Набрел случаем на такую заметку с картинкой
http://www.unige.ch/sciences/terre/geologie/w_groups/carbonate_sedimentology/roduit/roduit.html Quantification of porosity types on thin sections using neural networks
Nicolas Roduit
Supervisors: E. Davaud, B. Caline (TotalFinaElf)
The identification and the quantification of various types of porosity are significant to define the reservoir properties of carbonate rocks. The algorithm developed in this project uses functions available in the Visilog package and integrates neural network tools to identify the various pore shapes. The observer must initially indicate on the screen the pores which he regards as moldic, intergranular, intragranular etc. Once this phase of training is completed, the program automatically assigns the pore shapes encountered with a specific pore type and computes its surface.
The method developed in this work was tested on core samples provided by TotalFinaElf.
--------------------------------------------------------------------------------
Questions and commentaries
Last update: Monday 18 April 2005 Make a bookmark
Print this page