© с сайта ТПУ
TOMSK, Nov 17 – RIA Tomsk.
Scientists of Tomsk Polytechnic University (TPU) develop algorithms of
automatic improvement of images quality which underwater robots use for
identification of objects, it is reported on Tuesday on the website of the
higher education institution.
It was reported earlier that TPU
implements the "Telecommunication Systems of Monitoring and Management for
Independent Underwater Robots" megaproject. Scientists create
hardware-software complexes for new generation of the independent uninhabited
submersibles (IUS). In 2016 scientists have to provide a complex of the equipment
for underwater tests in Vladivostok.
"We develop software for
automatic improvement of images quality, and now there is an opportunity to
introduce it in the megaproject on independent submersibles. Programs work
well, but water we didn't test them under therefore it is necessary to carry
out tests and, perhaps to correct something", - professor of TPU Vladimir
Spitsyn is quoted.
© сайт ТПУ
It is specified that
"polytechnicians" adjust artificial neural networks by means of
evolutionary algorithms by the principle of the organization of nervous cages
of a live organism. Computer vision is necessary for the robot to identify
objects round itself. This task is complicated by muddy water and a lack of
light. The development of TPU will allow to improve quality of the received
picture and speed of processing of images.
"Speed of our system is about
10 frames per second for images of 512x 512 pixels in size. The power of the
ordinary personal computer is enough for work", - Spitsyn specifies.
It is added on the website that
scopes of IUS application are from search of sunken submarines till sea mines
neutralization.