A Neural Network for Segmentation of Line Drawings into Lines of Different Orientations

A neural network is described which is designed for segmentation of binary line drawings into separate line segments according to their orientations. The network consists of several neural layers; each layer serves for extracting line segments of a certain orientation. Every neural layer has one-to-one correspondence with a retina. The task is performed by means of the interative procedure which includes interactions of neurons inside each layer through oriented excitatory connections and inhibitory interrelations between the corresponding neurons of different layers. The network is intended to produce orientation features as a result of the drawing partition. These orientation features should be used for recognition of line drawings. Computer simulation of the network was performed. Experiments showed that the network provides reasonable segmentation of separate handwritten characters