Adaptive target spray system based on machine vision for plant protection UAV

Abstract

Abstract: Aiming at solving drawbacks of traditional spraying systems such as uniform spraying mode which is prone to waste of chemical pesticides and cannot be mounted on UAV, this paper designed a machine vision based system to identify crop coverage for target spraying by UAV. This research built a hardware device of the system using Raspberry Pie as the main controller, and then analyzed the grayscale processing effect of the ultra-green method, ultra-green and Ultra-red method, standard deviation index method in sunny and cloudy days. The result is that the Ultra-green and Ultra-red method has a better grayscale effect. Therefore, a spraying decision model based on rice canopy coverage calculation was constructed in this research. According to the rice canopy coverage, the system adjusts the spray nozzles to full, half, and no-spray states. The rice canopy identification model is evaluated in this paper based on four indicators: relative error of coverage, grayscale time, segmentation time, and total time. The experimental results show that the comprehensive performance of the ultra-green and ultra-red-maximum entropy is excellent, with the performance indexes of 5.43ms, 11.356ms, 11.356ms, 4.409ms, and 15.765 ms. Water-sensitive paper droplet distribution experiments show that the system can reduce unnecessary agent waste and provide a reference for the application of machine vision technology to target spraying by plant protection UAV. Keywords: plant protection UAV, target spraying, raspberry pie, rice canopy coverage, threshold segmentation DOI:  10.33440/j.ijpaa.20200303.92  Citation: Jinbao Hong, Yubin Lan, Xuejun Yue, Zhenzhao Cen, Linhui Wang, Wen Peng, Yang Lu. Adaptive target spray system based on machine vision for plant protection UAV.  Int J Precis Agric Aviat, 2020; 3(3): 65–71. Â