Drone based image processing for precision agriculture
Loading...
Date
2019-06
Authors
Muhammad Arif Syafiq Bin Md Sharif
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
In today’s world, with an advent of technological advancements, the use of automated monitoring in agriculture is gaining increase in demand. In the agricultural field, yield loss occurs primarily due to widespread disease. Most of the disease is detected and identified when the disease progresses to a severe stage. A specially equipped UAV can perform several important tasks in agriculture, including monitoring the agriculture land and perform disease detection for several plants at an early stage. Currently, disease traits in agriculture are visually assessed, which can be time-consuming, less accurate and more subjective. Hence, in this project, image processing is used for the detection of plant disease. Detection of plant disease using automated image processing method is beneficial as it can reduce huge work of monitoring in big farms comprising of numerous crops. Moreover, in order to monitor big farms it is a viable option to use unmanned aerial vehicle on specific drone (UAV) to take the snap shots of various diseased plants from multiple angles. This study proposes a parallel image segmentation algorithm in order to detect the diseased leaf in Coconut, Palm, Banana, Dwarf Palmetto and Sapodilla plants acquire using Parrot PF728000 Anafi Drone with 4K HDR Camera. At first, the parallel K-means clustering algorithm was applied on the acquired image to segregate various components acquired using UAV. Post K-means clustering, the diseased portions of the plants were assessed using Hue-Saturation-Value (HSV) based image segmentation algorithm. Moreover, a comparison for image segmentation was also done on non-K-means clustered image and K-means clustered image for which a difference of 18394𝑚𝑚2 , 80931𝑚𝑚2 , 43361𝑚𝑚2 , 16293𝑚𝑚2 and 77542𝑚𝑚2 corresponding to Coconut, Banana, Dwarf Palmetto, Palm and Sapodilla was obtained sequentially. The outcome of this project reflects that high-throughput phenotyping techniques will potentially improve the throughput and objectivity of detecting healthier plants and other crops, and will subsequently contribute to the development of new cultivars in breeding programs and yield estimation in precision agriculture.