Trichromatic sorting of in vitro regenerated plants of gladiolus using adaptive resonance theory

A machine vision system is described to sort the regenerated plants of gladiolus into groups using trichromatic features of leaves. The machine vision system consisted of a scanner, image analysis software and an adaptive resonance theory neural network. Leaf attributes extracted from the image histograms and used for network classification are the mean brightness, grey-scale level and the maximum pixel count. The system was able to sort the regenerated plants into two distinct groups based on the photometric behaviour. Vigilance parameter had a significant effect on grouping. The approach may provide a means of selecting plants suitable for ex vitro transfer and also helps in quality control of commercial micropropagation.