关键词: 视频压缩, 车牌自动识别, 神经网络训练

Abstract:

           We design a vehicle borne electronic police system that includes radar speed detection, video capture, PTZ and lens control and data management of illegal evidence. The system employs serial communications, video compression, license plate recognition and neural network training to monitor, capture, record and punish such illegal behavior as vehicles over speed, driving on single and double yellow lines, illegal parking and Uturn, reverse traffic, illegal overtaking, etc. This system therefore provides great convenience for the forensics of illegal traffic behavior. Experimental results show that the system has high reliability, high recognition rate, better realtime performance and higher comprehensive efficiency.

Key words: video compression, license plate recognition, neural network training