1

Driver drowsiness identification by eye detection

Zafar khan, R.R.Patane
Terna Engineering College, Nerul, Navi Mumbai.
Khan9226@gmail.com, rrpatane@yahoo.co.in

Abstract This paper presents an automatic vision-base

drowsy driver detection system. A new automatic eye detection
method to detect eye region is first presented. The reason is that
the performance of the drowsiness detection system depends on
the robustness of the eye detection system to obtain a set of
visual features e.g. eye region. The proposed method also takes
account for the case when the drivers head has completely
dropped down or not looking forward and alarm is given to alert
the driver. An innovative image processing algorithm as Drowsy
driver detection methods can form the basis of a system to
potentially reduce accidents related to drowsy driving. In this
method Region of Interest (ROI) is going to be an eye, whose
statistical property (average blob area) is considered.Fatigue is
detected based on the statistical value calculated for the ROI.
This project is focused on the localization of the eyes which
involves looking at the entire image of the face and determining
the position of the eyes by a self developed image-processing
algorithm ROI. Once the position of the eyes are located, the
system is designed to determine whether the eyes are opened or
closed to detect fatigue.
Key words eye detection,. Proposed method for detecting
driver drowsiness,result,conclusion.

I. INTRODUCTION
Drivers drowsiness or fatigue has been recognized as one of
the main causes of accidents. It is also an important cause of
combination-unit truck crashes. Intensive research has been
undertaken to try to assess the part of the total accident that
could be a consequence of drowsiness. Two major areas are
investigated. The first area intends to have better
understanding of drowsiness mechanisms in relation with
various factors in order to propose recommendations to
reduce the risk. The second area aims at proposing in-vehicle
alerting devices to counteract driver drowsiness. Devices can
be grouped into two categories. In this 1st category, the
devices are designed to sustain alertness by some form of
activation of the driver throughout the trip. The 2nd category
includes devices, which detect drowsiness followed by a
warning or an intervention. These devices are based on the
continuous measurement of indicators used for assessment of
drowsiness or of impaired driving performance. Several
researchers considered the video-based system. A
scientifically supported measure of drowsiness associated with
slow eye closure.
In this vision-based drowsy driver detection system. A
block diagram of the major functions of the proposed drowsy
driver detection system is shown in Fig. 1. The system has
two detection functions: eye detection system and drowsiness

detection system. The performance of the drowsiness

detection system clearly relies on a robust set of visual
features. Eye area obtained from the accurate eye detection
system. Therefore a new automatic eye detection method to
detect eye region is also presented. Besides eye detection, the
method can determine the drivers head is tilted or not
looking
forward and give an alarm to alert the driver subsequently.
After the image capturing, the image is processed by the eye
detection system that perform the detection of eye region. In
this project as per the block diagram continuous image is
taken by camera and continuous image is converted into gray
scale from the gray scale image only upper part of face is
capture that is eye part from the face on which mask is
applied and other part is discarded .On mask part (which is
detected by camera) converted into binary image that is black
and
white
image
by
using
thresholding
technique.Thresholding is depend upon the intensity of
surrounding. If black pixel count is more than predefine value
then it consider or detect as a drowsy for continuous frame
otherwise it go to check state of eye.
Rest of the paper is organized as follows: in Section II
the eye detection approach based region of interest (ROI),
while Section III deals with proposed method for detecting
driver drowsiness Section IV show flow chart. Section V
future work VI advantages VII deals with results and VIII
concluding remarks.

II. Eye detection on the image

In this method first reference image is capture by camera
which is 30cm away in front of driver face .From the
image system convert RGB to gray of extracted part by
rectangular Region of interest(ROI) and because both eyes
have same information that why it only detect and perform
operation on left eye only. Now by using histogram
equalization we adjusting image intensities to enhance
contrast of image. Let f be a given image represented as a
matrix of integer pixel intensities ranging from 0 to L 1.
L is the number of possible intensity values often 256.
Let p denote the normalized histogram of f with a bin for
each possible intensity. So
No. Of pixel with intensity n
Pn= ---------------------------------------

n = 0, 1, ..., L

1.
Total No. Of pixel

VESIT , International Technological Conference-2014 (I-TechCON), Jan. 03 04, 2014

249

2
The histogram equalized image G will be defined by
fi,j

Input Video Frame

by Frame

Gi,j=floor((L-1)pn)
n=0

where function floor() rounds down to the nearest integer.

Find Location of Eye

in Face

Check the Status of Eye

(open or close)

Drowsy Detected

Convert to Gray
Scale

Threshold

Drowsy NotDetected

Alarm
Fig. 1: Block diagram of Eye detection system
After histogram equalization thresholding is use to converting blob properties such as shape, size, area, centroid &
a gray scale input image to a bi-level image by using an orientation remain constant and varied within short duration
optimal threshold.global thresholding is used to threshold to in effect of normal eye blinks & turning back. If suppose the
get binary image global thresholding a new binary image can average blob area decays gradually and remains for a short
then be produced, in which objects are painted completely period then it is detected as fatigue. It will be evaluated, by
black, and the remaining pixels are white. Denote the original continuously checking for the average blob area in each
image f(x, y), then the thresholded product is achieved by incoming frames. Once the fatigue is detected the system
scanning the original image, pixel by pixel, and testing each gives some special warning/alarm to keep the drivers
pixel against the selected threshold: if f(x, y) > T, then the attention out from drowsy state.
pixel is classified as being a background pixel, otherwise the
pixel is classified as an object pixel. This can be summarized
in the following definition, where b(x, y) denotes the
thresholded binary image.

IV. FLOW CHART

After that system will calculate only black pixel in image and
save it as a reference image.Now camera will capture
continuous image frame by frame and convert each frame
into gray image then perform histogram equalization and
thresholding like reference image. Then system will count
blake pixel in binary image and compere continuous
frame(image) with reference image. As we know close eyes
are more blake pixel then open eyes.so if blake pixel count is
more then reference pixel count for continuous four frame
then system will detect it as drowsiness otherwise it go bake
to detect another frame.
III. PROPOSED METHOD FOR DETECTING DRIVER
DROWSINESS

A special non-intrusive monitoring system which measures

eye closure,more specifically; defined by the statistical
approach. The video frames are separated from the captured
video file.Then to locate the drivers eye (ROI),Still the preprocessed frame has noises and other different illumination
level causes different blobs will be removed by thresholding.
The properties of eye remain constant with same distance,

Fig. 2: Flow chart

In this project real time image from the camera is compare
with reference image which is save in data base initially
camera will capture open reference image of eye convert it
into gray scale because of both the eyes are same feature only
one eye is going to save as reference image for comparison

VESIT , International Technological Conference-2014 (I-TechCON), Jan. 03 04, 2014

250

3
that is half part of reference image is cut thresholding is
applied on that to convert the gray scale image into binary
image this binary image is actual reference to real time image
to compare.
After camera will capture video frame by frame each frame is
compare with reference image.After capturing image of both
eyes that is converted into gray scale image as like reference
image and then one eye is cut (left eye basically) from the
both eyes that cut eye image is converted into black and white
pixel that is binary image by Applying threshold on it. If
black pixel is less then reference image black pixel after
comparing it will consider as driver is not drowsy then next
image is going to compare this is repeated again again until
and unless black pixel of continuous image is not more than
black pixel of reference image if it is greater than reference
image black pixel continuously four time or continuously four
frames are above the given threshold of reference image
system will detect as drowsiness of driver and generate alarm
to alert driver.
V. FUTURE WORK
Currently there is not adjustment in zoom or direction of the
camera during operation. Future work may be to
automatically zoom in on the eyes once they are localized.
This would avoid the trade-off between a wide field of view in
order to locate the eyes, and a narrow view in order to detect
fatigue. This system only looks at the number of consecutive
frames where the eyes are closed. At that point it may be too
late to give the warning. By studying eye movement patterns,
it is possible to find a method to generate the warning sooner.
This will also allow adaptability to changes in ambient light.
The system does not work for dark skinned individuals. This
can be corrected by having an adaptive light source. Adaptive
light source would measure the amount of light being
reflected back from face. If little light is being reflected,
intensity of the light is increased. Darker skinned individual
need much more light so that when the binary image is
constructed, the face is white, and the background is black.
VI. ADVANTAGE OF PROPOSED SYSTEM
Drowsy driver detection system over comes the drawback of
the existing system in an effective way by using non-intrusive
monitoring system. In proposed systems the eye closure
measures effectiveness of this methodology in detecting the
drowsiness and also we are going to find the drowsiness by
head moments, is more when compared to other existing
methodologies. Drivers distraction by using sensors is
reduced.This is effective and efficient monitoring system, to
insure accuracy to detecting drowsiness is implemented.
VII. RESULTS
TABLE 1
Reference Continuous

image
reading
7935

image
readings
:
8180
8075
8667
8457
8522
8521
8484
8028

Fig. 3.Graph
If continuous four frame blake pixel count is 300 more then
reference blake pixel count then and then only system will
detect drowsiness last 16 to 19 frame are continuous 300
more blake pixel count then reference so this situation is
drowsy for the system.
VIII. CONCLUSION
An automatic vision-based drowsy driver detection system is
presented in this paper. This system has eye detection and
drowsiness detection functions. This paper mainly focuses on
a new automatic eye detection method to detect eye region to
avoid accident due to drowsiness.

REFERENCE
[1] Eriksson, M and Papanikolopoulos, N.P. Eye-tracking
for Detection of
DriverFatigue, IEEE Intelligent
Transport System Proceedings (1997), pp 314-319.
[2] Gonzalez, Rafel C. and Woods, Richard E. Digital
Image Processing, Prentice Hall: Upper Saddle
River,
N.J., 2002.
[3] Takehiro Ito, Shinji Mita, Driver blink measurement
by
the motion picture
processing
and its application
to
drowsiness
detection,
IEEE
5th
International
Conference on Intelligent Transportation Systems ,
Singapore, 2002.
[4] Takchito Hayami, Katsuya M., Detecting Drowsiness

VESIT , International Technological Conference-2014 (I-TechCON), Jan. 03 04, 2014

251

4
while driving by measuring eye movement a pilot
study,
IEEE 5th International Conference on
Intelligent Transportation Systems , Singapore, 2002.
[5] Andrea Cristiani, Marco Porta, Davide Gandini Driver
drowsiness identification by means of passive
techniques for eye detection and tracking 2010 Fourth
IEEE International Conference.
[6] Yuriy Kurylyak, Francesco Lamonaca Detection of the
Eye Blinks for Humans Fatigue Monitoring
IEEE 978-1-4673-0882-3/12, June 2012.
[7] C. Papageorgiou, M. Oren, and T. Poggio, A general
framework for object detection, International
Conference on Computer Vision, 1998.
[8] Performance Evaluation of Statistical Approach for
Drowsiness Detection for Drivers with and without
Spectacles (IEEE 2010)

VESIT , International Technological Conference-2014 (I-TechCON), Jan. 03 04, 2014

252