Ontology for the Object Recognition in Images

Object reconstruction from images is an actual research topic in computer vision, photogrammetry and spatial information science. Ontology can provide all necessary support we needed for its natural language appearance and flexible hierarchy structure. However, most current relative studies on field ontology are deduced from philosophical theme and not applicable for the raster expression-which is actually a kind of field-like phenomenon but does not physically coincide to the general concept of philosophical field. That's why we specifically discuss the pixel-based ontology in this paper. The discussion starts at the investigation of the physical characteristics of the pixel-based objects along with the processing of its compare to the philosophical field ontology terms. Then a unified pixel-based ontology framework for the recognition of the pixel-based objects is introduced, from which the connection of the human epistemology and the computer world is built for better knowledge discovery and sharing.