Complex Networks Applied To Image Classification Of 2019-nCoV Lung Infection

The chest X-ray image of a person infected with the novel coronavirus shows geometrical distortions,such as the formation of filamentous textures.In response to this problem,this article proposes the application of fractal dimensions to characterize the complexity of COVID-19 images.In addition,actual data often contains complex patterns other than physical characteristics.Since complex networks have the ability to capture the spatial,topological,and functional relationships between data,they are a suitable tool for characterizing data patterns.This paper also improves the classification technology based on the data-linked structure and applies it to the classification task of human chest X-ray images.The experimental results show that this method can obtain higher classification accuracy on X-ray image classification tasks;at the same time,this paper also compares this research with the existing classification techniques and obtains better results.In real-world data classification tasks,we always encounter such a situation.Under normal circumstances,the characteristics of data samples show regular data patterns,while the characteristics of abnormal data samples are highly dispersed,that is,there is no regular pattern..So far,the general data classification requires that the data features in each class show a certain degree of similarity.This reality is different from general data classification conditions.Therefore,in the real world,data classification is still a very difficult task.In response to such problems,a novel solution is proposed by using a complex network method,that is,using a training data set to construct a core/periphery network.In this network,the core node set is composed of normal data samples,while the edge node set is composed of training data.Set of abnormal sample composition.The test data samples are classified by calculating the core degree of the test data samples.The main contribution of this paper is the classification of new coronavirus images based on complex networks,and good results have been achieved.