Development of a Simple Radiometric Calibration System for Multispectral Images of an Unmanned Aerial Vehicle

摘要

An Unmanned Aerial Vehicle (UAV)-based remote sensing (RS) technique is a useful tool to observe spatial variation in crop growth conditions. To ensure the reliability of UAV-based RS images and acquire surface reflectance, the radiometric calibration using pseudo-invariant tarns has been generally used. While calibration tarps are used in general, it is difficult to apply them on multiple sections of a large field requiring multiple images. In addition, using calibration parameters from the tarps measured one time for multiple images can cause errors due to differences in an image acquisition time. Therefore, a calibration method integrated into a UAV is required to perform practical calibration. In this study, we aim to develop a standalone UAV-based calibration system without the calibration tarps through the quantification of sensor responses of a multispectral camera, which varies with an intensity of incident light. To develop the standalone UAV-based calibration system, we assembled a UAV with four propellers and a dimension 680 mm in wing's length and 250 mm in height. The system was composed of a multispectral camera (Tetracam, Inc., CA, USA) with green, red, and near-infrared filters to acquire spectral images. A line quantum sensor (LI-COR, Inc., NE, USA) was used to measure the intensity of the incident light to obtain input parameters of the calibration algorithm. We first constructed a library of sensor responses using pseudo-invariant tarps according to the intensity of incident light, and the relationship between the two factors was detennined. The calibrated images will be validated using the reflectance measured in crop fields. We report the evaluation outcome of the UAV-based calibration system. We also present spatial infonnation on the crop growth conditions based on spectral indices. Our study result suggests that the formulated standalone UAV-based RS system would be useful for practical and applied image processing for UAV-based RS images.