Performance Comparison of Image Restoration techniques using CNN and their applications
Manish Yadav, Namita Tiwari · 2021
Many Image applications depend on the image processing or image visualizations. Noisy image is a major concern due to the increase in pixels per inch(ppi), Nowadays, ppi are increasing making images clearer and pleasing, but also increasing the possibilities of noise addition and pixel corruption. However, image denoising methods requires a lot of computational effort and are degraded with noise which makes images visually unpleasing. Therefore, some denoising methods are required to lower the noise levels without losing the image features including textures, contrasts, edges and structures. As of now many methods have been proposed to decrease noise, including spatial domain, transform domain and convolutional based methods on image denoising problem. In this paper we will be focusing on the CNN based state-of-art denoiser methods and perform a comparative study on the various methods. After studying about all the popular image denoising techniques we are using it as a base network and use it for fingerprint enhancement . The efficiency and accuracy of model is it learns only dependent over the image dataset descriptions which was demonstrated by experiments ,also performed visual and metric comparisons of application using the trained model.