An Effective Approach for Image Denoising Using Wavelet Transform Involving Deep Learning Techniques
Abhay Shukla, K.S. Seethalakshmi, P Hema, Jitendra Chandrakant Musale · 2023
In contemporary years, image denoising has developed as an important investigation topic in the arena of image processing due to the increasing demand for high-quality images. Deep learning approaches have recently demonstrated outstanding results in a variety of image processing applications. In this research, a wavelet transform and deep learning-based method is suggested for image denoising. To successfully remove noise from an image, the suggested method is based on a hybrid network that combines wavelet transform with convolutional neural networks (CNNs). The suggested method is divided into two steps, one based on wavelet transforms and the other on deep learning. The input image is divided into wavelet coefficients at various scales and orientations in the wavelet transform-based stage. The noise is subsequently filtered out of the wavelet coefficients using a thresholding method. A CNN is taught to acquire the plotting among the noisy image and the denoised image during the deep learning-based step. The experimental results show that, in terms of both quantitative metrics and visual quality of the denoised images, the suggested strategy outperforms a number of state-of-the-art image denoising methods. The suggested method is also independent of the prior knowledge of the noise statistics and resilient to various types of noise. Overall, the suggested method successfully tackles image denoising by fusing the compensations of deep learning and wavelet transform. The suggested method could be used for a variety of computer vision applications that call for high-quality image denoising.