Real-Time Driver Attention Monitoring and Recording system

• Main Authors: Cai,Ming-Sin, 蔡明信
• Other Authors: 吳建中
• Format: Others
• Language: zh-TW
• Published: 103
• Online Access: http://ndltd.ncl.edu.tw/handle/86911709222858637254

碩士 === 南台科技大學 === 資訊工程系 === 102 === Because the driver is fatigued or distracted, to enhance the traffic safety to avoid the accident.The main contribution of this paper is A Real-Time Driver onitoring and Recording System, a driver-assisted system that improved the traffic safety to use the smartphone based on Android and NeuroSkyMindWaveMobile headset. A Real-Time Driver Attention Monitoring and Recording System detect a face in the system through a front-camera by a smartphone. Through the capture to image detection face, using the projector positioning method to detect the position of the eyes, the eyes after the opening of the vertical projection of the calculation to determine the status of the driver's eyes open, closed, and the use of concentration indices defined in this paper to determine the driver's concentration, when this index is greater than 3, indicating that the driver is in a serious state without concentration, when the index fell 1 to 3, are mild non-state concentration. An attention of driver by monitoring via NeuroSkyMindWave Mobile headset measure the signal of the electroencephalogram (EEG) and recorded the EEG informationinto SD card, Because the initial system development and testing requires a lot of driver behavior data, so in addition to develop a real-time traffic simulation system to assist in driving the development of monitoring and recording the concentration by a large number of repeated experiments to verify the accuracy of the system. According to different eye detection methods to test and analyze images in 480x640 resolution for static and dynamic experiments, projection positioned correctly in the benchmark rate proposed in this paper is better than Haar-like features eye classifier this paper fixed area prediction method, the correct rate of 85.7%, while in terms of speed detection method is better than the projection positioned to detect the speed of the other two methods, an image processing time is 114ms.