Camera/Video sensor networks

molassesitalianΤεχνίτη Νοημοσύνη και Ρομποτική

6 Νοε 2013 (πριν από 3 χρόνια και 5 μήνες)

65 εμφανίσεις

SMART
CAMERAS


Intelligent cameras


Self
-
contained

stand
-
alone vision
system


Built
-
in image
sensors

Architecture into “CONTROL” & “DATA
NETWORK”


The control network deals with low
-
level
sensor specific processing.


Both the networks share the same
Computational & Sensing resources.


CONTROL NETWORK:
Low bit
-
rate sensing
such as an RFID tag or Fire
-
alarm.


DATA NETWORK:
Video streaming


CONCERNS:


How to assign priorities
to different data
streams???


How to adjust their
Spatial & Temporal
fidelities (to maximize the
“Measure of
Effectiveness”)???


In general “Control”
network packets are given
higher priority than
“Data” network packets.


REASON:
Control network
packets are much smaller
than the Data network
packets.


EFFECT:
Supporting a
cluster of Sensor Agents
would not overload the
Communication
infrastructure.




A Camera sensor network may be a type of
wireless
sensor network
.


The network generally consists of the cameras
themselves, which have some local image
processing, communication and storage capabilities,
and possibly one or more central computers, where
image data from multiple cameras is further
processed and fused .


Camera/Video sensor networks also provide some
high
-
level services to the user so that the large
amount of data can be distilled into information of
interest using specific queries.


The data dissemination also depends upon the order
of the query.

The major difference between visual
sensor networks and other types of
sensor networks is the nature and
volume of information the individual
sensors acquire.

Unlike most sensors, cameras are
directional in their field of view, and
they capture a large amount of visual
information which may be partially
processed independently of data
from other cameras in the network.

STRINGENT CONSTRAINTS OF VIDEO
SENSOR NETWORKS


Limited Communication
Bandwidth


Processing Capability


Power Supply



OTHER CONSTRAINTS


1) Task
-

mapping


2) Scheduling problem

OTHER CHALLENGES OF CAMERA
SENSOR NETWORKS

1) Barrier Coverage


In a wireless sensor network, a barrier is formed by a set of sensors whose
sensing ranges are contiguous and span (usually a strip area) across the
monitored field. Every object traversing the field from one side to another is
detected by the sensors on the barrier.

However, the barrier coverage of camera sensors is much different and more
complicated than the traditional barrier coverage problem. Simply combining the
sensing range of a series of cameras across the monitored field does not pro
-

vide
effective barrier coverage. This is because an intruder may cross the barrier without
being identified, for instance its face image could be missed





PROPOSED SOLUTIONS



A camera barrier with the full
-
view coverage property
guarantees that no matter where the intruder faces, it will
always be detected by an active camera whose viewing
direction is close enough to the intruder’s facing direction and
hence the face image can be identified effectively.



2. Intrusion detection in an
Energy efficient way



This deals with how to detect
an intruder moving through a
polygonal space that is
equipped with a camera
sensor network.


SOLUTION :Camera takes
frames with an assigned
probability rather than
continuously monitor the
space. The camera
probabilities can be optimized
over a desired convex cost
function with relatively few
constraints. Once the
probabilities are set, no
further coordination is
required between the nodes.


3. traditional vision
-
based
techniques for accurately
calibrating cameras are not
directly suitable for ad
-
hoc
deployments of sensors
networks in remote locations.



SOLUTION: Relative locating
and orientations of camera
sensors without the use of any
landmarks or positioning
technology and by randomly
sampling the environment
with a reference object, the
degree and range of overlap
for each camera can be
exploited for duty cycling and
triggered wakeups.


Q1. Elaborate on the proposed
solution for Barrier Coverage

Ans.) A camera barrier with the full
-
view coverage property
guarantees that no matter where the intruder faces, it will
always be detected by an active camera whose viewing direction
is close enough to the intruder’s facing direction and hence the
face image can be identified effectively. Thus providing a solution
to this Barrier Coverage constraint.

Q2.Why are Control network packets
given higher priority than Data
network packets???


Ans.) Control network packets are much smaller
than the Data network packets. Supporting
these cluster of Sensor Agents would not
overload the Communication infrastructure.



Q3)Name a few Applications of
Camera & Video Sensor Networks.

Answer!!!


Unattended surveillance
(detection of intruders).


Detection of
Defects/Flaws.


Web Inspection


Biometric Recognition &
Access Control(Iris, Finger
print, Face recognition).


Detection of Missing
parts.


Inspection of Wires, Coils,
tubes, extruded Plastic.


Robot guidance.