Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa, 2020, Vol. 17, No. 7, pp. 155-163
Automated calculation of plant density based on unmanned aerial vehicle photos
A.A. Zlatopolsky
1 , S.A. Yarovoy
2 1 Space Research Institute RAS, Moscow, Russia
2 OOO Central Black Earth Agro-Industrial Company, Voronezh, Russia
Accepted: 09.12.2020
DOI: 10.21046/2070-7401-2020-17-7-155-163
This paper concerns the automatic determination of the agricultural plants density. Knowledge of the seedling density is necessary to estimate the quality of work performed, to predict the amount of harvesting, and to avoid unnecessary costs in case of poor seedlings. Currently, in production, the density of plants is calculated manually. We have considered the possibility of automatically identifying individual plants and counting their number in the field photo. To do this, sections of the sugar beet field were photographed from an unmanned aerial vehicle shortly after the emergence of seedlings. The paper describes the image analysis algorithm that includes color analysis, spatial filtering, and morphological analysis. This algorithm allowed us to get an estimate of the number of plants with good accuracy. It shows how algorithm parameters are set and how critical their choice is. The results of testing the proposed algorithm on images of a sugar beet field taken from an unmanned aerial vehicle at an altitude of about 30 meters are presented. The actual resolution of the image is estimated based on the known distance between rows, which is maintained very accurately when using modern sowing equipment. The technology for determining the density of plants, which can be built on the basis of the proposed algorithm, is described.
Keywords: plant density, sugar beet, UAV photography, color analysis, spatial filtering, morphological analysis, connectivity
Full textReferences:
- Serra J., Image analysis and mathematical morphology, Orlando, USA: Academic Press Inc., 1982, 621 p.