Author Archives: facciolo

This project is only for hackers, proficient with unix and python. The 2019 IEEE GRSS Data Fusion Contest track 3 [2] concerns multi-view stereo: Given multi-view images for each geographic tile, the objective is to predict semantic labels and a … Continue reading →

A naive way to create a “fake” orthophoto from a single satellite image consists in projecting the photo into a horizontal surface.  This is correct when the ground is flat, and the ground is often nearly flat, however it does … Continue reading →

The famous and elegant Patch Match algorithm [1], seen on the course, finds a dense set of correspondences between two images, by iteratively updating an initial random disparity chart.  These disparities are then “triangulated” to obtain a cloud of points … Continue reading →

The projection and localization functions of satellite images are often [1,2] represented by rational functions of degree 3 in 3 variables.  This gives a very precise approximation of the physical model. The advantage of this representation is that it provides … Continue reading →

The objective of this project is to evaluate the state-of-the-art on stereo matching algorithms for satellite images. The starting point will be the analysis of a recent CNN-based stereo matching algorithm proposed in [1], or a more recent article judged … Continue reading →

The optical stereo pipeline described in the course is a simplified version of the S2P pipeline [1], which was developed for high resolution pushbroom satellites such as Pleiades or WorldView. One of the particularities of these satellites is that the … Continue reading →