Primal access recognition of visual objects

We present PARVO, a computer vision system which addresses the problem of fast and generic recognition of unexpected 3D objects from single 2D views. After more than twenty years, the field of computer vision has still not produced any clear understanding of how this complex high-level visual capability is even possible. On the other hand, the human visual system is an existence proof that such a competence is attainable, as demonstrated informally by everyone's daily experience, and more formally, by the results of various psychological studies. Recently RBC, a new human image understanding theory, has been proposed on the basis of some of these psychological results. However, no systematic computational evaluation of its many aspects has yet been reported. Such an evaluation is essential if the theory is ever to play a role in the progress of the computer vision field. The PARVO system discussed in this thesis is a first step towards this goal since its design respects and makes explicit the main assumptions of the proposed theory. It analyses single-view 2D line drawings of 3D objects typical of the ones used in human image understanding studies. It is designed to handle partially occluded objects of different shape and dimension in various spatial orientations and locations in the image plane. The system is shown to successfully compute generic descriptions and then recognize many common man-made objects.