A theory of shape identification (Q931259)

The book deals with the latest achievements in the field of shape recognition, a hot topic for various current applications ranging from image retrieval to stereo vision or image mosaics. The authors have decided to focus the presentation on recent methods that, starting from a query image and an image database, return a list of the images in the database containing the query shapes. Solid shape descriptors, like scale-invariant feature transform, maximally stable extremal regions, or level line descriptor, are described and tested on several sets of images. The similarity of two shapes is treated with a statistical approach based on multiple hypothesis testing. While the book presents innovative approaches for shape identification at a high theoretical level that requires a good mathematical background, students and engineers interested in the applications of the shape identification theory will find this book very attractive due to the detailed descriptions of algorithms and experiments that are sustaining the theory.