Comparative Study on Different PCNN Models in Plant Leaf Classification

Plant classification is an important area of botany research. Plant classification based on image processing is also currently one of the hot spots in the research. Pulse coupled neural network(PCNN) has good performance in image processing, especially in the feature extraction. There are a lot of improved models. In this paper, three kinds of important improved PCNN models such as pulse-coupled neural network and intersecting cortical model, spiking cortical model and two-output pulse coupled neural network are applied to plant leaf classification based on image processing so that the network model suitable for leaf classification is selected. We have improved the extraction method of entropy sequence features, and then conducted a comparative experiment. Experimental results show that each model has its own characteristics. However, due to its complete functions, the standard pulse-coupled neural network has slightly better feature extraction ability, followed by intersecting cortical model, spiking cortical model and dual-output pulsed coupled neural network.