An analog CMOS pulse coupled neural network for image segmentation

A novel CMOS pulse coupled neural network (PCNN) circuit based on Integrate and Fire (IAF) model is proposed in this work for image segmentation. The network consists of IAF neurons and weight adaption circuit which represents the interaction between neurons. The IAF neurons exhibit the electrochemical dynamics of natural biological neurons. According to achieve the adaption of both weights between two neurons, the weight adaption circuit can adjust the frequency and phase of the pulse stream generated by the neurons. Then the network can implemented for image segmentation. The HSPICE simulation results show that the frequency and phase of the pulse stream generated by the neurons with similar inputs are able to be synchronized, which indicates that this network may provide substantial advantages for image segmentation.