As part of the COVID-19 Custom Script Contest, while collecting ideas on how to reduce the effects of the COVID-19 pandemic, Henrik Fisser developed a method for detecting moving trucks on roads using Sentinel-2 satellite imagery. In this thesis, we used the Fisser's method for detection of moving trucks which is based on spectral bands ratio, and exploits a temporal sensing offset of the Sentinel-2 multispectral instrument. On the satellite images with a 10 meters spatial resolution, moving trucks create an iridescent reflection trail. During the detection, we focused on the blue and red part of the coloured trail, which represent the start and end of the truck. We detected them on the basis of the raster layers arithmetic operations of the Sentinel-2 spectral bands and by determining the absolute thresholds of individual results.
We connected the detected red and blue part of the coloured trail with our own algorithm as well as an algorithm based on phase cross-correlation. Using the parameters of the sensor and the position of the pixels belonging to the detected truck, we then calculated their approximate velocity. From that we then calculated the average velocity of the trucks on the highway section. The results were analysed with the help of two web applications, Poligram and Wialon, both used for tracking trucks. The approximate velocity of the detected trucks and the average velocity of the trucks on the highway section were consistent with the reference speed of the trucks that were actually moving at the time of the satellite images being taken over a certain area.