Classification of Soya Beans Based Image Processing Techniques and Artificial Neural Network
Umar Faruk Abdulhamid *
Department of Mathematical Sciences, Kaduna State University, Kaduna, Kaduna State, Nigeria.
Simon Daniel
Department of Mathematical Sciences, Kaduna State University, Kaduna, Kaduna State, Nigeria.
Usman Babawuro
School of Informatics Kazaure, Jigawa State, Nigeria.
*Author to whom correspondence should be addressed.
Abstract
The benefits of using technology in agriculture cannot be overemphasised because of its impact that results in an increase in the quality and quantity of crops produced, minimising cost of farming, and providing suggestions for prompt action among others. Traditionally, to know the state of soya beans, farmers rely on observation to note the change in colour of the leaves so as to provide appropriate action to the crop. This process cannot be fully reliable as colour is subjective to human impression; and failure to act when there are changes in the state of the soya beans especially when affected by diseases can reduce the expected yield. The goal of this study is to classify soya beans leaves into various categories such as healthy, unhealthy/disease, ripe not ready for harvest and ripe ready for harvest so that prompt action can be taken. The work has employed the use of colour and texture features of leaves through image processing techniques in the pre-processing phase and artificial neural network for the classification with the aid MATLAB. An accuracy of 95.7% is obtained in the classification of the various categories of soya beans leaves.
Keywords: Image processing, artificial neural network, soya beans, MATLAB, machine learning, object recognition.