A Transformer-Guided Multi-Scale CNN Approach to Single-Channel Image Restoration
IEEE Access · 2026 · 14 · 26416-26430
DOI: 10.1109/access.2026.3665651 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Rinkal Jain und 2 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 45 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
For accurate biometric system identification, particularly palm print identification, which necessitates fine features, high-resolution images are required. Identification capacity is often compromised by picture quality reductions caused by environmental factors or low-resolution capture equipment. In this research, we present PalmNetPlus, a hybrid deep learning framework designed for grayscale palm print picture repair and improvement. The architecture combines multi-scale convolutional encoding to capture rich spatial features, a Swin Transformer bottleneck for global context modelling through self-attention, and a detail refinement module based on channel attention to preserve fine structural details. Contrary to conventional super-resolution models, PalmNetPlus emphasizes restoring clarity, contrast, and feature integrity from noisy or degraded inputs. Trained on standard palmprint datasets, the model exhibits better performance in improving image quality based on Peak Signal-to-Noise Ratio (PSNR) upto 31.89 dB and Structural Similarity Index Measure (SSIM) upto 0.9809. Experimental outcomes verify PalmNetPlus to be a solid and effective solution for palm print improvement with support for better downstream biometric recognition.
Zitieren
APA
Jain, R., Mishra, S., & Bhatt, C. (2026). A Transformer-Guided Multi-Scale CNN Approach to Single-Channel Image Restoration. IEEE Access, 14, 26416-26430. https://doi.org/10.1109/access.2026.3665651
Vancouver
Jain R, Mishra S, Bhatt C. A Transformer-Guided Multi-Scale CNN Approach to Single-Channel Image Restoration. IEEE Access. 2026;14:26416-26430. doi:10.1109/access.2026.3665651
BibTeX
@article{Jain2026,
author = {Jain, Rinkal and Mishra, Shakti and Bhatt, Chintan},
title = {A Transformer-Guided Multi-Scale CNN Approach to Single-Channel Image Restoration},
journal = {IEEE Access},
year = {2026},
volume = {14},
pages = {26416--26430},
doi = {10.1109/access.2026.3665651}
}
EUVIMED