Abstract
The analysis of petrographic thin sections is a widely used process for the identification of the mineralogy of rocks and the analysis of their textural and microstructural characteristics. Despite the benefits of petrographic thin section analysis, it remains a time-consuming process with the results produced being subjective and relying on the interpretation of the researcher. In this study the benefits of integrating rotational data for petrographic thin section images in plane and cross polarised light to improve the outcomes of image analysis were investigated. This was done using image stacks representing a full 90° rotation of the thin section because the angle between the polarisation of light and the mineral orientation alters the set of observed properties for each grain. Since the range of colour values upon rotation are more indicative than a single value at one orientation these are then attributed as ranges to each pixel. Grain segmentation is performed on these images and the pixel values are extracted and attributed to their associated grain. This is done with the assistance of Micro-fabric analyser, a software developed to identify grain boundaries, and produces a data set of objects representing their associated grains along with the corresponding shape data. Using the extracted pixel values, the aim is to determine a number of optical properties such as birefringence and extinction angle (both properties that rely on orientation of the mineral grains and therefore could not be extracted without integrating rotational data). These calculated values provide more diagnostic indicators of the minerals present as well as providing microtextural data for the minerals observed. Textural data for mineral groups are extracted such as grain shape, size and orientation. To achieve this, workflows were established for the acquisition and compositing of data to optimise image analysis.