ABSTRACT
From all that has been presented and discussed in this book, the following important characteristics and problems in shape analysis and classification should have become clear:
the inherent inter-and multidisciplinary nature of research in these areas, involving computer science, mathematics, electronic engineering, physics, artificial intelligence and psychology, among many others;
the often incomplete representations of scenes as images and objects in terms of features, implying several distinct objects to be mapped into the same representation;
the importance of pattern recognition not only for identifying specific objects, but also even in the particularly challenging task of segmenting an image into its constituent parts;
the lack of precise guidelines for choosing not only the shape measures (features)— be it for shape analysis or classification-but also the classification methods to be used in general problems;
the importance of the specific aspects of each problem and our knowledge about them for defining suitable measures and selecting effective classification methods;
the importance of 2D shapes;
the importance of properly validating not only the algorithms for image processing and shape analysis, but also the obtained classification results;
the broad application potential of shape analysis and classification in a variety of areas, ranging from biology to spatial research;
the relationship between continuous and discrete approaches.