It is well-known that active selection of fixation points in humans is
highly context and task dependent. It is therefore likely that succes
sful computational processes for fixation in active vision should be s
o too. We are considering active fixation in the context of recognitio
n of man-made objects characterized by their shapes. In this situation
the qualitative shape and type of observed junctions play an importan
t role. The fixations are driven by a grouping strategy, which forms s
ets of connected junctions separated from the surrounding at depth dis
continuities. We have furthermore developed a methodology for rapid ac
tive detection and classification of junctions by selection of fixatio
n points. The approach is based on direct computations from image data
and allows integration of stereo and accommodation cues with luminanc
e information. This work form a part of an effort to perform active re
cognition of generic objects, in the spirit of Malik and Biederman, bu
t on real imagery rather than on line-drawings.