During stereo reconstruction algorithm development it is necessary to have benchmark dataset with ground truth and evaluation tool. The benchmark dataset allows to make important design decisions, tune the algorithm and compare it against competitors. There are a lot of publicly available stereo datasets such as Middlebury and KITTI , however all these datasets consist of images that are not representative of satellite stereo images of Mars.
Context camera (CTX) is a camera carried on-board of Mars Reconnaissance Orbiter (MRO) satellite. Stereo image pairs acquired by this camera are freely available on-line and can be used to derive Digital Elevation Models (DEM). In this project we will create a convenient benchmark dataset that consist of satellite stereo images of Mars taken by CTX camera.
The main task of the project is to create dataset from 20 CTX image pairs and compute ground truth distance maps for these images, using data from Mars Orbiter Laser Altimeter (MOLA) or MRO HIRISE camera (instruments with higher precision). The secondary task is to code scripts that will compute different score metrics for given distance map and ground distance map