UDC 004.93
MODIFIED METHOD FOR AUTOMATED POSITION ESTIMATION OF A SATELLITE IMAGE ON A PANORAMIC IMAGE USING KEYPOINTS
D. M. Kostina, Master student, Department of Computer Software and Information Technology, Bauman
Moscow State Technical University, Moscow, Russia;
orcid.org/0009-0004-1047-7360; e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it..
K. L. Tassov, Senior lecturer, Department of Computer Software and Information Technologies, Bauman
Moscow State Technical University, Moscow, Russia;
orcid.org/0000-0002-4928-2964; e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it..
D. V. Gorbunov, Postgraduate student, Department of Computer Software and Information Technologies,
Bauman Moscow State Technical University, Moscow, Russia;
orcid.org/0000-0002-4646-2636; e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it..
The problem of automated position estimation of a satellite image on a panoramic image using keypoints
is considered. The aim is to develop and evaluate a modified method for position estimation of a satellite
image on a panoramic image based on SIFT (Scale-Invariant Feature Transform) method. Standard
methods implementing this approach are shown to encounter difficulties in the presence of repetitive textures
and noise in images which leads to a decrease in position estimation accuracy. In this regard, a method optimized for remote sensing imagery is proposed. The set of method modifications including cross-matching of
features, rotation-angle filtering, and iterative estimation of transformation parameters using M-estimators
is defined. The research results show that, compared with standard SIFT method, the method proposed provides the increase of about 26% in the number of consistent matches, the increase of nearly 48 % in the proportion of correct matches, and the reduction of about 27 % in root mean square error. The authors conclude that the method proposed ensures more accurate and robust position estimation of a satellite image on a panoramic image and can be applied in practical remote sensing tasks
Key words: satellite images, panoramic image, keypoint detection, SIFT, false match filtering,
M-estimators, homography matrix, remote sensing