2D Sub-Pixel Disparity Measurement Using QPEC / Medicis

Abstract

Commission V, WG V/4KEY WORDS: 2D Stereo Matching, Sub-pixel Disparity, Geometric Calibration, QPEC, Medicis ABSTRACT:In the frame of its earth observation missions, CNES created a library called QPEC, and one of its launcher called Medicis. QPEC / Medicis is a sub-pixel two-dimensional stereo matching algorithm that works on an image pair. This tool is a block matching algorithm, which means that it is based on a local method. Moreover it does not regularize the results found. It proposes several matching costs, such as the Zero mean Normalised Cross-Correlation or statistical measures (the Mutual Information being one of them), and different match validation flags. QPEC / Medicis is able to compute a two-dimensional dense disparity map with a subpixel precision. Hence, it is more versatile than disparity estimation methods found in computer vision literature, which often assume an epipolar geometry. CNES uses Medicis, among other applications, during the in-orbit image quality commissioning of earth observation satellites. For instance the Pléiades-HR 1A & 1B and the Sentinel-2 geometric calibrations are based on this block matching algorithm. Over the years, it has become a common tool in ground segments for in-flight monitoring purposes. For these two kinds of applications, the two-dimensional search and the local sub-pixel measure without regularization can be essential. This tool is also used to generate automatic digital elevation models, for which it was not initially dedicated. This paper deals with the QPEC / Medicis algorithm. It also presents some of its CNES applications (in-orbit commissioning, in flight monitoring or digital elevation model generation). Medicis software is distributed outside the CNES as well. This paper finally describes some of these external applications using Medicis, such as ground displacement measurement, or intra-oral scanner in the dental domain.

Contribution of Photogrammetry for Geometric Quality Assessment of Satellite Data for Global Climate Monitoring

“…The tool is planned to be integrated into The overall workflow for the absolute, band-to-band registration, and multi-temporal GQA of SEVIRI that is implemented in the final GQA Tool is given in [25]. The algorithm validation with respect to the MEDICIS Tool of CNES [52] and the traditional EUM online and offline tools of EUMETSAT are also provided in the same publication [25], along with the long-term quality monitoring results. Debacker et al [21] compared The overall workflow for the absolute, band-to-band registration, and multi-temporal GQA of SEVIRI that is implemented in the final GQA Tool is given in [25].…”

Sentinel-2 Global Reference Image Validation and Application to Multitemporal Performances and High Latitude Digital Surface Model

“…This tool uses local stereo-rectification to increase matching performance and is built with a modular architecture that allows to change the matching (or disparity estimation) algorithm. In this work, matching has been carried out by the in-house correlation tool Medicis (Cournet, 2016), which seems the most appropriate given the observed landscape and allows efficiently filtering bad matches over shadows, cloud or watering areas. Some modifications were made by CNES to the S2P software.…”