Plant-Stand Count and Weed Identification Mapping Using Unmanned Aerial Vehicle Images

Abstract

Modern agriculture encounters several challenges these days. There is a vital need for spatial data about plant and weed distributions. Obtaining accurate knowledge of the plants and weeds distribution in the field with manual methods are time-consuming. In this research, image processing programs were developed from the unmanned aerial vehicle (UAV) digital images to obtain the plant-stand count and weed identification and mapping in the field. Algorithms using pixel-march with search-hands criterion for the plant-stand count and shape-based features for weed identification were developed. Results were found to be accurate in the cropped UAV stitched images (>99 %) in manual image-based validation. User-friendly message windows, labeled images, textual results, and distribution maps were produced as outputs. The outcomes of this study will enable farmers to determine the plant and weed distributions in the field and will be helpful in deploying precision agriculture measures.