Design Of A High Order Neural Network Motion Sensor
Luis R. Lopez · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 1989
Designs that combine simple high order neural circuits, Feed Back Shunting (FBS)1 and Feed Forward Shunting (FFS)2 are presented. A cascaded FFS-FBS design approach results in two new networks. These networks are capable of extracting edge features from moving images in noisy environments. The use of FFS to detect motion gives these networks robustness in the presence of noise. FBS gives them the ability to extract edge features.