Stereoprocessing
Overview
The module 5 of the course is dedicated to an introduction on the stereo processing from fixed-time-lapse cameras.
Learning outcomes
In this module you will learn:
- how to perform a feature matching between two images by using Deep Learning algorithms, that are necessary for wide baseline stereo reconstruction;
- how to perform a stereo reconstruction from a stereo pair of images acquired in a single epoch;
- how to perform a multitemporal stereo reconstruction from a set of stereo pairs of images acquired in different epochs.
Table of contents
-
Getting started: exaplin how to install ICEpy4D that will used for the processing.
-
Data preparation: explains how to download the data and how to organize it for the processing.
-
Feature matching: explain how to perform a feature matching between two images by using Deep Learning algorithms, that are necessary for wide baseline stereo reconstruction.
-
Single epoch stereo reconstruction: explain how to perform a stereo reconstruction from a stereo pair of images acquired in a single epoch.
-
Multitemporal stereo reconstruction: explain how to perform a multitemporal stereo reconstruction from a set of stereo pairs of images acquired in different epochs.
Software
We will use ICEpy4D Python toolkit for the stereo processing. ICEpy4D is an open-source project for multitemporal photogrammetry developed by the Geodesy and Geomatics Laboratory of Politecnico di Milano and it is available on GitHub at https://github.com/franioli/icepy4d.
For the Bundle Adjustment and the dense reconstruction, you also need Agisoft Metashape Professional Edition and the Metashape Python API (see Getting started for more details).
A Linux environment is strongly recommended for the processing, but also Windows will work.