Volume 15, Number 5/6

The X-ray Euclidean Synthetic Image

  Authors

Halah Ahmad AbdAlmeneem, Jazan University, Saudi Arabia

  Abstract

One of the most popular medical diagnostic tools ever is X-ray imaging, and that's saying something considering how far biomedicine has come.The amount of information that may be retrieved from images is significantly impacted by the characteristics that are already there.We propose to apply the Euclidean distance transform technique for image preprocessing to take the advantage of the image feature technology that makes it easier to identify pneumonia cases.The paper concurrently applya number of image preprocessing techniques, such as binarization, thresholding, scaling, normalization, and others, to the sampled image before the features are obtained. Then we extract the characteristics for image classification and recognition.The suggestion is verified and examined using the publicly accessible COVID-19, Pneumonia, and Normal image datasets.

The objective is to acquire the maximum level of identification that is practically possible.This would create an accessible and affordable diagnostic and illness management tool that medical professionals might utilize in the event of the COVID-19 pandemic and other future events.

The practical results showed that, given any query image of the three cases of X-ray images; Covid-19 case, Normal case and X-ray image of Pneumonia case the high precision is evident when correlating with the normal cases, which it negates the infection.

  Keywords

Image recognition, Deep Learning, CNN, MobileNetV2, DenseNet121, Binary Classification, Strawberry Images