Research and Application of PCNN in Underground Infrared Image Segmentation

Underground infrared image segmentation has practical significance for underground auxiliary operations, intelligent mining, and intelligent excavation. Due to the limitations of infrared imaging machines, underground infrared image segmentation still poses challenges. Pulse coupled neural network is the third-generation artificial neural network, which has the characteristics of a single-layer network structure, no need for any training, and synchronous pulse discharge. It is widely used in the field of image processing and has derived various variants of models. In order to analyze and compare the segmentation performance of different pulse coupled neural network models on underground infrared images, this paper uses full pulse coupled neural network, simple pulse network model, intersecting cortical model, spiking cortical model, and simplified pulse coupled neural network model to segment infrared images collected underground. The results show that the simple pulse network model has the best segmentation performance, The segmentation performance of standard pulse coupled neural networks is poor, while the segmentation performance of other models is similar.