Abstract

We present and compare two new techniques for Learning Relational Structures (RS) as they occur in 2D pattern and 3D object recognition. These techniques, Evidence-Based Networks (EBS-NNet) and Rulegraphs (RG) combine techniques from Computer Vision with those from Machine Learning and Graph Matching. The EBS-NNet has the ability to generalize pattern rules from training instances in terms of bounds on both unary (single part) and binary (part relation) numerical features. It also learns, the compatibilities between unary and binary feature states in defining different pattern classes. Rulegraphs check this compatibilitybetween unary and binary rules bycombining Evidence Theory with Graph Theory. The two systems are tested and compared using a number of different pattern and object recognition problems. 2 1 Introduction In the context of Computer Vision, Relational Structures (RS) refer to the representation of patterns or shapes in terms of attributes of parts and part relations (s...

Citations