Digital Classification of Naturally Colored Cotton Fiber by RGB Imagery

Abstract

Image-based phenotyping represents an innovative approach in breeding research, allowing rapid and reliable measurements that aid in decision-making. This study used image analysis to quantify the intensity of fiber coloration in Gossypium hirsutum L. genotypes and investigate the impact of bolls from the upper, middle and lower thirds of the plants. Seventeen cotton genotypes (5 cultivars and 12 F2:3 generation genotypes) were evaluated in relation to the intensity of fiber coloration. Three representative bolls were manually harvested per plot, per third development, and were photographed, including a colorcard, and these were processed with ImageJ software for correction of brightness and coloration. Data analysis was performed in R software, and the FIELDimageR package was used to calculate the BI, SI and NGRDI indices. Calibration of the images with colorcards resulted in higher averages but did not change the grouping of the observed averages. Compared with the other thirds, the upper third presented lower fiber staining intensity, with a variation of 6.25%, but there was no interaction between the genotypes and thirds. The SI allowed the distinction of genotypes by the saturation of the fiber color. The BI was correlated with the fiber characteristics UI, STR and UHML, with coefficients of determination of 0.77, 0.80 and 0.88, respectively. This digital cotton fiber quality evaluation method streamlines breeding decisions, enhances data accuracy and reduces costs.

Plant phenotyping relevance

綿花繊維の着色強度という植物器官形質を、RGB画像、色カード補正、ImageJ、FIELDimageR指標で定量化する方法が研究の中心であり、育種判断への適用と指標の評価も行っている。

Code and data availability

The paper describes a 96-image RGB dataset of cotton fiber, ImageJ/FIELDimageR analysis, and supplementary tables, but no public deposit of images, phenotype data, or author analysis code is stated. Supplementary tables are only described in-text with no availability URL. FIELDimageR and R are generic third-party tools

No evidence-backed public reproduction asset is currently recorded.