Modeling of natural objects including fuzziness and application to image understanding

Presents an object recognition method which takes into account fuzziness both in the object model and the matching operation between the model and the image processing results. Objects are described by a hierarchical model. Attributes of components such as color, shape, and size are described by fuzzy sets. The importance and correlation between attributes and components are described using a fuzzy measure. Methods of determining this fuzzy measure are examined. Considering the relative importance and these correlations, objects are recognized by integrating the outcome of the matching between the results of image processing and the attributes of the model. At the experimental level, the method was applied to recognition of an abstract painting and a real photograph.<<ETX>>