Recognition of object position in two-D space, rotation and size by spreading associative neural networks
Keisuke Nakamura, T. Yoshikawa, L. Kinoshita, S. Miyamoto · 2003
Paying attention to the spatial recognition system in the brain, we proposed a "two-D spreading associative neural network (two-D SAN net)" and an "orientation and size spreading associative neural network (OS-SAN net)". The two-D SAN net not only recognizes the shape of the object (e.g. Arabic numerals and human faces) in the two dimensional space irrespective of its position, but also recognizes its position irrespective of its shape in the input pattern. On the other hand, the OS-SAN net simultaneously recognizes the size of the object irrespective of its orientation and shape, the orientation irrespective of its size and shape, and the shape irrespective of its size and orientation. The nonlinear transformation using spatial spreading in both SAN nets are crucial for the simultaneous recognition of both the shape and the spatial factors (position, orientation and size) of an object.