Drowsy Driver Detection System: A Novel Approach Using Haar Like Features

Abstract

There are many instances of road accidents round the world due to driver's lack of attention in driving. One of the prime reasons can be drowsiness. In this paper, a drowsiness detection system using Haar like feature technique is implemented for eye detection and a simulator for testing the system was built. An audible alert is connected with the system to acknowledge the drowsiness of the driver. A simulator is made by SDL (Simple Direct Media Layer) which is a cross platform multimedia development API library to provide low level access to joystick, mouse, etc. Whenever the driver is outside the lane in the simulator then also it generates an audible warning signal. Keywords - Eye Detection, Face Detection, Haar like features, USART. I. INTRODUCTION One of the main reason for the accidents round the world is the lack of attention of the driver. A driver may get distracted by many things. But out of all, the main reason is the drowsiness which may be due to anything like night shifts, heavy workload, etc. There are many methods used to detect a driver's drowsiness. Some of them are: Image processing based techniques such as PERCLOS (1), Haar like features, Physiological signal detection approach using Electroencephalogram (EEG) (2), Electrocardiogram (ECG) (3) and Monitoring the lateral position of the vehicle. This paper presents one such technique for drowsiness detection of a driver. This paper is divided in following parts: Section II gives the basic idea of the system, Section III presents a detailed approach for the system, Section IV shows the results and the simulator, Section V concludes the paper by its conclusion. II. BASIC CONCEPT The idea behind the implementation of this system is to use a webcam to detect a driver's eyes. A steering wheel is also interfaced with the PC to drive the car in the simulator. Image processing algorithms are used to detect eye's movement. Serial communication is used to get the acknowledgement of the drowsiness. A buzzer is interfaced via Atmega8 to get an audible sound alert. SDL library is used to make the simulation for driving conditions. The basic block diagram is shown in figure 1.