Early Cognitive Vision

In this monograph we present an Early Cognitive
Vision framework that aims at providing a rich and
reliable scene representation from visual
information. This framework preserves conflicting
hypothesis in its early stages, and makes use of
feedback mechanisms between different visual
processes and layers of representation to achieve
disambiguation. The first part presents how
symbolic local image descriptors are extracted from
the responses of early vision filters, and how
perceptual grouping constraints can be applied to the
resulting image representation. A second part
discusses the use of stereopsis to reconstruct an
equivalent 3D representation of visual information.
Interactions between perceptual grouping, stereopsis
and 3D reconstruction processes are discussed. The
third part integrates visual information across time
to further disambiguate the representation. This
framework has been used successfully in several
contexts that are discussed in the conclusion of this
book.