Automatic generation of significant and local feature groups of complex and deformed objects

Automatic generation of significant feature groups out of a given set of basic features, i.e., creation of abstract characteristics, is a fundamental problem to be solved in pattern recognition. With the presented method it is possible to detect automatically local and significant feature groups of 2D objects resulting in meaningful class memberships. In object recognition, contours and colors are of great importance. Local contour parts and color combinations need to be found that safely assign the objects to their classes. Here, discrete object contours are analyzed. Significant contour parts, i.e., feature groups, that are very different or closely resemble each other, are detected. Also, global similarity based on local similarities is derived and the quality of the obtained significant contour parts is assessed with standard cluster analysis methods. The method was designed for the development of general classifiers, relying on automatically generated, local and significant feature groups, that were derived from basic features