Multi-Scale Deep Neural Networks for Efficient and High-Quality Image Super-Resolution

DOI

10.2991/978-94-6463-738-0_98How to use a DOI?

Keywords

Image Super-Resolution; Deep Learning; Multi-Scale Neu- ral Networks; Convolutional Neural Networks; Generative Adversarial Networks; Feature Fusion; Adaptive Upsampling; Computational Effi- ciency; High-Quality Image Reconstruction

Abstract

Super-resolution (SR) is a crucial task in computer vision, aimed at generating high-resolution (HR) images from low-resolution (LR) inputs. Traditional techniques, such as interpolation-based and reconstruction-driven methods, often struggle to recover fine details, re- sulting in blurred and low-quality images. Recent advances in deep learn- ing, particularly convolutional neural networks (CNNs) and generative adversarial networks (GANs), have significantly enhanced SR perfor- mance. However, most existing models face a trade-off between com- putational cost and image quality, making them less viable for real-time applications.

This paper introduces a Multi-Scale Deep Neural Network (MS-DNN) designed to optimize both efficiency and image fidelity in super-resolution tasks. The model integrates multi-scale feature extraction, hierarchi- cal fusion, and adaptive upsampling to enhance fine-grained textures while preserving global structures. Extensive experiments on benchmark datasets show that MS-DNN achieves superior PSNR and SSIM scores compared to state-of-the-art approaches, with reduced inference time. These results highlight the model’s potential for real-time applications, including medical imaging, remote sensing, and high-quality video pro- cessing.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Prashant Kumar Tamrakar
AU  - Virendra Kumar Swarnkar
PY  - 2025
DA  - 2025/06/22
TI  - Multi-Scale Deep Neural Networks for Efficient and High-Quality Image Super-Resolution
BT  - Proceedings of the International Conference on Advances and Applications in Artificial Intelligence (ICAAAI 2025)
PB  - Atlantis Press
SP  - 1275
EP  - 1285
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6463-738-0_98
DO  - 10.2991/978-94-6463-738-0_98
ID  - Tamrakar2025
ER  -