Ontology of Optical Microscopic Imaging and Analysis

Abstract:

We are building the GeoMicrographOnt ontology to enable the annotation of optical photomicrographs and improve their sharing for petrological and structural analyses. The ontology is being built over four interconnected levels. The reality level represents the knowledge of mineralogy, petrology, and structural geology, and the methods and processes applied during optical microscopic investigations. Modeled entities at this level include rock, sample, thin section preparation procedure, mineral, fossil, matrix, shear zone, investigator, microscope, and camera. The captured micrograph level formalizes concepts such as image, pixel, cluster, shape, color, position, and patterns of pixels as information content entities (ICE), i.e., concepts that can be copied multiple times in different media (e.g., paper, database). The segmented micrograph level deals with the manual or algorithmic processing and segmentation of the image into one or more regions of interest (ROI), which like the captured image are information artifacts. The ROI, i.e., clusters of pixels that resemble a mineral megacrysts, a fossil, or a shear band, in this level are classified as candidates that are interpreted to correspond to the objects (minerals, structures) in the reality level by the domain expert. The model level, maps the shape, size, and relations (e.g., topological, distance, angular) among the parts of the segmented candidate clusters to those in real objects, and builds a validated geological model (also an ICE) that verifiably depicts the object based on geological knowledge. The GeoMicrographOnt ontology keeps track of the provenance of the image and its sub-images, the time they were taken, the investigator who captured or segmented them, and the processing protocols. It also allows annotating the attributes to the ROI (e.g., size, orientation, position in the image).