Economical thermal-RGB imaging system for monitoring agricultural crops

Abstract

In this study, a low-cost thermal-RGB imager was developed for use in agricultural crop monitoring applications. It is weatherproof, and has a geo-referencing capability along with a power management panel that allows unattended field deployment of the systems for crop monitoring over extended period of time. The imager is made up of single-board Linux-based computer integrated with RGB and thermal imaging modules. The imager can be configured as FTP server to allow data transfer to/from a client computer. Developed was also the custom image-processing algorithm which overlays, aligns thermal and RGB images, and mask for the thermal image to remove the soil background and shaded leaves. The algorithm outputs are the average temperature of sunlit leaves and canopy coverage. Prior to field validation, the performance of ten thermal modules and four fully assembled RGB-thermal imagers were assessed under laboratory conditions. In the spring of 2017, two imagers were mounted on a center pivot retrofitted with Medium Elevation Spray Application (MESA) and Low Elevation Spray Application (LESA) systems in a mint field near Toppenish, WA. The thermal modules showed an accuracy of ±2.4 °C on average over a range of 0–50 °C of a blackbody target. Although accurate for larger canopies, the imperfect alignment of RGB and thermal images introduced significant errors in the calculations of sunlit leaves surface temperature in images with small canopy coverage. Further investigations revealed that the first peak of thermal image relative frequency histogram could be a more accurate representative of sunlit leaf surface temperature. Overall, the amended image-processing algorithm was able to successfully extract canopy surface temperature and percent canopy cover from a wide range of images captured during the crop growing season. The current design of imager allows for creating a network of imaging units in the field to obtain real-time surface temperature data from plant canopies. The system has the potential to be used for creating evapotranspiration and prescription maps in real-time, and irrigation scheduling.