Image Classification of Wheat seeds using Supervised Learning Techniques

Abstract

The fundamental approach of the undertaking was to build up a strategy to classify the different types of wheat seeds.It describes the application to differentiate between various types of wheat seeds including different size, shape and color features. The sys- tem involved the process of collecting datasets used for performing image processing and image classification steps by using supervised learning techniques. The framework included the way toward gathering datasets for image processing and image classifica- tion ventures by utilizing machine learning methods.Dataset of Wheat seeds were taken for classification. Varieties of wheat like Bhalia, Sonalika, Tukdi, Daudkandi, Sharbati, Lokvan were taken in which some of the samples were taken for training and some for the testing phase.Each variety having 50 samples so total 300 samples were taken into consideration.Features like Area,Perimeter,Compactness, Length of kernel,Width of ker- nel,Asymmetry coeficient,Length of kernel groove are selected as Predictors while all six classes are imported as Responses. According to study and experiments, Radial basis function(RBF) Support vector machine (SVM) classification provided the best accuracy results compare to the FineKNN classification algorithm. For implementation point of view, image pre-processing steps and classification were done in MATLAB while feature extraction steps were done by using Digimiser image analysis software.