Search

In some applications, such as surveillance and biometrics, image enlargement is required to inspect small details on the image. One of the image enlargement approaches is by using convolutional neural network (CNN)-based super-resolution construction from a single image. The first CNN-based image super-resolution algorithm is the super-resolution CNN (SRCNN) developed in 2014. Since then, many researchers have proposed several versions of CNN-based algorithms for image super-resolution to improve the accuracy or reduce the model’s running time. Currently, some algorithms still suffered from the vanishing-gradient problem and relied on a large number of layers. Thus, the motivation of this work is to reduce the vanishing-gradient problem that can improve the accuracy, and at the same time, reduce the running time of the model. In this paper, an enhanced dense space attention network (EDSAN) model is proposed to overcome the problems. The EDSAN model adopted a dense connection and residual network to utilize all the features to correlate the low-level feature and high-level feature as much as possible. Besides, implementing the convolution block attention module (CBAM) layer and multiscale block (MSB) helped reduce the number of layers required to achieve comparable results. The model is evaluated through peak signal-to-noise ratio (PSNR) and structural similarity index measure (SSIM) metrics. EDSAN achieved the most significant improvement, about 1.42% when compared to the CRN model using the Set5 dataset at a scale factor of 3. Compared to the ERN model, EDSAN performed the best, with a 1.22% improvement made when using the Set5 dataset at a scale factor of 4. In terms of overall performance, EDSAN performed very well in all datasets at a scale factor of 2 and 3. In conclusion, EDSAN successfully solves the problems above, and it can be used in different applications such as biometric identification applications and real-time video applications.

To improve the quality of ghost image, we propose an efficient computational ghost imaging method in this article. The primary idea is to estimate the noise value of the ghost image by analyzing and eliminating the source of the noise. The innovativeness of this work lies in analyzing a new means of noise by dissecting the qualitative relationship between transmittance in different objects and speckle patterns. While using a scale factor to describe the change of transmittance at different points of the object. The simulation and experimental results prove the effectiveness and feasibility of the proposed method through two parallel experiments. Compared to other methods, the peak signal-to-noise ratio and contrasts both have significantly increased.

In this paper, we constructed a new rational fractal interpolation model with scale factors and shape parameters. The model obtains different expressions by changing the scale factor, according with the diversity of image features. On the basis of the constructed model, this paper presents a novel adaptive rational fractal magnification (ARFM) algorithm based on local fractal feature analysis. Firstly, the image is divided into different regions according to an adaptive threshold determined by a calculated fractal dimension, and the scaling factor is calculated based on the relationship between global fractal dimension (GFD) and local fractal dimension (LFD). Secondly, for different regions, different interpolation forms are selected according to regional characteristics. Thirdly, parameter optimization and sub-block selection are studied to further enhance the quality of the magnified image. The experimental results illustrate that the performance of the proposed ARFM algorithm is very competitive compared with the latest magnification algorithms.

The instability of scale factor and null shift are closely related to the change of the light intensity of a total reflection prism laser gyro (TRPLG). To improve the light intensity stability, the light intensity stabilization principle of TRPLG was analyzed, and a mathematical model of light intensity stabilization system was established. Aiming at the problem that original system has a long adjustment time, a series correction method of adding a proportional integral differential (PID) link in the forward channel was adopted. According to the third-order optimal design, the original I-type third-order overdamped system was optimized to the new II-type third-order underdamped system, which ultimately improved the system's dynamic and steady-state performance. Feedforward compensation was used to introduce the light intensity disturbance during the longitudinal mode hopping into the closed-loop system, which realized the full compensation of the light intensity error and improved the light intensity characteristics during the longitudinal mode hopping. Combined with the above two methods, experimental results showed that the new system with feedforward compensation improved the light intensity stability by 32% at constant temperature and by 40% at variable temperature, with the TRPLG's bias stability improved by more than 16% at constant temperature and by 22% at variable temperature.

In this study, the hourly power output of Wind Power Plants (WPP) to be used as input in long term Transmission System Planning (TSP) has been examined. In this context, Weibull Distribution approach, which is one of the probabilistic approaches generally used in the wind speed analysis, is used. Using historical hourly wind speed data; average wind speed, most probable wind speed, maximum wind speed, probability density function, cumulative distribution function, power density and wind power output for the Nordex N60 model are obtained for each hour of the planning period. Maximum Likelihood Method(MLM) and Energy Pattern Factor Method(EPFM) are used to obtain the parameters (Weibull Shape Factor, K and Weibull Scale Factor, C) belonging to the Weibull Distribution. In addition, the Root Mean Square Error (RMSE),one of the error analysis method, error analysis is done to find out which of the Weibull parameters’ calculated methods will be suitable for the wind speed analysis.

In order to improve the radio frequency stealth capability of satellite communication signals, a multiple parameters weighted fractional Fourier transform (MP-WFRFT) signal design method based on joint optimization of target signal characteristics and key parameters is proposed. This method takes the constellation characteristics of target signal as the optimization objective, selects the fission number, fission configuration, fission pattern and ambiguity degree of constellation as the key parameters, takes the modulus scale factor characteristics and the transformation order threshold as the signal characteristic constraints, and introduces genetic algorithm for iterative calculation to obtain the optimal control parameters of MP-WFRFT. In order to verify the effectiveness of the proposed method, the MP-WFRFT joint optimization method is simulated, and the software radio hardware platform is built to generate the real target modulation characteristic analog signal. The simulation and experimental results show that the MP-WFRFT signal design method based on joint optimization of target signal characteristics and key parameters proposed in this section can achieve the optimal simulation of target signal modulation characteristics, and verify the effectiveness and correctness of the method.

MEMS being the most crucial part in advancement of today's technology plays as the backbone for decrement in size of new electronic devices. They are widely used in pedometry, quad-copters, stabilizing platforms, Inertial Measurement Units (IMUs), Inertial Navigation Systems(INS), robotics and medical field etc. In this research paper we have focused on; bias offset, scale factor error, non-orthogonality, drift due to temperature. Because these errors are the major error sources in MEMS sensors. Different techniques like two position method, six position direct method and six position weighted least square method are used to calibrate MEMS sensors for these errors. Calibration data was collected by the help of CNC machines and a platform made of aluminum to mount the sensors easily. Calibration techniques are implemented on ADXRS649z (single axis gyroscope), MPU-3050 (Three axis digital gyroscope) and ADXL377 (single axis accelerometer) sensors. Temperature compensation is also done to compensate the temperature effects on sensors using controlled temperature chamber and finally, a comparison is made between calibration techniques.

In order to improve the ability of BBPSO to jump out of the local optimal solution and avoid it falling into the local optimal solution, this paper proposes the BBPSO with linear variable weights. First, a linearly changing adjustment factor β is introduced into the iterative formula to balance the exploration and exploitation. Secondly, the scale factor D is introduced for the fast convergence of BBPSO, so that the adjustment factor will decrease to its minimum in the first half of the iterative process. The experimental results show that BBPSO with linear variable weight has better global optimality and can effectively balance the exploration and exploitation.

By processing the wind tunnel virtual flight test data, the movement process of the fly-by-wire civil aircraft can be realistically simulated. This paper establishes the dynamic model of the fly-by-wire civil aircraft wind tunnel virtual flight, designs the aircraft's longitudinal and lateral heading control laws, and verifies the correctness of the test results by comparing the simulation data with the flight test data. The digital filter is used to eliminate the noise of the test data, and the data compatibility test is performed by the output error method to eliminate the constant error and the scale factor error in the flight test data. A set of virtual flight data processing methods for fly-by-wire civil aircraft in wind tunnels has been formed.

Many applications in our daily life need accurate orientation and motion information. The Inertial Measurement Unit (IMU) is a perfect sensor for that purpose, because of its small size, low power consumption, and affordable price. These low-cost IMUs suffer from several sources of noise, that cause large divergence over time during the absence of an absolute positioning system. In addition to its poor performance, low-cost IMUs do not have reliable noise metrics data information in their vendor specification. Multiple IMUs can be utilized to enhance the overall system performance through fusion. As gyroscope errors have the main contribution in deciding the navigation system performance, this paper proposes two adaptive weighting least square approaches. During fusion, the weights of the LS filter are changed online according to two windowing methodologies, these windowing approaches estimate the bias, noise density, and scale factor level for each sensor measurements inside this window of time, which is then used to estimate the variance accompanied by each sensor that will be used as a weight inside the LS algorithm, these weights are adaptively changing with each period. The second methodology is also based on two different size window techniques, with cascaded LS filter topology. MATLAB navigation toolbox is utilized to generate two gyroscope IMUs row data with different noise configurations like noise density, random walk, and bias instability rates…etc. to imitate real-life raw measurements. These raw measurements are then used to validate and evaluate the two proposed algorithms, by comparing their RMSE to a standard LS method RMSE. The simulation was repeated 100 times and the RMSE of the proposed methodologies showed enhancement in the estimated measurements by more than 20% when compared to the standard LS, under the condition that the noise figures of the sensors are completely random.

Switched reluctance motor (SRM) is usually in magnetic saturation state and has strong nonlinearity. For some traditional control algorithms, fixed PID parameters can not adapt to all kinds of working conditions, which will also lead to large torque ripple. Therefore, an improved fuzzy control based on torque sharing function (TSF) is presented to improve the dynamic response performance of the speed controller and reduce the torque ripple. This method combines the fuzzy control with the PID control method, and adjusts the proportion factor automatically. Firstly, when the speed error is big, the system adopts fuzzy control; when the speed error is within the set error range, the system adopts PID control, which can eliminate the blind area of the speed controller. Secondly, the speed scale factor and torque scale factor in fuzzy control are self-tuning, according to the characteristics of output torque, the fuzzy rules of torque scale factor are designed. Finally, in order to test the correctness of the presented method, it is compared with the PID and traditional fuzzy methods. Simulation results confirm that the presented scheme can not only improves the response speed of the system, but also decreases torque ripple.

The principle of vibration mode localization has been been researched in recent years as an approach towards the high accuracy transduction of inertial forces. This transduction scheme comprises of two or more weakly coupled resonators and a low coupling factor is critical in order to enhance parametric sensitivity and consequently resolution. In this paper, we report a new mode-localized accelerometer topology which enables a high amplitude ratio scale factor using a differential acceleration perturbation method. Further, by employing a parametric pump, an adjustable ultra-weak virtual coupling factor is realized where the coupled resonators can operate in the weak coupling regime. In this regime, modal overlap is apparent and the amplitude ratio sensitivity is much improved, by decreasing the pump strength. The experiments show that the amplitude ratio sensitivity of the proposed accelerometer is 9.2/g, and it can be enlarged to as high as 6300/g by employing a parametric pump signal with 100 mV strength, indicating a scale-factor improvement factor of 684. Consequently, the input-referred bias instability is improved from 6.7 µg for the case of no parametric pump to 420 ng while employing a parametric pump for operation in the modal-overlap regime, showing an improvement by a factor of ~16.

We report, for the first time, a robust angular rate sensor that is operated at 2:1 Autoparametric Resonance (AR) with a wide frequency bandwidth at 3dB amplitude drop of 320.7 Hz. The sensor utilizes inherent forcing and inertial or elastic nonlinearities arising from electrostatic forces and fabrication imperfections respectively, to excite the sense mode via 2:1 AR causing a wider frequency response of the sense mode. The sensor is actuated electrostatically, and its output is sensed using variable gap capacitive electrodes. The sensor is tested on a rate table and a maximum scale factor of $38.99\ \mu \mathrm{V}/^{\circ}/\mathrm{s}$ with a full-scale nonlinearity of 1.2%, dynamic range of ±270 °/s, and noise density of 0.042 °/s/√Hz are measured.

We propose a method for improving the temperature stability of the scale factor for silicon resonant accelerometer (SRA) based on crystallographic orientation optimization for the first time. The approach is a passive compensation per se that changes the temperature drift of Young's modulus via rotating the vibration axis of the device to a specific crystal orientation angle, thereby significantly reducing the temperature coefficient of the scale factor (TCS). Experimental results show that a TCS of 3ppm/°C is obtained with a 33.75° orientation SRA without any post-processing compensation, which is ∼8.6-fold reduced compared with the conventional 0° orientation device and show good agreement with the simulation results. Moreover, the overall variation of the scale factor can be further decreased to only 2.98ppm from −10 °C to 70 °C after simple polynomial compensation, which is a considerably competitive result among the previously reported results in literatures.

The paper presents an effective method for scale-factor stabilization and modulation depth control in a low-cost all-fiber interferometric gyroscope utilizing a piezoelectric phase modulator to perform modulation for biasing as well as compensation of the Sagnac phase shift in a closed-loop. The technique uses a virtual-sensor approach consisting in deriving the actual compensation shift from comparison of the real and the modeled gyroscope output intensity signals. The first experimental results are provided, suggesting that the model-base approach can decrease the scale-factor drift by more than one order of magnitude to reduce significantly the effect of the modulator parameter instability.

1-bit detectors show great promise for resource-constrained embedded devices but often suffer from a significant performance gap compared with their real-valued counterparts. The primary reason lies in the error during binarization. This paper presents a layer-wise searching (LWS) strategy to generate 1-bit detectors that maintain a performance very close to the original real-valued model. The approach introduces angular and amplitude loss functions to increase detector capacity. At 1-bit layers, it exploits a differentiable binarization search (DBS) to minimize the angular error in a student-teacher framework. We also learn the scale factor by minimizing the amplitude loss in the same student-teacher framework. Extensive experiments show that LWS-Det outperforms state-of-the-art 1-bit detectors by a considerable margin on the PASCAL VOC and COCO datasets. For example, the LWS-Det achieves 1-bit Faster-RCNN with ResNet-34 backbone within 2.0% mAP of its real-valued counterpart on the PASCAL VOC dataset.

SLAM systems can retrieve their metric scales and depth information using RGB-D cameras. However, limited by the sensing range and objects structure, RGB-D cameras can not always work well, resulting in failures sometimes. In this work, we present initialization and localization methods based on maximum-a-posteriori estimation. Our system endows monocular keypoints with valid depth values and introduce them into bundle adjustment. Depth bias coefficient and scale factor are also optimized in the local window, obtaining robustness in large scale environments and long-running operations. The experimental results indicate that our system provides the best robustness compared with other excellent methods in the literature, being able to process the most challenging sequences in the TUM RGB-D dataset.

The impulse current is a transient current waveform, which has a short duration and cannot be repeated, so it is difficult to evaluate the influencing factors of measurement errors. This paper studies the traceability method and uncertainty evaluation model of impulse current measurement device, and the influencing factors of measurement error are analyzed. The impulse calibrator which can output standard double exponential impulse current waveform with specific time parameters (8/20) is developed. The peak value and time parameters can be traced back to the national standards of DC voltage and capacitance, inductance and resistance. The measuring system consists of standard impulse shunt and measuring instrument, whose peak current and time parameters can be traced respectively. The impulse scale factor of the impulse current measuring system is the product of the impulse scale factors of shunt and digital recorder. Finally, the performance of the measuring system is evaluated, and the expanded uncertainty of scale factor and time parameter is 0.4% (k=2) and 0.8% (k=2) respectively.

This paper determines the scale factor for an electromagnetic groundwater flowmeter using an air-cored square pancake coil as the electromagnet. It also proposes the use of an array of electrodes to provide a better estimate of the average water speed between the electrodes.

The new circuit of fourth-order band-pass filter (BPF) has been developed with independently adjustable parameters - non-uniformity of the amplitude-frequency characteristic (AFC), bandwidth and transmission scale factor. The basic mathematical expressions for transfer function coefficients of the BPF structure and their parameters of AFC are presented. The AFCs was show from program Micro-Cap.

Aiming at the problem that the positioning accuracy of the KCF tracking algorithm is difficult to reach the pixel level and cannot adapt to the target scale variation well, the correlation filter tracking algorithm based on the scale pyramid achieves higher tracking accuracy, but the tracking speed is greatly reduced. Introducing the logarithmic polar coordinate transformation of the image, an object tracking algorithm based on the logarithmic polar coordinate transformation is proposed. First, the target template is transformed into the logarithmic polar coordinate, and the scale variation of the target is converted into a displacement signal, then extract the HOG features before and after the target template transformation, and the filter model of displacement and scale is established. Finally, the displacement and scale factor of the object are tracked synchronously under the framework of correlation filtering, and the two are merged to obtain the target tracking frame. The experimental results show that: The average overlap precision of the algorithm in this paper is high, and the tracking effect is better (Experiment 1). The algorithm in this paper can track rigid objects stably and can adapt to scale variation accurately very well (Experiment 2). The overall accuracy and success rate are in the first place (Experiment 3). The algorithm in this paper can approximately achieve pixel-level positioning accuracy, and the tracking speed can reach twice the traditional algorithm (Experiment 4).

Using sound features to complete the monitoring of automobile engine condition is not only less in operation process and cost, but also less affected by weather and other factors. However, the collected signals will inevitably be polluted by other sounds and become noisy signals, which will affect the output of the research results. Therefore, the purpose of this paper is to extract and identify all kinds of signals in the sound signal, so as to provide effective data for a signal to be used in the acoustic monitoring of engine state. In this paper, VMD algorithm is used to decompose two kinds of recorded mixed sound signals. The number of decomposition layers of the two signals is determined by calculating the mean of center frequency and instantaneous frequency respectively, and multiple intrinsic modal components are obtained. The type of each component is determined by calculating the correlation coefficient method. Then, MPE value was calculated by selecting the appropriate embedded dimension and scale factor by combining MPE. Observe the result of feature extraction. Finally, each component is classified and recognized by SVM, and the experimental results show that the method mentioned in this paper is accurate and effective for separating and recognizing mixed sound event signals.

In this paper, a novel image watermarking method is proposed which is based on discrete wave transformation (DWT), Hessenberg decomposition (HD), and singular value decomposition (SVD). First, in the embedding process, the host image is decomposed into a number of sub-bands through multi-level DWT, and the resulting coefficients of which are then used as the input for HD. The watermark is operated on the SVD at the same time. The watermark is finally embedded into the host image by the scaling factor. Fruit fly optimization algorithm, one of the natural-inspired optimization algorithms is devoted to find the scaling factor through the proposed objective evaluation function. The proposed method is compared to other research works under various spoof attacks, such as the filter, noise, JPEG compression, JPEG2000 compression, and sharpening attacks. The experimental results show that the proposed image watermarking method has a good trade-off between robustness and invisibility even for the watermarks with multiple sizes.

A data processing method for correcting the frequency sweeping nonlinearity of a tunable laser in an optical frequency domain reflectometry system is presented. The proposed method calibrates the positions of all the data points by the time-scale factor determined by the zero-crossing points of the auxiliary interferometer. Then, the signal of the main interferometer is interpolated with the calibrated results. In experiments, a spatial resolution of 0.17 mm over a measurable range of 155 m has been obtained even when the optical path difference of the auxiliary interferometer is shorter than that of the main interferometer. This method could resolve the problem that measurable range is limited by the length of delay fiber in the auxiliary interferometer and significantly decrease the requirement of sampling rate.

Due to the diversity of redundant configuration schemes, the existing calibration algorithm for the redundant inertial measurement unit (RIMU) cannot be applied generally. In addition, the linearized calibration model utilized in the traditional calibration algorithms also leads to low calibration accuracy. In this paper, a high-precision calibration scheme is proposed for the RIMU, which can be applied generally. At first, a general calibration model for the RIMU is established. Then, a zero-space amplification algorithm is applied to improve the observability of the system. At last, the extended Kalman filter (EKF) based on the second-order Taylor series expansion is proposed, which can estimate the biases, scale factor errors, and installation errors simultaneously. The simulation and experimental results demonstrate that the proposed calibration algorithm calibrates the system accurately and improves the performance of the RIMU effectively. Furthermore, the proposed algorithm can be applied generally in all kinds of RIMU.

The paper presents a miniaturized pitch/roll gyroscope with the lowest angle random walk (ARW of 600 $\mu \textbf{dps}/\sqrt{\textbf{Hz}})$ and lowest bias instability (BI of 2.8 dph) ever recorded on planar MEMS sensors for in-plane rate capture. The device is based on NEMS resistive sensing, features a 1.2 mm2footprint, a 100-Hz bandwidth, and can be combined to yaw gyroscopes for ultra-low-noise 3-axis systems. The key advancement over previous implementations is represented by a 10-fold increase in the scale factor, obtained through a novel architecture which, combined with quadrature compensation, optimizes at the same time the energy transfer to the NEMS gauges and the robustness to vibrations.

This paper reports a new Quartz vibrating sensor developed by iXblue to respond to new challenges for navigation applications as high dynamic range (100g), navigation grade performance (< 100μg and 30 ppm bias and scale factor respectively) according with the challenge of Size, Weight and Power (SWaP) consumption reduction (less than 10 grams and 50mW).

This paper reports on the design, implementations and preliminary characterization of an ultra-sensitive frequency modulated (FM) resonant accelerometer. The accelerometer consists of a pair of piezoelectrically actuated clamped-clamped beam resonators implemented in the device layer of an SOI wafer, and a proof-mass with mechanical coupler and force amplifier defined in both the device and handle layer of an SOI substrate. The device is designed to detect minimum acceleration in the range of 10s of nano-g while having a linear upper range of 25g with 0.5% of scale factor nonlinearity. Fabricated devices measure 660 nano-g/✓Hz VRW and 2µg BI using discrete electronics with projected bandwidth of 260Hz.

This work reports a novel design for high-frequency piezoelectric gyroscopes using support transducer which enables an efficient energy coupling from a length-extensional mode into a degenerate secondary elliptical mode (n = 3). The gyroscope structure is made by a single crystalline silicon (SCS) disk supported by eight-transducer arms. Large drive mode displacement is excited through piezoelectric AlN thin films on transducer arms. The simulation of the proposed gyroscope shows a high scale factor of 12.24 nA/°/s at resonator Q of 10,000 in a mode-matched condition. The device was fabricated using the PiezoMUMPs process with 10 μm thick silicon structural layer. The fabricated gyroscope shows that the frequency split between drive and sense mode is only 600 Hz at the mean resonant frequency of about 9.976 MHz in air (equivalent to 60 ppm), verifying our design efficacy. The measured scale factor of the proposed gyroscope is around 54.01 μV/°/s before demodulation.

A mode-matching and alignment algorithm for bulk acoustic wave (BAW) resonator gyroscopes via femtosecond laser ablation is presented in this paper and verified on a piezoelectrically-actuated AlN-on-Si disk resonator. Mode-matching is essential to operate piezoelectric BAW resonators as gyros with high sensitivity in air but challenging due to the lack of electrostatic spring-softening effect. In this work, a laser trimming algorithm is implemented to mechanically mode-match and align the two resonant modes of a disk resonator guided by novel numerical maps. After selective laser ablation trimming, the as-born frequency split of 191Hz between 1.4MHz modes of a disk due to fabrication imperfections is reduced to 12Hz, corresponding to 3.8% of its -3dB resonant bandwidth while maintaining a Q-factor of 4600 in air, and 30dB improvement in mode isolation. The laser-trimmed frequencies remain closely matched across the temperature range of -30°C to 90°C, verifying the temperature robustness of laser trimming results. The mode-matched disk resonator is operated as a gyroscope in air and exhibits a large scale factor of 4.12nA/°/s.

Stereo super-resolution (SR) aims to enhance the spatial resolution of one camera view using additional information from the other. Previous deep-learning-based stereo SR methods indeed improved the SR performance effectively by employing additional information, but they are unable to superresolve stereo images where there are large disparities, or different types of epipolar lines. Moreover, in these methods, one model can only super-solve images of a particular view, and for one specific scale factor. This paper proposes a cross parallax attention stereo super-resolution network (CPASSRnet) which can perform stereo SR of multiple scale factors for both views, with a single model. To overcome the difficulties of large disparity and different types of epipolar lines, a cross parallax attention module (CPAM) is presented, which captures the global correspondence of additional information for each view, relative to the other. CPAM allows the two views to exchange additional information with each other according to the generated attention maps. Quantitative and qualitative results compared with the state of the arts illustrate the superiority of CPASSRnet. Ablation experiments demonstrate that the proposed components are effective and noise tests verify the robustness of CPASSRnet.

Recently, self-supervised method has become an increasingly significant branch of depth estimation task, especially in the realm of autonomous driving applications. However, current per-pixel depth maps yielded from RGB images still suffer from uncertain scale factor generated by the nature of monocular image sequences, which further leads to the insufficiency in practical use. In this work, we first analyze such scale uncertainty both theoretically and practically. Then we perform scale recovery utilizing geometric constraint to estimate accurate scale factor, RANSAC(Random sample consensus) outlier removal is introduced into pipeline to obtain accurate ground point extraction. Adequate experiments on KITTI dataset(dataset generated by an autonomous driving platform built up by KIT and TRINA comprising stereo and optical flow image pairs as well as laser data, distributed to train set and test set on account of deep learning), show that, using only camera height prior, our proposed method, though not relying on additional sensors, is able to achieve accurate scale recovery and outperform existing scale recovery methods.

Nowadays, multimedia elements such as image, video, animation, audio, and software can be distributed in a short time anywhere in the world using internet technology. Content owners are concerned about copyright protection and authentication. Watermarking is one of the method to protect patient information in medical images such as magnetic resonance and ultrasound imaging. In the literature, there are several methods proposed using both frequency-domain (etc. digital cosine transforms (DCT), digital wavelet transforms (DWT), digital Fourier transforms (DFT), digital radon transform (DRT)) and spatial domain (least significant bits (LSB)) methods. Mostly digital wavelet transform based watermarking methods give very promising results after common attacks. In the DWT algorithm, scaling factor is used in embedding and extraction. A scaling factor is a number between 0 and 1 which has been determined by the user which is not efficient. We can use the brute force method which solves a problem by checking all the possible cases, but it is slow. In this work, we use simulated annealing heuristic search algorithm to find out the best scaling factor to reach more robust, transparent, high data capacity and resistant watermarking method in medical images. Experimental results show that simulated annealing-based scaling factor determination in frequency domain watermarking gives more robust, resistant, and transparent watermarked images.

In this paper, we study the influence of ill-conditioned regression matrix on two hyper-parameter estimation methods for the kernel-based regularization method: the empirical Bayes (EB) and the Stein’s unbiased risk estimator (SURE). First, we consider the convergence rate of the cost functions of EB and SURE, and we find that they have the same convergence rate but the influence of the ill-conditioned regression matrix on the scale factor are different: for upper bounds, the scale factor for SURE contains one more factor cond(ΦTΦ) than that of EB, where Φ is the regression matrix and cond(•) denotes the condition number of a matrix. This finding indicates that when Φ is ill-conditioned, i.e., cond(ΦTΦ) is large, the cost function of SURE converges slower than that of EB. Then we consider the convergence rate of the optimal hyper-parameters of EB and SURE, and we find that they are both asymptotically normally distributed and have the same convergence rate, but the influence of the ill-conditioned regression matrix on the scale factor are different. In particular, for the ridge regression case, we show that the optimal hyperparameter of SURE converges slower than that of EB with a factor of 1/n2, as cond(ΦTΦ) goes to ∞, where n is the FIR model order.

Inertial Navigation System (INS) is used in a variety of applications such as missile and marine navigation. INS is comprised of an inertial measuring unit (IMU) and a processor unit that performs the navigation mathematics calculations. In order to get accurate navigation data via INS, calibration of IMU sensors are necessary. Mechanical sensors are used in high-precision navigation, such as strategic missiles and are characterized by their low random noise. However, deterministic errors can cause very significant errors in positioning such as biases, scale factors, non-orthogonality errors, and g-sensitive and non-g-sensitive drifts. Therefore, determining an effective and accurate method of calibration is necessary to estimate and compensate for these errors. The suggested methodology accurately calculates the calibration parameters to minimize those errors. The traditional calibration techniques of the accelerometer are restricted in precision, since the estimation of scale factor and bias obtained from a limited number of positions depending on the gravity direction. These calibration techniques are also limited in the estimation of g-sensitive and non-g-sensitive drifts impacting gyro performance. This paper presents a new technique of calibration that overcomes these drawbacks, based on an enhancement of multi-position technique. Experimental results for the proposed technique was carried out to confirm its efficiency.

With the development of deep convolutional networks in target detection, deep convolutional networks are developing in the direction of deeper parameters with greater amounts. In order to reduce the parameters and inference time of the target detection network model, a pruning algorithm combining channels and layers is proposed and used for pedestrian detection. Through L1 regularization of the channel scale factor, the channels of the convolutional layer become sparse. The channels and shortcut layer with less information are pruned, so that the model is compressed. Pruning the YOLOv4 model based on this principle. Experiments prove that sparse training is important for model pruning. Through comparative experiments, the pruned YOLOv4 model can better detect targets on the INRIA dataset. Compared with the original YOLOv4, although there is a small loss of accuracy, the volume, parameter amount and reasoning time of the pruned model are greatly reduced.

A 500kV wide-band high-voltage divider for transient overvoltage measurement was developed. To improve the accuracy of the overvoltage measuement, the high-voltage arm of the developed voltage divider is based on a standard capacitor. By arranging equalizing electrodes reasonably inside the standard capacitor, the field strength distribution inside the voltage divider is optimized, and the overall size of the voltage divider is reduced. In order to optimize the dynamic performance of the voltage divider, it is equipped with excellent low-voltage arm components. The consistency of scale factor under AC voltage and impulse voltage for the divider is good. It can measure AC voltage with linearity of 0.2‰ under 500kV and measure lightning impulse voltage with linearity of 3‰ under 1200kV. The difference between AC voltage scale factor and impulse voltage scale factor is less than 0.01%. It can be used to measure transient overvoltage in substation and calibrate the noncontact transient voltage sensor and electric field sensor for field test.

In recent years, self-supervised methods for monocular depth estimation has rapidly become an significant branch of depth estimation task, especially for autonomous driving applications. Despite the high overall precision achieved, current methods still suffer from a) imprecise object-level depth inference and b) uncertain scale factor. The former problem would cause texture copy or provide inaccurate object boundary, and the latter would require current methods to have an additional sensor like LiDAR to provide depth ground-truth or stereo camera as additional training inputs, which makes them difficult to implement. In this work, we propose to address these two problems together by introducing DNet. Our contributions are twofold: a) a novel dense connected prediction (DCP) layer is proposed to provide better object-level depth estimation and b) specifically for autonomous driving scenarios, dense geometrical constrains (DGC) is introduced so that precise scale factor can be recovered without additional cost for autonomous vehicles. Extensive experiments have been conducted and, both DCP layer and DGC module are proved to be effectively solving the aforementioned problems respectively. Thanks to DCP layer, object boundary can now be better distinguished in the depth map and the depth is more continues on object level. It is also demonstrated that the performance of using DGC to perform scale recovery is comparable to that using ground-truth information, when the camera height is given and the ground point takes up more than 1.03\% of the pixels. Code is available at https://github.com/TJ-IPLab/DNet., Comment: 8 pages, 10 figures, accepted by 2020 IEEE/RJS International Conference on Intelligent Robots and Systems(IROS)

With the development of quantum control technology, atomic gyroscopes based on the Spin Exchange Relaxation Free (SERF) have been expected to be the development trend for the next generation of high-precision gyroscopes. As a crucial parameter of the SERF atomic spin gyroscope, the temperature stability of the alkali vapor cell determines the stability of the gyroscope scale factor. Therefore, high-precision temperature control of the alkali vapor cells is important to achieve an accurate measurement by an atomic gyroscope. Besides, since the atoms in the cell are very sensitive to the low-frequency magnetic field noise, it is necessary to ensure that the low-frequency magnetic field noise is not introduced during the heating process. In order to meet these requirements, a non-magnetic temperature control system of the alkali vapor cells based on the STM32 and Labview is designed in this paper. The proposed system is verified by experiments. The experimental results show that the equivalent interference magnetic field of the proposed system is less than 50 fT/ Hz 1/2, temperature measurement accuracy is 2 mK; and near the operating temperature of 180°C, the temperature stability is ±3 mK.

Laser gyroscope is a kind of high precision angular sensor which has been widely used in strapdown inertial navigation system. The cavity length of laser gyroscope varies with the external temperature changes, which have influence on the scale factor stability and bias stability. In this regard, to control cavity length of laser gyroscopes with higher precision and reliability, a precision cavity length control system with self-adaption temperature compensation method is designed. The main motivation of this contribution is to design the higher accurate cavity length control system with self-adaption temperature compensation algorithm embedded. Experimental tests show the significant performance and effectiveness of cavity length control system proposed.