Automatic Seed Classification by Shape and Color Features using Machine Vision Technology

Abstract

In this paper the proposed system uses content based image retrieval (CBIR) technique for identification of seed e.g. wheat, rice, gram etc. on the basis of their features. CBIR is a technique to identify or recognize the image on the basis of features present in image. Basically features are classified in to four categories 1.color 2.Shape 3. texture 4. size .In this system we are extracting color, shape feature extraction. After that classifying images in to categories using neural network according to the weights and image displayed from the category for which neural network shows maximum weight. category1 belongs to wheat and category2 belongs to gram. Experiment was conducted on 200 images of wheat and gram by using Euclidean distance(ED) and artificial neural network techniques. From 200 images 150 are used for training purpose and 50 images are used for testing purpose. The precision rate of the system by using ED is 84.4 percent By using Artificial neural network precision rate is 95 percent.