A Smart Target Detection System using Fuzzy Logic and Background Subtraction

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395 -0056
Volume: 04 Issue: 03 | Mar -2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 5.181 | ISO 9001:2008 Certified Journal | Page 2264
A smart target detection system using Fuzzy Logic and Background Subtraction
Madhurika Ubhe1, Deepak Parashar2, Prof. U. A. Jogalekar3, Vaishali Godse4, Kshitija Jarhad5
1,2,4,5 Students, Computer Department, Smt. Kashibai Navale College of Engineering, Pune-411041
3Assistant Professor, Computer Department, Smt. Kashibai Navale College of Engineering, Pune-411041
---------------------------------------------------------------------***---------------------------------------------------------------
Abstract - In the proposed system, fuzzy logichasbeenused
to model a robust Background Subtraction method for object
detection. We first separate the background from the input
static video using fuzzy logic. For this, the first few frames of
the input video are used as a template for the background.
After this, the objects in motion are identified and then,
whether the moving object is a car or human is inferred based
on the aspect ratio of the moving object. We implement target
detection in static videos using background subtraction and
fuzzy logic. In this project, we are using humans and cars as
our targets. We first identify our targets - cars/bikes and/or
humans if they are present in the video and then discriminate
between the two to show accuracy i.e. whether the system can
differentiate between cars/bikes and human. Thus, target
detection as well as identification can be implemented using
the proposed system.
Key Words: Detection, Background Subtraction, Fuzzy
Inference System, Fuzzy Logic, Background Modeling,
Moving Object Detection, Aspect Ratio.
1. INTRODUCTION
1.1 About Target Detection and our System
Target detection has become increasingly important
with respect to various security applications like bank
surveillance, traffic surveillance, and residential area
surveillance to check for intruders, etc.
Target detection has many applications apart from the
security perspective but we chose to exploit its uses for
security related applications like surveillance and
remote monitoring, because it a crucial need of the
present society where bank robberies, residential thefts,
illegal hunting, industrial espionage, etc. are rampant.
Target detection in images and videos involves image
processing. Various image processing techniques such
as denoising, edge detection, background elimination
and many others come into play when target detection
in videos and images is to be performed.
The requirement of all these techniques makes target
detection a bit complicated to implement for mediocre
developers and even if a successful implementation is
found, the speed of execution is an issue of concern.
Since the complex image processing operations take an
acceptably large amount of time to be performed,
software’s which are robust tothecomplexitiesofimage
processing and at the same time meet the “instant
result” criterion of today’s users , are the need of the
present scenario.
We present a novel approach to detect targets in static
videos using Fuzzy Logic and Background Subtraction.
The intricacies of the proposed system will makeitclear
that we have wisely exploited the nature of surveillance
applications. We are workingonlywithhumansandcars
as our targets, but animals, or other immobile objects
can also be used in place or in conjunction with humans
and cars.
1.2 Overview of our Target Detection System
As is seen the architecture, first a video is given as input to
our system. The video may be captured through a live
camera or it may be an already recorded static video.
First, the frames are extracted from the video using a
suitable frame extractingsoftware.Denoisingtechniques are
then applied to the frames to reduce blurring. This helps in
accurate pixel intensity calculation so that there is little
confusion in classification of pixels in the later stages.
Then, RGB (Red Green Blue) values of each pixel arethrough
RGB channel separation.
The first N frames of the video are used as a template for
background modeling. The larger the value of ‘N’, the more
accurate will be the classification of background pixels. The
redundancy of the intensities of the pixels inthe‘N’frames is
captured and it is used as a model for the background. But
the major requirement here is that the target object should
not be present in the first ‘N’ frames that we have chosen for
background modeling.Thebackgroundmodel ispreparedby
applying fuzzy logic on the RGB values obtained. The output
after applying fuzzy logic on the pixel intensities is a
fuzzified value for each pixel. We can then set thresholds

based on the fuzzified values for classifying each pixel as
background or foreground one.
Fig.1.2 Proposed System Architecture
Once the background model is prepared, the subsequent
frames are assessed for performingbackgroundsubtraction.
This is done by again performing fuzzification on the pixels
in the frames, then comparing the fuzzified values with the
threshold values set during the background modeling stage.
This will be explained in detail in section
After the background is subtracted, the moving objects are
detected in the foreground. Initially, after background
elimination, only the silhouettes of the moving objects are
obtained. The aspect ratio of the silhouettes is calculated by
drawing a rectangular blob around the silhouettes i.e. the
aspect ratio is the ratio of width of the rectangletotheheight
of the rectangle. If the aspect ratio is less than 1, the target is
identified as human, otherwise it is assumed to be a car. The
thresholds to be referred for car and human classification
are also specified as there can be moving objects other than
cars or humans too in the video like street dogs, cats, etc.
Thus, our targets are detected successfully. However, there
are certain limitations of our application.
2. Limitations
2.1 Challenges to Smart Eye : If there is not enough
contrast between the background and foreground,
or between the target and background, our
application cannot separate the background and
detect those targets. Ex. If The human is wearing
clothes of those colors which match some part of
the background, there are chances that the human
may not be detected successfully. This is the only
glitch in our application as of now.
2.2 Requirements to be satisfied by the input videos :
1. The input videos must be of a high resolution.
2. There should be enough contrast between the
foreground and background.
3. The videos must be shot by a stationarycamera
i.e. the position of the camera should remain
constant throughout the video shoot.
4. The target should appear after 3 seconds at
least after the start of the video.
2.3 Use of Fuzzy Logic in our system :
Fuzzy Logic is based on “degreesoftruth” whichcan
lie between 0 and 1 as opposed to Boolean Logic, in
which only two distinct values are used 0and 1.
3. CONCLUSIONS
We have implemented a smart target detectionsystemusing
fuzzy logic and background subtraction. Our system can
detect targets accurately given that contrast is maintained
between the target and background. This works well for
static as well as real videos being shot by a live camera.
Background separation using fuzzy logic is a robustandnew
method to detect targets .The only limitation being the
requirement of contrast between background and intensity
difference.
4. Future scope
1. We have used static videos for target detection. Our
system can also be used to detect targets in live videos being
captured by live camera. This has application in traffic
surveillance, detect hunters and keep track of endangered
animals in wildlife sanctuaries.
2. An android app can be built which will show the alerts
generated by our target detection system on mobile phones
this way, remote monitoring is possible The alert is
generated when a forbidden target is detected in the input
e.g. humans in a no entry server room

3. Advanced feature extraction methods can be applied to
detect targets of myriad categories. E.g. immobile thingslike
call phones, laptops, etc.
5. ACKNOWLEDGEMENT
It gives us great pleasure in presenting the report for our
dissertation on ECC Feature Based Target Detection Using
Sparse Representation. We would like to take this
opportunity to thank our internal guide Prof. U.A. Jogalekar
for giving us all the help and guidance we needed. Her
perceptual motivation, patience and excellent expertise in
discussion during progress of the project work have
benefited us to an extent, which is beyond expression.
REFERENCES
[1] I. Haritaoglu, D. Harwood, and L. S. Davis, “W4: Real-
Time Surveillance of People and their Activities,” IEEE
Transactions on Pattern Analysis and Machine Intelligence,
vol. 22, no. 8, pp. 809–830, 2000.
[2] M. Lalonde, S. Foucher, L. Gagnon, E. Pronovost, M.
Derenne, and A. Janelle, “A System to Automatically Track
Humans and Vehicles with a PTZ Camera,” in SPIE Visual
Information Processing XVI, vol. 6575, 2007.
[3] Z. Zhang, Y. Cai, K. Huang, and T. Tan, “Real-Time Moving
Object Classification with AutomaticSceneDivision,”in IEEE
International Conference on Image Processing, vol. 5, 2007,
pp. V149–V152.
[4] J. See, S. W. Lee, and M. Hanmandlu, “Human Motion
Detection using Fuzzy Rule-Base Classification of Moving
Blob Regions,” in International Conference on Robotics,
Vision, Information and Signal Processing, 2005, pp. 398–
402.
[5] R. D. de L´eon and L. E. Sucar, “Human Silhouette
Recognition with Fourier Descriptors,” in 15th International
Conference on Pattern Recognition, vol. 3, 2000, pp. 709–
712.
[6] L. M. Brown, “View Independent Vehicle/Person
Classification,” in 2nd international workshop on Video
surveillance & sensor networks, 2004, pp. 114–123.

A Smart Target Detection System using Fuzzy Logic and Background Subtraction

More Related Content

What's hot (20)

Similar to A Smart Target Detection System using Fuzzy Logic and Background Subtraction (20)

More from IRJET Journal (20)

Recently uploaded (20)

A Smart Target Detection System using Fuzzy Logic and Background Subtraction