Perceptrons for image recognition
Chung-Nim Lee, Seung-Cheol Goh · 1991
The authors describe perceptrons with two hidden layers for recognizing convexity, connectedness and simple connectedness of a digital image. In fact, the perceptron for connectedness is designed to recognize the connectedness of an arbitrary graph. The perceptrons have been implemented in the C language on a Solbourne 5/600 workstation in a UNIX environment. They have been tested on many input digital images in a 16*16 grid.>