Uniform Approach to Concept Interpretation, Active Contour Methods and Case-Based Reasoning

摘要

Active contours are methods for data analysis originally developed for image segmentation. They can be treated as contextual classifiers that use expert knowledge and operate in supervised or unsu-pervised mode. Recently there have been developed many generalizations and extensions of those methods. One of them, proposed by the authors of this paper, reveals that they can be interpreted as methods capable of identification of more complicated structures (concepts) basing on simpler ones. In the present paper, a simple model for concept identification is presented and elucidated both in terms of active contour methods and case-based reasoning approach. The application area is any kind of data (e.g. medical images or image sequences, or even the web source data [4]) assuming they fulfill weak formal requirements.