Towards Classification of Weeds through Digital Image

Maximum production of crops mostly depended on proper management of weeds. In this paper we proposed an automated weed control system which can differentiate the weeds and crops from the digital image. The images were segmented to separate plant from soil. This paper demonstrates the classification of weeds and crops according to twelve extracted features. Four hundreds sample images over five species were taken where each and every species contains 80 images. Minimizing the computation cost and achieving high accuracy rate, Naïve Bayes classification algorithm has been proposed as it gains 98.9% accuracy over 400 sample image.