Corresponding author: Insarov Wiljam V., D. Sc., Professor, Deputy Head of a Subdivision, State Research Institute of Aviation Systems, State Scientific Centre of the Russian Federation, Moscow, 125319, Russian Federation, e-mail: wiliam@gosniias.ru

Received on January 23, 2017
Accepted on February 04, 2017

In the modern world there are strict requirements to the unmanned aerial vehicles (UAV) in terms of the range of their tasks and the operating conditions. In order to solve this kind of tasks, it is necessary to produce technical resources for obtaining and processing of the information about the environment. One of these technical means for obtaining and processing of information about the environment is the machine vision system (MVS). The input information in the MVS is a current halftone image (CI) of the stationary ground multi-object scenes produced in an air flight. Quality estimation is one of the first operations done with this image. This operation allows one to take decisions about the possibility of image processing and to choose the procedures for this processing. Such kind of operations made with the use of a metric of quality can be divided into three groups: full-reference, reduced-reference and non-reference metrics of quality. The representative set of the non-reference metrics of quality, revived in the tasks like those, was divided into three groups: derivative-based, statistical-based and transform based metrics (wavelet transform, discrete cosine transform and Fourier transform (FT). The set of the non-reference metrics considered in this work requires further tests connected with the use of the banks of the real halftone images of the stationary ground multi-object scenes.
Keywords: image processing, image quality estimation, non-reference metrics, multi-object stationary scenes, time of day effects, weather effects, scene composition effects, correlation relationships

Acknowledgements: This work was supported by the Russian Foun­dation for Basic Research, project no. 17-08-00584 а.