Date of Award
Doctor of Philosophy
Under the petrographic microscope, most rocks in thin section appear as an assemblage of tightly interlocked mineral grains, inclusions, port spaces etc. The geometrical characteristics of these features, including their apparent sizes, shapes, orientations and distributions, define the texture of a rock.;In this study the optical image from a petrographic microscope is processed by a micro-computer. The image is converted into electrical signals by a Sony AVC-D5 monochrome video camera. These signals are digitized by an analog to digital converter in an Imaging Technology PCVISION plus frame grabber board which stores and manipulates the resulting digital image in its frame memory. The image is stored in 256-colour PCX format.;The extraction of geological information from the digital image requires that the features in the image be identified and their edges defined. Feature identification is accomplished by manipulation of the digital image which is referred to as image processing. This involves three sequential operations: digital filtering, image segmentation and feature extraction. In the present study fourteen digital filters are evaluated for their abilities to reduce normally distributed additive noise while preserving linear features and image texture. The Sigma Filter is shown to be most suitable for application to petrographic images. The edges of the features of interest are extracted using zero-crossing edge finders with varying window sizes. The sequential capturing of multiple images from one microscope field of view allows a thin section to be analyzed in a manner analogous to the procedure followed in manual petrography. Interactive manipulation of the image containing the detected edges is possible using an image editor.;The feature extraction process identifies and selects features of interest from the detected edges of the petrographic image. This information, the original prey level image and the intermediate segmented edge image, are all used to provide data which are not available from traditional petrography. Salient features of the image processing system developed here are illustrated by application to selected geological problems for which data obtained by conventional techniques are available.;Image processing can provide an initial step in expert systems developed to solve specific petrographic problems. This could allow the processing to be automated using knowledge banks interactively at each stage of image analysis.
Samantaray, Abani Kanta, "Computer-assisted Petrographic Image Analysis And Quantization Of Rock Texture" (1993). Digitized Theses. 2317.