UDC 004.932.2, 629.735
DETERMINING OPTIMAL UAV ALTITUDE FOR ON-BOARD OBJECT RECOGNITION BY IMAGE
B. V. Kostrov, Dr. in technical sciences, professor, Department of Electronic Computers, head of the
department, RSREU, Ryazan, Russia;
orcid.org/0000-0002-2984-4480, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
R. V. Khrunichev, Ph.D. (in technical sciences), associate professor, Department of Electronic Computers,
RSREU, Ryazan, Russia;
orcid.org/0009-0006-1442-0649, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
The article considers the problem of selecting the flight altitude of an unmanned aerial vehicle (UAV)
for on-board recognition of static objects from a video sequence. The relevance of the study is due to the increasing use of UAVs for monitoring certain territories and the need to ensure high-quality object recognition
with limited computing resources of on-board systems. The aim of the study is to develop a mathematical
model and methodology for selecting UAV flight altitude, ensuring the required quality of recognizing
the objects of various sizes during on-board video data processing. The main objectives are: constructing a
geometric model of image formation, determining the dependence of object size in the image on flight altitude
and camera parameters, accounting for the effect of aircraft motion on image quality, and constructing
a probabilistic detection model. Analytical relationships between flight altitude, image spatial resolution,
and object size in pixels are obtained. Based on the obtained models, the problem of optimally selecting the
flight altitude is formulated, taking into account the requirements for the probability of object detection and
computing performance of on-board system. Based on open aerial photography dataset, the dependence of
permissible flight altitude on the characteristic size of detected objects and camera parameters is shown.
Key words: Unmanned aerial vehicle, object recognition, video surveillance, image processing, flight altitude,
image spatial resolution, YOLO, VisDrone