Abstract
Abstract. Radiometric calibration has become important pre-processing with increasing use of unmanned aerial vehicle (UAV) images in various applications. In order to convert the digital number (DN) to reflectance, vicarious radiometric calibration is widely used including relative radiometric calibration. Some UAV sensor systems can measure irradiance for precise relative radiometric calibration. However, most of UAV sensor systems cannot measure irradiance and therefore precise relative radiometric calibration is needed to produce reflectance map with vicarious calibration. In this study, an optimal image selection method is proposed to improve quality of relative radiometric calibration. The method, relative calibration by the optimal path (RCOP), uses filtered tie points acquired in geometric calibration based on selection optimal image by Dijkstra algorithm. About 100 multispectral images were acquired with a RedEdge-M camera and a fixed-wing UAV. The reflectance map was produced using RCOP and vicarious calibration using ground reference panels. A validation data was processed using irradiance for precise relative radiometric calibration. As a result, the RCOP method showed root mean square error (RMSE) of 0.03–0.10 reflectance to validation data. Therefore, the proposed method can be used to produce precise reflectance map by vicarious calibration.
Highlights
Advancing technologies are making sensors smaller, more precise, and more popular
Seven ground reference panels were deployed on the grass of a soccer field for vicarious radiometric calibration of the unmanned aerial vehicle (UAV) images (Figure 3a)
(b) Figure 2. (a) The KD-2 fixed-wing UAV, and (b) the multispectral camera and irradiance sensor (b) Figure 3. (a) Ground reference panels installed for vicarious radiometric calibration of UAV images, and (b) spectral reflectance measured using a spectro-radiometer
Summary
Advancing technologies are making sensors smaller, more precise, and more popular. For UAV cameras without an irradiance sensor, conversion coefficients have been estimated through regression analysis between the DNs of pixels on overlapping regions between two images (Suomalainen et al, 2018) This method is prone to geometric distortions and pixels with radiometric anomalies (Xu et al, 2019), and they often result in visual discontinuity between adjacent scenes (Liu et al, 2011). An optimal image selection method is proposed to improve quality of relative radiometric calibration It uses filtered tie points acquired in geometric calibration based on optimal image selection by Dijkstra algorithm. It can minimize error accumulation by reducing the step of reaching the last image located at the region boundary.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have