Intelligent Video Surveillance System PDF

Case Study Report on Intelligent Video Surveillance System
Akoma et al

Module Leader: Dr S. Mori Module Assist: Sharmama M.S Page 1
TITLE PAGE

CASE STUDY

INTELLIGENT VIDEO
SURVEILLANCE SYSTEM
CASE STUDY ON HUAWEIS INTELLIGENT VIDEO
SURVEILLANCE SYSTEM

AKOMA, CHIEMELE O. (11065981)
IGBETA, OLUSHOLA D. (05055868)
NLEMADIM, ONYINYECHI (11047106)
MODULE CODE: CTP 148N
MODULE NAME: ADVANCED COMMUNICATION SYSTEMS
MODULE LEADER: DR MORI S. VAEZI-NEJAD

FACULTY OF COMPUTING
LONDON METROPOLITAN UNIVERSITY
166-220 HOLLOWAY ROAD N7 8DB
LONDON, UK
Akoma et al

TABLE OF CONTENT
TITLE PAGE 1
TABLE OF CONTENT 2
CHAPTER ONE 3
1.0 INTRODUCTION 3
1.1 HUAWEI'S INTELLIGENT SURVEILLANCE SYSTEM 3
CHAPTER TWO 6
2.0 EVOLUTION OF VIDEO SURVEILLANCE 6
2.1 ANALOG CCTV SYSTEM USING DVR: 7
2.2 NETWORK DVR-BASED ANALOG CCTV SYSTEMS 7
2.3 VIDEO ENCODER-BASED NETWORK VIDEO SYSTEMS 8
CHAPTER THREE 9
3.0 ARCHITECTURE OF AN INTELLIGENT VIDEO SURVEILLANCE SYSTEM 9
3.1 ACQUISITION 9
3.2 TRANSMISSION 12
3.3 COMPRESSION 13
3.4 PROCESSING 14
3.5 ARCHIVING 15
3.6 DISPLAY 16
CHAPTER FOUR 17
4.0 VIDEO ANALYTICS 17
4.1 REASONS FOR DEPLOYING VIDEO ANALYTICS 17
4.2 ADVANTAGES OF VIDEO ANALYTICS 18
4.3 LIMITATIONS AND DISADVANTAGES OF VIDEO ANALYTICS 18
4.4 DESCRIPTION OF VIDEO ANALYTICS TECHNIQUES 18
CHAPTER FIVE 23
5.0 SECTORS OF APPLICATION 23
5.1 RETAIL TRADE 24
5.2 TRANSPORTATION 24
5.3 EDUCATION 25
5.4 GOVERNMENT AND PUBLIC SECURITY 25
5.5 AIRPORTS 26
5.6 STATIONS AND PUBLIC TRANSPORTATION 26
5.7 BANK SETTING 26
5.8 OTHER 27
CONCLUSION 28
REFERENCES 29

Akoma et al

CHAPTER ONE
INTELLIGENT VIDEO SURVEILLANCE SYSTEM

1.0 INTRODUCTION
The goal of this report is to briefly describe an Intelligent Video Surveillance System, using The
Huaweis Intelligent Video Surveillance System.

Intelligent video surveillance system can be defined as the real-time monitoring within a specific
environment of both persistent and transient object. Top priorities of this system is to maintain access
control to secure area, monitor perimeter for intrusion and important personalities, watch public areas
for suspicious behaviors, unattended object, etc [2-3].

The evolution of intelligent video surveillance systems, architecture (acquisition, transmission,
processing, compression, archiving and display), video analytics (object detection, recognition,
tracking and behavior), and applications of intelligent video surveillance system are briefly covered in
this case study report, with a brief introduction to the Huaweis Intelligent Video Surveillance System.

1.1 HUAWEI'S INTELLIGENT SURVEILLANCE SYSTEM
The Huaweis Intelligent Video Surveillance System is designed to provide high definition, high
reliability, high performance, manageable small and medium video surveillance services to meet the
needs of its clients. To over the limitations posed by traditional video surveillance systems, Huawei
uses new technologies in digital video coding, compression, digital networks and digital storage, with
real-time monitoring, recording video playback, control centers, alarm detection and sharing, and
intelligence analysis. It adopts an all-IP implementation, using network cameras, encoders and storage
devices [4-5].

High points of the Huaweis Intelligent Video Surveillance System include [4-5]
a. Unified Management: it supports automatic deployment, remote maintenance, reports and alarm
management. It can monitor the entire systems temperature and fan speed in real time.
b. The system incorporates the Smart NVR 3000, which consists of a flexible, highly efficient and
reliable storage system. It uses RAID 5 to maximize hard disk usage and provide up to 16TB
storage from the recordings of 100 cameras.
Akoma et al

c. The system integrates several intelligence analysis modules which can be used for license plate
recognition, behavior analysis, demographics, and quality video analysis. All of these
incorporated together reduces operation cost, improves efficiency and increases accuracy in
monitoring.
d. The system adopts various energy-saving techniques to reduce costs, especially when its is
lightly loaded. It also incorporates an enhanced heat dissipation design and fan rotation speed
control based on the temperature of heat prone components.

Figure 5: Huawei Intelligent Video Surveillance System

Akoma et al

The Huaweis Symnatec Surveillance system consists of four (4) parts, namely: Video storage
management platform system, surveillance centre system, transmission network system and front end
system [3]. This is shown in figure 5 above.

The video management platform basically provides platform management, intelligent storage
management and client management. The transmission system performs the transmission of video
services using any TCP/IP supported network including Ethernet, PON networks and wireless
networks. The front end system consists of video encoders, cameras and alarm probes. The surveillance
system is made up of the large display system, software/hardware decoder, console and surveillance
management for professional users [4].
Akoma et al

CHAPTER TWO
2.0 EVOLUTION OF VIDEO SURVEILLANCE
Video surveillance is an industry that is more than 30yrs old and also had its share of technology
changes. The demands in a video surveillance market includes;
Simplified installation and maintenance
Longer retention of recorded video
Better image quantity
Reduction in costs
Size and scalability
More secure and reliable technology
Remote monitoring capabilities
More limit in system intelligence
In order to meet these requirements, video surveillance has experienced a number of technology shifts.
The latest is the shift from analog cctv surveillance to fully digital, network based video surveillance
systems. Video surveillance systems started out as 100 percent analog systems and are gradually
becoming digital [1, 4-5].

Video surveillance industry saw its first digital revolution with the introduction of the DVR in the mid-
1990s. Most DVR had several video imputs, With early DVRS, hard disk space was limited. The
introduction of the DVR system provided the following major advantages [1, 4-5]
No need to change tapes
Consistent image quality
Ability to quickly search through recorded video
The sections below outline the evolution of video surveillance systems. Different system
configurations, from fully analog to fully digital, are explained, along with the benefits of each
configuration. The solution comprises of the number of digital components but do not represent fully
digital systems. The video is constantly being transported over an IP network and which are fully
scalable and flexible in network video systems [1, 4-5].

Akoma et al

2.1 ANALOG CCTV SYSTEM USING DVR:
The diagram below represents an analog CCTV system using DVR(Digital video recorder).This is an
analog system with digital recording, In a DVR, the video is digitized and compressed so as to store as
as many videos as possible, The videotape is replaced with hard drives for the video recording.

Fig1:Analog CCTV using DVR (Digital video recorder)

ANALOG CCTV USING VCR: An analog cctv using VCR(video cassette recorder)represents a fully
analog system consisting of analog cameras with coax output connecting to the VCR from recording.
The video of the VCR is not compressed. It uses the same type of cassettes as a home VCR.An analog
monitor is used to monitor the video [1, 4-5].

Fig 2 showing a classical analog video surveillance system

2.2 NETWORK DVR-BASED ANALOG CCTV SYSTEMS
DVR was equipped with an Ethernet port for network connectivity. This introduced network DVRs to
the market and enabled remote video monitoring using PCs . Some network DVR systems in use today
enable the monitoring of both live and recorded video, whereas some allow the monitoring of only
recorded video. therefore, some systems require a special Windows client to monitor the video,
whereas others use a standard Web browser; the latter makes remote monitoring more flexible [1, 4-5].
Akoma et al


Figure 3. This system shows how analog cameras can be networked using a network DVR for remote
monitoring of live and recorded video.
The network DVR system provides the following advantages:
Remote monitoring of video via a PC
Remote operation of the system
2.3 VIDEO ENCODER-BASED NETWORK VIDEO SYSTEMS
A video encoder connects to analog cameras and digitizes and compresses the video. It then sends the
video over an IP network via a network switch to a PC server that runs video management software for
monitoring and recording (Figure 4). This is a true network video system because the video is
consistently sent over an IP network. In essence, the tasks previously performed by the DVR are now
divided up - with the digitization and compression being done by the video encoder and the recording
by the PC server - thus providing better scalability [1, 4-5].

Figure 4. A network video system, where video is continuously transported over an IP network. It uses
a video encoder as the cornerstone to migrate the analog security system into an open IP-based video
solution.
A video encoder-based network video system has the following advantages:
Scalability in steps of one camera at a time
Possibility to record off site
Use of standard network and PC server hardware for video recording and management
Scalability in steps of one camera at a time
Future-proof because the system is expanded easily by incorporating network camera
Akoma et al

CHAPTER THREE
3.0 ARCHITECTURE OF AN INTELLIGENT VIDEO SURVEILLANCE SYSTEM
Based on the distribution of components, an Intelligent Video Surveillance system can be classified in
Centralized and Distributed platforms. In centralized intelligent video system, video content analysis is
done at the back-end, using DVR to process and analyze data from all cameras. But in a distributed
platform, intelligent network cameras are used to analyze video content in real time for prompt
response. Advantages of the distributed platform include enables easy expansion in adding new
cameras and scalability, deployment cost, decrease in server workload, storage and bandwidth usage
and content accuracy, reduced labour cost [6-7]. The Huaweis system uses a distributed platform.

Figure 6: Video content analysis of (a) Centralized and (b) distributed architecture [6]
The components of a video surveillance system are briefly discussed below.
3.1 ACQUISITION
This refers to the various camera models used to acquire the videos/sounds into the surveillance
system. The camera models could analogue or digital, and some may be power-operated. Before the
advent of IP or network cameras, analogue or CCTV was in used, in which recordings was stored on
tapes. This was then replaced by digital video recorder (DVR) which converts the analogue videos to
digital. IP surveillance system uses cameras that are Internet Protocol (IP) enabled to transmit digital
data over an Ethernet or wireless network. An IP or network camera is completely a digital system.
New intelligent video surveillance systems deploy IP/network cameras alongside the analogue cameras.
The advantages of using IP cameras include, standard-based open architecture, function capabilities
(remote and shared viewing via internet or wireless), facilitation of intelligent surveillance (facial
recognition, object recognition, license plat recognition, automated alert/response, etc), scalability, cost,
recording, transmission and monitoring, better resolution [6,8-10]. This can be summarized in table 1.

Akoma et al


Table 1: Comparison between CCTV and IP surveillance system
CCTV (ANALOGUE) IP SURVEILLANCE (DIGITAL)
Cost Tend to use coaxial cables
which are expensive to deploy
and large cable runs involved.
Cameras and operating costs
can be expensive
Reduced system cost and added functionality
Less cabling required
IP cameras are inexpensive

Resolution CCTV cameras have improved
image quality but lack support
for megapixel image resolution
Support for variety of image resolutions
including both standard analogue CCTV and
megapixel resolutions

Scalability Can be complex to deploy as
cables runs with CCTV
cameras can be thousands of
feet
Expanding requires adding
camera capacity in units of 16
and additional cabling
Easy convergence onto new existing IP
cabling infrastructure
Easy and flexible to deploy-PoE allows the
camera to be placed just about anywhere
Very scalable-can be configured to meet your
precise requirements.

Functionality Many of the analogue cameras
lack some of the more
advanced features such as
digital zoom
Integration of video surveillance with other
systems and functions such as access control,
alarm systems, building management, traffic
control management etc. offers automated
real-time alerting via email, text or file
transfer in response to video motion detection
Allows for secure, remote access from any
device on the network-or outside the network
via VPN or https
Allows for secure, unlimited data storage.

Source [9]

Figure 7: Comparison between DVR and IP camera operations [10]

Akoma et al

An IP camera consists mainly of a lens, image sensor and processor, video compression (SoC) and an
Ethernet chip for network connectivity and data transmission [6].

Figure 8: An IP camera

IP camera can be divided in the following classes
a. Fixed: it Points in a single/fixed direction. It is used to cover a defined area such as an entrance,
part of a parking lot, etc. It is the camera of choice if the camera is to monitor a direction and be
visible, but it is prone to vandalism [6-7]
b. PTZ (Pan-Tilt-Zoom): it is Power-operated and can be manually or automatically configured for
pan/tilt/zoom movements to cover a wider field such as lobbies or parking lots. It can be used to
follow objects or individuals moving or even to zoom into particular areas of interest, such as a
license plate [6-7
c. Dome: it can be fixed or speed dome type. It is covered by a case, making it less obvious to
vandalism and weatherproof. The power-operated versions can do a 360 horizontal sweep and
180 vertical sweeps [6-7].
d. Megapixel: it offers a higher resolution compared to standard cameras, ranging from 1 to 16
megapixels. It is used to capture more detailed images or to cover a wider field, thereby
reducing the number of cameras needed for that scene. It offers the possibility of zooming into
the picture using the software as an alternative to the PTZ mechanical camera (which is subject
to wear in its parts). Its high resolution enables the detection and recognition algorithms
requiring a high level of detail, such as license plates and face recognition [7].
e. Infrared and thermal: it is sensitive to infrared radiation (IR) and can produce a good quality
image in the dark especially for night surveillance. Thermal cameras do not need light source,
since it record the thermal radiation from objects [7].
Akoma et al


Figure 9: Types of IP cameras, (a) fixed (b) PTZ (c) Fixed dome (d) Speed dome [6]

3.2 TRANSMISSION
The video captured by the camera has to be transmitted over the network for recording, processing and
viewing. Transmission is done using either one or more of the following: cable (coaxial, fibre optic
cables, stranded copper wire, RJ45) or by air (infrared signal, RF). Wired transmission offers greater
bandwidth and reliability over wireless connections at a lower cost while wireless transmission is best
in situations where a large perimeter is in surveillance or when the area under surveillance cannot be
reached using cables [6-7].

Irrespective of either wired or wireless or both being deployed in a surveillance system, the use of IP
has replaced previous systems. In a hybrid system consists of both IP cameras and analogues cameras
(a DVR is used to convert the analogue to digital) [6-7, 11].

IP network protocols that enable secure connection between devices include Domain Name Server
(DNS), Dynamic Domain Name Server (DDNS), Point-to-Point over Ethernet (PPPoE) and Universal
Plug and Play (UPnP). Protocols used in data transmission include User Datagram Protocol (UDP),
Transmission Control Protocol (TCP), Hypertext Transfer Protocol (HTTP). Real Time Streaming
Protocol (RTSP) is the media protocol used to stream live multimedia over the internet including
mobile phones [8, 10].

Wireless technologies used in IP video surveillance networks include WiFi, WiMAX and 3GPP.
WiMAX is preferred for long distance over WiFi because of its wider coverage, higher transfer rate,
Akoma et al

higher transmission volume and better QoS. WiFi standards include 802.11b (11Mbps for 100m range
max), 802.11g (54Mbps for 75m range max) and 802.11n (600Mbps for 300m range max). WiMAX
standard in use is the 802.16, with a transfer rate of 70Mbps per 70km [6].

Methods used in transmitting video over the IP network are Unicast, Broadcast and Multicast.
Multicast is a One-to-Many or Many-to-Many transmission method, where the data is transmitted a
group of PCs requesting for the same data. The Multicast method reduces bandwidth usage and is best
for web video applications [6, 11].

3.3 COMPRESSION
Processing of digital signals is easier compared to analogue signal. Methods in compression techniques
include eliminating irrelevant portions of images or eliminating redundant parts in the media stream [7,
11-12].

Compression entails that the data is packed in a smaller space by a process of removing irrelevant bits
of the information. In compression, care should be taken not to lose too much data. A trade off should
be made between file size and image quality [7, 11]

There are two main reasons why video should be compressed are:
To reduce the amount of storage space used in the storage of the video so that storing capacities of
hard disk, solid state memory, CDs and DVDs can be maximized.
To reduce the bandwidth required to transmit the video IP networks, the Internet, SCSI connections
and USB connections etc.

Tables 2 show the comparison between different compression algorithms used.
Compression Strengths Weakness Ideal Applications
JPEG An industry standard, meaning that a camera
they can generate JPEGs is likely to be viewable
by the largest range of video management
systems.
No noticeable degradation at 10-20xcompression.
Effective with very active scenes.
Particularly easy to implement trick-play(play/
pause /rewind / forward wind/ step fwd/rev
etc.)
Under high compression ratios the
video appears blocky.
Inefficient at compressing quiet-
moderate activity scenes.
Bitrates of 8 Mbps are very common,
which makes it hard to transmit but
almost unaffordable to store
Storing photographs or snapshots
from a video
Very low frame rate CCTV (<5 IPS)
but with high scene-activity level.
E.g. Boschs DiBos v7 Hybrid DVR.
Akoma et al


M-JPEG Same as JPEG, except the JPEG images are
played in rapid succession.
As per JPEG. As per JPEG, but particularly
common in CCTV due to its simplicity.
Its popularity has caused the CCTV
market to accept low frame rates
(3.75 IPS) as the norm.

MPEG-1 VHS quality video on extremely cheap CD media. VHS quality is considered by many as
an unreasonable quality restriction.
Low cost video CDs (VCDs).
The audio component (so-called Audio
Layer 3) of MPEG-1 has become
known as MP3.

MPEG-2 The best quality video modern CCTV systems can
buy.
Bandwidth-intensive, typically 2.5-15
Mbps per camera.
Broadcast quality video, DVDs, LAN
TV, high-fidelity stereo audio.
E.g. Boschs VideoJet 8000 and VIP
1000 encoder and decoder.

MPEG-4
(Part 2)
Very efficient at high frame rates when the
difference between subsequent frames
diminishes.
Bandwidth is typically 100-1,000 kbps per
camera.
Low efficiency at very low frame
rates or extremely high scene
activity.
When the bitrate is limited the video
artifacts are speckling and a blocky
effect.
CCTV, especially when a high frame
rate is used, or when the majority of
scene activity is low to medium.
E.g. Boschs VIP X1 and VIP XD
encoder and decoder.
E.g. Boschs DiBos v8 Hybrid DVR.

H.264
(a.k.a. MPEG-4
Part 10)
Promising broadcast-oriented technology that is
more efficient than current MPEG-4 Part 2
High power processing hardware
required and higher lag-times
Next generation MPEG-4, to be used
on HD DVDs, HDTV and pay-TV.
Microsofts new XBOX 360 games
console is expected to use HD DVD,
offering 15 GB DVDs.

Wavelet Video looks great even under high compression
because the human eye accepts a fuzzy picture
more than discrete blocks.
Under extreme compression the
video becomes fuzzy and diffused,
not blocky.
Incompatibility between different
manufacturers.
CCTV recording. E.g. Boschs Divar.
Source [12]

The most widely used standards in video surveillance are MJPEG and MPEG-4. Various improvements
in terms of compression ratio, latency and error resistance is enabling the H.264 standard to replace
MPEG-4. The H.264 encoder comes with an 80% reduction in size compared to the MJPEG and 50%
compared to MPEG-4 compression [6-7].

3.4 PROCESSING
The processing of Video surveillance images is handled by video management systems. This includes
video flows, viewing, recording, analyzing, and searching recorded footage. Four major categories of
video management systems are

a. Digital Video Recorded (DVR): it has an internal hard disk which it uses in the digital recording
Akoma et al

of video and built-in video processing software. It only takes video from analogue cameras and
digitize the video. It is phasing out due to IP video cameras [6].
b. Hybrid Digital Video Recorded (HDVR): It is similar to the DVR just that it can accept input
from IP camera models [6].
c. Network Video Recorder [NVR]: It is designed for IP network architecture, and accepts inputs
from only IP cameras or DVRs.
d. IP video surveillance software: It is a purely software-based managing system of an IP network.
A web browser can be used to manage surveillance system of few cameras, while in a large
surveillance system, a special software has to be designed to manage the system [7].

3.5 ARCHIVING
In a typical surveillance system, the following criteria must be taken into account, with regards to
storage, the number of cameras deployed, recording time, recording criteria (such as constant,
scheduled or event triggered recording), other factors (such as codec, image quality and frame rate) [6-
7,13].

Figure 10: Calculation of storage space required for 30 days recording (400kb) [6]
Types of storage media include
a. Internal buffer: Network camera models come integrated with flash memory or DRAM to
enable temporal storage of images. Some other models can make use of a memory card or USB
drive for storage. DVRs have up to 4TB storage capacity. Internal storage is good for systems
comprising of 50 cameras or less [6-7].
Akoma et al

b. Attached: Network Attached Storage (NAS), or Storage Area Network (SAN) both offer shared
storage between network hosts/clients. This is best suitable for large video surveillance systems;
they provide expandability, flexibility and redundancy [6-7].

RAID (Redundant Array of Independent Disks) is used to handle redundancies in network storage
media. It is an array of several hard disk used as a single hard disk. Data is usually spanned over
multiple disk drives, making it possible for data to be recovered even if one of the disks goes bad.
RAID gives the advantage of higher reading/writing speed, redundancy and processing capability,
better recovery capability and stronger data integration [6].

3.6 DISPLAY
A most of the video from the surveillance camera are not viewed, they are stored or archived to storage
for use in case of incident or investigation. However, the video can be viewed on mobile devices,
computer, etc. large security operation centres use an array of large screens to view some surveillance
videos, which could also be switching between several cameras [6-7].

Figure 11: Monitoring on different platforms [6]

Akoma et al

CHAPTER FOUR
4.0 VIDEO ANALYTICS
Intelligent video surveillance is also referred to as video analytics which involves the use of software to
automatically fish out objects of specific interest based of their appearance and or behaviours. The
video stream collected is transformed into metadata which is then transmitted or archived so that the
video surveillance system can act accordingly. Video analytics uses mathematical algorithms to detect
objects that are moving in an image and then filters out the irrelevant movements (due to natural
environmental changes). A database which holds all the information and attributes of all detected
objects as well as the properties of their movements is created. Events of interest are searched in stored
footage based on set rules such as if a person crosses a controlled territory or if the objects behavioral
movements are suspected, then an alarm is set off.
[2,7,14,15].

Figure 11: Architecture of Video Analytics [2]

4.1 REASONS FOR DEPLOYING VIDEO ANALYTICS
In recent times, video surveillance networks have become much larger which means more cameras are
deployed, especially in large infrastructure settings such as train stations, airports and seaports. Due to
so many cameras being installed to cover the entire area, very large video footage are transmitted,
viewed and archived and it becomes impracticable for system administrators to analyse the entire
Akoma et al

footage in cases of when a security threat or activity has occurred. This is why video analytics has been
developed and deployed where only the footage where an event of threat is recorded [6-7].

4.2 ADVANTAGES OF VIDEO ANALYTICS
Constant and continuous operation, 24 hours a day
It is able to trigger an alarm to inform the operator or causes the camera to zoom in to focus for
a more accurate surveillance of the monitored object or event
Bandwidth reduction due to transmission of only video footage that contains threats
Reduction in continuous human surveillance
Faster search operations through archived video footage due to reduced and relevant data being
stored only
Automatic identification of objects in a scene and continuous tracking [7]

4.3 LIMITATIONS AND DISADVANTAGES OF VIDEO ANALYTICS
There is difficulty in detecting the difference between usual and suspicious behaviour of objects and
events. This means that there are a lot of false alarms in cases where the objects intensions are legit,
such as running to catch a flight, where the speed of the object becomes faster than the threshold
configured in the system. However, if the threshold speed is increased, then there is a possibility that a
real threat may go unnoticed. Also the environmental changes in the background image can lead to
errors in analyzing threats with video analytics. These environmental changes include moving lighting
effects from the sun and clouds, causing shadows and also wind blowing on trees, etc. This is why
intelligence video is still not very effective in uncontrolled (outdoor) environments but remains very
effective in controlled (indoor) environments due to elimination of the adverse environmental effects
stated above [7, 15].

4.4 DESCRIPTION OF VIDEO ANALYTICS TECHNIQUES
Video analytics uses computer software programs to analyze video footage in just the way a human
analyzer would do. The software processes video images and automatically detects objects such as
humans, vehicles, baggage and events of interest for security purposes. Once detection is achieved, the
actions and interactions of the object are analyzed, classified and tracked in order to identify if there is
a threat or not. The different stages of video analytics are explained below [7, 15].
Akoma et al


Object or Event Detection
The basis for video analytics is object or event detection. This involves movements in the image view
being surveyed and also camera obstructions such as dust, stickers, moisture and tampering. Detection
of movement or changes in the image is achieved by comparing image frames. The previous frame is
compared with the new frame to identify changes in pixels, thereby detecting a movement in the video.
The image is being segmented in order to separate the identified change from the background scene in
the video which leads to the next step [7].

Figure 12: Architecture of Video Analytics [7]

Image Segmentation
Image segmentation is a big problem in video analytic due to environmental factors earlier stated. It
involves separating the moving image from the entire background. These results in efficient bandwidth
usage and lesser storage space required for archiving only relevant video footage. Three segmentation
techniques are briefly discussed below [7-14].

Akoma et al


Figure 13: Segmentation of an individual in a video footage [7]
Background Subtraction
This involves comparing the image in each frame in a sequence to a reference image, called the
background. The difference in pixel values and intensity allows the image region to be detected and
then removed from the entire background. Subtraction of images using the pixel-by-pixel method is
highly sensitive to the slightest environmental change so this method of image segmentation works best
in controlled environments [2, 7, 14-15].

Figure 14: Background Subtraction [2]

Time-based Difference
Another method of object segmentation is based on time differences between consecutive frames of a
video footage. The changes in the image due to the environment are slow compared to the object of
focus. Therefore these variations are catered for in the frames, using this time based method. Outdoor
operation using this method performs better than the previous method [7].

Akoma et al

Optical Flow
This method involved monitoring the consistent flow of pixel changes in a particular direction,
knowing that optical flow in environmental changes do not flow consistently in one direction. This
method however, requires complex calculations and is difficult to deploy on real time surveillance [7].

Object Tracking
After detecting and segmentation of the moving objects, tracking of the object is the next stage in video
analytics. Tracking is important in order to monitor the movement of the object across the areas it
covers. The object is tracked from frame to frame in the video footage. Tracking can be done in 2D
with a single camera, or 3D when combining two camera views and using their geometric relationship.
Some tracking techniques are based on mathematical prediction methods to obtain an objects position
over consecutive frames when taking into considerations, the objects previous positions in a frame [7,
15].

Object Classification and Identification
Detected objects are classified into different categories such as human, equipment, vehicle, etc. The
system is able to recognize the object based on its appearance and size in terms of its geometric
structure as well as its movement patterns. For example, vehicles appear wider than in height and
humans appear taller than in width [7-15].

Figure 15: Object classification (humans vs. vehicles)

Akoma et al

For humans, face recognition and gait recognition are used. Gait recognition is used predominantly
when the object (human) is far from the monitoring camera while face recognition comes into use
mostly when the human is in close proximity to the monitoring camera. 3D face recognition techniques
appear to improve the performance of object identification than 2D techniques. Also, license plate
recognition by video analytics is being is a huge security feature. It requires high resolution and fast
calibration cameras for proper capturing and detection of license plates especially when the vehicle is
in motion. For the camera to read a license plate, the system firstly, locates the rectangle of the license
plate and then optically recognizes the characters of the plate. Due to adverse environmental effects,
license plate recognition is better done with specialized video surveillance systems that concentrate on
the position of the camera and lighting quality [7].

Figure 16: License plate detection [7]

Classification of Activities and Behaviours
A big importance of video analytics is in the interpretation of behaviours of objects and their
interaction within a scene. It is also the most challenging part of video analytics in the sense that it
involves complex analysis and calculations. In order to analyse and interpret object behaviours, the
movement pattern of the object must be known. The objects interactions and actions are defined
typically and fed into the system as baseline for generic behaviours of such object, such as a vehicle
moving in a particular direction in a known road (background) and also the speed at with the vehicle is
moving. Therefore, in any case where a vehicle moves faster than the maximum speed threshold, then
the behavior of the vehicle will be seen as abnormal and an alert is flagged on the software for the
monitoring team to be aware of [2, 7, 15].
Akoma et al

CHAPTER FIVE
5.0 SECTORS OF APPLICATION
Video surveillance was originally used by public services (police, transportation, administration). It
was then adopted by companies looking to protect strategic assets, such as refineries, nuclear power
plants, dams, agri-food plants and pharmaceutical complexes. Casinos also appear as pioneers in the
deployment of large video surveillance systems. Today, surveillance cameras can be found in different
public and private places, such as apartment buildings, shops, parking lots, railroad/bus stations,
airports, roads, mass transit, banks, etcVideo surveillance systems are deployed on different scales. For
monitoring minor crimes (e.g., assaults, vandalism, theft), video surveillance is used primarily for post-
incident investigation. This level of surveillance requires simple, often analogue technologies that do
not involve video intelligence. This type of system can be found at the corner store, in small shops or in
a private home [5-6].
Monitoring of apartment buildings or larger shops often requires a more extensive camera system. The
goal of surveillance is mainly to monitor access paths, parking lots, and in the case of stores,
departments and points of sale. Video surveillance is used in these locations mainly for investigation
purposes. However, these users would like to obtain surveillance systems able to generate alerts in real
time for immediate intervention. Video analytics products are therefore a breakthrough in these sectors.
For this clientele, return on investment is a determining factor in buying intelligent video surveillance
equipment [5-6].
Large-scale video surveillance is found in cities and neighbourhoods, transportation systems, university
campuses, at major events (festivals, economic summits, Olympic games, etc.), with extensive security
parameters. It requires dozens, even hundreds of cameras to be deployed. These cameras must
sometimes be accessible to hundreds of security responders from different government agencies, police
forces or emergency [5-6].
Many sectors of the government require intelligent video surveillance for: identifying individuals and
vehicles, counting people and monitoring crowds, recognizing suspicious or violent behaviour (fights,
misdemeanours), detecting intrusions, and monitoring roads. More advanced analyses, such as
computer recognition of emotions or if an individual is telling a lie, are also anticipated.

Akoma et al

5.1 RETAIL TRADE
Retail trade is a growing market for video surveillance, which is used for both internal (store,
warehouse) and external (parking lot) security. Even the smallest shops have cameras to at least keep
video evidence in case of theft or an incident. In chain stores, much more sophisticated video
surveillance systems are set up for centralized monitoring of different locations. For the entire sector,
video surveillance is aimed primarily at:
monitoring registers and transactions (employee theft and fraud);
protecting material goods and infrastructures;
monitoring inventory and wares (deliveries);
protecting staff and clients;
controlling access to locked areas;
checking emergency situations (fire, alarms, etc.);
monitoring parking lots, vehicles, entries and exits.
Intelligent video surveillance is also increasingly used for non-security purposes, such as managing
operations and market launch. In this context, video analytics is used in particular to count clients,
analyze their behaviour and in-store movements, and compile statistics on consumer habits [5-6].
5.2 TRANSPORTATION
The security and smooth operation of airports, railroad/bus stations, ports and mass transit systems are
critical for a countrys economy. A security incident can seriously upset operations and result in
significant losses. Given the large flows of passengers that use transportation systems and the extent of
their infrastructures, these systems face extraordinary security challenges. Terrorist acts committed in
different transportation systems around the world have exacerbated these challenges [5-6].
Analytical software targets the transportation sector by offering different adapted functions: detecting
an intrusion in a controlled perimeter or area, detecting people entering an exit ramp, detecting
abandoned luggage, recognizing faces, counting people, recognizing license plates for monitoring
access to parking lots, detecting suspicious behaviour (loitering, vandalism, graffiti), detecting people
in lanes. However, the transportation sector poses major technical challenges for intelligent video
surveillance systems given the number of people who pass within the cameras field, the diversity of
passenger behaviours and unfavourable conditions for cameras (vibrations, dust, etc.). Moreover, for
Akoma et al

outside [5-6].
5.3 EDUCATION
Video surveillance is increasingly found in academic institutions. It is used to oversee the safety of
teachers and students, as well as to protect assets from vandalism and theft. video surveillance is used
in particular to:
monitor access to the institutions perimeter, which may be extensive, such as in the case of a
university campus;
monitor equipment and data;
detect and follow acts of vandalism, theft, misdemeanours and inappropriate behaviour;
recognize license plates;
support criminal investigations;
control access.
Since educational institutions often have an IP network infrastructure, it may be beneficial for
them to set up digital video surveillance systems [5-6].
5.4 GOVERNMENT AND PUBLIC SECURITY
The different levels of government must ensure the safety of the population and of public
infrastructures. On the national level, it will be used, for example, to monitor the following:
Sensitive infrastructures
Borders
Government buildings and sites
Laboratories
Military bases
Prisons
Locally, video surveillance is set up in several cities around the world to monitor crimes and for use as
an emergency intervention tool. It also helps to ensure security during large gatherings (shows,
demonstrations, sporting events, etc.). London is the city the most often cited for the number of
cameras deployed in its streets. Video surveillance is also used to manage parking, especially to
monitor parking permits, the application of rules, the detection of theft, vandalism or misdemeanours,
Akoma et al

and to control access. Monitoring, cameras and analytical algorithms must be able to operate despite
changing weather conditions (fog, snow, rain, etc.) [5-6].
5.5 AIRPORTS
Following the events of September 11, 2001, security measures were tightened, especially at large
airports, and new technologies were deployed. For airports, the priority consists in controlling access to
secure areas, in particular access to airplanes, and also in ensuring the safety of passengers, personnel
and the property within the perimeter (runways, parking lots, access routes, etc. The intelligent
functions already in place consist of counting people to assess, for example, processing times at
customs, as well as for license plate recognition in order to take stock of vehicles in parking lots and to
detect theft. Aroports de Montral will soon be adding a new component to its video surveillance
system that will be entirely IP and include intelligent cameras and video processing software [5-6].
5.6 STATIONS AND PUBLIC TRANSPORTATION
The purpose of video surveillance in railroad/bus stations and public transportation is mainly to
monitor access routes, platforms, rails and tunnels, parking lots and all auxiliary service structures. It is
connected to other access monitoring systems and security devices to oversee the protection of
passengers, personnel and infrastructures. The video is viewed live from the security control centre, or
at a later time for investigations following security incidents. . The new subway system, entirely on an
IP network, is an IP video surveillance flagship project in Quebec. Controlled centrally, the video
surveillance system is connected to the different monitoring and maintenance systems. The STM has
begun to deploy video analytics for automated detection of people on the tracks and states that it is one
of the first in Canada to use intelligent video in its mass transit. In this field, Canadian public
transportation systems are behind those of European countries, such as the United Kingdom, France
and Germany, which have much better video surveillance systems. In the long term, if using video
analytics proves effective, the STM anticipates some other applications for it, such as detecting
suspicious packages, monitoring doors and recognizing suspicious behaviour [5-6].
5.7 BANK SETTING
Video surveillance is widely used for bank security. The presence of cameras first acts as a deterrent to
committing armed robbery and assault. Should a crime occur, archived video footage is used to
investigate and identify the perpetrators. Automated bank machines are prime targets for criminal acts.
Surveillance cameras help to detect fraud, such as, for example, the installation of a device to read the
Akoma et al

magnetic information on bank cards [5-6].
In a bank setting, intelligent video surveillance can increase monitoring effectiveness. It provides for
monitoring of all branches in order to detect suspicious individuals or behaviour. It also makes it
possible to find, among other things, all video footage from all branches where a certain individual
appears using face recognition techniques [5-6].
Since casino monitoring requires watching the behaviour of many people in a crowded environment,
intelligent video surveillance is an interesting way of helping security personnel do their job. Face and
gesture detection and recognition are part of the useful analytical functionalities for this environment
[5-6].
5.8 OTHER
Video surveillance is used in many other environments, such as in stores and to monitor buildings. It is
most often used to film access routes and parking lots, monitor material valuables and ensure employee
and client protection. Video surveillance can also be used in healthcare to help with interventions. It
can even be found in ambulances and used in combination with other measuring instruments to monitor
a patient remotely. The American company IndigoVision has developed analytical solutions installed on
intelligent cameras specifically to detect the movement of helicopters landing on the site of an
American hospital [5-6].

Akoma et al

CONCLUSION
Overall, video surveillance has improved over recent years with the use of video analytics which the
use of intelligent system software. Benefits such as continuous 24 hours a day operation, reduction in
bandwidth, reduction in cost overhead in human monitoring are amongst the business benefits of
intelligent video surveillance. The Huawei Intelligent Video Surveillance System is an advanced
system and provides one of the best surveillance applications. It is popular in the China, and can find
application in various sectors of the economy for good intelligent video surveillance.

Akoma et al

REFERENCES
1. T. Hulsey, Evolution of video surveillance, retrieved from
http://www.slideshare.net/TomHulsey/evolution-of-video-surveillance on 21-04-2012

2. D.Y. Patil, S. Joshi, M. Bhagar, S. Aundhkar, Survey on Wireless Intelligent Video Surveillance
System Using Moving Object Recognition Technology, IISTE Journal of Computer Engineering
and Intelligent Systems, vol. 2, No. 7, ISSN 2222-2863, pp 25-30, Oct 2011.

3. Huawei Technologies, Huawei Intelligent Video Surveillance Solution SME, Huawei
Technologies Co. Ltd, Sept. 2011.

4. C. Dongdond, Huawei Symantec Industrial Surveillance Solution, Huawei Symantec Solution,
Issue 4, Feb 2009.

5. S. Gao, An Intelligent Video Surveillance System, IEEE Conf. on E-Product E-Service and E-
Entertainment (ICEEE), pp 1-4, Nov 2010.

6. VIVOTEK, IP Surveillance Handbook, VIVOTEK Inc, P/N 012000105, 2009.

7. V. Gouaillier, Intelligent Video Surveillance: Promises and Challenges, Technological and
Commercial Intelligent Report, CRIM, Apr 2009.

8. RMSI, The Benefits of IP-Surveillance over CCTV+DVR, retrieved from
http://www.teamrmsi.com/objects/the-benefits-of-ip-surveillance-over-cctv-and-dvr.pdf on 2-04-
2012

9. D-Link, IP Surveillance Seeing is Believing, retrieved from www.dlink.co.uk, on 24-03-2012

10. TNS, White paper From CCTV to IP Surveilance, TNS, 2010

11. PELCO, IP Video Security Guide, PELCO-Schneider Electric, 2010

12. BOSCH, White Paper: Video Compression for CCTV, BOSCH Security Systems, 2011

13. Seagate, Surveillance Solutions Guide, Seagate Technology LLC, 2007

14. Tu P., Doretto G., Krahnstoever N., Perera A., Wheeler F.,Liu X., Rittscher 1., Sebastian T., Yu
T., and Harding K., Intelligent video framework for homeland protection" Proceedings of SPIE
Defence and Security Symposium Unattended Ground, Sea, and Air Sensor Technologies and
Applications IX, Orlando, FL, USA, April 9-13, 2007

15. Chang, E.Y.; Yuan-Fang Wang; I-Jeng Wang; , "Toward building a robust and intelligent video
surveillance system: a case study," IEEE International Conference on Multimedia and Expo, 2004.
ICME '04. 2004, vol.2, pp.1391-1394, June 2004

Intelligent Video Surveillance System PDF

Transféré par

Informations du document

Description originale:

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Intelligent Video Surveillance System PDF

Transféré par

Droits d'auteur :

Formats disponibles

Case Study Report on Intelligent Video Surveillance System

Vous aimerez peut-être aussi