Using image analysis and ArcGIS® to improve automatic grain boundary detection and quantify geological images

Geological images, such as photos and photomicrographs of rocks, are commonly used as supportive evidence to indicate geological processes. A limiting factor to quantifying images is the digitization process; therefore, image analysis has remained largely qualitative. ArcGIS®, the most widely used Geographic Information System (GIS) available, is capable of an array of functions including building models capable of digitizing images. We expanded upon a previously designed model built using Arc ModelBuilder® to quantify photomicrographs and scanned images of thin sections.In order to enhance grain boundary detection, but limit computer processing and hard drive space, we utilized a preprocessing image analysis technique such that only a single image is used in the digitizing model. Preprocessing allows the model to accurately digitize grain boundaries with fewer images and requires less user intervention by using batch processing in image analysis software and ArcCatalog®.We present case studies for five basic textural analyses using a semi-automated digitized image and quantified in ArcMap®. Grain Size Distributions, Shape Preferred Orientations, Weak phase connections (networking), and Nearest Neighbor statistics are presented in a simplified fashion for further analyses directly obtainable from the automated digitizing method. Finally, we discuss the ramifications for incorporating this method into geological image analyses.

:► We use an integrated method to rock texture image analysis and spatial statistics. ► Semi-automatic digitalization of images uses readily available software. ► Software used includes Adobe Creative Suite® 5, Arc ModelBuilder®, and ArcMap®. ► Rock textures digitized into ArcMap® allow textural quantification and statistics. ► The method is quick, economical, user-friendly, and pertinent to many applications.