High-resolution Module Research Articles

An improved lightweight high-resolution network based on multi-dimensional weighting for human pose estimation

Human pose estimation is one of the key technologies in action recognition, motion analysis, human–computer interaction, animation generation etc. How to improve its performance has become a current research hotspot. Lite-HRNet establishes long range connections between keypoints and exhibits good performance in human pose estimation tasks. However, the scale of this method to extract features is relatively single and lacks sufficient information interaction channels. To solve this problem, we propose an improved lightweight high-resolution network based on multi-dimensional weighting, named MDW-HRNet, which is implemented by the following aspects: first, we propose global context modeling, which can learn multi-channel and multi-scale resolution information weights. Second, a cross-channel dynamic convolution module is designed, it performs inter-channel attention aggregation between dynamic and parallel kernels, replacing the basic convolution module. These make the network capable of channel weighting, spatial weighting and convolution weighting. At the same time, we simplify the network structure to perform information exchange and information compensation between high-resolution modules while ensuring speed and accuracy. Experimental results show that our method achieves good performance on both COCO and MPII human pose estimation datasets, and its accuracy surpasses mainstream lightweight pose estimation networks without increasing computational complexity.

BalanceHRNet: An effective network for bottom-up human pose estimation

In the study of human pose estimation, which is widely used in safety and sports scenes, the performance of deep learning methods is greatly reduced in high overlap rate and crowded scenes. Therefore, we propose a bottom-up model, called BalanceHRNet, which is based on balanced high-resolution module and a new branch attention module. BalanceHRNet draws on the multi-branch structure and fusion method of a popular model HigherHRNet. And our model overcomes the shortcoming of HigherHRNet that cannot obtain a large enough receptive field. Specifically, through the connecting structure in balanced high-resolution module, we can connect almost all convolutional layers and obtain a sufficiently large receptive field. At the same time, the multi-resolution representation can be maintained due to the use of balanced high-resolution module, which enable our model to recognize objects with richer scales and obtain more complex semantics information. And for branch fusion method, we design branch attention to obtain the importance of different branches at different stages. Finally, our model improves the accuracy while ensuring a smaller amount of computation than HigherHRNet. The CrowdPose dataset is used as test dataset, and HigherHRNet, AlphaPose, OpenPose and so on are taken as comparison models. The AP measured by BalanceHRNet is 63.0%, increased by 3.1% compared to best model — HigherHRNet. We also demonstrate the effectiveness of our network through the COCO(2017) keypoint detection dataset. Compared with HigherHRNet-w32, the AP of our model is improved by 1.6%.

CT synthesis from MRI with an improved multi-scale learning network

Introduction: Using MRI to synthesize CT and substitute its function in radiation therapy has drawn wide research interests. Currently, deep learning models have become the first choice for MRI—CT synthesis because of their ability to study complex non-linear relations. However, existing studies still lack the ability to learn complex local and global MRI–CT relations in the same time , which influences the intensity and structural performance of synthetic images.Methods: This study proposes a hybrid multi-scale model to explore rich local and global MRI—CT relations, relations, namely, the hybrid multi-scale synthesis network (HMSS-Net). It includes two modules modelling different resolution inputs. In the low-resolution module, the Transformer method is applied to build its bottleneck part to expand the receptive field and explore long-range MRI—CT relations to estimate the coarse distribution of widely spread tissues and large organs. In the high-resolution module, residual and dense connections are applied to explore complex local MRI—CT relations under multiple step sizes. Then, the feature spaces of two modules are combined together and utilized to provide synthetic CT. HMSS-Net also introduces the multi-scale structural similarity index measure loss to provide multi-scale supervision during training.Results: The experimental results on head and neck regions of 78 patients showed that HMSS–Net reduced the average of 7.6/3.13 HU on the mean absolute error and increased the average of 2.1/1.8% on the dice coefficient of bone compared with competing image–to–image synthesis methods.Conclusion: The results imply that HMSS–Net could effectively improve the intensity and structural performances of synthetic CT.

Extension of the open-source TIGRE toolbox for proton imaging

Proton irradiation is a well-established method to treat deep-seated tumors in radio oncology. Usually, an X-ray computed tomography (CT) scan is used for treatment planning. Since proton therapy is based on the precise knowledge of the stopping power describing the energy loss of protons in the patient tissues, the Hounsfield units of the planning CT have to be converted. This conversion introduces range errors in the treatment plan, which could be reduced, if the stopping power values were extracted directly from an image obtained using protons instead of X-rays. Since protons are affected by multiple Coulomb scattering, reconstruction of the 3D stopping power map results in limited image quality if the curved proton path is not considered. This work presents a substantial code extension of the open-source toolbox TIGRE for proton CT (pCT) image reconstruction based on proton radiographs including a curved proton path estimate. The code extension and the reconstruction algorithms are GPU-based, allowing to achieve reconstruction results within minutes. The performance of the pCT code extension was tested with Monte Carlo simulated data using three phantoms (Catphan® high resolution and sensitometry modules and a CIRS patient phantom). In the simulations, ideal and non-ideal conditions for a pCT setup were assumed. The obtained mean absolute percentage error was found to be below 1% and up to 8 lp/cm could be resolved using an idealized setup. These findings demonstrate that the presented code extension to the TIGRE toolbox offers the possibility for other research groups to use a fast and accurate open-source pCT reconstruction.

High-Resolution Ex-Vivo Imaging of Retina with a Laptop-Based Portable Endoscope.

Purpose Ophthalmic endoscopy is useful in vitreoretinal surgery with opaque anterior segments or anatomically challenging structures. However, standard ophthalmic endoscopy devices are usually large and costly. Thus, the development of a portable endoscope is necessary. Methods The portable endoscope consisted of a probe, an illumination system, a high-resolution camera module, and a universal serial bus (USB). It was connected to a laptop and applied for intraocular imaging of porcine eyes in vitro. Basic parameters and pictures of the same tissue target were compared with those of the standard Endo Optiks E4 system. Results The retinal images were displayed on the laptop screen, which clearly showed the overall appearance of the central and peripheral retina, and the details of the retinal vasculature and ciliary body. Compared with a standard endoscope, our portable endoscope is smaller and more affordable. It can be taken anywhere for intraocular imaging and vitreoretinal surgery. Conclusion A laptop-based portable endoscope is a promising device in vitreoretinal surgery. It provides high-resolution images of intraocular tissues that should make a noticeable difference in intraocular surgery with disordered anterior segments. With its portability and high-resolution imaging, it may promote the application of endoscopes in ophthalmology.

Research on an Intelligent Driving Algorithm Based on the Double Super-Resolution Network

Semantic segmentation plays a very important role in image processing, and has been widely used in intelligent driving, medicine, and other fields. With the development of semantic segmentation, the model has become more and more complex and the resolution of training pictures is higher and higher, so the requirements for required hardware facilities have become higher and higher. Many high-precision networks are difficult to apply in intelligent driving vehicles with limited hardware conditions, and will bring delay to recognition, which is not allowed in practical application. Based on the Dual Super-Resolution Learning (DSRL) network, this paper proposes a network model for training high-resolution pictures, adding a high-resolution convolution module which improves segmentation accuracy and speed while reducing computation. In a CamVid dataset, taking the road category as an example, IOU is 95.23%, which is 4% higher than DSRL, the real-time segmentation time of the same video is reduced by 46% from 120 s to 65 s, and the segmentation effect is better and faster, which greatly alleviates the recognition delay caused by high-resolution input.

Evaluation of high-resolution and depth-encoding PET detector modules based on single-ended readout with TOFPET2 ASIC

Evaluation of high-resolution and depth-encoding PET detector modules based on single-ended readout with TOFPET2 ASIC

Detection of Small Changed Regions in Remote Sensing Imagery Using Convolutional Neural Network

As a crucial application in remote sensing, change detection can detect semantic change of the land use in multi-temporal aerial imagery. Especially, with the improvement of the spatial resolution of remote sensing imagery, more and more small objects have been taken into account. Although several methods based on deep learning have been proposed to solve change detection in satellite imagery, most of existing methods cannot properly handle the small changed regions and the class imbalance, which are extremely serious in satellite imagery. In this paper, we propose a Siamese High-Resolution Network (Si-HRNet) to detect the small changed regions in remote sensing imagery. To our best knowledge, this is the first time to transfer High Resolution (HR) module into the Siamese network, so that our network can maintain high-resolution representation throughout the whole process and repeatedly fuse multi-resolution representations to obtain rich feature representations, especially can reduce information loss for small objects. In addition, to handle the class imbalance issues, we combine weighted binary cross entropy (BCE) and inverse volume weighted generalized dice loss (GDL) in small objects change detection. Experimental results show that the proposed Si-HRNet achieves state-of-the-art performance in both DigitalGlobe dataset and WHU Building change detection dataset, and the F1 score is improved by 3.35% ∼ 3.43%.

Application of high-resolution CT local augmentation reconstruction to improve the accurate diagnosis of small pulmonary nodule structure

Objective To explore the accurate display method of the local structure of small CT nodules in lung. Methods Close the automatic radiation dose adjustment technique. Keep the thickness of the phantom was 2 mm, the interval between the phantom layers was 2 mm, the pitch was 1, and the reconstructed convoluted nucleus was B80s. Group A controlled the scanning matrix FOV to be 500 mm×500 mm, 400 mm×400 mm, 300 mm×300 mm, 200 mm×200 mm and 100 mm×100 mm, while group B controlled the scanning matrix FOV to be 500 mm×500 mm. Use these parameters to scan the catphan 500 phantom, a routine chest CT conditioned scanning quality control model. The original data raw data were used to reconstruct the FOV to be 400 mm×400 mm, 300 mm×300 mm, 200 mm×200 mm and 100 mm×100 mm under the scanning FOV to be 500 mm×500 mm. All other conditions are consistent. Observe the high contrast resolution module of Catphan 500 phantom, and compare the line logarithms of two groups of images under different scanning FOV or different reconstruction FOV. Thirty-five patients with small pulmonary nodules from February 2018 to March 2018 of the Second Affiliated Hospital of Zhejiang University Medical College were retrospectively collected. The raw data were used to reconstruct the images. The FOV was 320 mm×320 mm in the conventional reconstruction group and 100 mm×100 mm in the local magnification reconstruction group. The subjective score data of the two groups were compared by using rank-sum test. Results When catphan 500 phantom was used, the number of lines in group A and group B increased gradually with the decreasing of FOV. The subjective score of local magnification reconstruction group (4.77±0.35) was higher than that of conventional reconstruction group (3.86±0.50) and the difference was statistically significant (Z=-5.763, P<0.05). Conclusion Local magnification and reconstruction of high-resolution CT images can achieve the same image quality as local magnification, local magnification and reconstruction of image quality is significantly better than simple image magnification. Key words: Tomography,X-ray computed; Radiation dosage; Imaging quality; Comparative study

Warm Molecular Hydrogen in Nearby, Luminous Infrared Galaxies

Abstract Mid-infrared molecular hydrogen (H2) emission is a powerful cooling agent in galaxy mergers and in radio galaxies; it is a potential key tracer of gas evolution and energy dissipation associated with mergers, star formation, and accretion onto supermassive black holes. We detect mid-IR H2 line emission in at least one rotational transition in 91% of the 214 Luminous Infrared Galaxies (LIRGs) observed with Spitzer as part of the Great Observatories All-sky LIRG Survey. We use H2 excitation diagrams to estimate the range of masses and temperatures of warm molecular gas in these galaxies. We find that LIRGs in which the IR emission originates mostly from the Active Galactic Nuclei (AGN) have about 100 K higher H2 mass-averaged excitation temperatures than LIRGs in which the IR emission originates mostly from star formation. Between 10% and 15% of LIRGs have H2 emission lines that are sufficiently broad to be resolved or partially resolved by the high-resolution modules of Spitzer’s Infrared Spectrograph (IRS). Those sources tend to be mergers and contain AGN. This suggests that a significant fraction of the H2 line emission is powered by AGN activity through X-rays, cosmic rays, and turbulence. We find a statistically significant correlation between the kinetic energy in the H2 gas and the H2 to IR luminosity ratio. The sources with the largest warm gas kinetic energies are mergers. We speculate that mergers increase the production of bulk inflows leading to observable broad H2 profiles and possibly denser gas.

Preparing for hard x-ray microscopy with Multilayer Zone Plates

With Pulsed Laser Deposition, Multilayer Zone Plates can be fabricated to focus hard x-ray beams into 2D spots smaller than 10 nm. To put these optics into use for imaging applications, we have commissioned a new dedicated sample tower as a high-resolution module for the GINIX instrument, stationed at the P10 beamline at PETRA III. Here we summarise the motorisation and show first imaging benchmark results obtained with a “traditional” Fresnel Zone Plate. The first 2D continuous STXM scan using the new EigerX 4M detector at full 750 Hz speed is shown: a field of view of roughly about 1 μm squared has been recorded with 255 × 255 images within 96 seconds.

An efficient algorithm for polarization in the SASKTRAN radiative transfer framework

Techniques for remote sensing of atmospheric composition often involve the minimization of the residual between spectrally-resolved atmospheric radiometry and the output of a radiative transfer model which solves the equations of radiative transfer in a candidate atmosphere. Here, SASKTRAN, a framework of climatologies, optical property calculators, and scalar radiative transfer models specifically designed for modeling limb scattering observations, is upgraded to solve the vector radiative transfer equations. In particular, the Monte Carlo and High Resolution Successive Orders modules of the SASKTRAN framework are extended to handle polarized scattering of light in a fully spherical atmosphere, with the assumption of a Lambertian earth underneath. The fast, polarized, High Resolution module is validated against the statistically exact polarized Monte Carlo module for limb-scatter observations in the UV–visible–NIR spectral range and is found to be accurate to within 0.2% in its fully polarized mode. An approximate method to solve the vector radiative transfer equations is introduced; in this mode, the high resolution engine is accurate for limb-scatter geometries to within about 0.2% in almost all cases with minimal degradation in performance as compared to the scalar mode. The models are used to show that scalar simulations of limb-observed UV–vis radiances may contain absolute radiometric errors of up to 5% for typical stratospheric aerosol loads. Furthermore, limb-observed radiances may have nearly any linear polarization, and this is in general altitude and wavelength dependent.

Development of a high resolution module for PET scanners

Positron Emission Tomography (PET) scanners require high performances in term of spatial resolution and sensitivity to allow early detection of cancer masses. In small animal and organ dedicated PET scanners the Depth of Interaction (DOI) information has to be obtained to avoid parallax errors and to reconstruct high resolution images. In the whole body PET, the DOI information can be useful to correct for the time jitter of the optical photons along the main axis of the scintillator, improving the time performances. In this work we present the development of PET module designed to reach high performance as compared to the current scanners while keeping the complexity of the system reasonably low. The module presented is based on a 64 LYSO (Lutetium-yttrium oxyorthosilicate) crystals matrix and on a 4×4 MPPC (Multi Pixels Photon Counter) array as detector in a 4 to 1 coupling between the crystals and the detector and a single side readout. The lateral surfaces of the crystals are optically treated to be unpolished. The DOI and the energy resolution of the PET module are presented and a fast method to obtain the DOI calibration is discussed.

203 - Development of a High Resolution Module for PET scanners

203 - Development of a High Resolution Module for PET scanners

New Electromagnetic Design of Miniature AF VCM Actuator with Low Cost

AbstractRecently, compact and high-resolution camera modules with auto-focusing (AF) function have been integrated into cell phones in order to capture sharp photographs. Consumer demands AF camera modules in cell phones to have high performance with low cost. Accordingly, the present study proposes a new electromagnetic design of miniature AF voice coil motor (VCM) actuator with closed-loop control for cell phone camera modules to satisfy the requirements. The structure of the proposed AF VCM actuators was designed by using simulation methods. The performance of the proposed AF VCM actuators was demonstrated by a laboratory-built prototype. The experimental results have shown that the proposed AF VCM actuator has excellent performance with lower power consumption, higher positioning repeatability, and lower cost, when compared to previous AF VCM actuators with open-loop control or closed-loop control.

CASSIS: THE CORNELL ATLAS OF SPITZER /INFRARED SPECTROGRAPH SOURCES. II. HIGH-RESOLUTION OBSERVATIONS

The Infrared Spectrograph (IRS) on board the Spitzer Space Telescope observed about 15,000 objects during the cryogenic mission lifetime. Observations provided low-resolution (R~60-127) spectra over ~5-38um and high-resolution (R~600) spectra over ~10-37um. The Cornell Atlas of Spitzer/IRS Sources (CASSIS) was created to provide publishable quality spectra to the community. Low-resolution spectra have been available in CASSIS since 2011, and we present here the addition of the high-resolution spectra. The high-resolution observations represent approximately one third of all staring observations performed with the IRS instrument. While low-resolution observations are adapted to faint objects and/or broad spectral features (e.g., dust continuum, molecular bands), high-resolution observations allow more accurate measurements of narrow features (e.g., ionic emission lines) as well as a better sampling of the spectral profile of various features. Given the narrow aperture of the two high-resolution modules, cosmic ray hits and spurious features usually plague the spectra. Our pipeline is designed to minimize these effects through various improvements. A super sampled point-spread function was created in order to enable the optimal extraction in addition to the full aperture extraction. The pipeline selects the best extraction method based on the spatial extent of the object. For unresolved sources, the optimal extraction provides a significant improvement in signal-to-noise ratio over a full aperture extraction. We have developed several techniques for optimal extraction, including a differential method that eliminates low-level rogue pixels (even when no dedicated background observation was performed). The updated CASSIS repository now includes all the spectra ever taken by the IRS, with the exception of mapping observations.

An ultrasound-based liquid pressure measurement method in small diameter pipelines considering the installation and temperature.

Liquid pressure is a key parameter for detecting and judging faults in hydraulic mechanisms, but traditional measurement methods have many deficiencies. An effective non-intrusive method using an ultrasound-based technique to measure liquid pressure in small diameter (less than 15 mm) pipelines is presented in this paper. The proposed method is based on the principle that the transmission speed of an ultrasonic wave in a Kneser liquid correlates with liquid pressure. Liquid pressure was calculated using the variation of ultrasonic propagation time in a liquid under different pressures: 0 Pa and X Pa. In this research the time difference was obtained by an electrical processing approach and was accurately measured to the nanosecond level through a high-resolution time measurement module. Because installation differences and liquid temperatures could influence the measurement accuracy, a special type of circuit called automatic gain control (AGC) circuit and a new back propagation network (BPN) model accounting for liquid temperature were employed to improve the measurement results. The corresponding pressure values were finally obtained by utilizing the relationship between time difference, transient temperature and liquid pressure. An experimental pressure measurement platform was built and the experimental results confirm that the proposed method has good measurement accuracy.

Psychophysical evaluation of Catphan®600 CT image quality findings using human observers and software analysis

PurposeA comparison of image quality obtained from human observers and software analysis of CT phantom images. Methods and materialsA Catphan®600 CT QA phantom was scanned for: posterior fossa; cerebrum; abdomen and chest on three CT models, as part of a dose optimisation strategy. CT image data sets (n = 24) obtained pre and post optimisation were blindly evaluated by radiographers (n = 8) identifying the number of distinct line pairs and contrast discs for each of the three supra-slice sets within the phantom's high and low contrast resolution modules. The same images were also reviewed using the web based service – Image Owl for automatic analysis of Catphan®600 images. ResultsInter-observer reliability measured using Cronbach's α between human observers and again including software analysis as the 9th observer gave α = 0.97 for both instances, indicating comparable internal consistency with and without software analysis.Results of a paired sample t-test showed no significant difference (p ≥ 0.05) between human observers and software analysis in 37.5% of observations for line pairs and 37.5%; 12.5% and 50% for the sets of contrast discs representing nominal contrast of 1.0%, 0.5% and 0.3% respectively. Software analysis findings improved compared to observer readings as contrast levels reduced. ConclusionCombined use of human observers and software analysis for evaluation of image quality in CT using phantoms is recommended. However the sole use of software analysis may provide more detail than that obtained by human observers. Further research to investigate the clinical relevance of such image quality findings is recommended.

High-resolution Optical and Electro-optical Microscopy for PV-modules

Non-destructive methods for high-resolution inspection are of great importance for quality assurance, reliability assessment and for research and development. Especially for samples like solar modules or large specimens high-resolution microscopy measurements are not used and therefore observable details are strongly limited by the optical resolution. Small defects can grow and cause a reduction of the module's lifetime. A portal-based inspection system has been set up for large samples like a PV-module or any other large sample. It can be equipped with several optical systems like cameras or a microscope. The optical properties are optimized for visual or for electroluminescence imaging on a macroscopic and microscopic scale and complements an existing Dark Lock-In Thermography (DLIT) setup for modules.The inspection system has been exemplarily used on two samples with defects and degradations to elucidate the wide range of applications. An investigation on a module with snail trail discolorations has been conducted. Cell cracks, finger interruptions and finger discoloration are measured using light microscopy and micro-electroluminescence imaging (μ-EL). Cracks within EL-inactive cell parts could be revealed only with light microscopy. On another sample the correlation on a microscopic scale of visible cell cracks, local delamination and the corresponding micro-electroluminescence (μ-EL) will be demonstrated. The correlation of light microscopy, μ-EL and DLIT illustrates the potential of a high-resolution module inspection for quality control, reliability assessment and for research and development.

The Development of the Durable Test-Bed of the Hydraulic Pump

An energy recovery method during the durable test is put forward in this essay, and the durable test-bed of the hydraulic pump is developed. The efficiency and accuracy data acquisition is realized by using High-Speed High-Resolution sensor and data collection module. Control automation is come true by using PLC, industrial tablet PC and electric command device. The instrument panel constructed by LabWindows/CVI made the operation more concision, flexible and easy-to-use. This test bed has been put into use for a year and shows that the system can realize duration and efficient test of hydraulic pump easily and reliable.