Search

Seismic data are fundamental for understanding subsurface geological structures and geological events. Previous studies have primarily focused on interpolating missing traces in seismic data, overlooking the reconstruction of complex geological events and patterns in 3D post-stack seismic images. A novel approach for the predictive reconstruction of missing geological events and patterns using our developed U-Net neural network model is proposed in this study. In contrast to existing methods that primarily interpolate traces, our proposed approach aims to predict and reconstruct complex missing geological events, including faults and unconformities, by leveraging the inherent patterns in seismic data. The impact of the different architectural choices, including various downsampling layers (max pooling, strided convolution, and dilated convolution) and loss functions such as SSIM, VGG19 perceptual loss, MSE, SNR, and PSNR, on the reconstruction process was comprehensively investigated in this study. The effectiveness of our proposed approach in generating realistic reconstructions of missing geological events and patterns was demonstrated by the experimental results. Specifically, the validation scores were found to be 0.86 for SSIM, 16.09 dB for SNR, and 16 dB for PSNR. Additionally, the validation losses were 0.0007 for VGG19 perceptual loss and 0.025 for MSE loss. These results underscore the ability of the model to capture the intricate geological and seismic features. The performance of the model in capturing the complexities of seismic data was further highlighted by a qualitative interpretation of the generated images. By addressing the limitations of existing methods and focusing on reconstructing geological events, our work advances the field of seismic data reconstruction. Our findings underscore the critical importance of a predictive reconstruction in seismic imaging and provide valuable insights for optimizing the architectural design choices. This study lays the groundwork for future research in this area, emphasizing the need for a nuanced understanding of architectural choices in seismic data reconstruction. [ABSTRACT FROM AUTHOR]

This study proposes an intelligent impulse denoising technique that utilizes sequential region expansion to address the limitations of traditional denoising methods such as mean and median filters. The proposed algorithm selectively identifies noisy pixels and adaptively expands the filtering region based on local noise density. This adaptive approach minimizes the processing of non-noisy pixels, thereby preserving important image details and maintaining visual fidelity. Extensive experiments on grayscale and color images validate the effectiveness of the proposed method. The algorithm shows improved performance in noise reduction and detail preservation compared to conventional techniques, including adaptive and weighted median filters. Quantitative evaluations using Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM) metrics demonstrate reliable performance across varying noise intensities, maintaining high image quality. [ABSTRACT FROM AUTHOR]

The demand for real-time 4K video streaming has introduced technical challenges due to the high bandwidth, low latency, and minimal jitter required for high-quality user experience. Traditional single-path networks often fail to meet these requirements, especially under network congestion and packet loss conditions, which degrade video quality and disrupt streaming stability. This study evaluates Multipath tunnel- Generic Routing Encapsulation (MPT-GRE), a technology designed to address these challenges by enabling simultaneous data transmission across multiple network paths. By aggregating bandwidth and adapting dynamically to network conditions, MPT-GRE enhances resilience, maintains quality during network disruptions, and offers throughput nearly equal to the sum of its physical paths' throughput. This feature ensures that even if one path fails, the technology seamlessly continues streaming through the remaining path, significantly reducing interruptions. We measured key video quality metrics to assess MPT-GRE's performance: Structural Similarity Index Measure (SSIM), Mean Squared Error (MSE), and Peak Signal-to-Noise Ratio (PSNR). Our results confirm that the MPT-GRE tunnel effectively improves SSIM, PSNR, and reduces MSE compared to single-path streaming, offering a more stable, high-quality viewing experience. Our results indicate that analyzing the SSIM, MSE, and PSNR values for 4K video streaming using the MPT tunnel demonstrates a significant performance improvement compared to a single path. The improvement percentages of the SSIM and PSNR values for the MPT tunnel are (29.05% and 29.04%) higher than that of the single path, while MSE is reduced by 81.17% compared to the single path. [ABSTRACT FROM AUTHOR]

Introduction: Wavelet thresholding techniques are crucial in mitigating noise in data communication and storage systems. In image processing, particularly in medical imaging like MRI, noise reduction is vital for improving visual quality and accurate analysis. While existing methods offer noise reduction, they often suffer from limitations like edge and texture loss, poor smoothness, and the need for manual parameter tuning. Methods: This study introduces a novel adaptive wavelet thresholding technique for noise reduction in brain MRI. The proposed method utilizes a linear prediction factor to adjust the threshold adaptively. This factor leverages temporal information and features from both the original and noisy images to determine a weighted threshold. This dynamic thresholding approach aims to selectively reduce or eliminate noise coefficients while preserving essential image features. Results: The proposed method was rigorously evaluated against existing state-of-the-art noise reduction techniques. Experimental results demonstrate significant improvements in key performance metrics, including mean squared error (MSE), peak signal-to-noise ratio (PSNR), and structural similarity index (SSIM). Discussion: The proposed adaptive thresholding technique effectively addresses the limitations of existing methods by providing a more efficient and accurate noise reduction approach. By dynamically adjusting the threshold based on image-specific characteristics, this method effectively preserves image details while effectively suppressing noise. These findings highlight the potential of the proposed method for enhancing the quality and interpretability of brain MRI images. [ABSTRACT FROM AUTHOR]

Purpose: The structural similarity index measure (SSIM) has become a popular quality metric to evaluate QSM in a way that is closer to human perception than RMS error (RMSE). However, SSIM may overpenalize errors in diamagnetic tissues and underpenalize them in paramagnetic tissues, resulting in biasing. In addition, extreme artifacts may compress the dynamic range, resulting in unrealistically high SSIM scores (hacking). To overcome biasing and hacking, we propose XSIM: SSIM implemented in the native QSM range, and with internal parameters optimized for QSM. Methods: We used forward simulations from a COSMOS ground‐truth brain susceptibility map included in the 2016 QSM Reconstruction Challenge to investigate the effect of QSM reconstruction errors on the SSIM, XSIM, and RMSE metrics. We also used these metrics to optimize QSM reconstructions of the in vivo challenge data set. We repeated this experiment with the QSM abdominal phantom. To validate the use of XSIM instead of SSIM for QSM quality assessment across a range of different reconstruction techniques/algorithms, we analyzed the reconstructions submitted to the 2019 QSM Reconstruction Challenge 2.0. Results: Our experiments confirmed the biasing and hacking effects on the SSIM metric applied to QSM. The XSIM metric was robust to those effects, penalizing the presence of streaking artifacts and reconstruction errors. Using XSIM to optimize QSM reconstruction regularization weights returned less overregularization than SSIM and RMSE. Conclusion: XSIM is recommended over traditional SSIM to evaluate QSM reconstructions against a known ground truth, as it avoids biasing and hacking effects and provides a larger dynamic range of scores. [ABSTRACT FROM AUTHOR]

Recently, Siamese-like trackers have performed very well. Most of the methods exploit classification scores and quality assessment scores to estimate a target. However, their classification scores have a low correlation with target locational estimation, and quality scores by a simple strategy benefit the correlation limitedly, which damages the tracking ability of a tracker. To alleviate this problem, we propose a simple Siamese target estimation with image contents (luminance, contrast, and structure) method for object tracking. Specifically, we first employ image contents involving the target to generate similarity scores by SSIM (Structure Similarity Index Measure) in the similarity branch, aiming to aid the classification branch in improving target estimation by considering the whole target context information in our model. Secondly, we give different weights of the classification branch and similarity branch during inference to ease the low correlation, which shows more flexibility for the target locational estimation. Our tracker achieves competitive performance on three challenging benchmarks like OTB100, GOT-10k, and TrackingNet over a real-time speed, proving the effectiveness of our method. Particularly, our tracker outperforms the leading baseline by over 6.0% in SR 0.75 score on GOT-10k benchmark still running at 67 FPS. [ABSTRACT FROM AUTHOR]

Underwater image enhancement and processing have gained prominence in the field of image processing due to marine scientists' interest in uncovering new species and environments. This work presented a lightweight, attention-based, deep hybrid convolutional neural network (LW-AB-DHCNN) architecture to enhance overall efficiency. The traditional CNNs use subnetworks to enhance model depth and achieve the same functionality, but the proposed method employs multiple depth-wise separable convolutions, thereby reducing the computational complexity of the system. The proposed approach integrates Deep CNN with CBAM to provide an enhanced U-Net model. CBAM employs a self-attention method to acquire both local and global data in underwater images, thereby augmenting their semantic interpretation. This work also employed a unified error function to direct the training and optimization of the model. When the presented scheme was evaluated against benchmark datasets, it achieved an average PSNR of 25.69 dB and an average SSIM of 0.8624. [ABSTRACT FROM AUTHOR]

Copyright of Baghdad Science Journal is the property of Republic of Iraq Ministry of Higher Education & Scientific Research (MOHESR) and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Hyperspectral remote sensing images obtained from cameras are characterized by high-dimensions and low quality, which makes them unfavorable for various analytics purposes. This is due to the presence of visible and invisible frequencies of the reflected light making it poorly reveal the spectral signatures of the image. Visual communication advancement has paved the need for Image Super-Resolution (SR) which recovers high-resolution images from low-resolution images. Several works were carried out earlier on image SR using variants of supervised and unsupervised models that still lack accuracy. In this paper, we propose an unsupervised learning model titled Gompertz Function–based Convergence War Accelerometric Optimization–GAN framework for generating of High-Resolution (HR) images. The framework comprises a pre-processing stage, where the incoming Low-Resolution (LR) image is preprocessed for noise removal by applying Shannon-Gaussian Filter (S-GF). Following is the Gradient Domain Approach based Tone-Mapping (TM). Skew correction is done to remove distortion and maintain original resolution that may change during TM stage. The next stage comprises the boundary and edge enhancement of the resulting preprocessed image generated by the method of Inverse Gradient Mapping (IGM) followed by patch extraction to extract minute low-frequency information from the resulting boundary and edge-enhanced image. The contrast of the enhanced patches is improved by removing blurriness effect. The preprocessed image patches are then fed into the Gompertz Function-based Convergence War Accelerometric Optimization – GAN for feature mapping on the trained SR Image features that are clustered using Krzanowski and Li- Kantorovich Metric-K-Means clustering Algorithm (KL-KM-KMA) for effective generation of SR image. The developed model is validated for both qualitative and quantitative measurements. Comparisons are made with several other state-of -the-art methods for accuracy of 98.05%, precision of 97.98%, inception score of 8.71, Fréchet Inception Distance of 36.4 with reduced clustering and training time proving the efficiency of the proposed model. [ABSTRACT FROM AUTHOR]

In this paper, a watermarking technique based on discrete curvelet transform and discrete cosine trans- form is proposed to protect the color document images. The six layers of the document image are created using the Discrete Curvelet Transform (DCuT). The sixth layer is chosen for the embedding process, while the remaining layers are discarded. Discrete Cosine Transform (DCT) is applied on the 8 × 8 blocks of the real part of sixth layer. DCT coefficients from low to medium are selected for the embedding process, excluding the first one. To choose the Direct Current (DC) coefficients of DCT to embed the watermark bits, a zigzag function is used. The same process is applied on the receiver side watermark image is extracted. Since this is a non-blind method, we require both a cover image and a watermarked image. DCT of watermarked image and the cover image are compared to extract the watermark image. The resistivity of the watermarked image against image processing attacks is measured using the Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index Measure (SSIM), Mean Square Error (MSE), Normalization Correlation (NC), Bit Error Rate (BER) and Universal Quality Index (UQI). [ABSTRACT FROM AUTHOR]

This study presents a multi-exposed fusion algorithm aimed at enhancing the quality of retinal images captured under variable illumination conditions. Retinal imaging devices frequently struggle with inconsistent lighting, which can lead to low-contrast images where critical vascular details may be lost. The proposed algorithm combines multiple exposures, preserving the best features from each - improving both clarity and detail. Using the database of 40 retinal images, the method evaluates image quality through the structural similarity index measure (SSIM). Results indicate high structural similarity between fused images and input images across different illumination levels, with SSIM values above 0.9 for medium and high exposure. Furthermore, incorporating Contrast-Limited Adaptive Histogram Equalization (CLAHE) enhances contrast, facilitating clearer vessel visualization against the background. The improved contrast and detail retention achieved by the algorithm support accurate retinal vessel analysis, which is crucial in diagnosing conditions like diabetic retinopathy and glaucoma. This approach provides a robust, enhanced imaging solution for medical diagnostics, significantly improving readability and reliability in retinal assessments. [ABSTRACT FROM AUTHOR]

Video analytics can achieve increased speed and efficiency by operating directly on the compressed video format, thereby alleviating the decoding burden on the analytics server. The encoded video streams are rich in semantic binary information and this information can be utilized more efficiently to train the classifiers. Motivated by the same notion, a deep learning-based video compression-cum-classification network has been proposed. In the proposed work, the binary-coded semantic information is extracted by using an auto encoder-based video compression component and the same fed to the MobileNetv2-based classifier for the classification of the given video streams based on their content. Using large-scale user-generated content provided by YouTube UGC dataset, it has been demonstrated that using deep neural networks for compression not only provides on-par compression results to traditional methods, it makes analytical processing of these videos faster. Video content tagging of YouTube UGC dataset has been used as the analytics task. The proposed DLVCC approach performs 10 × faster with 30 × fewer parameters than MobileNetv2 in video tagging of compressed video with no loss in accuracy. [ABSTRACT FROM AUTHOR]

With the increasing use of online meetings, there is a growing need for efficient tools that can automatically generate meeting minutes from recorded sessions. Current solutions often rely on proprietary systems, limiting adaptability and flexibility. This paper investigates whether various open-source models and methods such as audio-to-text conversion, summarization, keyword extraction, and optical character recognition (OCR) can be integrated to create a meeting minutes generation tool for recorded video presentations. For this purpose, a series of evaluations are conducted to identify suitable models. Then, the models are integrated into a system that is modular yet accurate. The utilization of an open-source approach ensures that the tool remains accessible and adaptable to the latest innovations, thereby ensuring continuous improvement over time. Furthermore, this approach also benefits organizations and individuals by providing a cost-effective and flexible alternative. This work contributes to creating a modular and easily extensible open-source framework that integrates several advanced technologies and future new models into a cohesive system. The system was evaluated on ten videos created under controlled conditions, which may not fully represent typical online presentation recordings. It showed strong performance in audio-to-text conversion with a low word-error rate. Summarization and keyword extraction were functional but showed room for improvement in terms of precision and relevance, as gathered from the users' feedback. These results confirm the system's effectiveness and efficiency in generating usable meeting minutes from recorded presentation videos, with room for improvement in future works. [ABSTRACT FROM AUTHOR]

Transmission of images over a long distance is a need of the day for different applications. As a lot of information is carried by high-resolution images, huge bandwidth is required to transmit them. So, to make efficient transmissions, compression of images is very important. Again, images involved in various applications carry confidential information and need to be highly secured before transmission. For the purpose of secured transmission, images must be encrypted with robust encryption schemes. In this communication, an effective and secured method for frequency-domain compression and transmission of infrared images is developed which is equally applicable for any other images. It is shown in this communication that the compression ratio can be improved by the proposed technique without compromising the PSNR value. In this technique, the compression ratio is improved from 2.79 to 11.13 without degradation in PSNR value of retrieved images (around 34). Our compression technique involves the modulation of images using amplitude grating. Different orientation angles and grating frequencies are selected for modulation. Security of transmitted images is ensured by 6-D hyper-chaotic encryption. This method is efficient enough to encrypt, transmit, and recover multiple infrared images without the occurrence of aliasing errors. The algorithm shows satisfactory performance for applications where huge image data are to be transmitted in a highly secure way. [ABSTRACT FROM AUTHOR]

Currently, evaluators assess the quality of content transmitted by computing devices, such as smartphones and computers, based on the success or failure of the human visual system. However, many failures go unnoticed because of the flow of frames and the large volume of information transmitted. This paper proposes automatic monitoring of the quality of multimedia content transmitted between a smartphone and a digital TV via a wireless transmission channel. The methodology combines two tools: the peak signal-to-noise ratio (PSNR) and the structural similarity index (SSIM). To receive the transmitted data, we establish intermediation through a hotspot and monitor it using the Wireshark tool. The results indicate a 99.42% accuracy rate for local media and 99.02% accuracy in processing media consumed from streaming channels. Based on test and analysis results, this research concludes that the proposed architecture allowed a better measure of imperceptibility in aspects where human vision is more sensitive, such as color changes. [ABSTRACT FROM AUTHOR]

The constant increase in multimedia Internet traffic in the form of video streaming requires new solutions for efficient video coding to save bandwidth and network resources. HTTP adaptive streaming (HAS), the most widely used solution for video streaming, allows the client to adaptively select the bitrate according to the transmission conditions. For this purpose, multiple presentations of the same video content are generated on the video server, which contains video sequences encoded at different bitrates with resolution adjustment to achieve the best Quality of Experience (QoE). This set of bitrate–resolution pairs is called a bitrate ladder. In addition to the traditional one-size-fits-all scheme for the bitrate ladder, context-aware solutions have recently been proposed that enable optimum bitrate–resolution pairs for video sequences of different complexity. However, these solutions use only spatial resolution for optimization, while the selection of the optimal combination of spatial and temporal resolution for a given bitrate has not been sufficiently investigated. This paper proposes bit-ladder optimization considering spatiotemporal features of video sequences and usage of optimal spatial and temporal resolution related to video content complexity. Optimization along two dimensions of resolution significantly increases the complexity of the problem and the approach of intensive encoding for all spatial and temporal resolutions in a wide range of bitrates, for each video sequence, is not feasible in real time. In order to reduce the level of complexity, we propose a data augmentation using a neural network (NN)-based model. To train the NN model, we used seven video sequences of different content complexity, encoded with the HEVC encoder at five different spatial resolutions (SR) up to 4K. Also, all video sequences were encoded using four frame rates up to 120 fps, presenting different temporal resolutions (TR). The Structural Similarity Index Measure (SSIM) is used as an objective video quality metric. After data augmentation, we propose NN models that estimate optimal TR and bitrate values as switching points to a higher SR. These results can be further used as input parameters for the bitrate ladder construction for video sequences of a certain complexity. [ABSTRACT FROM AUTHOR]

Facial recognition systems are pivotal in modern applications such as security, healthcare, and public services, where accurate identification is crucial. However, environmental factors, transmission errors, or deliberate obfuscations often degrade facial image quality, leading to misidentification and service disruptions. This study employs Generative Adversarial Networks (GANs) to address these challenges by reconstructing corrupted or occluded facial images with high fidelity. The proposed methodology integrates advanced GAN architectures, multi-scale feature extraction, and contextual loss functions to enhance reconstruction quality. Six experimental modifications to the GAN model were implemented, incorporating additional residual blocks, enhanced loss functions combining adversarial, perceptual, and reconstruction losses, and skip connections for improved spatial consistency. Extensive testing was conducted using Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index (SSIM) to quantify reconstruction quality, alongside face detection validation using SFace. The final model achieved an average PSNR of 26.93 and an average SSIM of 0.90, with confidence levels exceeding 0.55 in face detection tests, demonstrating its ability to preserve identity and structural integrity under challenging conditions, including occlusion and noise. The results highlight that advanced GAN-based methods effectively restore degraded facial images, ensuring accurate face detection and robust identity preservation. This research provides a significant contribution to facial image processing, offering practical solutions for applications requiring high-quality image reconstruction and reliable facial recognition.

Computerized evaluation of valve anomalies from cardiac sound is a well-tried endeavor in cardiology. Conversely, automated methods for the diagnosis of cardiovascular diseases mainly depend on the features collected from the cardiac signal. Analyzing phonocardiogram (PCG) signals can yield useful information into the mechanics of the heart. A machine learning technique for detecting and classifying murmurs is proposed, which takes into account the statistically significant features derived from comparing spectrogram images obtained by the Short-Time Fourier Transform (STFT) of the PCG signals. The spectrograms are compared by Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index Matrix (SSIM). Finally, these similarity index features are fed into various decision trees, both with and without PCA to classify like normal heart sound and murmurs like systolic, diastolic, and continuous. The SSIM and PSNR alone offer accuracy of 88.23% and 87.94%, respectively for distinguishing normal and murmur and are differ with a P-value of 2.05 × 10−19. The PCA enabled coarse tree performs better in terms of classification accuracy of 85% and 92.50% during training and testing, respectively. The results show that this method can accurately detect and classify heart murmurs, outperforming conventional methods. [ABSTRACT FROM AUTHOR]

Cryptography and steganography are employed to secure digital data transfers. We introduced an efficient region-based steganography pipeline to enhance security by concealing confidential information within an image. Our approach involves isolating the blue channel from the cover image, partitioning it into blocks, identifying smooth blocks, and embedding the message in the Least Significant Bit (LSB). Smooth blocks were determined using the Pixel Value Differencing (PVD) method, which compares a specific pixel value to the block's average pixel value (M) of the particular block. Concealed areas exhibit greater imperceptibility in smooth regions than in rough ones. We performed experiments on a carefully chosen image set and assessed the performance of the region-based steganography method using widely recognized metrics such as PSNR, MSE, and SSIM. These metrics were applied to a widely recognized benchmark dataset for comparison. Results indicate significantly improved PSNR and SSIM levels for selected images, confirming the suitability of smooth, edge-free regions for concealing hidden messages with greater imperceptibility. We compared our method with recently published steganography methods and observed a significant enhancement in its ability to conceal information effectively. [ABSTRACT FROM AUTHOR]

Copyright of Chinese Journal of Liquid Crystal & Displays is the property of Chinese Journal of Liquid Crystal & Displays and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

In current years, underwater wireless sensor networks (UWSNs) have attracted huge attention from researchers. Generally, UWSN encompasses a huge quantity of sensors, and underwater vehicles are collaboratively arranged for performing the collection of data, interpretation, and, processing. However, its difficult nature makes position updates or including new devices more challenging. Moreover, owing to the limitations of UWSN energy storage of end devices, restricted bandwidth, and its difficulty in recharging or repairing the underwater device, this is extremely important to improve the energy performance of UWSN. The power consumption imbalance can cause restricted network lifetime and less performance. In order to overcome these issues, an optimal threshold-driven image compression approach is proposed in UWSN. The run length encoding model is applied for improving the compression performance in UWSN. The performance of the designed optimal image compression approach is assessed by means of various performance measures, such as compression time (s), decompression time (s), space-saving (%), and compression ratio, Peak Signal to Noise Ratio (PSNR), Signal to Noise Ratio (SNR), Mean Square Error (MSE), Structural Similarity (SSIM), and Mean Absolute Error (MAE). Thus, the devised image compression model attained less, compression time, compression ratio, decompression time, MSE, and MAE of 5.426, 2.0294, 5.064, 0.0550, and 0.141, respectively. [ABSTRACT FROM AUTHOR]

A multi-image compression technique compresses multiple images of the same or various sizes together to generate a common codebook. In multi-image compression, the size of the common codebook or code vector matrix formed from multiple images is crucial to the algorithm's compression ratio performance. This codebook comprises codewords created by the multi-Image compression technique after various tuning settings have been modified. The compression ratio of the multi-image compression approach can be improved even more by lowering the size of the common code vector matrix. The common codebook or code vector matrix is reduced in size in this study using deep learning based auto-encoder technology. The encoded matrix is substantially smaller than the matrix formed using standard encoding techniques. For decoding purposes, information on the number of neurons and layers employed during encoding is also stored. The suggested approach is tested on a large number of standard photos and images from the UCID version 2 database. The experimental results are examined using compression ratio, PSNR, and SSIM. The results demonstrate that the suggested technique decreases the size of the common code vector matrix or codebook by 20%, improving overall algorithm performance by about 1.5% while maintaining the visual quality of the decompressed images. [ABSTRACT FROM AUTHOR]

Copyright of Journal of the College Of Basic Education is the property of Republic of Iraq Ministry of Higher Education & Scientific Research (MOHESR) and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

The accurate diagnosis of medical conditions from low-light images, particularly black-and-white x-rays, is impeded by challenges such as noise, constrained visibility, and a lack of detail. Existing enhancement methods often exacerbate these issues by introducing detail loss, color oversaturation, or higher noise levels. This paper proposes a novel U-Net-based Convolutional Neural Network (CNN) specifically developed to address these challenges in low-light black-and-white medical images. Our designed architecture employs skip connections within the U-Net framework to effectively balance noise reduction with detail information preservation. This makes it possible for the network to learn hierarchical image representations while retaining important features for diagnosis. The trained network accomplishes real-time image enhancement, enabling immediate visual improvement during diagnosis and perhaps assisting radiologists in making faster and more accurate findings. Our approach illustrates a significant improvement in image quality and outperforms traditional methods in terms of noise reduction and detail preservation. This study holds significant potential to improve medical image analysis and diagnosis, potentially leading to enhanced patient care and earlier interventions. [ABSTRACT FROM AUTHOR]

In the process of face attribute transfer, non-frontal and occluded face images often suffer from low generation quality, missing facial edges, and a lack of diversity. To address these challenges, we present the FES-Star- GANv2, an unsupervised multi-domain face attribute transfer network. In the feature extraction phase, we incorporate an attention-guided feature fusion module aimed at enhancing image details while preserving the overall integrity of the transferred images. Moreover, a style code extraction module is presented, refining the style code of the target domain, enhancing the learning capabilities of the generator. To further augment image diversity and authenticity, a face image optimization module and a structural diversity loss function are integrated. Experimental results reveal that, in comparison with the baseline StarGANv2, our approach achieves substantial improvements of 23% and 3.9% in FID and LPIPS metrics, respectively, attaining optimal 13 and 0.453. Notably, in terms of visual quality, significant enhancements were observed, particularly in addressing issues of low image quality and edge deficiencies. The FES-StarGANv2 approach effectively addresses the challenges associated with non-frontal and occluded facial images. [ABSTRACT FROM AUTHOR]