A biologically inspired system for the detection of partially occluded objects

Thls paper briefly describes the recent development of a robotic vision system, which is based in part on biologically inspired elements and in particular on a saccading space-variant pseudo-sensor. The current system was developed for the detection of Partially Occluded Objects (POOs) in indoors' environments. It is based on a mixed color-shape local featuring approach. The sensed features are fIltered and organized in unique starlike graphs termed Local Panorama Features (LPFs). In addition to the system brief description a duality between POQs' detection and object discrimination is used to strengthen the philosophy that quality comes before quantity in evaluating the detection of POOs by LPFs.