MACHINE VISION TECHNIQUES FOR SOMATIC COFFEE EMBRYO MORPHOLOGICAL FEATURE-EXTRACTION

Machine vision algorithms were developed to examine somatic coffee emb ryo morphological features between maturation and germination stages. A skeleton morphological feature was extracted and recognition rules w ere established to identify ''Y'' shaped skeletons. An improved thinni ng algorithm was developed to obtain single-pixel-width skeletons that simplified the recognition task. The ''Y'' shaped skeleton was found to be a promising feature to represent torpedo stage embryo. For the 1 27 embryos tested, the success rate was 73% in identifying torpedo-sta ge somatic coffee embryos from the selected morphological feature.