Search

Color constancy is one of the fundamental tasks in computer vision. Many supervised methods, including recently proposed Convolutional Neural Networks (CNN)-based methods, have been proved to work well on this problem, but they often require a sufficient number of labeled data. However, it is expensive and time-consuming to collect a large number of labeled training images with accurately measured illumination. In order to reduce the dependence on labeled images and leverage unlabeled ones without measured illumination, we propose a novel semi-supervised framework with limited training samples for illumination estimation. Our key insight is that the images with similar features from different cues will share similar lighting conditions. Consequently, three graphs based on three visual cues, low-level RGB color distribution, mid-level initial illuminant estimates and high-level scene content, are constructed to represent the relationship among different images. Then a multi-cue semi-supervised color constancy method (MSCC) is proposed after integrating these three graphs into a unified model. Extensive experiments on benchmark datasets demonstrate that our proposed MSCC method outperforms nearly all the existing supervised methods with limited labeled samples. Even with no unlabeled samples, MSCC still obtains better performance and stableness than most supervised methods.

Stretchable electronics‐based surface electromyography (sEMG) evaluation devices are expected to play a big role in clinical diagnosis. However, the stability and quality of the signals collected by existing stretchable electronics are too poor, especially when muscle movement is involved, making them inappropriate for high standard clinical diagnosis. Here, a physiology‐based design method for stretchable electronics and a novel airbag‐type stretchable electrode array (ASEA) device for assessment of the complex female pelvic floor muscle (PFM) is proposed. Clinical trials show that the ASEA device is able to provide stable contact interface and multi‐channel accurate data acquisition. The stability and quality of the sEMG signal are much better than those obtained by the existing stretchable electronics‐based PFM electrode devices. Furthermore, a muscle‐unit evaluation method (MUEM) to assess the PFM complex state is proposed, especially its cross‐interaction between muscles. Clinical trials show that MUEM can accurately and comprehensively assess PFM state and the correlations between main muscles, which unveils the mechanisms of some special muscle states that are not possible using traditional methods. This proof of concept research holds the promise for the development of new diagnostic strategies for muscle pathological research, and has the potential for clinical application and general implication.

In order to compact coal logs which could be transported smoothly in the hydraulic pipeline, the polyvinyl alcohol was used as the binder and mixed with anthracite, bituminous coal and lignite, res...

Efforts are made to improve the design process of Axiomatic Design (AD) in this paper. Rooted in logic and rational thought, AD provides systemic tools that helps facilitate design process. However, the mapping process and operability of AD remain to be clearly defined. The research on AD theory shows that AD lacked the description of elements in domains and the specific method during mapping process, which could lead to confusions of the designers. To remedy the issue, the Extenics theory is integrated with AD to provide improved guidance for design activity. The integration of the two theories are established in the form of mathematical expression and flow chart. Examples are presented to illustrate the integration of the two theories. The resulted mapping of the decomposition process is shown to be properly guided.

The matched filter is often used for the radar receiving terminal, but such application in wideband radar may cause complex system and high cost. The adoption of compressed sensing can not only remove the matched filter and the high-speed analogdigital converter, but also realize high-resolution radar imaging. In consideration of the sparseness of the radar target scene, a compressed sensing method used for chaotic radar is proposed in this article, wherein the chaotic frequency-modulated signal is generated by Bernoulli mapping and the randomness of such signal can be directly used for constructing the observation matrix. Meanwhile, the under-sampled echo signal can be used for radar scene reconstruction through the optimization technology. According to the simulation of stationary and non-stationary target radar scenes, compared with the matched filter processing technology, the proposed method has simple processing procedure and optimal performance, and the feasibility of this method is also verified by the simulation result.

The pressure distribution around two coal logs in tandem (a coal log train) flowing in a horizontal pipe (inner diameter of 50 mm) is experimentally studied. Experiments are conducted at three water velocities (0.92, 1.10, and 1.50 m/s; corresponding Reynolds numbers 45 771, 54 726, and 74 627, respectively). The hydraulic characteristics of the coal log train are analyzed, and the results are used to calculate the lift and drag forces together with the lift and drag coefficients. The pressure at the faces and surfaces of the coal log train increases at rising water velocity, enhancing the lift and drag forces. The forces acting on the first coal log are the largest, and forces on other coal logs decrease stepwise. The hydraulic characteristics of the first coal log are more complex than that of other coal logs. The lift and drag coefficients decrease with increasing water velocity. Both the lift and drag forces on individual coal logs are smaller than those on a single coal log in a pipe, when the coal logs follow one another. These results show the great difference in hydraulic characteristics between a coal log train and a single coal log in a pipe.

This article reports the performance of coal logs made using cationic-emulsified asphalt as the binder. Anthracite, bituminous, and lignite coals were used to compact coal logs. The strength, water resistance, and wear resistance of coal logs were evaluated in detail. The results showed that cationic-emulsified asphalt improved both the water and wear resistance of coal logs. Coal logs made of anthracite were strong with the best water resistance and wear resistance properties, followed by coal logs made of bituminous coal and lignite. Both the weight and external dimensions of coal logs increased during water absorption tests. Anthracite coal logs were able to achieve less than 3% of wear loss with a binder addition of 5% or more, while bituminous coal and lignite coal logs were unable to meet this requirement only by adding a binder. Nevertheless the coal logs have the enough compressive strength to bear the pressure provided by water to ensure that they are complete during transportation except...

This study proposes a numerical simulation of the unsteady turbulent flow of a single coal log in a pipe using a dynamic mesh method. An eccentric two-dimensional dynamic model is developed in FLUENT 6.2. The model contains a straight section and a bent section. Comparisons of the coal log velocity and the pressure distributions of the pipe and the coal log surface are made between the calculated results and the experimental data. The shear force and the friction coefficient are also calculated to verify changes of the influence of water on the coal log. The results of this simulation display the dynamic motion of the coal log for startup and turning. Water drag force moves the coal log, making the coal log velocity greater than the water velocity. The lifting force lifts the coal log and moves its tail slightly upward. The pressure drop of the whole pipe is respectively larger in the presence of the coal log. The pressure along the coal log surface slowly declines. The appearance of a bend impedes the motion of the coal log and enhances abrasion. The coal log has difficulty being suspended due to decreased lifting force and increased friction. The pressure on the coal log surface rises.

For the issue of being difficult to carry out various experiments of vehicle braking systems, a real-time simulation platform which is an vehicle braking pressure based on xPC Target is proposed. In the design of platform, the method is studied through writing CS-function driver development. Under MATLAB RTW, the problems of hardware simulation in the loop to support I / O device is resolved in xPC Target toolbox. Upon the analysis of the basic principles and structure of the driver, the solution to the key issues such as register operation and base seeking address are studied. And PCI-1710 driver is developed. The verifications about the driver validation of the data acquisition card and closed-loop regulation to the vehicle braking pressure are completed with the platform. The experimental results show that the A / D sampling frequency of acquisition card can be up to 10 kHz, signals can be displayed in real time and the test platform can be long-running. The development process is stable. Pressure waveforms consistent with the waveforms measured by the oscilloscope. The braking pressure can be adjusted by the real-time simulation platform. A new experimental method is provided for research on the vehicle braking performance.

The utility model provides a three-phase static var generator based on a Z source network. The three-phase static var generator comprises the Z source network, a main inversion circuit and a detection control module, wherein output ends of the Z source network are connected to a direct current side of the main inversion circuit; output ends of the main inversion circuit are connected to a three-phase power grid in parallel; and the detection control module is used for detecting voltages and currents of the three-phase power grid, the output ends of the main inversion circuit, and the Z source network, controlling the on-off state of a switch device of the main inversion circuit through driving signals, and further realizing the dynamic reactive compensation of the three-phase power grid. Unique X-type combination of small inductors and capacitors of the Z source network is used to replace the topological structure on the direct current side of an existing SVG, the defect that the SVG device is liable to be influenced by the insufficient voltage on the direct current side is overcome, a dead-zone circuit is effectively eliminated, and the output noises caused by the dead time are avoided, so that the SVG provided by the utility model is lower in cost, the reactive power compensation range is wider, the compensation effect is better, and the working reliability of the SVG is substantially improved.

Aiming at the MTF evaluation limitations in visible imaging system, this paper introduces the minimum resolvable contrast. On the basis of MRC theoretical, the paper obtains subjective and objective methods of MRC measurement. Then MRC measurement based on neural network is put forward, which does not depend on subjective judgment of person. BP neural network is established and trained. Therefore, the network can replace human eyes to judge test patterns with different spatial frequencies and contrasts. Sony camera with 500 megapixels is selected in the experiment. Results show that MRC values of the objective measurement at all frequencies are less than those of the subjective measurement. The MRC Measurement has good stability.

Dimensionality reduction is a problem of fundamental importance in both machine learning and data mining. In this paper, we develop a new approach that can process labeled datasets and accurately reduce their dimensionalities. The approach is based on a new objective that contains information from both the global and local structures of a data set. An iterative approach is used to compute the directions along which the objective can be optimized. Experiments on benchmark data sets show that both the global and local information in a dataset can be effectively captured by this approach and it is thus able to provide more accurate results for dimensionality reduction than some existing approaches.

The Independent Component Analysis (ICA) method has applied in the field of blind source separation. On the basis of analyzing ICA, the study ameliorates the FastICA. The conventional FastICA has only a second order convergence rate and combination of static model and batch optimization algorithm. An improved ICA algorithm is therefore proposed to reduce the iteration steps and dynamic the algorithm. The experimental results show that the improved algorithm achieved satisfactory results.

In order to solve the problem of serious harmonic pollution and power factor decline caused by uncontrolled rectification of the power electronic devices, an active power factor correction method based on UC3854 chip is proposed. The simulation model on the Saber software simulation platform is set up. Simulation results show that the power factor reaches more than 0.99. The system hardware circuit is constructed and the experimental results are basically the same as those of the simulation. Moreover the rationality and superiority of the circuit are verified.

Single-phase shunt active power filter (APF) is able to compensate harmonics in grid. This paper uses PSCAD/EMTDC to build a simulation model of system based on single-phase shunt APF with one-cycle control under bipolar modulation. In the meantime, it simulates compensation performance for system. Single-phase APF is designed and tested on the basis of simulation. The simulation and experimental results show APF has a good application prospect because it can sufficiently compensate for grid, control circuit has simple structure and dynamic characteristic is fine.

Aiming at the limitation of the xPC target toolbox in driving I/O devices for hardware-in-the-loop simulation under MATLAB RTW, the driver is written with a CS function development so as to solve the problem of computer supporting I/O devices. Based on the analysis of the basic principles and structure of the drivers, the solutions to key problems, such as register operations and base address looking for, are put forward and the PCI-1710 driver is developed. The experiments are done on the two assembly personal computers, where A/D sampling frequency can be up to 10 kHz, the signal can be displayed in real-time and driver can run for a long time. Experimental results show that the development driver is stable and portability.

Traditional image during MRC measurement needs eyes to judge, which exists the subjective error. This paper describes theoretical framework, design and testing process of objective measurement MRC. By using BP neural network model, MRC image can be recognized automatically and reduces subjective error. The experimental results show that three layers BP neural networks can effectively identify the image by choosing appropriate characteristic value.

Aiming at the effect of environment to MTF measurement, the MTF measurent at different contrasts is presented. In order to get different contrasts, two integrating spheres are used to illuminate the face and back of the test target uniformly. The target luminance and background luminance of the test target are regulated by luminance control parts conveniently. The MTF measurement system is designed and developed. Many experimental results of show that the luminance differences between the values by the system and those by L88 standard level luminance meter are within ±0.3 cd/m 2 so that the MTF measurement precision can be ensured. MTFs of Sony camera and Cannon camera at different contrasts are measured. The measurement values imply that MTFs at different contrasts can evaluate the imaging quality fully and objectively. This study provides an effective method to assess the imaging quality of visible imaging systems. DOI: http://dx.doi.org/10.11591/telkomnika.v11i2.1996

In the light of the DC component existing in active power filter (APF) based on traditional one-cycle control strategy, one-cycle control under double-direction compensation control strategy is presented. One-cycle control under double-direction compensation shunt APF is designed, which is used in three-phase three-wire system. APF with resistive inductive load is simulated by PSCAD/EMTDC. Simulation results show that the proposed control method not only can eliminate the harmonic and reactive current effectively but also can eliminate the DC component. This control method provides theoretical and experimental guide for the application of APF.

Aiming at serious harm caused by reactive power of power grid, the reason why SVG can compensate the reactive power is analyzed. Direct current control based on the detection method of source current is selected. SVG with resistive inductive load is simulated by Simplorer. Operational performance is analyzed in detail. The simulation result verifies that if SVG and power grid are connected in parallel and the appropriate algorithm is adopted, there is only active power in power grid. SVG can provide needed reactive current of working load. Compensation time can change by alternating the certain system parameters. The result further supports the feasibility and efficiency of this control method.

In order to make the inverter system has a better control flexibility and stronger interference immunity, an inverter based on digital control is created. This design can improve the output stability through controlling the instantaneous value of voltage loop. Meanwhile considering dynamic output performance, digital PI control is applied to the closed-loop system. A simulation and an operation are designed to confirm that. Firstly a simulation model of the single-phase digital inverter is established on Saber simulation platform to make a simulation for PI closed-loop strategy which is used in the inverter. Besides that a hardware circuit is set up to bring about the closed-loop design of single-phase digital inverter. The results of simulation and operation show that output waveform is better and this system has stronger interference immunity. It proves that single-phase digital inverter by PI control can be easily controlled. What's more it's convenient to upgrade and maintain this inverter.

Enhanced ground proximity warning system (EGPWS) is very important airborne system for flight safety. The structure of awareness alerting module in EGPWS, the design of terrain database and alerting algorithm are researched in this paper. The mathematical model of awareness alerting is built and algorithm based on terrain database is proposed. Additionally, Simulation system of EGPWS is designed and fulfilled based on the software package with high fidelity flight tool. Awareness alerting algorithm is test by the simulation system. Those discussed in the paper are expected to lay a good foundation for the development in EGPWS.

This paper has proposed an automatic precise alignment coupling algorithm of planar optical waveguide devices based on polynomial fitting, and the experiment shows that for 1 × 8 planar optical waveguide splitter, the time it takes for alignment is within 2.5 min and for 1 × 16 planar optical waveguide splitter, the required time is within 3 min. For 1 × 8 planar optical waveguide splitter, many repeated experiments can guarantee that the insertion loss of the device channels is less than 10.5 dB, with the maximum uniformity of 0.67 dB; for 1 × 16 planar optical waveguide splitter, many repeated experiments can guarantee that the insertion loss of the device channels is less than 13.7 dB, with the maximum uniformity of 0.66 dB. The alignment velocity, precision, and reliability of this algorithm are all greatly increased.

With the increasing digitization of the healthcare industry, a wide range of devices (including traditionally non-networked medical devices) are Internet- and inter-connected. Mobile devices (e.g. smartphones) are one common device used in the healthcare industry to improve the quality of service and experience for both patients and healthcare workers, and the underlying network architecture to support such devices is also referred to as medical smartphone networks (MSNs). MSNs, similar to other networks, are subject to a wide range of attacks (e.g. leakage of sensitive patient information by a malicious insider). In this work, we focus on MSNs and present a compact but efficient trust-based approach using Bayesian inference to identify malicious nodes in such an environment. We then demonstrate the effectiveness of our approach in detecting malicious nodes by evaluating the deployment of our proposed approach in a real-world environment with two healthcare organizations. Refereed/Peer-reviewed