An invariant pulse-coder for 2-D shape recognition

We present some novel schemes for (i) pulse coding for invariant representation of shape; and (ii) a neural architecture for recognizing the encoded shape. The proposed shape encoder utilizes the properties of complex logarithmic mapping (CLM) which transforms rotation and scaling (in its domain) to shifts (in its range). In order to handle this shift, in an attempt to provide increased speed of operation, the encoder converts the CLM output to a sequence of pulses. These pulses are fed to a novel multilayered neural recognizer which (i) invokes template matching with a distinct architecture; and (ii) achieves robustness (to noise and shape deformation) by virtue of its overlapping strategy for code classification. The proposed encoder-recognizer system, which is hardware implementable by a high-speed electronic switching circuit, and can add new shapes on-line to the existing ones, is illustrated by examples.