Low-quality Images Research Articles

Automatic crack tip localization in enormous DIC images to in-situ characterize high-temperature fatigue crack growth behavior

Digital Image Correlation (DIC), with its advantages of full-field, non-contact and applicability to extreme environments, is a powerful deformation measurement technique to in-situ monitor the high-temperature fatigue crack propagation behavior. For the DIC-based fracture mechanics analyses of cracking, crack tip position in DIC images, which is usually determined visually and manually, plays a significant role. However, the long-term fatigue-DIC experiment at both room-and high-temperatures poses a real challenge due to the huge amount of DIC images, lower image quality at high-temperature, and even the invisibility of cracks. In this paper, a fully automated algorithm for localizing the crack tip is proposed based on the crack tip displacement field and the immune algorithm. Compared to traditional algorithms, it does not necessitate iterative initial values of crack tip location, and permits precise and fast determination of critical fracture mechanics parameters, such as the Stress Intensity Factors (SIFs). The algorithm reliability is validated by both numerical simulation and fatigue testing of nickel-based superalloy GH4169 at room temperature and 650℃. It is shown that crack tip position and SIF values can be precisely and automatically determined within 7 s per image using an ordinary low-cost computer, which significantly improves the analyzing efficiency.

Deep convolutional neural network for weld defect classification in radiographic images

The quality of welds is critical to the safety of structures in construction, so early detection of irregularities is crucial. Advances in machine vision inspection technologies, such as deep learning models, have improved the detection of weld defects. This paper presents a new CNN model based on ResNet50 to classify four types of weld defects in radiographic images: crack, pore, non-penetration, and no defect. Stratified cross-validation, data augmentation, and regularization were used to improve generalization and avoid over-fitting. The model was tested on three datasets, RIAWELC, GDXray, and a private dataset of low image quality, obtaining an accuracy of 98.75 %, 90.255 %, and 75.83 %, respectively. The model proposed in this paper achieves high accuracies on different datasets and constitutes a valuable tool to improve the efficiency and effectiveness of quality control processes in the welding industry. Moreover, experimental tests show that the proposed approach performs well on even low-resolution images.

Identification of Rare Wildlife in the Field Environment Based on the Improved YOLOv5 Model

Research on wildlife monitoring methods is a crucial tool for the conservation of rare wildlife in China. However, the fact that rare wildlife monitoring images in field scenes are easily affected by complex scene information, poorly illuminated, obscured, and blurred limits their use. This often results in unstable recognition and low accuracy levels. To address this issue, this paper proposes a novel wildlife identification model for rare animals in Giant Panda National Park (GPNP). We redesigned the C3 module of YOLOv5 using NAMAttention and the MemoryEfficientMish activation function to decrease the weight of field scene features. Additionally, we integrated the WIoU boundary loss function to mitigate the influence of low-quality images during training, resulting in the development of the NMW-YOLOv5 model. Our model achieved 97.3% for mAP50 and 83.3% for mAP50:95 in the LoTE-Animal dataset. When comparing the model with some classical YOLO models for the purpose of conducting comparison experiments, it surpasses the current best-performing model by 1.6% for mAP50:95, showcasing a high level of recognition accuracy. In the generalization ability test, the model has a low error rate for most rare wildlife species and is generally able to identify wildlife in the wild environment of the GPNP with greater accuracy. It has been demonstrated that NMW-YOLOv5 significantly enhances wildlife recognition accuracy in field environments by eliminating irrelevant features and extracting deep, effective features. Furthermore, it exhibits strong detection and recognition capabilities for rare wildlife in GPNP field environments. This could offer a new and effective tool for rare wildlife monitoring in GPNP.

Development of learning-based online C-arm CT imaging technique for image-guided adaptive brachytherapy

Existing protocols for image-guided adaptive brachytherapy (BT) pose distinct limitations, which include: (1) the possible misalignment of the application position as a consequence of patient translocation between the treatment room and the computed tomography (CT)/magnetic resonance imaging (MRI) suite, (2) an inherent inability to continually monitor physiological and anatomical changes during the course of 3–10 fraction treatment regimens, and (3) insufficient tracing of the radioactive source movement within the patient's body. In response to these challenges, this study proposes an innovative deep-learning-based C-arm CT/SPECT imaging system tailored specifically for image-guided adaptive BT. This technological advancement aims to enable the execution of high-precision online adaptive BT, thereby amplifying the adaptability and effectiveness of oncological care. An image dataset of 66 non-contrast pelvic CT studies was acquired and limited-angle cone-beam CT (CBCT) images were generated by mathematically calculating sinograms for voxel phantoms based on these CT images. Using a Generative Adversarial Network (GAN), low-quality CBCT images obtained using a 110° limited-angle rotating C-arm system were transformed into high-quality diagnostic CT images. The performances of these networks were evaluated by comparing them with CBCT images reconstructed using an iterative reconstruction method and ground-truth images. Synthetic CT images were successfully transformed from low-quality CBCT images, considerably reducing streaking artifacts and preserving the anatomical structures. Our deep-learning-based image reconstruction technique could successfully improve the image quality of limited-angle CBCT, similar to the ground-truth image quality, with a faster speed than the latest iterative reconstruction algorithm. We expect that this will be effectively implemented in online image-guided adaptive BT and patient dose verification.

Low-quality image object detection based on reinforcement learning adaptive enhancement

The task of object detection for low-quality images in multiple scenarios remains a challenging task. To alleviate this problem, a novel approach to improve the detection performance of the model on low-quality images using deep reinforcement learning is proposed. The approach differs from methods that involve data labeling and retraining, as the detection results of the model are utilized to formulate a reward function. This enables deep reinforcement learning to transform low-quality images into high-quality images through image enhancement, ultimately leading to an improvement in the model's recognition performance on low-quality images. An image enhancement tool chain (IETC) has been prepared, offering a flexible and lightweight algorithm selection for the Dueling Deep Q-Network-based Tool Selector (DDQN-TS). By adding a “Pass” option, Enable DDQN-TS to learn the relative relationship between performing and not performing image enhancement. To mitigate the potential inconsistency between the output of the detection model and the actual situation caused by negative samples, which can lead to incorrect decisions made by DDQN-TS, the introduction of a “thresh” parameter regulates the level of intervention in target detection by DIE. The experimental results demonstrate the effectiveness and portability of the DIE model in environments with noise, fog, and uneven brightness.

Multitask Adversarial Networks Based on Extensive Nonlinear Spiking Neuron Models.

Deep learning technology has been successfully used in Chest X-ray (CXR) images of COVID-19 patients. However, due to the characteristics of COVID-19 pneumonia and X-ray imaging, the deep learning methods still face many challenges, such as lower imaging quality, fewer training samples, complex radiological features and irregular shapes. To address these challenges, this study first introduces an extensive NSNP-like neuron model, and then proposes a multitask adversarial network architecture based on ENSNP-like neurons for chest X-ray images of COVID-19, called MAE-Net. The MAE-Net serves two tasks: (i) converting low-quality CXR images to high-quality images; (ii) classifying CXR images of COVID-19. The adversarial architecture of MAE-Net uses two generators and two discriminators, and two new loss functions have been introduced to guide the optimization of the network. The MAE-Net is tested on four benchmark COVID-19 CXR image datasets and compared them with eight deep learning models. The experimental results show that the proposed MAE-Net can enhance the conversion quality and the accuracy of image classification results.

Clinical feasibility of post-contrast accelerated 3D T1-Sampling Perfection with Application-optimized Contrasts using different flip angle Evolutions (SPACE) with iterative denoising for intracranial enhancing lesions: a retrospective study.

Post-contrast T1-Sampling Perfection with Application-optimized Contrasts using different flip angle Evolutions (SPACE) is the preferred 3D T1 spin-echo sequence for evaluating brain metastases, regardless of the prolonged scan time. To evaluate the application of accelerated post-contrast T1-SPACE with iterative denoising (ID) for intracranial enhancing lesions in oncologic patients. For evaluation of intracranial lesions, 108 patients underwent standard and accelerated T1-SPACE during the same imaging session. Two neuroradiologists evaluated the overall image quality, artifacts, degree of enhancement, mean contrast-to-noise ratiolesion/parenchyma, and number of enhancing lesions for standard and accelerated T1-SPACE without ID. Although there was a significant difference in the overall image quality and mean contrast-to-noise ratiolesion/parenchyma between standard and accelerated T1-SPACE without ID and accelerated SPACE with and without ID, there was no significant difference between standard and accelerated T1-SPACE with ID. Accelerated T1-SPACE showed more artifacts than standard T1-SPACE; however, accelerated T1-SPACE with ID showed significantly fewer artifacts than accelerated T1-SPACE without ID. Accelerated T1-SPACE without ID showed a significantly lower number of enhancing lesions than standard- and accelerated T1-SPACE with ID; however, there was no significant difference between standard and accelerated T1-SPACE with ID, regardless of lesion size. Although accelerated T1-SPACE markedly decreased the scan time, it showed lower overall image quality and lesion detectability than the standard T1-SPACE. Application of ID to accelerated T1-SPACE resulted in comparable overall image quality and detection of enhancing lesions in brain parenchyma as standard T1-SPACE. Accelerated T1-SPACE with ID may be a promising replacement for standard T1-SPACE.

Wasserstein-GAN-based noise reduction for preserving anatomical structures in low-dose CT

The purpose of low-dose computed tomography (LDCT) is to reduce patient's radiation dose. However, LDCT imaging often suffers from low image quality due to computed tomography (CT) noise patterns. While conventional methods such as iterative reconstruction and image processing have been studied to address this issue, their clinical application remains limited due to the degradation of critical anatomical structures. Recently, deep-learning-based noise reduction approaches have shown promising results in improving the image quality of LDCT. Particularly, generative adversarial network (GAN)-based noise reduction methods have produced outstanding results, approaching the quality of normal-dose CT (NDCT) images. In this study, we propose a hybrid framework based on Wasserstein-GAN with the non-subsampled contourlet transform for enhanced noise reduction in LDCT. The aim is to meticulously address CT noise patterns and generate CT images similar to NDCT. We implemented the proposed algorithm and conducted an experiment using clinical abdominal CT datasets to demonstrate its feasibility. According to our results, the proposed method effectively reduced noise components in LDCT images while preserving the anatomical structures of NDCT. This was confirmed through both qualitative and quantitative evaluations. The implications of our findings for clinical practice are significant, as the proposed method can improve the accuracy of LDCT and reduce radiation dose to patients, ultimately enhancing patient safety.

Hybrid regularization inspired by total variation and deep denoiser prior for image restoration

Image restoration is a fundamental problem in computer vision, with the goal of restoring high-quality images from degraded low-quality observation images. However, the ill-posedness of restoration problem often leads to artifacts in the results. It inspires us to study the combination of prior, effectively restrict the solution space and improving the quality of the restored image. In this paper, a novel hybrid regularization method for image restoration is proposed, which adopts both internal and external image priors into a unified framework. Specifically, a new hybrid regularization model is designed. The TV model and a deep image denoiser are inserted into the restoration model using the plug-and-play framework, protecting image details while implicitly using external information for deep denoising. Moreover, in order to make the proposed hybrid regularization operable, an adaptive parameter method is proposed to adaptively balance the TV model and learned model in iteration. Experiments show that the proposed method surpasses the performance of existing image restoration techniques.

Imaging Assessment of Peripapillary Vessel Diameters in Postmortem Eyes.

Proof of concept of ex vivo retinal vessel diameter measurements in human postmortem eyes. En face near-infrared (IR) images and optical coherence tomography (OCT) of the optic nerve head (ONH) were captured ex vivo with a Heidelberg Engineering Spectralis (Spectralis, version 7.0.4, Image Capture Module, version 1.2.4, Heidelberg Heidelberg, Germany) device, using a custom-made eye chamber holding and positioning the eyes during the image process. Thirty-two formaldehyde-fixated eyes of 16 patients were imaged. In the IR images, two independent graders measured retinal vessel diameters at the intersection of a drawn circle centered on the ONH with diameters of 2.0 mm and 3.4 mm, respectively. The anatomically corresponding measurements between both graders were statistically analyzed using a Wilcoxon signed-rank test. A total of 246 matched measurements of both graders were analyzed across all 32 imaged eyes. Statistically significant differences between the graders were found for arterioles at 2 mm from the ONH. The other measurements did not show statistically significant intergrader differences. The mean values for arteriole diameters were 72.2 µm at 2.0 mm and 61.5 µm at 3.4 mm for grader 1, and 66.4 µm at 2.0 mm and 63.2 µm at 3.4 mm for grader 2. The mean diameter for venules were 75.5 µm at 2.0 mm and 79.3 µm at 3.4 mm for grader 1, and 67.4 µm at 2 mm and 79.1 µm at 3.4 mm for grader 2. To the best of our knowledge, this is the first study to present IR image-based retinal vessel diameters in ex vivo postmortem eyes. Retinal IR/OCT imaging is possible, and measurements are reproducible in formaldehyde-fixated human eyes. Fixation artefacts result in lower image quality, and this can impose challenges in correctly detecting, classifying, and measuring retinal vessels.

Topology Regulated Background Extraction (TRBE) method for eye fundus images

One of the initial steps in the preprocessing of digital fundoscopy images is the identification of pixels containing relevant information. This can be achieved through different approaches, one of them is implementing background extraction, reducing the set of pixels to be analyzed later in the process. In this work, we present a background extraction method for digital fundoscopy images based on computational topology. By interpreting binarized images as cubical complexes and extracting their homological groups in 1 and 2 dimensions we identify a subset of luminescence values that can be used to binarize the original grayscale image, obtaining a mask to achieve background extraction. This method is robust to noise and suboptimal image quality, facilitating the analytical pipeline in the context of computer aided diagnosis approaches. This method facilitates the segmentation of the background of a digital fundoscopy image, which allows further methods to focus on pixels with relevant information (eye fundus). This tool is best suited to be implemented in the preprocessing stages of the analytical pipeline by computational ophthalmology specialists.•It is robust to noise and low-quality images.•Output provides an ideal scenario for down-the-line analysis by facilitating only relevant pixels in a digital fundoscopy.

Artificial-Intelligence-Enhanced Analysis of In Vivo Confocal Microscopy in Corneal Diseases: A Review.

Artificial intelligence (AI) has seen significant progress in medical diagnostics, particularly in image and video analysis. This review focuses on the application of AI in analyzing in vivo confocal microscopy (IVCM) images for corneal diseases. The cornea, as an exposed and delicate part of the body, necessitates the precise diagnoses of various conditions. Convolutional neural networks (CNNs), a key component of deep learning, are a powerful tool for image data analysis. This review highlights AI applications in diagnosing keratitis, dry eye disease, and diabetic corneal neuropathy. It discusses the potential of AI in detecting infectious agents, analyzing corneal nerve morphology, and identifying the subtle changes in nerve fiber characteristics in diabetic corneal neuropathy. However, challenges still remain, including limited datasets, overfitting, low-quality images, and unrepresentative training datasets. This review explores augmentation techniques and the importance of feature engineering to address these challenges. Despite the progress made, challenges are still present, such as the "black-box" nature of AI models and the need for explainable AI (XAI). Expanding datasets, fostering collaborative efforts, and developing user-friendly AI tools are crucial for enhancing the acceptance and integration of AI into clinical practice.

Automatic breast mass segmentation in ultrasound images with U-Net and resolution enhancement blocks

This research proposes a deep learning method for breast mass segmentation in 2D ultrasound images. The low quality of ultrasound images and indistinct boundaries of masses cause difficulties in the automatic segmentation of breast masses. To solve this problem, we introduce an autoencoder-based model that is equipped with resolution enhancement blocks. The proposed model is composed of U-Net, along with residual blocks and attention blocks. Its objective is to increase the accuracy of breast mass segmentation by enhancing the feature maps and paying more attention to the region of interest, and preventing the loss of features in the encoder part of U-Net, this helps to rebuild an image with better quality in the decoder section. The proposed model achieved an accuracy of 98.84% and a Dice score of 84.72% for image containing benign tumor and 82.14% for image containing malignant tumor. For the IoU metric, the model obtained scores of 85.95% and 82.21% for images containing benign and malignant tumors, respectively. These results showed an improvement in segmentation accuracy compared to the performance of previous models based on U-Net. This highlights the usefulness and advantage of using U-Net and resolution enhancement blocks for breast mass segmentation11The code for the implemented model can be found in the article's GitHub..

A versatile, semi-automated image analysis workflow for time-lapse camera trap image classification

Camera traps are a powerful, practical, and non-invasive method used widely to monitor animal communities and evaluate management actions. However, camera trap arrays can generate thousands to millions of images that require significant time and effort to review. Computer vision has emerged as a tool to accelerate this image review process. We propose a multi-step, semi-automated workflow which takes advantage of site-specific and generalizable models to improve detections and consists of (1) automatically identifying and removing low-quality images in parallel with classification into animals, humans, vehicles, and empty, (2) automatically cropping objects from images and classifying them (rock, bait, empty, and species), and (3) manually inspecting a subset of images. We trained and evaluated this approach using 548,627 images from 46 cameras in two regions of the Arctic: “Finnmark” (Finnmark County, Norway) and “Yamal” (Yamalo-Nenets Autonomous District, Russia). The automated steps yield image classification accuracies of 92% and 90% for the Finnmark and Yamal sets, respectively, reducing the number of images that required manual inspection to 9.2% of the Finnmark set and 3.9% of the Yamal set. The amount of time invested in developing models would be offset by the time saved from automation after 960 thousand images have been processed. Researchers can modify this multi-step process to develop their own site-specific models and meet other needs for monitoring and surveying wildlife, balancing the acceptable levels of false negatives and positives.

B-spine: Learning B-spline Curve Representation for Robust and Interpretable Spinal Curvature Estimation

Spinal curvature estimation is important to the diagnosis and treatment of the scoliosis. Existing methods face several issues such as the need of expensive annotations on the vertebral landmarks and being sensitive to the image quality. It is challenging to achieve robust estimation and obtain interpretable results, especially for low-quality images which are blurry and hazy. In this paper, we propose B-Spine, a novel deep learning pipeline to learn B-spline curve representation of the spine and estimate the Cobb angles for spinal curvature estimation from low-quality X-ray images. Given a low quality input, a novel SegRefine network which employs the unpaired image-to-image translation is proposed to generate a high quality spine mask from the initial segmentation result. Next, a novel mask-based B-spline prediction model is proposed to predict the B-spline curve for the spine centerline. Finally, the Cobb angles are estimated by a hybrid approach which combines the curve slope analysis and a curve based regression model. We conduct quantitative and qualitative comparisons with the representative and SOTA learning-based methods on the public AASCE2019 dataset and our new proposed JLU-CJUH dataset which contains more challenging low-quality images. The superior performance on both datasets shows our method can achieve both robustness and interpretability for spinal curvature estimation.

Iterative Token Evaluation and Refinement for Real-World Super-resolution

Real-world image super-resolution (RWSR) is a long-standing problem as low-quality (LQ) images often have complex and unidentified degradations. Existing methods such as Generative Adversarial Networks (GANs) or continuous diffusion models present their own issues including GANs being difficult to train while continuous diffusion models requiring numerous inference steps. In this paper, we propose an Iterative Token Evaluation and Refinement (ITER) framework for RWSR, which utilizes a discrete diffusion model operating in the discrete token representation space, i.e., indexes of features extracted from a VQGAN codebook pre-trained with high-quality (HQ) images. We show that ITER is easier to train than GANs and more efficient than continuous diffusion models. Specifically, we divide RWSR into two sub-tasks, i.e., distortion removal and texture generation. Distortion removal involves simple HQ token prediction with LQ images, while texture generation uses a discrete diffusion model to iteratively refine the distortion removal output with a token refinement network. In particular, we propose to include a token evaluation network in the discrete diffusion process. It learns to evaluate which tokens are good restorations and helps to improve the iterative refinement results. Moreover, the evaluation network can first check status of the distortion removal output and then adaptively select total refinement steps needed, thereby maintaining a good balance between distortion removal and texture generation. Extensive experimental results show that ITER is easy to train and performs well within just 8 iterative steps.

Suppressing Uncertainty in Gaze Estimation

Uncertainty in gaze estimation manifests in two aspects: 1) low-quality images caused by occlusion, blurriness, inconsistent eye movements, or even non-face images; 2) uncorrected labels resulting from the misalignment between the labeled and actual gaze points during the annotation process. Allowing these uncertainties to participate in training hinders the improvement of gaze estimation. To tackle these challenges, in this paper, we propose an effective solution, named Suppressing Uncertainty in Gaze Estimation (SUGE), which introduces a novel triplet-label consistency measurement to estimate and reduce the uncertainties. Specifically, for each training sample, we propose to estimate a novel ``neighboring label'' calculated by a linearly weighted projection from the neighbors to capture the similarity relationship between image features and their corresponding labels, which can be incorporated with the predicted pseudo label and ground-truth label for uncertainty estimation. By modeling such triplet-label consistency, we can largely reduce the negative effects of unqualified images and wrong labels through our designed sample weighting and label correction strategies. Experimental results on the gaze estimation benchmarks indicate that our proposed SUGE achieves state-of-the-art performance.

DICAM: Deep Inception and Channel-wise Attention Modules for underwater image enhancement

In underwater environments, imaging devices suffer from water turbidity, attenuation of lights, scattering, and particles, leading to low quality, poor contrast, and biased color images. This has led to great challenges for underwater condition monitoring and inspection using conventional vision techniques. In recent years, underwater image enhancement has attracted increasing attention due to its critical role in improving the performance of current computer vision tasks in underwater object detection and segmentation. As existing methods, built mainly from natural scenes, have performance limitations in improving the color richness and distributions we propose a novel deep learning-based approach namely Deep Inception and Channel-wise Attention Modules (DICAM) to enhance the quality, contrast, and color cast of the hazy underwater images. The proposed DICAM model enhances the quality of underwater images, considering both the proportional degradations and non-uniform color cast. Extensive experiments on two publicly available underwater image enhancement datasets have verified the superiority of our proposed model compared with several state-of-the-art conventional and deep learning-based methods in terms of full-reference and reference-free image quality assessment metrics. The source code of our DICAM model is available at https://github.com/hfarhaditolie/DICAM.

FaceCoresetNet: Differentiable Coresets for Face Set Recognition

In set-based face recognition, we aim to compute the most discriminative descriptor from an unbounded set of images and videos showing a single person. A discriminative descriptor balances two policies when aggregating information from a given set. The first is a quality-based policy: emphasizing high-quality and down-weighting low-quality images. The second is a diversity-based policy: emphasizing unique images in the set and down-weighting multiple occurrences of similar images as found in video clips which can overwhelm the set representation. This work frames face-set representation as a differentiable coreset selection problem. Our model learns how to select a small coreset of the input set that balances quality and diversity policies using a learned metric parameterized by the face quality, optimized end-to-end. The selection process is a differentiable farthest-point sampling (FPS) realized by approximating the non-differentiable Argmax operation with differentiable sampling from the Gumbel-Softmax distribution of distances. The small coreset is later used as queries in a self and cross-attention architecture to enrich the descriptor with information from the whole set. Our model is order-invariant and linear in the input set size. We set a new SOTA to set face verification on the IJB-B and IJB-C datasets. Our code is publicly available at https://github.com/ligaripash/FaceCoresetNet.

A multipath scheduler based on cross-layer information for low-delay applications in 5G edge networks

Virtual Reality (VR) applications that require extremely low delay and high image quality are widely used in online games and other 5G scenarios, becoming a hot research field in recent years. However, the limited bandwidth in 5G edge networks fails to meet the peak rate requirements for multiple VR flows. MPTCP is suitable for 5G edge networks, supporting the simultaneous use of multiple networks on mobile devices. Nevertheless, accurately scheduling VR data blocks to different subflows to satisfy their low delay requirements is challenging due to their microburst characteristic. In this paper, we propose a novel MPTCP scheduler for cloud VR applications in 5G edge networks, called the Cross-layer information-based one-way delay Predictive Scheduler (CPS). CPS accurately predicts one-way delay by incorporating cross-layer information from both the application and edge wireless sides, and adaptively schedules VR data blocks to the optimal subflow. Experimental results show that CPS outperforms existing strategies, supporting 125% more users for VR applications in the typical scenario. CPS successfully meets the quality of experience needs of more users, providing a promising solution for large-scale deployment of cloud VR services in 5G edge networks.