Karjalainen Mika
Helsinki University of Technology
Department of Surveying
This thesis describes possibilities of using SAR images in crop species interpretation. Researches made in England, Netherlands and Germany proved that the interpretation should be done area-based. In this work a statistical pattern recognition method so called Maximum Likelihood Classifier was used.
SAR (Synthetic Aperture Radar) is an active radar which uses the microwave part of the electromagnetic radiation. SAR works independently of solar illumination and it can achieve images through clouds. SAR image is formed of the echoes of the transmitted pulses. Intensity value of a SAR image pixel is proportional to the target's ability to return microwave radiation which is called backscattering of the target.
The interpretation is based on the crop growth which changes the backscattering of the target parcels. When SAR images are chosen at the right moments of the growing season every crop should have distinctive backscattering profile.
The results are represented in the end of the thesis. Interpretation accuracy of the individual classes varies between 38% and 84% and overall accuracy is around 70% when four SAR images were used.
Return to the page
Master's Theses
Return to the home page of
HUT Institute of
Photogrammetry and Remote Sensing