VLSI structure for static image processing with pulse-coupled neural network

An electronic hardware model of a pulse-coupled neural network (PCNN) and an effective image processing technique using the PCNN are proposed in this paper. The behavior of the presented neuron circuit mimics actual biological neuron in such a way that a frequency-modulated pulse train is generated and its pulsing frequency depends upon the magnitude of an input excitation. The presented electronic neuron model finds a number of useful applications in the field of image processing. Among several features of the PCNN, an effective filtering technique is particularly exploited and its performance is compared with traditional filtering methods in this paper. Furthermore, an innovative way to determine one of the image quality elements, "image sharpness", is proposed using the PCNN. Particularly, a neuron coupling behavior which is one of the main characteristics of the PCNN is analyzed and shown to demonstrate the determination of "image sharpness" for a given input image.