Advanced Image Processing Techniques

Image denoising aims to recover a clean image from noisy observations while preserving structural and edge information. Classical pixel-wise losses, particularly mean squared error (MSE), are effective for noise reduction; however, their local structure may fail to capture nonlocal residual interactions and may cause…

This study proposes five lightweight image SuperResolution (SR) architectures, Restormer, SwinIR, HINet, and two hybrid ensemble models (Hybrid-1 and Hybrid-2), to restore heavily compressed JPEG images. The proposed Hybrid-2 model achieves state-of-the-art performance on the LIVE1 dataset with only 41,347 parameters,…

Bu çalışma, dijital görüntülerde sıkça karşılaşılan periyodik gürültülerin giderilmesine yönelik derin öğrenme tabanlı hibrit bir yöntem önermektedir. Periyodik gürültü, görüntü kalitesini düşürerek analiz ve yorum süreçlerini güçleştirir. Geleneksel filtreleme yöntemlerinin bu tür gürültüler üzerinde sınırlı başarı s…

In recent years, artificial intelligence-based Text-to-Image (T2I) models have demonstrated a strong capability to generate highly realistic outputs in architectural visualization. However, the extent to which these images accurately and consistently reflect site-specific spatial, cultural, and structural characterist…

No abstract yet.

No abstract yet.

Image Super-Resolution (SISR) is a vital, cost-effective solution for enhancing the spatial resolution of satellite imagery. However, a significant challenge in current SISR techniques is that while deep learning models excel at structural learning, they often struggle to accurately reconstruct intricate high-frequenc…

No abstract yet.

No abstract yet.

Low-light images often suffer from low contrast, noise pollution, and loss of detail information, significantly degrading the performance of subsequent image processing tasks. Existing Transformer-based image enhancement methods are characterized by large model parameters, high computational complexity, and poor real-…

This paper presents a hardware–software co-design framework for real-time super-resolution (SR) of low-quality video on resource-constrained edge platforms. At its core is a compact residual network obtained by once-for-all (OFA) neural architecture search over the Residual Channel Attention Network (RCAN) design spac…

In this study, three global tone mapping operators are built from classical polynomial families to enhance low contrast documentation images, their behaviour fixed in advance by symmetry rather than chosen by taste. Lucas, Faber and Gegenbauer polynomials are generated by a recurrence, a conformal map and an orthogona…