Complex Network-Based Image Classification Method
Zhuang Ma, Guangdong Huang · 2022
Complex networks are topologically invariant. Using complex network in image recognition and classification can greatly reduce the influence of image rotation, translation and scaling on recognition accuracy. Complex networks are modeled by mathematical graph theory and consist of numerous interconnected nodes. The topology of a complex network consists of “connections” and “disconnections” between nodes. The existence of connected edges between vertices depends on the weight of the connected edges. Based on this, we process the traditional image recognition dataset and construct the complex network model of the image. Then the neural network is used to train and classify the processed images. The results show that the accuracy of image classification based on complex network is higher than that of traditional methods.