Original Image Set Research Articles

Magnetic Flux Leakage Defect Identification Method for Small-Diameter Pipeline Elbow Based on the Improved YOLOv5

Abstract The elbow is an important constituent of oil and gas pipeline systems and plays a key role in changing the direction of pipelines. Corrosion defects pose a significant risk to the safe operation of elbows. Magnetic flux leakage (MFL) detection has been developed as a suitable technique for identifying defects in pipelines. To address the distortion of elbow defect signals in the images arising from variations in the liftoff value of the leakage detector, this paper proposed an image identification method based on an improved YOLOv5 network. The differences in defect images are simulated by analyzing the liftoff value of the magnetization unit. A defect image enhancement method of multiscale retinex with color restoration fusion homomorphic filtering (MSRCR-HF) is employed to enhance the features of defective MFL signal images. To further improve the accuracy of the model, the YOLOv5 network is optimized by integrating the convolutional block attention module (CBAM) and the space-to-depth-nonstrided convolution (SPD-Conv) module. The results show that the proposed image enhancement method effectively accentuates the features of defect images. Moreover, the suggested image identification method exhibits superior accuracy in identification. The mean average precision (mAP) values for the original image set and the enhanced image set are 85.0% and 91.4%, respectively. Consequently, the proposed method is shown to be highly viable for the automatic identification of MFL defects in small-diameter pipe elbows.

A CHERRY FRUIT RECOGNITION METHOD FOR CURRENT TASK BASED ON DEEP LEARNING

To improve the accuracy of cherry fruit recognition during picking operations and to enable automated cherry picking by robots, both original and complex sample image sets were collected at the orchard site in two phases. Concurrently, the basic data set, enhanced data set, and task data set were generated, wherein the recognition targets were all mature cherry fruits and the recognition targets were cherry fruits that became within the current operation range. First, the basic and enhanced data sets were used for comparative training on you only look once (YOLO) v3. The test results demonstrated that the ability of YOLO v3 to recognize occluded and overlapping cherry fruits can be greatly enhanced by increasing the quantity and percentage of samples in the training set that have high levels of occlusion and overlap. Second, the enhanced data set was used to train the Faster R-CNN, SSD, YOLO v3, and YOLO v5 networks. The test results indicated that YOLO v5 had the best recognition effect on mature cherries. Next, the enhanced data set and the task data set were used to train YOLO v5. Experiments revealed that YOLO v5 was more successful in identifying cherry fruits that needed to be picked within the current task range, while shielding mature cherries that should not be picked outside of it. 96% of the data was precise, and 99% of the data was recall.

Research on image acquisition and contour extraction algorithm of partial discharge in oil

Abstract Partial discharge in oil paper insulation is an important reason for insulation breakdown. The study of partial discharge in transformer oil is of great significance for ensuring the reliability of transformers. In this paper, a comprehensive observation platform of partial discharge with needle plate electrode and oil-paper insulation is built, and the original shape image set of partial discharge is collected and established. Through multi-channel convolution processing of the true color image, gray level equivalent distribution of 3D section, segmentation stretching of threshold interval and binary neighborhood search, a general technical method suitable for processing a series of low brightness discharge images with background interference is finally obtained.

Application of a light-weighted convolutional neural network for automatic recognition of coal workers' pneumoconiosis in the early stage

Objective: To construct and verify a light-weighted convolutional neural network (CNN), and explore its application value for screening the early stage (subcategory 0/1 and stage Ⅰ of pneumoconiosis) of coal workers' pneumoconiosis (CWP) from digital chest radiography (DR) . Methods: A total of 1225 DR images of coal workers who were examined at an Occupational Disease Prevention and Control Institute in Anhui Province from October 2018 to March 2021 were retrospectively collected. All DR images were collectively diagnosed by 3 radiologists with diagnostic qualifications and gave diagnostic results. There were 692 DR images with small opacity profusion 0/- or 0/0 and 533 DR images with small opacity profusion 0/1 to stage Ⅲ of pneumoconiosis. The original chest radiographs were preprocessed differently to generate four datasets, namely 16-bit grayscale original image set (Origin16), 8-bit grayscale original image set (Origin 8), 16-bit grayscale histogram equalized image set (HE16) and 8-bit grayscale histogram equalized image set (HE8). The light-weighted CNN, ShuffleNet, was applied to train the generated prediction model on the four datasets separately. The performance of the four models for pneumoconiosis prediction was evaluated on a test set containing 130 DR images using measures such as the receiver operating characteristic (ROC) curve, accuracy, sensitivity, specificity, and Youden index. The Kappa consistency test was used to compare the agreement between the model predictions and the physician diagnosed pneumoconiosis results. Results: Origin16 model achieved the highest ROC area under the curve (AUC=0.958), accuracy (92.3%), specificity (92.9%), and Youden index (0.8452) for predicting pneumoconiosis, with a sensitivity of 91.7%. And the highest consistency between identification and physician diagnosis was observed for Origin16 model (Kappa value was 0.845, 95%CI: 0.753-0.937, P<0.001). HE16 model had the highest sensitivity (98.3%) . Conclusion: The light-weighted CNN ShuffleNet model can efficiently identify the early stages of CWP, and its application in the early screening of CWP can effectively improve physicians' work efficiency.

Symbolic capital of the region in the context of the construction of an ethno-cultural brand (case of the Republic of Mordovia)

The purpose of the article is to identify the cultural and symbolic attributes that form the symbolic capital of the Republic of Mordovia. As part of the study, a representative survey of 384 residents was conducted in accordance with quotas by sex, age, place of residence, as well as a complete survey of the target group — employees of cultural institutions of the Republic of Mordovia (763 people) using an interactive questionnaire on the Google Forms platform. The study of cultural and symbolic attributes and the main components of the ethno-cultural image is due to the search for competitive advantages of the territory to determine and form the image of the region, taking into account the potential of national cultures. The results of the study made it possible to determine a set of authentic and original meanings and images that ensure fame and cause a stable positive association with Mordovia. It was revealed that natural-geographical, ethno-cultural and socio-economic components give the republic its uniqueness and recognition. The natural and geographical advantage of the territory is unique natural and recreational resources, favorable geographical position, proximity to the federal center and the presence of a network of high-quality roads of federal and republican significance. The national and cultural identity of the region acts as a resource for generating its positive reputation. The organization of interesting and non-trivial cultural events that attract local and external audiences increases the attractiveness of the republic. The socio-economic advantage of Mordovia is determined by the production of industrial products and food products that have a high reputation among consumers. For the inhabitants of the region, the places of concentration, preservation and promotion of the cultural heritage of the region, located in the regional centers, become important. The authors confirmed that the Republic of Mordovia has rich intangible assets that make it possible to construct its own unique ethno-cultural brand of the territory.

Hybrid computational pregnant female phantom construction for radiation dosimetry applications

The number of patients undergoing diagnostic radiology and radiation therapy procedures has increased drastically owing to improvements in cancer diagnosis and treatment, and consequently, patient survival. However, the risk of secondary malignancies owing to radiation exposure remains a matter of concern. We previously published three hybrid computational fetal phantoms, which contained 27 fetal organs, as a starting point for developing the whole hybrid computational pregnant phantom set, which is the final objective of this study. An International Commission on Radiological Protection (ICRP) reference female voxel model was converted to a non-uniform rational B-spline (NURBS) surface model to construct a hybrid computational female phantom as a pregnant mother for each fetal model. Both fetal and maternal organs were matched with the ICRP- 89 reference data. To create a complete standard pregnant computational phantom set at 20, 30, and 35 weeks of pregnancy, the model mother’s reproductive organs were removed, and fetal phantoms with appropriate placental and uterine models were added to the female pelvis using a 3D-modeling software. With the aid of radiological image sets that had originally been used to construct the fetal models, each fetal position and rotation inside the uterus were carefully adjusted to represent the real fetal locations inside the uterus. The major abdominal soft tissue organs below the diaphragm, namely the small intestine, large intestine, liver, gall bladder, stomach, pancreas, uterus, and urinary bladder, were removed from non-pregnant females. The resulting fetal phantom was positioned in the appropriate location, matching the original radiological image sets. An obstetrician-gynecologist reviewed the complete internal anatomy of all fetus phantoms and the pregnant women for accuracy, and suggested changes were implemented as needed. The remaining female anatomical tissues were reshaped and modified to accommodate the location of the fetus inside the uterus. This new series of hybrid computational pregnant phantom models provides realistic anatomical details that can be useful in evaluating fetal radiation doses in pregnant patients undergoing diagnostic imaging or radiotherapy procedures where realistic fetal computational human phantoms are required.

Deep Morphological Neural Networks

Mathematical morphology intends to extract object features such as geometric and topological structures in digital images. Given a set of target images and original images, it is cumbersome and time-consuming to determine the suitable morphological operations and structuring elements. In this paper, we propose deep morphological neural networks, which include a nonlinear feature extraction layer to learn the structuring element correctly and an adaptive layer to select appropriate morphological operations automatically. We demonstrate the applications of object recognition, including hand-written digits, geometric shapes, traffic signs, and brain tumor. Experimental results show the higher computational efficiency and higher accuracy of our developed model as compared against existing convolutional neural network models.

Composition of sharp-focused image by rotation of Scheimpflug camera

A Scheimpflug camera has an image sensor tilted intentionally with respect to the lens. As the sensor and the lens are not parallel, the sharp-focused subject plane is slanted with respect to the camera according to the lens formula; this is known as the Scheimpflug principle. Therefore, if the camera is rotated about the optical axis, the sharp-focused plane covers a volume of the subject space. In this study, an application of the Scheimpflug camera is proposed to obtain a sharp-focused image with an extended depth of field (DOF) everywhere in the image. Off-the-shelf optical parts are used to implement the camera. A set of images are acquired by rotating the camera about an optical axis and the sharp-focused image is composed by selecting the focused pixels in the set of original images based on the focus measure. The experimental results show that the final composed image has a sharp focus in every region although each original image has a regional blur in the image plane. The focus measure of the composed image is 3–4 times higher than that of the original images.

Evaluation of Multisource Adaptive MRI Fusion for Gross Tumor Volume Delineation of Hepatocellular Carcinoma.

PurposeTumor delineation plays a critical role in radiotherapy for hepatocellular carcinoma (HCC) patients. The incorporation of MRI might improve the ability to correctly identify tumor boundaries and delineation consistency. In this study, we evaluated a novel Multisource Adaptive MRI Fusion (MAMF) method in HCC patients for tumor delineation.MethodsTen patients with HCC were included in this study retrospectively. Contrast-enhanced T1-weighted MRI at portal-venous phase (T1WPP), contrast-enhanced T1-weighted MRI at 19-min delayed phase (T1WDP), T2-weighted (T2W), and diffusion-weighted MRI (DWI) were acquired on a 3T MRI scanner and imported to in-house-developed MAMF software to generate synthetic MR fusion images. The original multi-contrast MR image sets were registered to planning CT by deformable image registration (DIR) using MIM. Four observers independently delineated gross tumor volumes (GTVs) on the planning CT, four original MR image sets, and the fused MRI for all patients. Tumor contrast-to-noise ratio (CNR) and Dice similarity coefficient (DSC) of the GTVs between each observer and a reference observer were measured on the six image sets. Inter-observer and inter-patient mean, SD, and coefficient of variation (CV) of the DSC were evaluated.ResultsFused MRI showed the highest tumor CNR compared to planning CT and original MR sets in the ten patients. The mean ± SD tumor CNR was 0.72 ± 0.73, 3.66 ± 2.96, 4.13 ± 3.98, 4.10 ± 3.17, 5.25 ± 2.44, and 9.82 ± 4.19 for CT, T1WPP, T2W, DWI, T1WDP, and fused MRI, respectively. Fused MRI has the minimum inter-observer and inter-patient variations as compared to original MR sets and planning CT sets. GTV delineation inter-observer mean DSC across the ten patients was 0.81 ± 0.09, 0.85 ± 0.08, 0.88 ± 0.04, 0.89 ± 0.08, 0.90 ± 0.04, and 0.95 ± 0.02 for planning CT, T1WPP, T2W, DWI, T1WDP, and fused MRI, respectively. The patient mean inter-observer CV of DSC was 3.3%, 3.2%, 1.7%, 2.6%, 1.5%, and 0.9% for planning CT, T1WPP, T2W, DWI, T1WDP, and fused MRI, respectively.ConclusionThe results demonstrated that the fused MRI generated using the MAMF method can enhance tumor CNR and improve inter-observer consistency of GTV delineation in HCC as compared to planning CT and four commonly used MR image sets (T1WPP, T1WDP, T2W, and DWI). The MAMF method holds great promise in MRI applications in HCC radiotherapy treatment planning.

Abstract P3-02-09: Three timepoint pharmacokinetic modeling to incorporate within standard of care MRI breast exams

Abstract Background: Standard of care (SOC) breast MRI exams typically acquire 4-7 frames of dynamic contrast-enhanced MRI (DCE-MRI) for cancer screening and staging. Post-contrast images depict lesion spiculations and boundaries to identify and characterize tumors. Pharmacokinetic (PK) analysis of DCE-MRI involves modeling blood flow to the lesion and surrounding tissue and has shown promise in diagnosis and prediction of therapeutic response. Currently, SOC DCE-MRI requires ~60-90 seconds per volume for images with sufficient quality and spatial resolution. However, PK analysis of DCE-MRI requires faster time course sampling. For this reason, PK modeling is limited to research scans with lower spatial resolution and higher temporal resolution. PK modeling would improve feedback of treatment response, and implementation in the SOC exam would increase imaging trial participation. In this study, we tested the estimation of Ktrans, a mixed perfusion and permeability PK parameter, from three images at optimal time points after contrast agent (CA) injection, and compared it to the Ktrans estimation from analysis of the full-length time course.. Methods: Women (N=23) with newly diagnosed invasive breast cancers who were eligible for neoadjuvant therapy (NAT) were scanned with a research MRI protocol as part of a treatment-monitoring study. Images acquired prior to the start of NAT were used. MRI was performed on 3.0T Siemens Skyra scanners at two sites with bilateral breast coils. The research protocol included ten sagittal slices centered about the primary tumor. The DCE-MRI images came from a fast sequence with 1.3 × 1.3 × 5.0 mm resolution acquired at 7.3 seconds per frame (66 frames total,) with a gadolinium-based CA injected one minute into the scan. A population arterial input function was used to implement a mathematical graph-based search of possible tissue CA concentration curves from the expected range of PK parameters. The search results gave a set of three optimized sub-sampled timepoints, Topt, from the full set of sample times, Tfull, at which to best sample the CA concentration curves to optimally estimate PK values. The imaging data was analyzed to find one parameter map from image times Tfull, and another from the subset of images at times Topt. The difference in Ktrans was computed at each parameter map voxel, and the concordance correlation coefficient (CCC) was computed per patient to determine agreement. The median Ktrans values were also compared for each patient. Results: The graph-based search of CA concentration curves found optimal times Topt of 37, 66, and 153 seconds after injection. The averaged values over all patients for median and maximum Ktrans from the original Tfull image set were 0.07 and 0.5 (min)-1. The average difference in Ktrans values between the Topt and Tfull sets was 0.02 (min)-1. When the median Ktrans values for each patient were compared, the average difference in median Ktrans values was 15% +/- 9%. The concordance correlation coefficients comparing the Topt and Tfull -sampled parameter maps for each patient were 0.89 +/- 0.12, showing high agreement. Discussion: This retrospective analysis suggests that it is possible to estimate PK parameters from a few properly selected post-contrast images inserted into a SOC DCE-MRI exam. The combination of optimal timing with fast acquisition techniques for high-resolution imaging could be used to provide quantitative data while preserving post-contrast images with the necessary spatial resolution for clinical reading. Importantly, the test images were acquired in the community setting with widely available MRI hardware, further indicating the potential for integration with SOC exams. Funding: NCI U24 CA226110, NCI U01 CA174706, NCI U01 CA142565, CPRIT RR160005 Citation Format: Julie C DiCarlo, Angela M Jarrett, Anum S Kazerouni, John Virostko, Anna G Sorace, Kalina P Slavkova, Debra Patt, Boone W Goodgame, Sarah Avery, Thomas E Yankeelov. Three timepoint pharmacokinetic modeling to incorporate within standard of care MRI breast exams [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P3-02-09.

Exploring the effect of image enhancement techniques on COVID-19 detection using chest X-ray images

Computer-aided diagnosis for the reliable and fast detection of coronavirus disease (COVID-19) has become a necessity to prevent the spread of the virus during the pandemic to ease the burden on the healthcare system. Chest X-ray (CXR) imaging has several advantages over other imaging and detection techniques. Numerous works have been reported on COVID-19 detection from a smaller set of original X-ray images. However, the effect of image enhancement and lung segmentation of a large dataset in COVID-19 detection was not reported in the literature. We have compiled a large X-ray dataset (COVQU) consisting of 18,479 CXR images with 8851 normal, 6012 non-COVID lung infections, and 3616 COVID-19 CXR images and their corresponding ground truth lung masks. To the best of our knowledge, this is the largest public COVID positive database and the lung masks. Five different image enhancement techniques: histogram equalization (HE), contrast limited adaptive histogram equalization (CLAHE), image complement, gamma correction, and balance contrast enhancement technique (BCET) were used to investigate the effect of image enhancement techniques on COVID-19 detection. A novel U-Net model was proposed and compared with the standard U-Net model for lung segmentation. Six different pre-trained Convolutional Neural Networks (CNNs) (ResNet18, ResNet50, ResNet101, InceptionV3, DenseNet201, and ChexNet) and a shallow CNN model were investigated on the plain and segmented lung CXR images. The novel U-Net model showed an accuracy, Intersection over Union (IoU), and Dice coefficient of 98.63%, 94.3%, and 96.94%, respectively for lung segmentation. The gamma correction-based enhancement technique outperforms other techniques in detecting COVID-19 from the plain and the segmented lung CXR images. Classification performance from plain CXR images is slightly better than the segmented lung CXR images; however, the reliability of network performance is significantly improved for the segmented lung images, which was observed using the visualization technique. The accuracy, precision, sensitivity, F1-score, and specificity were 95.11%, 94.55%, 94.56%, 94.53%, and 95.59% respectively for the segmented lung images. The proposed approach with very reliable and comparable performance will boost the fast and robust COVID-19 detection using chest X-ray images.

Assistant Diagnosis of Insanity Based on Infrared Thermal Image Analysis and Deep Learning Algorithm

Considering the lack of in-depth research on the cases of insanity and other mental diseases in the context of traditional Chinese medicine, an deep learning based algorithm which can realize auxiliary judgment function of insanity are proposed in this paper. First, the original image set is screened and then an improved U-net network is used to realize the division of the trunk and limbs of the human body, thus preventing interference from disease-independent areas to affect subsequent disease judgment. Finally, based on the classification of insanity, the reference function of visual analysis is added. 1508 IR images are divided into two groups to test the proposed method. And experimental results show that accuracy of the classification of insanity diseases can reach 0.92, which has a high reference value for the clinical diagnosis of insanity.

A pre-processing method for digital image correlation on rotating structures

Inter-frame structural motions are required to be small for classic Digital Image Correlation (DIC) algorithms, which severely limits the application of the technique in large rotation cases. To solve the problem, multiple methods to obtain initial guesses of deformation parameters have been introduced over recent years. However, the success rate is still limited by rotation angles and the efficiency is low due to extra calculations with these methods. To further develop the application of DIC in large rotation cases, Phase Mapping Method (PMM) is introduced in this paper. The principle of the proposed method is to rearrange the obtained images to be sorted by the structural rotating phases, which are previously mapped into the range of 0 to 2π. As a result, for an image set of n frames, the average inter-frame rotating angle is reduced to 2πn-1, and the original image set is transferred into an equivalent semi-static one by PMM. Additionally, a fast phase-detection approach is introduced to automate PMM in cases that the structural rotating speeds are unknown. By performing multiple tests, it is proved that PMM, as a pre-processing method, can significantly improve the success rate and the efficiency of DIC, whether used alone or concurrently with other algorithms. The first natural frequencies and mode shapes of a rotating beam under different rotating speeds are extracted through vibration test, from which the hardening effect is observed.

Image registration and selection for unmanned aerial vehicle image stitching

In this study, a general transformation-based framework for unmanned aerial vehicle image stitching is proposed that can resist global distortion while improving the local registration accuracy. In the first step, with tie points as constraints, the global transformation function of each image is obtained in an optimization manner and no reference image is needed. In the second step, to reduce data redundancy, an image selection algorithm based on information entropy is proposed. The optimal image combination covering the entire scene is selected from the original image set. In the third step, a local optimization algorithm based on mesh deformation is proposed to further optimize the registration accuracy of the optimal image combination. Finally, all of the images are combined to obtain a high-resolution panorama. On challenging datasets, the proposed algorithm can not only reduce the global distortions caused by the accumulation of errors during image stitching, but can also reduce the redundant data, which will benefit the postprocessing. The local mesh optimization can greatly improve the registration accuracy and eliminate the obvious misalignment problems. Tested on a large number of challenging datasets, the proposed method substantially outperforms several state-of-the-art image-stitching methods and commercial software.

Automatic characterization of myocardial perfusion imaging polar maps employing deep learning and data augmentation.

To investigate a deep learning technique, more specifically state-of-the-art convolutional neural networks (CNN), for automatic characterization of polar maps derived from myocardial perfusion imaging (MPI) studies for the diagnosis of coronary artery disease. Stress and rest polar maps corresponding to 216 patient cases from the database of the department of Nuclear Medicine of our institution were analyzed. Both attenuation-corrected (AC) and non-corrected (NAC) images were included. All patients were subjected to invasive coronary angiography within 60 days from MPI. As the initial dataset of this study was small to train a deep learning model from scratch, two strategies were followed. The first is called transfer learning. For this, we employed the state-of-the-art CNN called VGG16, which has been broadly exploited in medical imaging classification tasks. The second strategy involves data augmentation, which is achieved by the rotation of the polar maps, to expand the training set. We evaluated VGG16 with 10-fold cross-validation on the original set of images performing separate experiments for AC and NAC polar maps, as well as for their combination. The results were compared to the standard semi-quantitative polar map analysis based on summed stress and summed difference scores, as well as to the medical experts' diagnostic yield. With reference to the findings of coronary angiography, VGG16 achieved an accuracy of 74.53%, sensitivity 75.00% and specificity 73.43% when the AC and NAC polar maps were incorporated into one single image set. Respective figures of MPI interpretation by experienced Nuclear Medicine physicians were 75.00%, 76.97% and 70.31%. The accuracy of semi-quantitative polar map analysis was lower, 66.20% and 64.81% for AC and NAC technique, respectively. The proposed deep learning model with data augmentation techniques performed better than the conventional semi-quantitative polar map analysis and competed with doctor's expertise in this particular patient cohort and image set. The model could potentially serve as an assisting tool to support interpretation of MPI studies or could be used for teaching purposes.

Reduction of respiratory motion artifacts in gadoxetate-enhanced MR with a deep learning\u2013based filter using convolutional neural network

ObjectivesTo reveal the utility of motion artifact reduction with convolutional neural network (MARC) in gadoxetate disodium–enhanced multi-arterial phase MRI of the liver.MethodsThis retrospective study included 192 patients (131 men, 68.7 ± 10.3 years) receiving gadoxetate disodium–enhanced liver MRI in 2017. Datasets were submitted to a newly developed filter (MARC), consisting of 7 convolutional layers, and trained on 14,190 cropped images generated from abdominal MR images. Motion artifact for training was simulated by adding periodic k-space domain noise to the images. Original and filtered images of pre-contrast and 6 arterial phases (7 image sets per patient resulting in 1344 sets in total) were evaluated regarding motion artifacts on a 4-point scale. Lesion conspicuity in original and filtered images was ranked by side-by-side comparison.ResultsOf the 1344 original image sets, motion artifact score was 2 in 597, 3 in 165, and 4 in 54 sets. MARC significantly improved image quality over all phases showing an average motion artifact score of 1.97 ± 0.72 compared to 2.53 ± 0.71 in original MR images (p < 0.001). MARC improved motion scores from 2 to 1 in 177/596 (29.65%), from 3 to 2 in 119/165 (72.12%), and from 4 to 3 in 34/54 sets (62.96%). Lesion conspicuity was significantly improved (p < 0.001) without removing anatomical details.ConclusionsMotion artifacts and lesion conspicuity of gadoxetate disodium–enhanced arterial phase liver MRI were significantly improved by the MARC filter, especially in cases with substantial artifacts. This method can be of high clinical value in subjects with failing breath-hold in the scan.Key Points• This study presents a newly developed deep learning–based filter for artifact reduction using convolutional neural network (motion artifact reduction with convolutional neural network, MARC).• MARC significantly improved MR image quality after gadoxetate disodium administration by reducing motion artifacts, especially in cases with severely degraded images.• Postprocessing with MARC led to better lesion conspicuity without removing anatomical details.

Robust Vehicle Detection on Multi- Resolution Aerial Images

Detecting vehicles in aerial images is one of the core components for intelligent transportation system. This task is challenging due to the comparably small size of the target objects, the complex background and multi-perspective views. It is particularly difficult for the real-time detection cases where only several tens of milliseconds delays are allowed. In this paper, we propose an approach to detect vehicles from aerial images with different resolutions and perspectives in an approximate real-time manner. The proposed model is robust and can detect vehicles in various detection scenarios. It can detect vehicles from an image within 47 milliseconds. We evaluate our method on a challenging data set of original aerial images over Munich and our data set collected using an unmanned aerial vehicle (UAV). The experimental results have demonstrated that our proposed method is superior to the state-of-the- art algorithms with respect to accuracy, recall and detection time.

Generative Principal Component Thermography for Enhanced Defect Detection and Analysis

Machine learning methods play an important role in the nondestructive testing field for quality assessment of polymer composites. As a popular deep learning branch, a generative adversarial network is introduced to the thermography field as an image augmentation approach to improve its defect detection performance. Specifically, a generative principal component thermography (GPCT) method for defect detection in polymer composites is proposed. By employing the data augmentation strategy, more informative images are generated to enlarge the diversity of the original set of images. The defect detection results can be visualized using a number of interpretable features. Consequently, the defect detection performance of thermographic data analysis can be enhanced to some extent. The experimental results on a carbon fiber reinforced polymer specimen demonstrate the feasibility and advantages of the GPCT method.

Entropy Based Data Expansion Method for Blind Image Quality Assessment.

Image quality assessment (IQA) is a fundamental technology for image applications that can help correct low-quality images during the capture process. The ability to expand distorted images and create human visual system (HVS)-aware labels for training is the key to performing IQA tasks using deep neural networks (DNNs), and image quality is highly sensitive to changes in entropy. Therefore, a new data expansion method based on entropy and guided by saliency and distortion is proposed in this paper. We introduce saliency into a large-scale expansion strategy for the first time. We regionally add distortion to a set of original images to obtain a distorted image database and label the distorted images using entropy. The careful design of the distorted images and the entropy-based labels fully reflects the influences of both saliency and distortion on quality. The expanded database plays an important role in the application of a DNN for IQA. Experimental results on IQA databases demonstrate the effectiveness of the expansion method, and the network’s prediction effect on the IQA databases is found to be improved compared with its predecessor algorithm. Therefore, we conclude that a data expansion approach that fully reflects HVS-aware quality factors is beneficial for IQA. This study presents a novel method for incorporating saliency into IQA, namely, representing it as regional distortion.

The China Image Set (CIS): A New Set of 551 Colored Photos With Chinese Norms for 12 Psycholinguistic Variables

Normative image sets are widely used in memory, perception, and language studies. Following the pioneering work of Snodgrass and Vanderwart (1980), a number of normalized image sets with various language norms have been created. However, original image sets that are carefully selected to accommodate Chinese culture and language are still in short supply. In the present study, we provided the China Image Set (CIS), a new set of photo stimuli with Chinese norms. The CIS consists of 551 high-quality colored photo stimuli that cover 21 categories and are normalized on 12 different variables, including name agreement, category agreement, familiarity, visual complexity, object manipulability, manipulation experience, color diagnosticity, shape diagnosticity, image variability, age of acquisition, image agreement, and within-category typicality. Of the 12 variables, shape diagnosticity and manipulation experience with the object depicted in a stimulus are the two newly introduced and normalized variables. Multiple regression analysis reveals that name agreement, age of acquisition, image agreement, shape diagnosticity, and image variability are the most robust determinants of picture naming latency. Our normative dataset of the high-quality photo stimuli offers an ecologically more valid tool to study object recognition and language processing within Chinese culture than has previously been available.