Analog VLSI and neural systems
Analog VLSI and neural systems
Analog implementation of shunting neural networks
Advances in neural information processing systems 1
Neural Networks - Special issue: automatic target recognition
Sensory Neural Networks: Lateral Inhibition
Sensory Neural Networks: Lateral Inhibition
Hi-index | 0.00 |
We designed circuits that have potential use as preprocessors of noisy image data, which vary in intensity over as much as four orders of magnitude. The circuits are based on equations representing a stage of the Boundary Contour System/Feature Contour System model. We compared the model performance (equations) using Gaussian and exponential filters with the simulated circuit performance. Our circuits achieved compression of up to four orders of input range, while maintaining contrast ratios in local regions. Edge enhancement is maintained down to a SNR of -20 dB, and at that noise level the circuit's RMS error performance is comparable to that of the mathematical formulation. For quantitative display purposes, we present responses to test stimuli from a linear array of circuits. A 15 x 15 array design is also presented.