Evaluation of Morphological Variation of Common Bean Seeds Through Image Analysis

Measurement of various morphological traits of common bean seeds and grouping of landraces collected from Uttarakhand was carried out using image analysis. Projected area, perimeter, major diameter, minor diameter and color of seeds were measured using image analysis. Colors were expressed in terms of RGB channels (red, green and blue) and HSL channels (hue, saturation and lightness). Images of 50—100 seeds per landrace in three replicates were captured. All image analysis was carried out using ImageJ, a Java based, multithreaded, freeware image processing and analysis program. Weight of 100 seeds was also taken into account. All data were subjected to principal component analysis, where 90.9% of the variability was expressed and used to group landraces into clusters. Clusters were further refined by hierarchical clustering using the Unweighted Pair Group Method with Arithmetic Mean (UPGMA), and the resulting dendrogram with cophenetic correlation coefficient of 0.97 grouped 26 landraces into four clusters. Analysis of variance revealed that only the color feature significantly discriminates these landraces.