TECHNIQUES BASED ON UAV IMAGES FOR WHEAT PRODUCTION ESTIMATION.

The aim of the study was to use a technique based on aerial images (UAV) to wheat production estimation. The images were taken with the drone (UAV, DJI Phantom 4) at different image capture heights (ICH), in the range of 1.5 - 50 m from the ground level, and resulted in 18 sets of images (trials, T). The resized images (crop image, 1,000 x 2,000 pixels) were analyzed and the information for RGB color parameters was obtained. Very strong correlations were recorded between color parameters R and G (r=0.993), as well as between these two color parameters and ICH (r=0.940 in the case of R, respectively r=0.918 in the case of G). Based on the trials distribution in the PCA diagram and in the CA dendrogram, 14 trials (T1, T3-T6, T8-T10, T12-T14, T16-T18) were selected for Training & model construction (Tmc), and four trials (T2, T7, T11 and T15) for were used for Testing & model validation (Tmv). Through the regression analysis, a wheat production estimation model was obtained based on RGB parameters under statistical safety conditions (R2=0.999, p<0.001). Flowchart diagram of process was proposed for the present study, and the work process, respectively the obtained models can be adapted for various other agricultural crops. [ABSTRACT FROM AUTHOR]