Jul 2026· 2026 5th International Conference on Distributed Computing and Electrical Circuits and Electronics (ICDCECE)· pp. 1-6· 0 citations· 17 references
Abstract
Image quality degrades under complex imaging environments due to factors such as mixed noise and illumination degradation. This paper proposes a denoising and enhancement method based on a deep convolutional neural network. The network employs an end-to-end fully convolutional architecture, designing dual-branch multi-scale dilated convolutional modules to collaboratively capture microscopic texture and global topological information; a dual attention mechanism of concatenated spatial and channel attention is used to achieve adaptive feature calibration; dense residual skip connections are introduced to enhance the reuse of shallow geometric information; and a composite supervision function is constructed by combining smoothing L1 loss, perceptual loss, and structural similarity loss. This method achieves a peak signal-to-noise ratio of 31.72 dB, a structural similarity of 0.905, and a perceptual distance as low as 0.101. Processing time for a single $256 \times 256$ image is 21.3 ms, and the generalization performance across datasets is stable. The proposed method effectively suppresses complex noise while preserving fine structural and texture information.
Image super-resolution is an important image processing task that aims to reconstruct a high resolution image from a low resolution input. Conventional interpolation methods often produce blurred edges, loss of texture, and pixelated outputs, particularly when images contain noise, compression artefacts, or complex real world degradations. This work proposes a hybrid CNN Transformer GAN framework for natural image super resolution. The CNN component extracts local features such as edges, patterns, and fine textures; the Transformer component captures long range dependencies and preserves global image structure; and the Generative Adversarial Network (GAN) component improves perceptual quality by generating sharper and more realistic visual details. The proposed model supports 2X, 4X, and 8X upscaling, with 4X selected as the primary evaluation scale because it provides a practical balance between image quality and computational complexity. The system is trained using degraded low-resolution images produced through blur, noise, JPEG compression, and downsampling operations. Performance is evaluated using Peak Signal to Noise Ratio (PSNR), Structural Similarity Index Measure (SSIM), Learned Perceptual Image Patch Similarity (LPIPS), Mean Squared Error (MSE), and visual quality analysis. The hybrid architecture is expected to improve edge sharpness, preserve structural details, reduce visual artefacts, and generate natural looking high-resolution outputs compared with conventional CNN and GAN based approaches. The proposed framework can support applications in photo restoration, mobile imaging, surveillance, e-commerce, digital archives, and web image enhancement.
S. B· International Journal of Lat...· 0 citations
Motion-coded blurred image restoration is an important research topic in computer vision and intelligent imaging systems, directly affecting the accuracy of scene perception and information extraction. To address the limitations of existing methods in complex motion blur kernel estimation, deep network optimization, and multi-channel information fusion, an optimized restoration framework based on deep residual networks is proposed. The method incorporates a multi-scale residual learning architecture, enhanced skip-connection mechanisms, and a channel attention module to improve blur feature representation, training efficiency, and color information utilization. Experimental results demonstrate that the proposed approach achieves superior restoration performance, reaching a PSNR of 32.15 dB and an SSIM of 0.949, while significantly improving pattern recognition accuracy in high-speed production scenarios. The multi-scale framework effectively enhances adaptability to spatially varying motion blur, and the channel attention mechanism improves color fidelity and visual quality. Beyond industrial inspection applications, the proposed method is applicable to image reconstruction and information recovery tasks in intelligent sensing systems, including optical-electromagnetic imaging, remote sensing observation, and antenna-assisted imaging platforms, where motion-induced degradation can reduce the reliability of feature extraction and target interpretation. The study provides an effective engineering solution for high-quality image restoration and robust visual information recovery under dynamic imaging conditions.
R.-Q. Tian, W.-J. Sun, L. Zhou et al.· Advanced Electromagnetics· 0 citations
In recent years, vision transformers have demonstrated remarkable superiority to CNNs in single image super-resolution (SISR), yet their heavy computational and storage costs hinder practical deployment. In this work, we revisit CNN-based residual learning for SISR and identify a specific yet overlooked problem: the residual signal is often biased by mapping errors during model training, and this bias can propagate and accumulate through layers. To this end, we revisit the commonly used residual learning and feature fusion in SISR and propose an adaptive residual correction network (AdaRCN) in this work. First, we introduce a residual correction mechanism that adaptively compensates for the bias in the residual, which is utilized to ease error accumulation and improve mapping accuracy. On the other hand, we generalize the standard identity shortcut to a weighted channel concatenation followed by a 1 × 1 convolution, which is a more versatile strategy for adaptive feature fusion. Our AdaRCN is built entirely upon a naive CNN without complex architecture design and training strategies, thus ensuring efficient inference and parallelization. Extensive experiments verify the benefits and effectiveness of residual correction and adaptive feature fusion in improving the representational capability of our model, enabling it to achieve impressive performance comparable to advanced SISR models with moderate overhead.
Jinsha Tian, Junyu Guo, Bing-Bin Feng et al.· Italian National Conference...· 0 citations
A comparative study of different deep learning architectures, including classical CNNs, deep hierarchical models, residual and dense networks, and compound-scaled architectures is presented, showing that deeper networks provide better representation, while residual connections and compound scaling improve training stability and efficiency.
Riyaz Mohammed· International Journal of App...· 0 citations
Digital images acquired through imaging systems often lose their original fidelity due to various types of noise introduced during acquisition, storage, or transmission, consequently reducing clarity and adversely affecting subsequent image-processing tasks. Conventional noise-filtering approaches, such as Wiener, median, and Gaussian filtering, have been widely used because they are easy to implement and provide reasonably consistent performance. At the same time, these techniques have inflexible limits in stabilizing edges and small pixel-based details, especially when noise levels are high. In contrast, recent evolution in adaptive learning-based techniques, including convolutional based neural networks and denoising autoencoders, has employed more productive results for image restoration. This research undertakes a systematic comparison between classical filtering and modern adaptive denoising models using various levels of image datasets affected with Gaussian, salt-and-pepper, and speckle noise. Evaluation metrics, such as peak signal-to-noise ratio, structural similarity index, mean squared error, and qualitative visual observation, are used to assess performance. The findings reveal that learning-based models always exceed traditional filters, specifically in preserving structural information, texture details, and overall image fidelity. While adaptive models achieve superior restoration quality, they demand greater computational resources, which may limit their applicability in low-power or real-time settings. Overall, the research results validate that adaptive learning based techniques are well chosen for high-precision imaging results where each and every detail preservation is important. The research reinforces the significance of selecting denoising strategies corresponding to the particular requirement and constraints of the proposed application.
S. M, M. S., Mohamed Ali E. A· AIP Advances· 0 citations
Hyperspectral imagery (HSI) is highly susceptible to severe contamination by mixed noise during acquisition and transmission, including Gaussian noise, impulse noise, and complex multidirectional (e.g., oblique or curvilinear) stripes. This article proposes an optimization-inspired deep network for complex noise removal in HSI (OIC-Net) that integrates the mathematical interpretability of model-based optimization with the representational power of deep learning for HSI mixed noise removal. The proposed framework deeply unrolls an iterative physical solver into an end-to-end architecture, seamlessly unifying variational optimization with deep neural networks. In particular, OIC-Net constructs an interactively coupled, dual-branch subnetwork mechanism dedicated to abundance map restoration and stripe elimination, achieving progressive noise suppression through efficient latent variable interaction in a single forward pass. Tailored to complex real-world degradations, an omnidirectional convolutional module is designed to explicitly encode structural priors for the effective isolation of oblique and multidirectional stripes. Concurrently, the abundance restoration subnetwork employs an encoder-decoder architecture to guarantee highly accurate spatial structure reconstruction. The proposed framework possesses explicit mathematical interpretability while demonstrating superior cross-sensor generalization and computational efficiency, enabling robust noise isolation and high-fidelity detail restoration under severe real-world degradations.
Dongyi Li, Huanfeng Shen, Xiaobin Guan et al.· IEEE Transactions on Geoscie...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
