THE PARTS OF PERCEPTION
ALEX P. PENTLAND Artificial Intelligence Center, SRI International CSLI, Stanford University
To support our reasoning abilities, perception must recover environmental regularities (e.g., rigidity, "objectness," axes of symmetry) for later use by cognition. Unfortunately, the representations that are currently available were originally developed for other purposes (e.g., physics, engineering) and have so far proven unsuitable for the task of perception. In answer to this problem we present a representation that has proven competent to accurately describe an extensive variety of natural forms (e.g., people, mountains, clouds, trees), as well as man-made forms, in a succinct and natural manner. The approach taken in this representational system is to describe scene structure at a scale that is similar to our naive perceptual notion of "a part," by use of descriptions that reflect a possible formative history of the object, for example, how the object might have been constructed from lumps of clay. One absolute constraint on any theory of shape representation is that it must be possible to recover accurate descriptions from image data. We therefore present several examples of recovering such a "part" description from natural imagery, and show that this recovery process is overconstrained and, therefore, potentially quite reliable. Finally, we show that by using this shape representation we can improve man-machine communication in several contexts; this provides further evidence of the "naturalness" of the representation.