J. Schreiter, J. Jens Döge, D. Matolin, and R. Schüffny, A. Heittmann, and U. Ramacher (Germany)
Pulse-coupled neural networks, weight adaptation, imagesegmentation, analog pcnn implementation
A cellular pulse-coupled neural network with adaptive weights for image processing is presented. The network performs feature preserving smoothing and segmentation. Segments are marked with synchronous neural activity. Weight adaptation rules to achieve the synchronization are motivated and explained. They result in a wave like activity pattern. Image processing is demonstrated by some examples. The network is currently being implemented in a mixed signal integrated circuit.
Important Links:
Go Back
