0% found this document useful (0 votes)
15 views2 pages

Computer Vision in Surveillance Systems

The CSIR has decades of experience in image processing, computer vision, and developing surveillance systems. They have strong capabilities in fields like image enhancement, low light imaging, object detection, tracking and classification that allow them to support applications like maritime surveillance. Some of their technical capabilities include real-time implementation using GPUs, deep learning networks for pattern recognition, augmented reality, and developing a wide area surveillance system.

Uploaded by

isak berk
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
15 views2 pages

Computer Vision in Surveillance Systems

The CSIR has decades of experience in image processing, computer vision, and developing surveillance systems. They have strong capabilities in fields like image enhancement, low light imaging, object detection, tracking and classification that allow them to support applications like maritime surveillance. Some of their technical capabilities include real-time implementation using GPUs, deep learning networks for pattern recognition, augmented reality, and developing a wide area surveillance system.

Uploaded by

isak berk
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Focus on CSIR

Services in Optronic Sensor Systems


Image Processing and
Computer Vision

World-class tools, sophisticated


techniques and knowledge workers
with decades of experience
positions the CSIR to adequately
satisfy the requirements of
the customer with regard to
surveillance systems.

A strong capability in various Image processing and computer vision is a broad research area that
fields of image processing allows involves specialists from various fields of expertise. These include:
the CSIR to support several
applications in surveillance. Photogrammetry Computational
Some of these include: The relationship between the image and photography
the world around it can be determined The CSIR looks at combining non-
• Maritime object detection and using photogrammetry - the science behind traditional camera designs (including
tracking taking measurements from photographs. coded apertures and plenoptic focal
Modern maps are typically created using plane arrays) with computer processing.
• 360 degree image stitching
photogrammetry and photographs. With This allows for advanced image
• Atmospheric turbulence this capability, a normal optical camera processing that includes digital refocus,
mitigation in images is turned into a precision measurement perspective shifts on images and foliage
instrument, which can be used to create penetration.
• Detection of moving targets images for real-time stitching, helmet
from aerial platforms tracker-based navigation and targeting,
see-through armour and image fusion.
• Distributed surveillance
Focus on CSIR Optronic Sensor Systems • Image Processing and Computer Vision

Low contrast image enhancement Low light contrast enhancement

Image enhancement Coarse to fine image matching strategies


Video frames can be distorted by several are used to align video frames. For
factors such as atmospheric turbulence particular applications, computer vision
(mist or rain), and platform motion (when models have been developed for robust
an optronic system is mounted on a ship tracking. Object trackers consist of target
in stormy weather). Several algorithms detection and segmentation modules for
were developed by the CSIR’s optronic automatic operation in real-time. Image
experts for tone mapping, low-light image segmentation refers to the extraction of
enhancement, atmospheric turbulence pixels of objects of interest from video.
mitigation, image stabilisation, motion
de-blur and super resolution.
Augmented reality
Augmented reality is a ‘real-world’
Real-time implementation is achieved
view of an environment with
using multiple central processing units
computer-generated sensory input.
(CPU) and/or graphics processing units
Graphics combined with see-through
(GPU). This capability is particularly
or transparent head mounted displays
useful when looking at real-time video
can be used to relay tactical information
during surveillance missions.
Classification and training simulations, or increasing
Pattern classification one’s situation awareness, for example.
and machine learning Wide area surveillance
The CSIR’s Optronic Sensor Systems
The Wide Area Surveillance System
group has conducted research into
(WASS) provides instantaneous wide
microdot processing, optical character
area situational awareness. With a
recognition (OCR), object detection and
visual switch panorama of 360° and a high-resolution
target classification. These rely on state
digital video at 20 frames per second,
of the art methods, such as deep learning
the sensor initially began operations in
and convolutional neural networks. These
the visual band. Ruggedised modules
networks are computing systems made
have been tested in the maritime domain
up of a number of highly interconnected
for the detection of small craft in support
processing elements that process
of anti-piracy operations. Multi-spectral
information through their dynamic state
versions of the system employ thermal
response to external inputs. The systems
cameras and active cooling. Calibration
LWIR switch are able to, for example, differentiate
of the cameras is done using the CSIR’s
between types of animals and
patented Automatic Photogrammetric
different makes of vehicles in cluttered
Camera Calibration System (APCCS).
environments

Object tracking Contact details


Tracking algorithms developed by the Email: dpss@[Link]
CSIR use techniques such as particle filters
to estimate a target’s state.
[Link]
Fused switch

Photogrammetrically stitched panoramas.

You might also like