Coarse to fine difference edge detection with binary neural firing model

In this paper, we propose an analog circuit for binary neural firing model that can extract various image features. Both computational and hardware models were designed for feature extraction algorithm that explores the dependency of firing rates on the pixel intensity in alignment with inhibition and excitation principles. The circuit for translating each pixel intensity into a series of pulses is implemented using a well timed circuit consisting of a series of difference circuits, comparators for thresholding, memory circuits and resistive networks for averaging. The circuit can be configured to select a required number of inhibition and excitation pixels, and can be used to generate a range of filtered signals from different sized filter windows. The difference between the features from different sized filter windows are used to extract the fine to coarse features from the images reflected as image edges, background features, and object textures.