On this page
Research Article | Open Access
Volume 14 2022 | None
LUNG CT IMAGE DENOISING USING BLOCK MATCHING 3D, CNN RESIDUAL LEARNING AND BATCH NORMALIZATION
Madhura J , S. Raviraja and Ramesh Babu D R
Pages: 1143-1152
Abstract
The intent of image denoising is to extract precise and accurate information for disease diagnosis, staging and treatment. A method based on block matching (BM3D) where convolution neural network with residual learning and batch normalization is proposed in this research article. The alliance of denoising and enhancement technique aims to denoise image and simultaneously enhance the edges while preserving morphological features. The techniques preserves morphological features and enhances the edges while focusing on denoising the image. Peak Signal-to-Noise ratio (PSNR) and Root Mean Square Error (RMSE) statistical quantity measures are used for measuring the results. Best performance is seen at σ = 25 which indicates the proposed method is best effective when the noise variance is at 25. Enhancement quality indicate that the proposed method was well performed when compared to other filtering techniques.The research work demonstrate an improved denoising effect along with retaining the image features and greater PSNR values over the many wellknown existing methods
Keywords
BM3D, Denoising, Neural network, CNN, RMSE, PSNR
PDF
56
Views
35
Downloads