Use of digital image analysis for the flower color evaluation in ornamental sunflower

Abstract

Sunflower (Helianthus annuus L.) is broadly used as an ornamental plant in landscape gardening, but also as a potted plant and a cut flower. Since aesthetic traits, including color, are most important for newly developed ornamental plants, sunflower petal (ray floret) color has a high value for the development of new genotypes and its position on the horticulture market. The most common methodology for the evaluation of sunflower petals is based on UPOV guidelines for sunflower. By the guidelines, the color of sunflower ray florets can be described as yellowish white, light yellow, medium yellow, orange yellow, orange, purple, reddish brown and multicolored. Although there are photographs to define these color categories, the provided material does not give clear information on the color definition, especially the multicolored category. The main obstacle of this methodology is its high subjectivity and necessity of high expertise for evaluators. In order to make the process of evaluation of sunflower petal color more objective, we propose a new methodology that combines image segmentation (pixel-based classification), and UPOV sunflower guidelines for the definition of color groups (classes). Images of six sunflower genotypes (Ring of Fire, CMS1-30, Heliopa, Dwarf, Neoplanta and Pacino Gold) were used in the software analysis. Visual results of this process of image segmentation presented different colors for the examined varieties. This visual presentation serves as a guideline for an evaluator to determine whether there is more than one dominant color in the examined genotypes.