ViT-Based Denoising For Enhancing Image Clarity
Bharatula Chyavan, Tirupathi Karthik, R. Sathya Bama Krishna · IOSR Journal of Computer Engineering · 2025
This venture is dedicated to raise the quality of the image by bringing the Vision Transformer (ViT) models for the denoising of images into the scene, a challenge in imaging processing that affects applications like medical imaging and surveillance systems. The proposed approach is to enhance the ability of the model to distinguish between noise and signal and thus to improve the quality of the image by optimizing attention mechanisms within ViTs, which capture spatial dependencies across image patches. Furthermore, the concepts of new architectures to cope with noise reduction are developed by merging principles from classic image processing and the newest deep learning techniques. A variety of experiments using benchmark datasets display the top performance of this method, which hence demonstrates the viability of introducing ViT-based denoising with optimized attention mechanisms into the list of image enhancement.