A Computer Vision‐Based Methodology to Estimate Fruit Colour Diversity in Ornamental Pepper (Capsicum spp.)

Abstract

Fruit colour diversity within different ripening stages confers ornamental value for pepper plants. Using images can be helpful in analysing the fruit colour-related genetic diversity and enable selecting accessions for ornamental purposes by avoiding subjectiveness. This study aimed to evaluate the phenotypic diversity among accessions of ornamental pepper using image processing techniques to identify fruit colours. Photos from 40 fruits from each of 12 accessions were captured by a camera from a smartphone. Separate channels and colour indices were extracted from the images to analyse the accessions using both dendrogram and principal components analysis. Dendrogram analysis allowed separating accessions with predominance of dark fruits from those whose fruits showed lighter and more vivid colours, with a cophenetic correlation coefficient of 0.811. The VARI and NGRDI vegetation indices were efficient in discriminating fruits with a predominance of green colour, and the BGI could be used to discriminate accessions with predominantly reddish-brown fruits. The proposed method can be used in small-scale breeding programs for accurately assessing the development of varieties regarding their fruit colour diversity.