BP神经网络来训练识别的车辆特征,以达到车型识别的目的。试验证明,该方法识别效果好,识别有效率达99%。更多还原

【Abstract】 Because of different span and imaging of the video camera,the distance of the vehicles in the image greatly affects the abstraction of vehicles’ features.A pre-procession method to extract the vehicle feature was presented in this paper,which combined the detection line and virtual loops.Firstly,detection line was used to detect the pre-set position.Then,if the vehicle arrived at the detection line,the virtual loops would work to extract the vehicle feature.At last,the feature vectors were trained to recognize the vehicle module with the BP neural network.Experimental results verify the effectiveness of the proposed method with a recognition effectiveness rate of 99%.更多还原