Influência da variação de limiar de segmentação em sistemas de mensuração de parâmetros radicular
Loading...
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universidade Estadual de Ponta Grossa
Abstract
In agronomy, measuring length, diameter and volume of the root system serves to indicate whether the crop is developing as planned, and to assist in the decision to apply fertilizers to the soil, making it possible to improve the health and productivity of the crop. However, this measurement must be meticulous since the separation of the samples, acquisition of the images and use of the software. The biggest computational challenge in this case is automation for separating the roots from the background of the image. Most existing software provides better results when the user participates in this process by setting the parameters, such as the threshold. The proposed work evaluates the sensitivity of the WinRhizo, RoMe and TMM software in the threshold variation, and aims to indicate the best threshold value for images of corn roots acquired by the scanner Epson Perfection V700 Photo, at 300 dpi, with the lid open, in colorful, professional mode, BMP format. First, the software was tested with nylon yarn images. They were tested with threshold values from 0 to 255 interval in 11 units, listing the mean relative error (RE) with the best results to be varied in 5 units above and below. The threshold value of 25 was the one with the best performance for the TMM and WinRhizo systems, and the value of 21 for the RoMe system. The RE of the WinRhizo, TMM and RoMe systems, for these threshold values were 3.99%, 4.12% and 69.72%, respectively. Applying this value to measure corn root images, REs of 49%, 79% and 87% were obtained for the WinRhizo, RoMe and TMM systems respectively. Repeating the procedure to find the best threshold value for the washed roots, REs of 13%, 17% and 34% were obtained with threshold values 42, 43 and 28, respectively for WinRhizo, TMM and RoMe systems. However, observing the determination coefficients (R²) and the standard deviation (SD) with the resulting REs, the conclusive analysis was the definition of an appropriate range of thresholds for the three systems. For root measurement, with SD<15%, RE<20%, and R² ranging from 0.933 to 0.965, the threshold range between 34 and 44 was defined as appropriate for the WinRhizo system. For the TMM system, for the same SD and RE as before, but with R² of 0.943 and 0.967, the appropriate threshold values are 25, 34, and 41 to 43. For the RoMe system, the threshold range adequate was listed with REs<35%: 29 to 30, with approximately SD=20%, and R²=0.82.
Description
Keywords
Medição de raízes, Limiar, Comprimento, Milho, WinRhizo, TMM, RoMe, Root Measurement, Threshold, Length, Corn, WinRhizo, TMM, RoMe
Citation
OLIVEIRA, Maria Carolina de. Influência da variação de limiar de segmentação em sistemas de mensuração de parâmetros radicular. 2020. Dissertação (Mestrado em Computação Aplicada) - Universidade Estadual de Ponta Grossa, Ponta Grossa, 2020.
Endorsement
Review
Supplemented By
Referenced By
Creative Commons license
Except where otherwised noted, this item's license is described as Acesso Aberto
