Image Segmentation Method Based on Spiking Neural Network with Adaptive Synaptic Weights

As the third-generation artificial neural network, the spiking neural network is characterized by using an accurate neural model to simulate and transmit spikes in biological neurons. In this paper, we present a new image segmentation method based on spiking neural network with adaptive synaptic weights. Using the time-to-first-spike encoding strategy and the excitability of neurons in a receptive field, we sort the first spike time of neurons. If the neurons in the receptive field are more excitatory than the center neuron, the connection weights are positive numbers. Otherwise, the center neuron of the field inhibits the pixel information input, the connection weights are inhibitory. According to the weight rules of the neurons in the receptive field, we implement synaptic weight adaptation based on spike firing times. In the experiment, maximum Shannon entropy is used to evaluate image segmentation results. The segmentation results of different noise images are shown to verify the effectiveness of the proposed algorithm. Experimental results show that the algorithm has a good performance in noisy image processing.