2017

The aim of the project was to develop and optimise new algorithms for the automatic registration of images. Compared to existing methodologies, that which was developed during the project has the peculiarity of evaluating a global transformation of the entire data set, in analogy to the bundle block adjustment of photogrammetry.
The experiment centred around the processing of time series captured using optical satellites, with the aim of overcoming the typical limitations due to occlusions (shadows, snow, cloud) and the change in land cover. The results have demonstrated how the multi-image methodology that has been developed is also capable of automatically processing sets of data that the traditional techniques are not capable of processing.