Low-cost multispectral imaging system for crop monitoring A. Montes de Oca, L. Arreola, A. Flores, J. Sanchez and G. Flores Abstract— This work presents the design and development of a multispectral imaging system to precision agriculture tasks. The imaging system features two small digital cameras controlled by a microcomputer embedded in a drone. One of the cameras has been modified to be sensitive to near-infrared radiation reflected by the vegetation, whereas the other one remains as a common RGB camera. In order to determine the health status of the crop, the Normalized Difference Vegetation Index (NDVI) is computed in a developed software. Once the aerial imagery is obtained by the drone, it is processed to eliminate image distortions and insert specific metadata needed for generating the orthomosaics with the health information of the plant or soil of interest. Finally, the vegetation index will be computed from the visible and near-infrared orthomosaics for a better interpretation of the user. Experiments are presented to show the effectiveness of the system. Index Terms—UAV; Drone; Precision Agriculture; Multi- spectral Imaging; Image georeferencing; Normalized Difference Vegetation Index; Aerial photogrammetry. I. I NTRODUCTION The use of Unmanned Aerial Systems (UAS) in precision agriculture applications has increased in the last three years. This is mainly due to the UAS capability to provide the farmers with important information related to crop health for a better input management. This allows the constantly growth resource optimization, a underlaying issue for farm- ers. Furthermore UAS are relatively cheap in comparison with manned aircraft or satellite-based systems. They are also small and easy to use. All these facts promote the growing popularization of agriculture UAS. In this paper an easy-to-implement and low-cost system is proposed for basic agriculture tasks, such as NDVI computation and crop imagery collection. There is an important fact about this technology: it has been estimated that 80% of the sold drones from 2015 to 2025 will be used in applications related to precision agriculture [1]. Aerial imagery collected by drones provide fast and reliable information of crop fields. This data allows to determine the health status of a crop by calculating the so-called vegetation indices. Among these indices, the Normalized Difference Vegetation Index (NDVI) has become popular due to the precise data that can be obtained analyzing it. Most of vegetation indices are computed using reflectance values of specific wavelengths along visible and infra-red spectra. To measure that radiations, certain projects propose Corresponding author: G. Flores (email: gflores@cio.mx). A. Montes de Oca, L. Arreola, A. Flores, J. Sanchez and G. Flores are with the Perception and Robotics Laboratory, Centro de Investigaciones en ´ Optica, Le´ on, Guanajuato, Mexico, 37150. This work was supported by the FORDECYT-CONACYT under agree- ment 000000000292399. Multispectral camera system Fig. 1: The unmanned aerial vehicle used in this work. This drone is equipped with the proposed multispectral imagery system presented in this paper. to use hyperspectral imaging by means of spectographs [2], [3]. A good understanding of these indices can help to save significantly the input sources in order to obtain an equal or even larger output from crop fields [4]. Also multispectral imagery is obtained using customized and modified digital cameras as in [5] [6], [7]. Relating the UAS configuration used in crop monitoring, in [8] and [9] fixed-wing platforms have been used, while in [10] and [11] quadrotor-based sys- tems has been implemented. The specific UAS is determined by the task that must be done. Due to its aerodynamics, fixed-wing aircrafts can travel more distance than a quadrotor with the same amount of power. This is traduced in a larger area coverage and also a larger amount of aerial imagery collected. A quadrotor instead, is more precise and easier to manipulate in smaller areas. This make it ideal for outdoor testing on a lower scale. In this paper an aerial imaging system was developed with the use of a quadrotor and a low-cost image acquisition system. The imaging system is composed by two digital cameras and a microcomputer. The microcomputer is used to develop a trigger system compatible with the flight controller and the cameras. This trigger system is responsible to activate the camera shot at a given distance traveled by the drone. The prototype shown at Fig. 1 is used to collect aerial imagery of the interest area. Once the flight is completed, the images are processed to generate orthomosaics. These orthomosaics correspond to near-infrared and visible radiation and will be used to compute the NDVI. Having this index will help identifying damaged areas in the vegetation. The remainder of this paper is organized as follows. Section II offers an overview of the elements involved in the 2018 International Conference on Unmanned Aircraft Systems (ICUAS) Dallas, TX, USA, June 12-15, 2018 978-1-5386-1353-5/18/$31.00 ©2018 IEEE 443