Abstract
Image restoration is a critical task in modern computer vision, aiming to reconstruct high-quality images from degraded inputs. Recent advances in deep convolutional networks (DCNs) have significantly outperformed traditional signal processing techniques by learning complex hierarchical features directly from data. This paper explores the integration of signal processing methods with deep learning frameworks for effective image restoration. We review key architectures such as U-Net, ResNet, and encoder-decoder models tailored for denoising, deblurring, and super-resolution tasks. Signal processing priors, including frequency-domain constraints and filtering, are incorporated to guide learning and enhance generalizability. Experimental comparisons highlight performance gains across benchmark datasets in terms of PSNR, SSIM, and perceptual metrics.
All articles published in the American Journal of Signal and Image Processing are licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0). This license allows others to copy, distribute, remix, adapt, and build upon the work, even commercially, as long as proper credit is given to the original author(s) and source. Authors retain full copyright of their work