Ostrikov V. N., Kirienko A. V.

For small unmanned aerial vehicles (UAV) with the optical equipment of formation of images, with reference to visible and especially infra-red range of a spectrum, the use of scanning means of registration is characteristic. Hence during formation of the video data in the image the significant geometrical perturbations, including influence of angular fluctuations of the carrier and mistakes of synchronization of the beginning of lines of records of scanning means of registration, are imposed. Since the result of solving the final task of decoding and bindings of images is influenced by casual mutual shifts of lines, they are usually compensated during processing taking account of the navigating data of angle of a roll. However used methods of registration of perturbations register these fluctuations with much lower frequency, than that of scanning of the device. To reduce high-frequency fluctuations the information received on the basis of the correlation analysis of displacement is used. The Lack of direct application of such method is in the fact that in most cases used correlation algorithm "withdraws" the image aside, bringing in image inadmissible distortions. In such cases, the filtered high-frequency part of the designed data, capable to compensate only local perturbations, is allocated. Using combined data, received on the basis of navigating measurements on which rather low-frequency part of perturbations is compensated with the data obtained on the basis of the correlation analysis, allowing to reveal a high-frequency component of mutual displacement of lines, is offered. Combined processing is based on a rational choice of parameters of the filter realizing allocation both low-frequency - of the navigating data, and high-frequency - from the data of the correlation analysis of components of resulting correction. The received method realizes the combined approach to compensation of considered distortions. Application of such a method allows us to increase the mean square value of accuracy in 1.5 times.