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