Stereo camera modeling and image correspondence for three-dimensional machine vision

An important task in stereo machine vision is to determine the three-dimensional location of objects given different views (images) of the object scene. The actual location of a scene element can be determined from the disparity of two (each in a different image) of its depicted entities. Prior to establishing disparity, however, the correspondence problem must be solved. Identification of image elements, each in a different image, as being the stimuli generated by the same physical object constitutes the solution to the correspondence problem.

Particular emphasis is placed on stereo camera modeling since it greatly influences the outcome of image matching. It is shown that for the motion-in-depth stereo camera model the probability of determining unambiguous correspondence assignments is significantly greater than that for other stereo camera models. However, the mere geometry of the stereo camera system does not provide sufficient information for uniquely identifying correct correspondences. Therefore, additional constraints derived from justifiable assumptions about the scene domain and from the scene radiance model axe utilized to reduce the number of potential matches. Finally, the measure for establishing the correct correspondence is shown to be a function of the geometrical constraints, scene constraints, and the scene radiance model.