A Driver Warning System for the Android System

Up until now, a lot of driving safety assistant systems have been developed. As to the low stability and precision, android system is confined in reality use. This paper aims at the real-time warning system based on android APP. In order to decrease the traffic accident rate, a warning system based on video detection and android system is developed. The real-time visual detection and tracking approach benefit from Camshift and Kalman filter with frame differential in region of interest (ROI) can detect the vehicles successfully. With the direct linear transformation (DLT) calibration method, motion parameters are easily extracted. Besides, current images are collected and shown to the drivers involved in danger by APP. With another APP the real latitude and longitude could be easily obtained in one mobile phone. Comparing with the actual road surveying data, the calibration accuracy could easily be obtained. When hazardous vehicle behavior is detected by PC, warning signals would be released to attract the related drivers’ attention. The experimental results indicate that the proposed system is of high precision and real time.