COVID-19 Detection on Chest x-ray Images by Combining Histogram-oriented Gradient and Convolutional Neural Network Features

Authors

Reda Elbarougy, Department of Information Technology, Faculty of Computer and Artificial Intelligence, Damietta University, New Damietta, EgyptFollow
Esmail Aboghrara, Department of Computer Science, Faculty of Sciences, Sebhau University, Sebhau, LibyaFollow
G. M. Behery, Department of Information Technology, Faculty of Computer and Artificial Intelligence, Damietta University, New Damietta, EgyptFollow
Y. M. Younes, Department of Mathematics, Faculty of Sciences, Damietta University, New Damietta, EgyptFollow
Noha M. El-Badry, Department of Mathematics, Faculty of Sciences, Damietta University, New Damietta, EgyptFollow

Abstract

The COVID-19 coronavirus epidemic has spread rapidly worldwide after a person became infected with a severe health problem. The World Health Organization has declared the coronavirus a global threat (WHO). Early detection of COVID 19, particularly in cases with no apparent symptoms, may reduce the patients mortality rate. COVID 19 detection using machine learning techniques will aid healthcare systems around the world in recovering patients more rapidly. This disease is diagnosed using x-ray images of the chest; therefore, this study proposed a machine vision method for detecting COVID-19 in x-ray images of the chest. The histogram-oriented gradient (HOG) and convolutional neural network (CNN) features extracted from x-ray images were fused and classified using support vector machine (SVM) and softmax. The proposed feature fusion technique (99.36 percent) outperformed individual feature extraction methods such as HOG (87.34 percent) and CNN (93.64 percent).

Recommended Citation

Elbarougy, Reda; Aboghrara, Esmail; M. Behery, G.; M. Younes, Y.; and M. El-Badry, Noha (2023) "COVID-19 Detection on Chest x-ray Images by Combining Histogram-oriented Gradient and Convolutional Neural Network Features," Information Sciences Letters: Vol. 12 : Iss. 5 , PP -.
Available at: https://digitalcommons.aaru.edu.jo/isl/vol12/iss5/56