No full text
Unpublished conference/Abstract (Scientific congresses and symposiums)
RockePedia - Towards a shared intelligence platform to support computer-assisted rock identification?
Pirard, Eric
2021International Rock Imaging Summit 2021
 

Files


Full Text
No document available.
Annexes
PUB_21_10_PPT_RockePedia.pdf
Publisher postprint (9.12 MB)
Download

All documents in ORBi are protected by a user license.

Send to



Details



Keywords :
pixel; roxel; rock imaging
Abstract :
[en] For over forty years, with the growing availability of frame grabbers and later digital scanners and cameras, geologists have been collecting, storing and often losing billions of pixels in a large diversity of storage devices. These pixels were simply picture elements and, as such, they carried very little information about the target material they represented. Most of the pictures obtained from microscopy or outcrop imaging contained at best a triplet of coordinates in an RGB colour space and most of the time a simple grey level intensity with no calibration test pattern. With the advent of scientific imaging and the generalisation of tracking and positioning techniques, we should now make sure that every pixel we collect is a roxel or, in full words, a rock element. A properly acquired roxel will be the necessary building block of a future digital twin of the Earth crust. It is the responsibility of an international rock imaging society to define the quality criteria that all scientific images of rocks and minerals should meet. In doing so, it is hoped that in the reasonably near future it will be possible to pool all the acquired images from geological surveys to individual scientists into a large collective visual intelligence database tentatively named hereafter RockePedia. Just like what is done in the medical world, we can imagine that RockePedia is developed from a sharing platform like Cytomine. Such a platform would not only allow the archiving of very large images but also their exploration and annotation in an interactive and collective manner. This collective intelligence, coupled with the pooling of all the images acquired by mineralogical mapping systems operating on the scale of microscopy, core scanners and even satellites, should make it possible to constitute a unique database that can feed powerful artificial intelligence algorithms. Visual experience in geology is very important. It is through repetition of this experience that geologists gradually acquire a skill that allows them to move from seeing to understanding the geological process that gave rise to a particular rock or texture. It is this same visual experience that allows us to recognise clues that are essential to guide exploration for both mineral resources and petroleum reservoirs.
Disciplines :
Geological, petroleum & mining engineering
Author, co-author :
Pirard, Eric  ;  Université de Liège - ULiège > Département ArGEnCo > Géoressources minérales & Imagerie géologique
Language :
English
Title :
RockePedia - Towards a shared intelligence platform to support computer-assisted rock identification?
Publication date :
11 November 2021
Event name :
International Rock Imaging Summit 2021
Event date :
9 - 11 Novembre 2021
Audience :
International
Available on ORBi :
since 09 November 2021

Statistics


Number of views
95 (7 by ULiège)
Number of downloads
32 (0 by ULiège)

Bibliography


Similar publications



Contact ORBi