An Efficient Methodology for Preprocessing of COVID-19 Images Using BM3D Technique

SARS-CoV-2 virus is an infectious virus that instigated a coronavirus illness (COVID) outbreak in 2019. In the current context, datasets pertaining to study the features of the post COVID-19 symptoms require Computerized tomography (CT) scan, chest X-ray, and statistical data such as oxygen levels and pulse rate. The work focusses on identifying the dataset and cleanse the data and arrange the scattered data to a form amenable to the machine learning module. The dispensation of the images necessitates an image processing technique superseded by engaging appropriate machine learning algorithms contingent on the anticipated distinctive feature. In the same way, for the mathematical dataset, a statistical built algorithm is favored so that computational stretch can be abridged and high data estimation precision can be accomplished. DWT technique was used for noise removal of images. The work has proposed a BM3D technique for effective preprocessing of images. Performance metrics such as PSNR, SSIM, CC, MSE, RMSE, and NCC have been calculated for DWT and BM3D techniques for Covid and non-Covid classes of images and analyzed. The BM3D technique promises better performance for all the metrics.