Efficient semantically-aware annotation of images

The collection of knowledge about an imaging domain usually requires input from experts on that domain. In a two-dimensional image space this input can be provided through graphic annotation of image regions that represent objects of interest. This is a necessary process which is both time-consuming and cost inefficient, especially when sequences of images are involved. In this paper we present a novel software tool for efficient annotation of images and image sequences. This tool has a simple graphical user interface through which the average user can use alternatively two graphic annotation protocols and combine them with a versatile snake-based framework for semi-automatic image annotation to significant speedup the annotation process. Graphic annotations can be stored and retrieved directly from the web, whereas they can be associated with semantic identifiers available on any OWL ontology. The efficiency of the proposed annotation tool is demonstrated for the annotation of sequences of chest radiographs. The results obtained promise a wide applicability of the tool for a variety of intelligent imaging applications extensible to the semantic-web.