Search

This work presents a new fine-grained transparent object segmentation dataset, termed Trans10K-v2, extending Trans10K-v1, the first large-scale transparent object segmentation dataset. Unlike Trans10K-v1 that only has two limited categories, our new dataset has several appealing benefits. (1) It has 11 fine-grained categories of transparent objects, commonly occurring in the human domestic environment, making it more practical for real-world application. (2) Trans10K-v2 brings more challenges for the current advanced segmentation methods than its former version. Furthermore, a novel Transformer-based segmentation pipeline termed Trans2Seg is proposed. Firstly, the Transformer encoder of Trans2Seg provides the global receptive field in contrast to CNN's local receptive field, which shows excellent advantages over pure CNN architectures. Secondly, by formulating semantic segmentation as a problem of dictionary look-up, we design a set of learnable prototypes as the query of Trans2Seg's Transformer decoder, where each prototype learns the statistics of one category in the whole dataset. We benchmark more than 20 recent semantic segmentation methods, demonstrating that Trans2Seg significantly outperforms all the CNN-based methods, showing the proposed algorithm's potential ability to solve transparent object segmentation.Code is available in https://github.com/xieenze/Trans2Seg.

Although deeper and larger neural networks have achieved better performance, due to overwhelming burden on computation, they cannot meet the demands of deployment on resource-limited devices. An effective strategy to address this problem is to make use of dynamic inference mechanism, which changes the inference path for different samples at runtime. Existing methods only reduce the depth by skipping an entire specific layer, which may lose important information in this layer. In this paper, we propose a novel method called Dynamic Multipath Neural Network (DMNN), which provides more topology choices in terms of both width and depth on the fly. For better modelling the inference path selection, we further introduce previous state and object category information to guide the training process. Compared to previous dynamic inference techniques, the proposed method is more flexible and easier to incorporate into most modern network architectures. Experimental results on ImageNet and CIFAR-100 demonstrate the superiority of our method on both efficiency and classification accuracy.

Practical face image quality assessment (FIQA) models are trained under the supervision of labeled data, which requires more or less human labor. The human labeled quality scores are consistent with perceptual intuition but laborious. On the other hand, models can be trained with data generated automatically by the recognition models with artificially selected references. However, the recognition scores are sometimes inaccurate, which may give wrong quality scores during FIQA training. In this paper, we propose a labour-saving method for quality scores generation. For the first time, we conduct systematic investigations to show that there exist severe contradictions between different types of target quality, namely distribution gap (DG). To bridge the gap, we propose a novel framework for training FIQA models by combining the merits of data from different sources. In order to make the target score from multiple sources compatible, we design a method called quality distribution alignment (QDA). Meanwhile, to correct the wrong target by recognition models, contradictory samples selection (CSS) is adopted to select samples from the human labeled dataset adaptively. Extensive experiments and analysis on public benchmarks including MegaFace has demonstrated the superiority of our in terms of effectiveness and efficiency.

Face recognition has achieved remarkable improvements with the help of the angular margin based softmax losses. However, the margin is usually manually set and kept constant during the training process, which neglects both the optimization difficulty and the informative similarity structures among different instances. Although some works have been proposed to tackle this issue, they adopt similar methods by simply changing the margin for different classes, leading to limited performance improvements. In this paper, we propose a novel sample-wise adaptive margin loss function from the perspective of the hypersphere manifold structure, which we call companion guided soft margin (CGSM). CGSM introduces the information of distribution in the feature space, and conducts teacher-student optimization within each mini-batch. Samples of better convergence are considered as teachers, while students are optimized with extra soft penalties, so that the intra-class distances of inferior samples can be further compacted. Moreover, CGSM does not require sophisticated mining techniques, which makes it easy to implement. Extensive experiments and analysis on MegaFace, LFW, CALFW, IJB-B and IJB-C demonstrate that our approach outperforms state-of-the-art methods using the same network architecture and training dataset.

This paper describes a sample-hold (SH) circuit for the front-ended pipelined 12-bit analog-to-digital converter (ADC). A differential OTA (operational transconductance amplifier) used in the sample-hold (SH) circuit is presented. The OTA is optimized for high-speed high-accuracy applications by using gain-boosted topology. By means of clamp circuit, the slewing rate of the OTA is greatly improved. The design uses the chartered 0.35-μm 2P4M CMOS process with a 3 V supply. The open-loop DC gain is over 110 dB and unity-gain bandwidth is 524.6 MHz. The slewing rate of the OTA is 1160 V/μs while the total load capacitance is 6pf. The SH circuit can settle within 10 ns to an accuracy of

This scheme makes the best use of structure of the cargo ship by integrating with air-breathing helicon plasma electrical propulsion. The normal operation of a manned space station requires regular replenishment of oxygen, fuel, food and instruments. Research on resupply spacecraft to supply oxygen and fuel to the space station will help reduce the replenishment needs of the station. A resupply spacecraft with air-breathing electrical propulsion as core technology will meet the replenishment needs of the space station during on-orbit operation at a low cost and bring high loading capacity and substantial economic benefits. The replenishment ship based on this scheme can collect 1000 kg working gas a year on 180 km orbit and reduce the cost of building and launching cargo ships. This paper carried out studies on mission process, resource consumption, orbit design and thrust Analysis. The ionization mode of air-breathing electrical propulsion is helicon discharge and the average power for orbit maintaining and air collecting is about 775 W.

Form understanding depends on both textual contents and organizational structure. Although modern OCR performs well, it is still challenging to realize general form understanding because forms are commonly used and of various formats. The table detection and handcrafted features in previous works cannot apply to all forms because of their requirements on formats. Therefore, we concentrate on the most elementary components, the key-value pairs, and adopt multimodal methods to extract features. We consider the form structure as a tree-like or graph-like hierarchy of text fragments. The parent-child relation corresponds to the key-value pairs in forms. We utilize the state-of-the-art models and design targeted extraction modules to extract multimodal features from semantic contents, layout information, and visual images. A hybrid fusion method of concatenation and feature shifting is designed to fuse the heterogeneous features and provide an informative joint representation. We adopt an asymmetric algorithm and negative sampling in our model as well. We validate our method on two benchmarks, MedForm and FUNSD, and extensive experiments demonstrate the effectiveness of our method., Accepted to EMNLP 2020 Findings

In this paper, we introduce an anchor-box free and single shot instance segmentation method, which is conceptually simple, fully convolutional and can be used by easily embedding it into most off-the-shelf detection methods. Our method, termed PolarMask, formulates the instance segmentation problem as predicting contour of instance through instance center classification and dense distance regression in a polar coordinate. Moreover, we propose two effective approaches to deal with sampling high-quality center examples and optimization for dense distance regression, respectively, which can significantly improve the performance and simplify the training process. Without any bells and whistles, PolarMask achieves 32.9% in mask mAP with single-model and single-scale training/testing on the challenging COCO dataset. For the first time, we show that the complexity of instance segmentation, in terms of both design and computation complexity, can be the same as bounding box object detection and this much simpler and flexible instance segmentation framework can achieve competitive accuracy. We hope that the proposed PolarMask framework can serve as a fundamental and strong baseline for single shot instance segmentation task.

In this paper, we consider the low-bit quantization problem of face recognition (FR) under the open-set protocol. Different from well explored low-bit quantization on closed-set image classification task, the open-set task is more sensitive to quantization errors (QEs). We redefine the QEs in angular space and disentangle it into class error and individual error. These two parts correspond to inter-class separability and intra-class compactness, respectively. Instead of eliminating the entire QEs, we propose the rotation consistent margin (RCM) loss to minimize the individual error, which is more essential to feature discriminative power. Extensive experiments on popular benchmark datasets such as MegaFace Challenge, Youtube Faces (YTF), Labeled Face in the Wild (LFW) and IJB-C show the superiority of proposed loss in low-bit FR quantization tasks.

This work presents an efficient yet effective online Knowledge Distillation method via Collaborative Learning, termed KDCL, which is able to consistently improve the generalization ability of deep neural networks (DNNs) that have different learning capacities. Unlike existing two-stage knowledge distillation approaches that pre-train a DNN with large capacity as the ''teacher'' and then transfer the teacher's knowledge to another ''student'' DNN unidirectionally (i.e. one-way), KDCL treats all DNNs as ''students'' and collaboratively trains them in a single stage (knowledge is transferred among arbitrary students during collaborative training), enabling parallel computing, fast computations, and appealing generalization ability. Specifically, we carefully design multiple methods to generate soft target as supervisions by effectively ensembling predictions of students and distorting the input images. Extensive experiments show that KDCL consistently improves all the ''students'' on different datasets, including CIFAR-100 and ImageNet. For example, when trained together by using KDCL, ResNet-50 and MobileNetV2 achieve 78.2% and 74.0% top-1 accuracy on ImageNet, outperforming the original results by 1.4% and 2.0% respectively. We also verify that models pre-trained with KDCL transfer well to object detection and semantic segmentation on MS COCO dataset. For instance, the FPN detector is improved by 0.9% mAP.

Therapeutic treatment has been revolutionized with the advent of implantable drug delivery devices. These devices hold immense potential to treat chronic ailments, such as cancer tumours that require high drug concentrations. A wireless system is an integral part of the device to regulate drug release according to prescribed dosing schedule. An antenna is important component of a wireless system for data telemetry and actuating the drug release mechanism. In this paper, we propose an ultrawideband conformal loop antenna around the capsule shaped device. The flexible layer of microchambers is also conformed around inner shell of the capsule. The CST Gustav voxel human body model was used to perform numerical study of the antenna. According to the simulation results, the proposed capsule antenna has shown reflection coefficient of -11.83 dB at the desired frequency of 900 MHz.

In this communication, the design procedure of a wideband antenna with omnidirectional radiation pattern is demonstrated based on the theory of characteristic modes. Consisting of a dipole and a loop antenna, the antenna has a very simple structure. A wide impedance bandwidth is obtained because of the simultaneous excitation of the antenna’s first two modes. Meanwhile, due to the fact that these two modes share a similar omnidirectional radiation pattern, a stable radiation pattern is also achieved across the operating frequency band. In order to identify the antenna’s different modes, a characteristic mode analysis of the antenna is carried out first. Then, a feed configuration is specifically designed to excite the desired modes. To validate the antenna design, a prototype was fabricated and tested. Measured results agree well with the simulated ones. Measurement shows that a wide impedance bandwidth of 44.2% with $\vert {\text{S}_{11}}\vert dB (1.85–2.9 GHz) and stable radiation patterns at both E-plane and H-plane were achieved over the operating frequency band.

A new wideband differentially fed dual-polarized antenna with stable radiation pattern for base stations is proposed and studied. A cross-shaped feeding structure is specially designed to fit the differentially fed scheme and four parasitic loop elements are employed to achieve a wide impedance bandwidth. A stable antenna gain and a stable radiation pattern are realized by using a rectangular cavity-shaped reflector instead of a planar one. A detailed parametric study was performed to optimize the antenna’s performances. After that, a prototype was fabricated and tested. Measured results show that the antenna achieves a wide impedance bandwidth of 52% with differential standing-wave ratio $\approx 8$ dBi) and a stable radiation pattern with 3-dB beamwidth of 65° ±5° were also found over the operating frequencies. Moreover, the proposed antenna can be easily built by using printed circuit board fabrication technique due to its compact and planar structure.

Helicon wave ionization source could provide abundant and excellent plasma for space electric propulsion. Now it is widely recommended and required to test and verify the technology. This paper designed and analyzed the structure of the helicon wave electric propulsion system and completed mechanics calculation with finite element method (FEM) software to validate the rationality and reliability. At last, it gave final conclusion and prospect.

Low-resolution text images are often seen in natural scenes such as documents captured by mobile phones. Recognizing low-resolution text images is challenging because they lose detailed content information, leading to poor recognition accuracy. An intuitive solution is to introduce super-resolution (SR) techniques as pre-processing. However, previous single image super-resolution (SISR) methods are trained on synthetic low-resolution images (e.g. Bicubic down-sampling), which is simple and not suitable for real low-resolution text recognition. To this end, we propose a real scene text SR dataset, termed TextZoom. It contains paired real low-resolution and high-resolution images which are captured by cameras with different focal length in the wild. It is more authentic and challenging than synthetic data, as shown in Fig. 1. We argue improving the recognition accuracy is the ultimate goal for Scene Text SR. In this purpose, a new Text Super-Resolution Network, termed TSRN, with three novel modules is developed. (1) A sequential residual block is proposed to extract the sequential information of the text images. (2) A boundary-aware loss is designed to sharpen the character boundaries. (3) A central alignment module is proposed to relieve the misalignment problem in TextZoom. Extensive experiments on TextZoom demonstrate that our TSRN largely improves the recognition accuracy by over 13% of CRNN, and by nearly 9.0% of ASTER and MORAN compared to synthetic SR data. Furthermore, our TSRN clearly outperforms 7 state-of-the-art SR methods in boosting the recognition accuracy of LR images in TextZoom. For example, it outperforms LapSRN by over 5% and 8% on the recognition accuracy of ASTER and CRNN. Our results suggest that low-resolution text recognition in the wild is far from being solved, thus more research effort is needed. The codes and models will be released at: github.com/JasonBoy1/TextZoom

With the development of electric propulsion technology, the measurement of plasma parameters has been widely used in the field of aerospace. In the process of electric propulsion performance examination and test, the plasma parameters are indispensable to the analysis of electric propulsion performance. Plasma parameters such as electron temperature, plasma density, plasma flux density play important roles in plasma technology, which are the important basis to determine plasma state of electric propulsion. There has a variety of diagnosis method for Plasma electric propulsion. But the radio frequency, electric and magnetic fields often existed in electric propulsion environment. Therefore, It is necessary to use the Retarding potential analyzer (RPA) and Langmuir probe (ALP) together to test the plasma parameters. Experiment data can be mutually verified to effectively improve the accuracy and reliability of test data. The test data of xenon and nitrogen plume obtained in this experiment is of great significance for the in-depth study of helical wave propulsion.

This paper puts forward a kind of self-excited RF power control system, which is more suitable for helicon plasma electric propulsion system than traditional RF generator based on a servo motor. Simple structure makes it smaller and lighter. It has no moving parts, and so that can be easily transformed into aerospace products. This system achieves the goal of impedance matching by using phase lock loop to trace load frequency. A verification circuit is designed, and the results prove that the design is feasible.

Distillation-based learning boosts the performance of the miniaturized neural network based on the hypothesis that the representation of a teacher model can be used as structured and relatively weak supervision, and thus would be easily learned by a miniaturized model. However, we find that the representation of a converged heavy model is still a strong constraint for training a small student model, which leads to a higher lower bound of congruence loss. In this work, we consider the knowledge distillation from the perspective of curriculum learning by teacher's routing. Instead of supervising the student model with a converged teacher model, we supervised it with some anchor points selected from the route in parameter space that the teacher model passed by, as we called route constrained optimization (RCO). We experimentally demonstrate this simple operation greatly reduces the lower bound of congruence loss for knowledge distillation, hint and mimicking learning. On close-set classification tasks like CIFAR and ImageNet, RCO improves knowledge distillation by 2.14% and 1.5% respectively. For the sake of evaluating the generalization, we also test RCO on the open-set face recognition task MegaFace. RCO achieves 84.3% accuracy on one-to-million task with only 0.8 M parameters, which push the SOTA by a large margin.

Background: Probiotics is a prevalence therapeutic method for irritable bowel syndrome (IBS), but there is lack of comparison in different protocols. We aim to differentiate the reasonable protocols by assessing the efficacy and safety through the combined way of traditional and network meta-analysis. Method: PubMed, Medline, EMBASE, Web of Science, and Cochrane Central Register of Controlled Trials databases were searched from January 2006 to April 2019. The relative risk (RR) with a 95% confidence interval (CI) was used to combine dichotomous data of responders. Result: Among 14 studies included 1695 patients were identified as suitable for inclusion. The proportion of responders was associated with the administration of multispecies probiotics (RR: 1.39; 95% CI: 1.19–1.61) and the dose of 109∼1010 (RR: 2.08; 95% CI: 1.59–2.71). In network meta-analysis, the protocol of DUO had a significant effect for diarrhea type of IBS compared with placebo (RR: 7.46; 95% CI: 2.00–32.23). In the rest of 4 protocols, no significant difference was found in each other except F19 which appears inferior when compared with Pro (RR: 0.16; 95% CI: 0.03–0.88). Meanwhile, Pro showed a superior effect for undifferentiated-type IBS compared with placebo (RR: 7.16; 95% CI: 1.72–29.89). No probiotics-associated severe adverse event was reported in included studies. Conclusion: Probiotics is a safety choice to improve the overall symptoms for IBS patient. The protocols with suitable dose combined of Lactobacillus and Bifidobacterium can have prepotent effects compared with single species or over-dosage protocols. Network meta-analysis shows that DUO may be the first recommendation for diarrhea-type IBS. In the remaining 4 regimes of this study, Pro has a high rank for undifferentiated-type IBS.

This paper proposes the dynamic recursive neural network (DRNN), which simplifies the duplicated building blocks in deep neural network. Different from forwarding through different blocks sequentially in previous networks, we demonstrate that the DRNN can achieve better performance with fewer blocks by employing block recursively. We further add a gate structure to each block, which can adaptively decide the loop times of recursive blocks to reduce the computational cost. Since the recursive networks are hard to train, we propose the Loopy Variable Batch Normalization (LVBN) to stabilize the volatile gradient. Further, we improve the LVBN to correct statistical bias caused by the gate structure. Experiments show that the DRNN reduces the parameters and computational cost and while outperforms the original model in term of the accuracy consistently on CIFAR-10 and ImageNet-1k. Lastly we visualize and discuss the relation between image saliency and the number of loop time.

In this paper, we raise a new problem, namely cross model face recognition (CMFR), which has considerable economic and social significance. The core of this problem is to make features extracted from different models comparable. However, the diversity, mainly caused by different application scenarios, frequent version updating, and all sorts of service platforms, obstructs interaction among different models and poses a great challenge. To solve this problem, from the perspective of Bayesian modelling, we propose R3 Adversarial Network (R3AN) which consists of three paths: Reconstruction, Representation and Regression. We also introduce adversarial learning into the reconstruction path for better performance. Comprehensive experiments on public datasets demonstrate the feasibility of interaction among different models with the proposed framework. When updating the gallery, R3AN conducts the feature transformation nearly 10 times faster than ResNet-101. Meanwhile, the transformed feature distribution is very close to that of target model, and its error rate is incredibly reduced by approximately 75% compared with a naive transformation model. Furthermore, we show that face feature can be deciphered into original face image roughly by the reconstruction path, which may give valuable hints for improving the original face recognition models.

A ±45° dual-polarized base station antenna with a planar structure is proposed for 2G/3G bands. The +45° antenna and the −45° antenna share the same radiator, which is composed of four folded dipoles and four coplanar striplines. The coplanar striplines work as $\lambda $ /4 impedance transformers. Two Y-shaped feeding lines is designed to obtain a good impedance matching and a planar reflector is employed to achieve a stable radiation pattern. The antenna element has a wide impedance bandwidth of 27.8% with SWR 30 dB), a stable radiation pattern with 3-dB beamwidth 65° ± 5° in the horizontal plane, a high XPD and a high front-to-back ratio were achieved.

A novel method of enhancing cross-polarization discrimination (XPD) of a ${\pm }{45}^{\circ}$ dual-polarized base-station antenna, by addition of four parasitic elements placed in a square contour, is presented. First, it was established that high XPD at the broadside can be obtained by addition of four parasitic elements around a simple ${\pm }{45}^{\circ}$ dual-polarized base-station antenna. Second, we investigated the principle of wide beamwidth high XPD after studying magnitude currents on parasitic element, spacing between parasitic elements and length of parasitic elements. Finally, a ${\pm }{45}^{\circ}$ dual-polarized base-station antenna was designed, fabricated and measured. Measurement results are in good agreement with the results of simulation in terms of radiation pattern, gain, and Voltage Standing Wave Ratio (VSWR). In particular, the enhanced XPD is more than 20 dB at the boresight and more than 10 dB within ${\pm }{60}^{\circ}$ of the main lobe at the horizontal plane.

Scene text detection and recognition have been well explored in the past few years. Despite the progress, efficient and accurate end-to-end spotting of arbitrarily-shaped text remains challenging. In this work, we propose an end-to-end text spotting framework, termed PAN++, which can efficiently detect and recognize text of arbitrary shapes in natural scenes. PAN++ is based on the kernel representation that reformulates a text line as a text kernel (central region) surrounded by peripheral pixels. By systematically comparing with existing scene text representations, we show that our kernel representation can not only describe arbitrarily-shaped text but also well distinguish adjacent text. Moreover, as a pixel-based representation, the kernel representation can be predicted by a single fully convolutional network, which is very friendly to real-time applications. Taking the advantages of the kernel representation, we design a series of components as follows: 1) a computationally efficient feature enhancement network composed of stacked Feature Pyramid Enhancement Modules (FPEMs); 2) a lightweight detection head cooperating with Pixel Aggregation (PA); and 3) an efficient attention-based recognition head with Masked RoI. Benefiting from the kernel representation and the tailored components, our method achieves high inference speed while maintaining competitive accuracy. Extensive experiments show the superiority of our method. For example, the proposed PAN++ achieves an end-to-end text spotting F-measure of 64.9 at 29.2 FPS on the Total-Text dataset, which significantly outperforms the previous best method. Code will be available at: https://git.io/PAN., Comment: Accepted to TPAMI 2021

In this study, an environmental site investigation was conducted of a typical lead (Pb)-related production plant. Based on a combination of manual and Geoprobe sampling, Pb concentrations in soils were measured using standard analytical methods to clarify the distribution of Pb concentrations in soils of the plant’s production workshops, open spaces within the plant, and in the adjacent farmland. The results show that the maximum depth of Pb pollution in the soil of the production workshop was 1.5 m. Pb pollution was found in the soil of the adjacent farmland within a range of 50 m from the plant and occurred at a maximum depth of 0.15 m. Probably as a result of the characteristics of Pb species, there was little horizontal movement of Pb observed in the sampled soils. Further, the solid ground surfaces in the plant (e.g., hardened concrete) effectively prevented the pollution of subsurface soil by Pb.

A broadband $\pm 45^{\circ}$ dual-polarized base station antenna is proposed for 2G/3G/LTE bands. The proposed antenna has Y-shaped feeding lines to impart a wide impedance bandwidth to it, and a rectangular box-shaped reflector to enhance its stability in radiation patterns over the operating frequencies. A prototype of the proposed antenna is designed, fabricated, and tested. The measured results show achievement of impedance bandwidth of 45% for $\hbox {SWR} at both ports, port-to-port isolation $>$ 25 dB , a stable radiation pattern with 3-dB beamwidth $68^{\circ}\pm 2^{\circ}$ at H-plane and V-plane, and a stable antenna gain of 8.2 $\pm$ 0.6 dBi from 1.7 to 2.7 GHz. A broadband dual-polarized antenna array with five elements for the band is developed for base station application. Measurements show achievement of array gain of 14.8 $\pm$ 1.4 dBi, 3-dB beamwidth $66.56^{\circ}\pm 2.22^{\circ}$ at H-plane at port 1 and $64.85^{\circ}\pm 4.96^{\circ}$ at port 2 with $\hbox {SWR} over the operating frequency band. Simulation results have a good agreement with measurement ones.

Incidental scene text spotting is considered one of the most difficult and valuable challenges in the document analysis community. Most existing methods treat text detection and recognition as separate tasks. In this work, we propose a unified end-to-end trainable Fast Oriented Text Spotting (FOTS) network for simultaneous detection and recognition, sharing computation and visual information among the two complementary tasks. Specially, RoIRotate is introduced to share convolutional features between detection and recognition. Benefiting from convolution sharing strategy, our FOTS has little computation overhead compared to baseline text detection network, and the joint training method learns more generic features to make our method perform better than these two-stage methods. Experiments on ICDAR 2015, ICDAR 2017 MLT, and ICDAR 2013 datasets demonstrate that the proposed method outperforms state-of-the-art methods significantly, which further allows us to develop the first real-time oriented text spotting system which surpasses all previous state-of-the-art results by more than 5% on ICDAR 2015 text spotting task while keeping 22.6 fps., 10 pages, 6 figures

A complex-shaped underwater robot with a device which can weld for a nuclear reactor pool in emergency and inspect in daily was developed. Hydrodynamic coefficients in the dynamic model were critical to control precisely. In this work, a numerical simulation of robot had been employed for hydrodynamic coefficients calculation based on the computational fluid dynamics (CFD). Hydrodynamic coefficients including inertial hydrodynamic coefficients, viscous hydrodynamic coefficients were respectively solved by simulating virtually a steady-state motion simulation test and an unsteady-state motion simulation test in CFD. A prototype test in a circulating water channel had been launched to validate simulation results. Furthermore, viscous hydrodynamic and inertia hydrodynamic coefficients were acquired. The accuracy and reliability of hydrodynamic coefficients was proved by comparing to the prototype test results. All work carried out was indicated to be benefit to hydrodynamic coefficient study and complex-shaped underwater robot control.

The calibration process of a digital Multi-meter (DMM) is complicated and tedious. The whole procedure needs to calibrate dozens of test points of several measuring ranges. Usually, the automatic calibration system is used to replace the manual operation. In metrology, the MAT/CAL which is a software produced by Fluke company has been widely used, but it can’t meet our special needs. This article is mainly on the introduction of design method of an automatic calibration system for DMM basing on Fluke MAT/CAL. It describes the setup of hardware, the improvement of ACC file in MAT/CAL and the basic principle of our software.

A new generation of negative tone and chemically amplified positive tone photoresists by TOK, JSR, Dow Chemical and others has gained momentum for advanced packaging applications. Resist thickness requirements are increasing to the 40-100 μm range as Cu pillars and micro-bumps are adopted, to accommodate the tighter pitches required in the newest multi-chip package designs. In order to form the pillars, the resist mask must be thicker than the height of the pillars to contain the entire bump structure.

Multi-nodes inertial data acquisition has the widespread applications in medical rehabilitation research and biomechanical analysis. In the traditional method, the inertial data of every child-node is sent to a center node by wired line firstly. Then the center node transfers the data of all child-nodes to PC through a single serial port. This approach has some obvious drawbacks, including the fact that the sampling rate is low, the communication protocols are complex and installation is inconvenient. This article proposed a multi-node and multi-base-station wireless data acquisition system. The communication between child-node and the base-station is performed in a point-to-point manner. The data of micro inertial sensors on the child-node circuit board is collected by the MCU then is sent to the matching base-station by the RF module. The base-station send data packet to the upper computer through serial port. The results showed that the upper computer software can collect data packets synchronously from three child-nodes (the data packet of each child-node contain 19 Bytes) with high-speed sampling rates up to 118 Hz. And there is no obvious deferring time under this condition. The technique has the certain applications in the data collecting system of motion capture. Key words: Pcomm; multi-thread; multiple serial ports; nRF905; MPU6000

Falls are the second leading cause of unintentional injury deaths worldwide, so how to prevent falls has become a safety and security problem for elderly people. At present, because the sensing modules of most fall alarm devices generally only integrate the single 3-axis accelerometer, so the measured accuracy of sensing signals is limited. It results in that these devices can only achieve the alarm of post-fall detection but not the early pre-impact fall recognition in real fall applications. Therefore, this paper aimed to develop an early pre-impact fall alarm system based on high-precision inertial sensing units. A multi-modality sensing module embedded fall detection algorithm was developed for early pre-impact fall detection. The module included a 3-axis accelerometer, a 3-axis gyroscope and a 3-axis magnetometer, which could arouse the information of early pre-impact fall warning by a buzzer and a vibrator. Total 81 times fall experiments from 9 healthy subjects were conducted in simulated fall conditions. By combination of the early warning threshold algorithm, the result shows that the detection sensitivity can achieve 98.61% with a specificity of 98.61%, and the average pre-impact lead time is 300ms. In the future, GPS, GSM electronic modules and wearable protected airbag will be embedded in the system, which will enhance the real-time fall protection and timely immediate aid immensely for the elderly people.

Nowadays falls are a serious problem for elderly people with the coming of aged society in the world. According to statistics,hip fracture accounts for the most of the deaths and costs of all the fall-induced injury. This paper presented an airbag system of hip protection, which included air source, sensors, microcontroller, gas circuit and airbags. A six-axial inertial sensor module that integrated an embedded three-axis MEMS accelerometer and three-axis MEMS gyroscope was used to collect human motion data, and a one-axis obliquity sensor was used to collect human angle data. The microcontroller was employed to recognize the activities of daily living (ADL) and falls based on fall detection algorithm and the collected data from sensors. The gas circuit was triggered once the event that the fall would be inevitable was confirmed by the microcontroller, then the compressed gas would fill into airbags through the gas circuit rapidly. Therefore, a buffer would appear between the human body and the ground before the body impacting the ground, which would reduce the impact of the human body. Compressed CO2 was selected as the air source after we tested several kinds of gas. A 16g CO2 pressurized cylinder could provide enough pressure and volume to inflate quickly the airbags. In order to improve the reliability of the gas circuit, a needle valve was optimized from the several designed structures by the experimental optimization methods. Finally, the airbag system was tested in various designed trials. The results indicated that the system gained the satisfaction for the design requirements and would be potential to apply to the protection of hip joint in the fall high-risk people in the future.

A smartwatch antenna with a novel high impedance surface (HIS) is investigated in this paper. In order to fit the all-metal smartwatch applications, a non-planar HIS is proposed instead of using a traditional planar one. With the presence of the HIS, the wrist phantom has a slight impact on the antenna's performance and a low SAR can also be obtained. The antenna can achieve a gain of >1.3 dBi and an efficiency of >40% from 2.4 to 2.484 GHz even when it is mounted above a wrist phantom. Hence, it can work effectively for WIFI or Bluetooth system.

In [1] , there are some mistakes in the antenna dimensions in Fig. 1 on page 2672 and Fig. 11 on page 2674. These two figures should be corrected as follows

This paper demonstrates an approach that negative uniaxial crystal has a relative anomalous dispersion effect which can compensate group velocity delay, and applies this approach to nonlinear frequency conversion of an ultrafast laser field. High efficiency of the third harmonic generation is experimentally fulfilled by adopting a collinear configuration of doubing-compensation-tripling system. Through finely adjusting the incident angle and optical axis direction of the compensation plate, it obtains ultraviolet (UV) output energy of 0.32 mJ centered at 270 nm with spectral bandwidth of 2 nm when input beam at 800 nm was 70 fs pulse duration and 6 mJ pulse energy which was extracted from Ti:sapphire laser system by a diaphragm, corresponding to an 800-to-270 nm conversion efficiency of 5.3% and a factor-of-1.6 improvement in the third harmonic generation of UV band in comparison with a general conventional configuration. Furthermore, when the full energy of 18 mJ from a Ti:sapphire laser system was used and optimized, the UV emission could reach 0.83 mJ.

In this paper, we study the bistatic re∞ection and transmission properties of random rough surface with large slope and large height. Method of Moment (MOM) is used to solve the surface integral equations for 2D rough surface scattering problem. The modeled rough surfaces are similar to random rectangular grating, so that there are large slopes on the surface. The motivation of the study is to analyze scattering by sastrugi surface in Polar Regions. The ridges on the sastrugi surface have heights of about 20cm. In microwave remote sensing of land at 5GHz, 10GHz, 19GHz and 37GHz, these heights are larger than wavelength. Next, we consider

A model for the microwave emissions of multilayer dry snowpacks, based on dense media radiative transfer (DMRT) theory with the quasicrystalline approximation (QCA), provides more accurate results when compared to emissions determined by a homogeneous snowpack and other scattering models. The DMRT model accounts for adhesive aggregate effects, which leads to dense media Mie scattering by using a sticky particle model. With the multilayer model, we examined both the frequency and polarization dependence of brightness temperatures (Tb's) from representative snowpacks and compared them to results from a single-layer model and found that the multilayer model predicts higher polarization differences, twice as much, and weaker frequency dependence. We also studied the temporal evolution of Tb from multilayer snowpacks. The difference between Tb's at 18.7 and 36.5 GHz can be 5 K lower than the single-layer model prediction in this paper. By using the snowpack observations from the Cold Land Processes Field Experiment as input for both multi- and single-layer models, it shows that the multilayer Tb's are in better agreement with the data than the single-layer model. With one set of physical parameters, the multilayer QCA/DMRT model matched all four channels of Tb observations simultaneously, whereas the single-layer model could only reproduce vertically polarized Tb's. Also, the polarization difference and frequency dependence were accurately matched by the multilayer model using the same set of physical parameters. Hence, algorithms for the retrieval of snowpack depth or water equivalent should be based on multilayer scattering models to achieve greater accuracy.

Dense media radiative transfer (DMRT) theory is used to study the multiple-scattering effects in active microwave remote sensing. Simplified DMRT phase matrices are obtained in the 1-2 frame. The simplified expressions facilitate solutions of the DMRT equations and comparisons with other phase matrices. First-order, second-order, and full multiple-scattering solutions of the DMRT equations are obtained. To solve the DMRT equation, we decompose the diffuse intensities into Fourier series in the azimuthal direction. Each harmonic is solved by the eigen-quadrature approach. The model is applied to the active microwave remote sensing of terrestrial snow. Full multiple-scattering effects are important as the optical thickness for snow at frequencies above 10 GHz often exceed unity. The results are illustrated as a function of frequency, incidence angle, and snow depth. The results show that cross polarization for the case of densely packed spheres can be significant and can be merely 6 to 8 dB below copolarization. The magnitudes of the cross polarization are consistent with the experimental observations. The results show that the active 13.5-GHz backscattering coefficients still have significant sensitivity to snow thickness even for snow thickness exceeding 1 m

The depth quality of a time-of-flight (ToF) camera is influenced by many systematic and non-systematic errors 1 . In this paper we present a simple method to correct and reduce these errors and propose a multi-phase approach to improve the depth acquisition accuracy. Compared with traditional calibration methods, we take the position of light source into account, and calibrate the light source together with the camera to reduce depth distortion. To ameliorate the sensor errors caused in the manufacturing process, a Look-up Table (LUT) is used to correct pixel-related errors. Besides, we capture images with multiple phases and apply FFT to get the true depth. By the proposed approach, we are able to reconstruct an accurate 3D model with RMSE of the measured depth belowing 1.2mm.

The methods to design base-station antennas with stable radiation patterns in broadband are presented. Through a rational place of two dipoles in parallel configuration or in orthogonal over a reflector, the antennas with a wide impedance bandwidth, stable radiation pattern and nearly identical E- and H-plane can be achieved due to the complementation of the radiation patterns. According to these method, multi-dipole antennas with about 60% impedance bandwidth (SWR

A polarization rotator-splitter is demonstrated using a Si3N4 waveguide atop a silicon waveguide. The device exhibits polarization crosstalk < −19 dB, loss < 1.5 dB, and PDL < 1.0 dB over an 80 nm bandwidth.

Logistic regression (LR) has been widely used in applications of machine learning, thanks to its linear model. However, when the size of training data is very large, even such a linear model can consume excessive memory and computation time. To tackle both resource and computation scalability in a big-data setting, we evaluate and compare different approaches in distributed platform, parallel algorithm, and sublinear approximation. Our empirical study provides design guidelines for choosing the most effective combination for the performance requirement of a given application.