|
Making a deep analysis of the
video received on the ground from for example an aerial ISR asset is
something like a forensic study. Many factors affect the video quality,
and leave a fingerprint. The multidimensional turbulence shake of the
airframe and engine vibrations. The camera/sensor itself if of course a
vital component affecting imaging quality, and gyro-stabilized gimbals
and their servos leaves residues; and obviously so does also
compressions, format conversions and the data links. Typically, each
component in the image capture chain gets tested carefully in lab
environments, and comes with complete specifications. But as every test
engineer can witness to; the end result of a system in operation is not
merely a sum (and maybe not even a linear function) of all its
components. Errors and deviations have a way to propagate in obscure
ways.
 Each image in the image capture chain adds its own fingerprint, affecting visual quality
With
so many airborne systems with expensive payloads up there, it seems the
time has come for a new set of metrics to simply and objectively assess
vision quality from full motion video. De-facto standards used today
has their origin from analysis of still images, historically high
resolution photos from “spy satellites”, like the NIIRS scale to assess
level of details identifiable. However, watching live full
motion video has somewhat other
purposes, and hence there is a need for other metrics. Factors that should be considered in video quality metrics are :
● Endurance – how do we manage to the keep an operator stay focused and alert over long period of
times?
● Time critical – how do we ensure a human observer extract the critical information from an event
they may only last a few seconds?
● Precision – how do get as accurate details as possible?
● Automated analysis – how can we make sure that any automated image processing (such as object tracking) can work at optimum with the incoming video stream?
At Imint, with our combined expertise in video processing and aerial surveillance, we have defined not
only a set of relevant metrics relevant to the factors above, but also a test product – VISQA – to measure a
system in operation and simply extract those metrics. So whether you are an UAV system provider, a component provider, a user of an UAV system or a
governmental body procuring these systems, we think you find it interesting to read more about our
VISQA product here . |
