Published in Scientific Papers. Series "Management, Economic Engineering in Agriculture and rural development", Vol. 23 ISSUE 4
Written by Andreea Lidia JURJESCU, Florin SALA

The study used imaging analysis to describe and characterize a soybean crop, in terms of plant uniformity and the presence of weeds, with the formulation of intervention decisions for crop management. Images were taken in digital format (cell phone camera), which were analyzed for spectral information (RGB) and fractal geometry (D). Based on the obtained data, normalized values (rgb), different ratios (R/G, R/B, G/R, G/B, B/R, B/G), specific indices (NDI – normalized difference index, INT - intensity) and degree of vegetation cover (DVC, %), as green canopy cover. A flow diagram was developed for the study carried out in the assessment of soybean culture. From the correlation analysis, the values of the correlation coefficient (r) were obtained, in conditions of statistical safety. The regression analysis led to obtaining some equations that described the variation of DVC in relation to the considered parameters. Very high levels of precision and safety were recorded based on the R/G ratio, the corrected g value (R₂=0.998, p<0.001), followed immediately by the equations based on the G/R ratio and the NDI index (R₂=0.997, p <0.001). 3D models and in the form of isoquants described the variation of DVC in relation to the considered parameters. According to PCA, PC1 explained 95.997% of variance, and PC2 explained 3.9978% of variance. Cluster analysis led to the grouping of the variants based on similarity, highlighting the variants with a high degree of weeding (s8, s9), the variant with a normal state (s7) and the other experimental variants.

[Read full article] [Citation]