DEEP LEARNING-BASED SUPER-RESOLUTION TECHNIQUES: A COMPARATIVE ANALYSIS WITH RECENT INSIGHTS
In computer vision, Super-Resolution (SR) is one of the most effective techniques for producing a high-definition visual from several disgraced Low-Resolution (LR) images. Current researchers have employed machine-learning techniques, neural networks, and Deep Learning based methods to enhance the q...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
University of Kragujevac
2025-03-01
|
| Series: | Proceedings on Engineering Sciences |
| Subjects: | |
| Online Access: | https://pesjournal.net/journal/v7-n1/55.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | In computer vision, Super-Resolution (SR) is one of the most effective techniques for producing a high-definition visual from several disgraced Low-Resolution (LR) images. Current researchers have employed machine-learning techniques, neural networks, and Deep Learning based methods to enhance the quality of LR images obtained from different field applications. However, DL-based super-resolution methods have outperformed the competition due to their efficient mapping and feature extraction abilities. This study strives to provide details of significant progress in DL-based SR approaches based on single and multiple input frames. The primary objective of this paper is to present the hierarchical technological advancements in the field of single and multiple SR along with a systematic characterization and review of various DL-based SR approaches. Afterwards, this paper discussed a group of standard frameworks and benchmark datasets with relevant features, performance evaluation metrics, and real-world applications to facilitate improved user experiences, analysis, and decision-making and open issues concerning single and multiple SR. This comprehensive review provides deep insights into SR techniques while emphasizing gaps, opportunities, and challenges in the application domain. Furthermore, this paper presents future directions for future researchers to follow SR trends with potential DL-based algorithms. |
|---|---|
| ISSN: | 2620-2832 2683-4111 |