Agricultural event detection from the satellite image zonal statistics
Roto, Mikael (2023-05-30)
Agricultural event detection from the satellite image zonal statistics
Roto, Mikael
(30.05.2023)
Julkaisu on tekijänoikeussäännösten alainen. Teosta voi lukea ja tulostaa henkilökohtaista käyttöä varten. Käyttö kaupallisiin tarkoituksiin on kielletty.
avoin
Julkaisun pysyvä osoite on:
https://urn.fi/URN:NBN:fi-fe2023060151715
https://urn.fi/URN:NBN:fi-fe2023060151715
Tiivistelmä
Satellite images have become an important tool for event detection and monitoring. The key advantage for the satellite based monitoring is their ability to cover large areas frequently which in turn makes them very cost-efficient solution for monitoring geographically large areas. Due to advances in the satellite technology and the image processing techniques, satellites are capable of providing high resolution data within real-time from the Earth’s surface.
In this thesis we provide a brief introduction to the satellite based remote sensing and how these methods can be used to model different agricultural events. We inspect theoretical satellite signal responses to a common agricultural events and try to detect these patterns from our own dataset.
We develop a method to process satellite images into signals and apply preprocessing methods to increase signal to noise ratio. We then train a gradient boosting classifier to the smoothened signals and process the individual predictions so that we can detect the start and end times for various agricultural events from the agricultural parcels.
In this thesis we provide a brief introduction to the satellite based remote sensing and how these methods can be used to model different agricultural events. We inspect theoretical satellite signal responses to a common agricultural events and try to detect these patterns from our own dataset.
We develop a method to process satellite images into signals and apply preprocessing methods to increase signal to noise ratio. We then train a gradient boosting classifier to the smoothened signals and process the individual predictions so that we can detect the start and end times for various agricultural events from the agricultural parcels.