Video Traffic Surveillance

This section discusses the automating of various tasks required by the

Department of transportation. Discussed below are some of the tasks
which when automated will reduce a lot of the overhead on the human
operators, make traveling safer, increase the cost effectiveness of fuel use,
can decrease travel time, and make the traffic control system more robust.
Algorithm application is discussed for each of these tasks. An evaluation
of the algorithms usability is reported for each of these tasks.
2.1.1 Video Surveillance :
An appliance that enables embedded image capture capabilities that
allows video images or extracted information to be compressed, stored or
transmitted over communication networks or digital data link. Digital
video surveillance systems are used for any type of monitoring. Broadly,
video surveillance is the image sequences which are recorded to monitor
the live activities of a particular scene. The importance of this digital
evidence is given the first priority for any kind of occurrence. This digital
information is recently become the field of interest to the researchers on
the field of AI, Robotics, Forensic Science and other major fields of
science.
Video images can be used to keep a watch on traffic in the various
roads of a town or on a highway. Several video cameras are installed on
the roads on which the traffic needs to be checked. These images could be
transferred to the operator working in the Traffic Management Center
through a broadband connection. A computer algorithm to look for certain
behavior in the traffic can analyze the images from these cameras. This
could include:
A. Intersection Control:

The video image from a busy traffic intersection could be

analyzed to find if the traffic at the intersection is causing congestion. If
this is the case, then the traffic at the previous intersections could be
alerted about the congestion at the approaching intersection so that the
vehicle owners could take precautionary measure, such as an alternate
route or just be mentally prepared about the traffic ahead. If this
information could be made available to them beforehand, i.e., before they
left their homes or offices then they could possibly delay their trip so as
to avoid the congestion ahead. Fig 2.1
B. Queue Length Measurement:
Similar to the above discussion, if there is significant amount of
traffic congestion on a certain road, then the traffic approaching the
congested area could be warned beforehand about the situation so that the
vehicle owners could take alternate measures in advance to handle the
situation. If they are warned well in advance of the congestion, then they
could avoid the congested road. And, if that is not possible, then they
could compensate for the delay, such as by rescheduling appointments
and meetings. Fig 2.1

Fig 2.1: Queue Length Measurement and Intersection Control [11]

C. Travel Time:
It would help the commuters, if existing software could
approximate their travel time depending on the route taken and the
amount of traffic on that route prior to the commencement of their
journey. Using software like this, the commuter could travel a route with
the least traffic and subsequently, least travel time.
I.

Algorithm Usability:
With some modifications, the algorithm could be adapted to the

above-mentioned applications. For example, instead of identifying just

individual vehicles, it needs to identify a group of vehicles and find how
dense they are. If the density is high, it suggests that the road is crowded
with chances of a big queue of vehicles and travel time can be higher on
this road. Fig 2.2

Fig 2.2 An algorithm of video traffic surveillance.

II.

Video Capturing System:

This section or module basically consists of a camera, which captures

a frame as the reference picture or the desired background condition

which thus forms the reference for any kind of motion tracking or
foreground extraction. After capturing a snapshot of the standard
background the camera is programmed to capture a video of user defined
number of frames. The captured video is to be processed for detecting
any motion with the reference frame snapshot as the reference condition.
Here in this project we go for a camera that is controlled by MatLAB
coding itself. When it comes to camera controlling through MatLAB
coding we make use of the Image Acquisition Toolbox available with

MatLAB, for the various options available for controlling a camera

connected to the personal computer. By using the various functions
provided by the image acquisition toolbox we can use the camera to take
snapshots and even a sequence of images which is a video. This module
can be separated into few blocks as shown bellow:
Install Your Image Acquisition Device
Follow the setup instructions that come with your image acquisition
device. Setup typically involves installing the frame grabber board in
your computer. Installing any software drivers required by the device.
These are supplied by the device vendor. Connecting a camera to a
connector on the frame grabber board. Verifying that the camera is
working properly by running the application software that came with the
camera and viewing a live video stream.
Retrieve Hardware Information
In this step, one get several pieces of information that the toolbox needs
to uniquely identify the image acquisition device one want to access. One
use this information when they create an image acquisition object.
The following table lists this information.
Device

Description

Information
Adaptor

An adaptor is the software that the toolbox uses to

name

communicate with an image acquisition device via

its device driver. The toolbox includes adaptors for
certain vendors of image acquisition equipment and
for particular classes of image acquisition devices.
See Determining the Adaptor Name for more

Device ID

information.
The device ID is a number that the adaptor assigns
to uniquely identify each image acquisition device

with which it can communicate. See Determining

the Device ID for more information.
Note: Specifying the device ID is optional; the
toolbox uses the first available device ID as the
Video

default.
The video format specifies the image resolution

format

(width and height) and other aspects of the video

stream. Image acquisition devices typically support
multiple video formats. See Determining the
Supported Video Formats for more information.
Note: Specifying the video format is optional; the
toolbox uses one of the supported formats as the

default.
The above information regarding the camera interfaced with the personal
computer is gathered for controlling the camera in acquiring the video
that has to be processed by the code. The set of code that collects and
displays all the required information regarding the camera is listed
bellow.
>> imaqInfo = imaqhwinfo % information regarding the adaptor is
acquired.
>>imaqInfo.InstalledAdaptors
>>matroxInfo = imaqhwinfo('winvideo')
>>matroxInfo.DeviceInfo
>>device1.DeviceName % device name
>>device1.DeviceID %device ID
>>device1.DefaultFormat % information regarding the default format of
the camera in use
>>device1.SupportedFormats %all the supported formats.

Thus to perform listed operations above there required code used is

shown bellow
>>vidobj = videoinput('winvideo', 1);% creating video object.
>>preview(vidobj)%starting the preview window
>>set(vid.source,Brightness,100);%brightness adjustment
>>r=input('ready to capture the standard background ( if yes press a
numeric key)');
>>pic = getsnapshot(vidobj);% acquire the standard background image.
>>r=input('ready to capture video ( if yes press a numeric key)');
>>set(vidobj, 'FramesPerTrigger', 50);%capture 50 frames
>>start(vidobj)%starting the video capture.
>>pause(5)
>>numAvail = vidobj.FramesAvailable
>>imageData = getdata(vidobj, numAvail);%imagedata carries the
acquired video.
These few lines of code allows us to control a camera that is an image
acquisition device to capture a video which is taken as the input for the
processing required to construct information regarding the foreground
moving object along with proper detection motion. Any motion detected
is forwarded to the hardware unit to intimate through an alarming or a
control system section.
Here in this project we make use of a web-cam which is installed with the
personal computer. This web-cam is the source of video for this project
controlled by the functions used in MatLAB made available by the
Instrument Control Toolbox in MatLAB.
References;
1. VEHICLE DETECTING AND TRACKING IN VIDEO FOR
INCIDENT DETECTION

Nitin Agarwal
Nicholas Andrisevic
Kiran Vuppla
Marian S. Stachowicz

2. http://www.matlabcodes.com/2010/08/video-survillience-robotmatlab-code.html