DescriptionUltra-tight source rocks such as shales and coals are extremely difficult to characterise due to their submicron pore sizes and complex compositions. This calls for multi-scale imaging and multi-scale characterisation. One of the key challenges in multi-scale characterisation is to establish a registration framework to allow images acquired at different field of views and resolutions to link up in order to re-construct models for subsequent modelling of physics. Due to constraints of imaging techniques, it is difficult, if not impossible, to establish a pixel-based registration framework on a fixed coordinate system. Hence, a registration framework that allows features of textures to be related to each other within the datasets would be advantageous. The authors have explored an approach based on advanced imaging processing and machine learning on a set of images of a tight rock sample, and obtained encouraging outcomes. We consider this approach applicable to multi-scale characterisation of ultra-tight source rocks too.
|Period||9 May 1800|
|Event title||9th International Conference on Porous Media and Annual Meeting|
|Degree of Recognition||International|
Activity: Participating in or organising an event › Participation in conference