Sie sind auf Seite 1von 7

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/257932015

Automatic Alert of Security Threat through Video Surveillance System

Conference Paper · July 2013

CITATIONS READS
0 1,502

3 authors, including:

Vipin Shukla Gaurav Kumar Singh


Pandit Deen Dayal Petroleum University Institute for Plasma Research
3 PUBLICATIONS   1 CITATION    10 PUBLICATIONS   1 CITATION   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Study of thwo phase flow in fusion magnets View project

All content following this page was uploaded by Gaurav Kumar Singh on 29 May 2014.

The user has requested enhancement of the downloaded file.


Automatic Alert of Security Threat through Video Surveillance System
Vipin Shukla1 , Gaurav Kumar Singh2, Dr. Pratik Shah3
1,2
Department of Nuclear Engineering
3
Department of Electrical Engineering
Pandit Deendayal Petroleum University, India
Email: vipin.ssne12@sne.pdpu.ac.in,gaurav.ssne12@sne.pdpu.ac.in,pratik.shah@sot.pdpu.ac.in

ABSTRACT

Close Circuit Television Camera (CCTV) has played very important role in many surveillance and
security systems. However, such system requires continuous monitoring by human and hence there
is possibility of failure because of boredom or fatigue. The requirements of continuous monitoring
can be avoided using sensor systems which can alert the human on the occurrence of undesired
event. Sensors only detect events and do not provide information about the threat. Hence use of
CCTV camera and sensor systems, independently or jointly, may not be sufficient for timeliness
detection of undesired events. When combined with humans and sensors, cameras can provide an
immediate method to assess a scene of interest. The human efforts in monitoring can greatly
reduced by using activity analysis of video captured by CCTV. By analyzing the captured video,
information about the threat and cause can be obtained very quickly and accurately in order to take
mitigating actions. However, in the absence of light, camera cannot detect such threat. Hence, some
hybrid techniques need to be implemented, which can work in different lighting conditions. So our
focus is to design a system that will detect threat in time under different lighting conditions using
camera and sensor networks. The system will be aided with smart, measureable, reliable and robust
algorithm for motion detection, tracking and activity analyses. This paper reviews different
approaches for detecting object and its motion, tracking of object and activity analysis in order to
prevent adverse consequences. It also proposes methodology to improve the security of Nuclear
Power Plant using existing approaches with appropriate modification.

1. INTRODUCTION

Automated video surveillance system [8, 9] has emerged as an important research topic in the
computer vision community. The growth in this area is being driven by the increased availability of
inexpensive computing power and image sensors, as well as by the inefficiency of manual
surveillance and monitoring system. Applications such as event detection, human action
recognition, and semantic indexing of video are being developed in order to partially or fully
automate the task of surveillance. Recently, there has been a large amount of work reported
addressing the motion detection and tracking of object [1, 2, 3, 4]. To be more effective, such
applications require real-time motion detection and tracking algorithms [10, 11, 12], which provide
low-level functionality upon which higher level recognition capabilities can be built.

1.1 Goals and Objectives

Abnormal motion detection is the key to effective and economical video transmission, recording,
alarm assessment in security systems. It can be used to identify the location of threat and hence
immediate mitigating actions can be taken before the occurrence of undesired event. The main
objective of this paper is to identify various approaches for motion detection, tracking and activity
analysis and to propose a method to improve threat detection in timely manner under various
conditions. Detection of abnormal motion will raise an alarm which will alert the security personnel
to take necessary action. This approach will not suffer from fatigue and boredom. The scope of this
paper will be limited to activity analysis. It is assumed that sensor network will provide input to the
algorithm to take appropriate actions for capturing under poor lighting conditions.

2. THEORY/METHODOLOGY

The video surveillance system refers to continuous monitoring activities in a specific area through
the CCTV cameras without benefit of intrusion sensor to direct attention to a specific area.
Assessment of scene takes place is followed by monitoring. Assessment refers to analysis of the
scene where undesired activity has been detected. This type of surveillance system can be
implemented in places where consequence of loss can be tolerated. In other words, such system
does not prevent occurrence of undesired event. In Nuclear Industry consequence of loss of asset is
unacceptably high so we have to go for better alternative for surveillance and alarm assessment
system. Effective protection system must incorporate some automated surveillance technology to
assist in the detection function and reduce the chances of occurrence of adverse event.

So our attempt will be to detect of abnormal activities using motion detection technique, object
tracking methods and activity analyses. The security personnel will be provided with accurate
location of such event in order to analyze and act as soon as possible. The system may be enabled to
automatically record the complete adversary action. Background subtraction [5, 6, 7] is considered
greatly as an intelligent algorithms for the same.

2.1Background Subtraction Algorithm

Detection of motion in many current tracking systems relies on the technique of background
subtraction. Background subtraction is a widely used approach for detecting moving objects in
videos recorded from static cameras. The background image must be such that it should not contain
any moving objects and must be kept regularly updated to adapt the varying lighting conditions and
geometry settings. By subtracting background image from the incoming video frames, the presence
of an object and its motion can be tracked. Graphical representation of background subtraction
technique is shown in the Figure 1.

Figure 1: Basic Background Subtraction Algorithm


The methods also exist for background estimation, which will establish the model of background for
the subtraction. There is no need to update background images. Various techniques for background
estimations are listed below.

 Estimating median over time - This algorithm will update the median value of the time series
data based upon the new data sample. The will increment or decrement the median by an
amount that is related to the running standard deviation and the size of the time series data. The
approach will also apply a correction to the median value if it detects a local ramp in the time
series data. Overall, the estimated median is constrained within Chebyshev's bounds, which are
square root (3/5) of the standard deviation on either side of the mean of the data.
 Computing median over time - This method will compute the median of the values at each pixel
location over a time window of 30 frames.
 Eliminating moving objects - This algorithm will identify the moving objects in the first few
image frames and labels the corresponding pixels as foreground pixels. Next, the algorithm
identifies the incomplete background as the pixels that do not belong to the foreground pixels.
As the foreground objects move, the algorithm estimates more and more of the background
pixels.

Once the background image is estimated, it is subtracted from each video frame to produce
foreground images. By thresholding and performing morphological closing on each foreground
image, the model produces binary feature images (Figure 2c). One such example is shown in Figure
2. The model locates the cars in each binary feature image draws a green rectangle around the cars
that pass beneath the white line. The counter in the upper left corner of Figure 2d tracks the number
of cars in the region of interest.

a b

c d
Figure 2: Example output from tracking cars using background estimation techniques
For human motion analysis also similar techniques are used [14]. For real time performance,
automatic surveillance needs to compute very complex algorithm very fast and accurately. For such
a system, we need to analyze not only the motion of people, but also the posture of human. The
postures of human can provide important clues for the understanding of their activities. A
background-based estimation and body-based detection are performed to capture the human motion
and analyze the human outline using various edge detection algorithms [13]. Activity analysis can
be carried out by analyzing human posture in subsequent frames and can be classified as normal or
suspicious behavior of human being present at the scene. We will analyze various such algorithms
and will come up with the new algorithm that will classify the behavior of the human in various
categories and appropriate actions can be taken if the behavior is found to be suspicious.

3. EXPERIMENTAL DESIGN

3.1 Hardware
i. High-end Computing Machine for fast computation.
ii. DM648 Digital Video Development Platform from TI.
iii. CCTV camera system for live video capture.

3.2 Software
i. MATLAB®, Simulink®.
ii. Code Composer Studio.
iii. Operating System: Windows XP or Higher.

4. EXPECTED RESULTS

This system will be successfully used to track people in indoor environments and will analyze their
activities in order to raise an alarm on the occurrence of abnormal event. This system will achieve
our goals of reducing or preventing occurrence of hazardous event by detecting any anomaly at the
scene of interest very quickly. A video monitoring, detection, tracking system is expected to be
developed successfully by using this approach. This system will provide an efficient method for
surveillance purposes and is expected to be highly beneficial.

By using this method, continuous observation of live video can be avoided as well as storage space
is reduced by continuous deletion of previous unwanted recordings, only desired recordings are
stored. One approach for reducing storage requirement is shown in Figure 3. The camera will
capture the video continuously, but the recording will be done only for duration of two consecutive
spans of 60 minutes. If in the second span of 60 minute abnormal activity is not reported then, the
recording of first span of 60 minutes will be deleted and the space will be available for third span of
60 minutes. This process will continue until abnormal activity is detected. If during any span,
abnormal activity is detected then, system will not delete the current span as well as previous span
of 60 minutes for further analysis of threat. Using such approach, we can greatly reduce the storage
requirement without losing information of undesired events.
(a) Initially A will be recorded. (b) B is being recorded and no abnormal activity is
detected during time slot B.

(c) Recorded Video during time slot (d) New recording will start for the (e) Abnormal activity is detected
A will be deleted. time slot C. during C. The system will not delete
recording of slot B and will keep
recording beyond slot C until operator
gives signal.
Figure 3: Mechanism for recording of video

5. CONCLUSION
Real-time monitoring of human movements provides an effective means of inferring a person’s
level of activity. This project initiates the development of a system for nuclear facility monitoring
and study of adversary behaviors. The advantage of this approach is that allows human operators to
use context-based alerts and the response to these alerts is much faster.

6. REFERENCES
[1] Nahum Kiryati, Tammy Riklin Raviv, Yan Ivanchenko, Shay Rochel , “Real-time Abnormal
Motion Detection in Surveillance Video”,19th International Conference on Pattern
Recognition,dec 2008.
[2] D. Gutchess, M. Trajkovi´c, E. Cohen-Solal, D. Lyons, A. K. Jain, “A Background Model
Initialization Algorithm for Video Surveillance”,Proc. of The Eighth IEEE International
Conference on Computer Vision, Vancouver, July 9-12, 2001.
[3] Jiingecin Lim, Hong K. Cho, and Jong Beom Rat, “An Improved Video Object Traking
Algorithm Based on Motion Re-estimation”, image processing,2000.
[4] M. F. Kazemi, A. H. Mazinan, A. Amir-Latifi, “A Knowledge-Based Objects Tracking
Algorithm in Color Video Using Kalman Filter Approach”, International Conference on
Information Retrieval & Knowledge Management , 13-15 March 2012.
[5] Massimo Piccardi,“Background subtraction techniques: a review”,IEEE International
Conference on Systems, Man and Cybernetics , 10-13 Oct. 2004.
[6] Niel Joubert,“Background modelling and subtraction” for object detection in video.
Stellenbosch University, (BEng Sci, 2009).
[7] Sorayya Panahi, Samira Sheikhi, Shahrzad Hadadan, Niloofar Gheissari, “Evaluation of
Background Subtraction Methods”, Digital Image Computing: Techniques and Applications
(DICTA), 2008.
[8] Mubarak Shah, “Automated Visual Surveillance in Realistic Scenarios”,MultiMedia,
IEEE (Volume:14 , Issue: 1 ), Jan.-March 2007.
[9] M. Valera and S.A. Velastin, “Intelligent distributed surveillance systems: a review”.
Vision, Image and Signal Processing, IEE Proceedings - (Volume:152 , Issue: 2 ),8 April
2005.
[10] R. Bogush, S.Maltsev, A. Kastryuk, N. Brovko, D. Hlukhau, “Motion detection and tracking
algorithms in video streams”, VNU Journal of Science, Mathematics - Physics 25 (2009) .
[11] Ashish Kumar Sahu, Abha Choubey, “A Motion Detection Algorithm for Tracking of Real
Time Video Surveillance”, International Journal of Computer Architecture and Mobilitym
(ISSN 2319-9229) Volume 1-Issue 6, April 2013.
[12] Wael Abd-Almageed, Mohamed Hussein, Mohamed Abdelkader Larry Davis, “Real-Time
Human Detection and Tracking from Mobile Vehicles” ,Proceedings of the 2007 IEEE
Intelligent Transportation Systems Conference Seattle, WA, USA, Sept. 30 - Oct. 3, 2007.
[13] Raman Maini, Dr. Himanshu Aggarwal, “Study and Comparison of Various Image Edge
Detection Techniques”. International Journal of Image Processing (IJIP), Volume (3) : Issue
(1),august 30,2010.
[14] Bhavana C. Bendale, Prof. Anil R. Karwankar, “Moving Object Tracking in Video Using
MATLAB”,International Journal of Electronics, Communication & Soft Computing Science
and Engineering,ISSN: 2277-9477, Volume 2, Issue 1,2013

View publication stats

Das könnte Ihnen auch gefallen