Certificate of Registration Media number Эл #ФС77-53688 of 17 April 2013. ISSN 2308-6033. DOI 10.18698/2308-6033
  • Русский
  • Английский

Contour analysis-based identification of nature objects at processing data of Earth remote sensing

Published: 19.11.2013

Authors: Domracheva A.B., Basarab M.A., Medvedeva A.L.

Published in issue: #10(22)/2013

DOI: 10.18698/2308-6033-2013-10-1093

Category: Information technology

The possibility of using the computer vision technology for solving problems of orientation, navigation, and control of unmanned aerial vehicles (UAV), as well as the problem of identifying the locations of the observed object by comparing its shape with the reference image on the digital map is described. The problem of extracting contours of objects in the processing of Earth remote sensing data is solved. The approach for isolation and comparison of contours of an object based on the contour analysis is proposed. The methodology for selection and comparison of contours is presented and parameters of the similarity of contours are defined. The problems arising in the construction and comparison of the isolated and reference vectors are analyzed and their solutions are specified. The choice of numerical methods for solving the given problems is justified. To close the contours, the method of cartographic generalization, based on the theory of fractals, is used and for simplifying the circuit, the method of de Berg is applied, allowing us to obtain circuits without self-intersections. The results of testing the contour analysis method to a fragment of satellite imagery, including, in particular, natural objects are presented.