Detecting the Weeds in Crop Fields Using Image Segmentation

Abstract

In image processing, even though there exist many advanced tools for classification and recognition of the images, some unrelenting challenges need to be faced when it transfers to different application requirements. In agricultural industry, discriminating weed coverings and crop rows under uncontrolled lighting on real-time remains a challenging task in image processing technique. The paper presents an image processing technique to get a knowledge and information within the crop field for the distribution of weeds, a prerequisite for site-specific treatments in agricultural sector, which is of boundless economic importance. An adjustable algorithm for segmentation of colour image using the Principal Component Analysis (PCA) and Otsu's thresholding method has been proposed for sorting and grading the weed in the crop field. It is carried out by constricting three-dimension vector of an image to one dimension using Principal Component Analysis method. We demonstrated here how image processing technique in MATLAB could be employed for weed classification in crop fields.