Image Segmentation based Methodology for Classification of various Seed varieties

The main objective of this research is to develop a methodology for identifying different varieties of paddy seeds using morphological features. Morphological features are powerful tool which could separate the different varieties of seeds. Three varieties of paddy seeds are taken that are classified according to their morphological feature range. Image of paddy seeds was acquired by a canon digital IXUS 132 camera. Images were stored in jpg format for further analysis Contour based feature extraction and traversal of seeds are carried out. This methodology consist of several steps image acquisition, segmentation using mean shift segmentation that is based on color merge over the space, binarization using otus’s threasholding method, feature extraction and classification. Six morphological features are extracted from paddy seed images and they are area, diameter, major axis, minor axis, compactness, roundness. All three verities that are classified in this paper vary according to all morphological features. More numerical results are presented in this paper.