IMAGE AND BRAIN SIGNAL PROCESSING BASED DROWSINESS DETECTING AND ALARMING SYSTEM

Saranya K, Assistant Professor, Department of Computer Science and Engineering, Kumaraguru College of Technology, Coimbatore, Tamil Nadu, India, Jayanthy S, Syed Ali Fathima S J, Professor, Department of Electronic Communication and Engineering, Sri Ramakrishna Engineering College, Coimbatore, Tamil Nadu, India, Assistant Professor, Department of Computer Science and Engineering, Kumaraguru College of Technology, Coimbatore, Tamil Nadu, India
2019 MATTER International Journal of Science and Technology  
Accidents due to drowsiness are considered as silent killers. In India, accident rates are more than 20% and it is increasing year by year. Loss of consciousness causes changes in human body which leads to drowsiness. Detecting drowsiness of drivers while driving is a big challenge but at the same time serious road accidents are mostly due to micro sleep and fatigue. Drowsiness detection of drivers uses modern technology that helps the drivers to prevent accidents due to drowsiness. A system
more » ... siness. A system that automatically detects drowsiness of the driver in real time using computer vision is proposed. This multi-model technique makes use of Signal and Image processing for effectively detecting drowsiness. The face is detected by using a camera and the facial landmarks were captured for analysis and the eye region is being extracted for calculating EAR (Eye Aspect Ratio) 191 which would help in understanding the open and close movement of eyes. Added to that the Electro Encephalography (EEG) signals of the brain are captured using non-invasive method and the processed signals are used to check the alertness of driver. This multi-model drowsiness detection system detects the difference between blinking and drowsiness easily. The value which is being fetched from both the proposed techniques is fused and given as input to the alert system for providing a caution to the drivers. The number of road accidents could be avoided on successful implementation of this system as accuracy of detection is expected to increase in decent scale compared to the existing systems.
doi:10.20319/mijst.2019.52.190201 fatcat:l66lix47rrflbgkuexxf2qbwye