Drowsy Driver Identification Using Eye Blink detection

As field of signal processing is widening in various security and surveillance applications, motivated the interest for implementing better application with less complications. A non-intrusive machine vision based concepts is used to simulate Drowsiness Detection System. The system is consisting of web camera which placed in a way that it records driver’s head movements in order to detect drowsiness. As drowsiness is detected, a signal is issued to alert the driver. The system deals with detecting face, eyes and mouth within the specific segment of the image. All the possible actions have been considered and output is generated accordingly. Drowsiness is determined by observing the eye blinking action of the driver. Other than drowsiness, driver’s attention while driving is also considered. The proposed algorithm is developed to minimize the complexity level from existing system while efficiency has given prime importance which was a main objective of the paper. The system is implemented using cascade object identifier from vision toolbox of Matlab, which detects face, eyes, nose and mouth from the image which is captured from web camera. For this system Region of Interest is location of eyes and mouth which are determined and indicated by rectangle. Logic has been used here to identify whether eyes are open or closed unlike general methods. From mouth portion yawning is determined and considered. Project is simulated for on line and off line video with all possible situations of a driver. Results are formulated under different categories like normal driver, driver with glass under different light intensities. It is concluded that proposed system can also be utilized for other application. Results obtained from the proposed system provide efficient system analysis and overall good efficiency with some precautions by using simple flow of programming. KeywordsDrowsiness, ROI, Surveillance, EEG, EOG.