Search

This paper introduces a novel method for accurately estimating the 2D coordinates of hand keypoints from single static images, utilizing a sequential convolutional neural network optimized with Maximum Likelihood Estimation Loss. Unlike traditional heatmap-based techniques, our approach eliminates the need to generate label heatmaps and sidesteps the direct optimization of model parameters based on noisy labels. Instead, it concentrates on modeling the distribution of the discrepancies between predicted results and ground truth, rather than the potential presence of noisy labels, thus enabling the direct prediction of hand keypoint coordinates. Furthermore, we propose a sequential training and inference framework that consists of a deep convolutional backbone network and a multi-stage sequential network. Each stage of this network features similar structures, facilitating the progressive and precise prediction of hand keypoint coordinates. Our extensive experimental results demonstrate that our approach is both highly accurate and robust, outperforming mainstream methods under the experimental conditions detailed in this paper.

FengYun-3E (FY3E) is the world's first dust-dawn orbiting meteorological satellite for civil use, which has filled the vacancy of global early-morning-orbit satellite observation by working together with FengYun-3C and FengYun-3D satellites. The Medium Resolution Spectral Imager-Low Light (MERSI-LL) sensor carried by FY3E can detect surface temperature variation. The radiometric calibration conditions of MERSI-LL thermal infrared bands were evaluated using the collected field measurements and atmospheric transfer simulations during 5–23 December, 2022, at Lake Erhai. A thermal infrared radiometer equipped on an unmanned surface vehicle was used to continuously collect surface-emitted radiance and water temperature. Atmospheric conditions, surface emissivity, and aerosol optical depth measured near-field experiment site were adopted by atmospheric radiative transfer code to calculate the influence of atmosphere on long-wave radiation during the propagation from land surface to satellite aperture. The good in-orbit operational status could be detected according to our calibration experiments, and the calibration analysis suggested that the differences between the simulated brightness temperature and satellite-based brightness temperature are 0.71 K with an RMSE of 0.79 K and 0.11 K with an RMSE of 0.47 K for channels 6 and 7, respectively, which achieved better or similar calibration accuracy by comparing with TIR channels of other FengYun satellites.

Flashover at gas/insulator interface acts as a vital challenge to the safe and stable operation of gas-insulated system especially when involved with surface metallic particle. In this paper, flashover characteristics at SF6/epoxy insulator interface were investigated concerning on the tracking property based on image analysis. DC flashover experiments were operated for basin-type spacer with metallic particle at various locations. Flashover voltage was obtained and tracking on the surface after successive flashover was captured. Improved image analysis method based on fractal theory was carried out to evaluate tracking pattern quantitatively. The relation between the fractal dimension of tracking and field distribution at particle tip was revealed. Obtained results showed that flashover voltage decreased due to severe field distortion with moving particle from ground to HV electrode. Meanwhile, fractal dimension of tracking covering the area from particle to ground electrode indicated centralized tendency. Simulation result confirmed this variation as the consequence of centralized field distribution at the tip of particle. The quantitative relation between fractal dimension and field distribution suggests new sight to estimate the influence of metallic particle on insulation property in the gas-insulated system.

The insulator in direct current gas-insulated transmission lines (DC-GIL) would suffer discharge risk due to surface charge accumulation under thermal-electric coupled fields. In this paper, the transient surface charge accumulation characteristics of a basin-type DC-GIL insulator is investigated via finite element method based on a three-dimension horizontally installed GIL model. The stationary temperature distribution of the model is obtained and then applied to the transient simulation of charge. Weak form partial differential equation is employed to deal with the ion transportation equation. Equations and parameters in the simulation are optimized to reduce the computing memory and time. Results indicate that the charge accumulation is accelerated due to the promotion of conduction through the insulator under thermal gradient. Higher charge density is obtained under thermal gradient. And the surface charge density of the convex surface is higher due to the promoted conduction. The highest field strength increases and the corresponding location moves along the convex surface during the transient process. This could attribute to the influence of transient charge behavior under thermal gradient on the electric field distribution. This study indicates that the thermal gradient and transient charge accumulation should be considered when dealing with the insulation characteristics of DC-GIL with insulators.

The goal of scene graph generation (SGG) is to classify objects and their pair-wise relationships in a visual scene. Object occlusion is a critical challenge when generating scene graphs in complex scenes. However, this issue has rarely been explored in recent works. Accordingly, in this paper, we propose a subset matching network (SM-Net) that handles the above problem. First, we decompose SGG into two types of subset matching problems: node subset matching and edge subset matching. Each node/edge subset handles the occlusion between one node/edge pair, thereby reducing the difficulty of SGG in a “divide and conquer” manner. Second, we introduce a node subset prediction module that utilizes a subset-based message passing module to refine the node subset representation and a matching loss to supervise node subset prediction. Third, we propose an edge subset prediction module that applies a feature selection-based fusion function to obtain edge subset features and a matching loss to supervise edge subset predictions. Experiments on three popular datasets show that our model achieves state-of-the-art performance. The code of SM-Net will be released.

With the popularity of video conferences, video calls and other activities, video transmission has been widely used. To meet a huge number of subscribers’ requirements, the mobile video transmission scheme needs to overcome some disadvantages, such as resources limitation and noise interference. The knowledge-enhanced mobile video broadcasting (KMV-Cast) is a scheme utilizing joint source-channel coding and correlated information in clouds. However, there is an item of noise that cannot be removed in the original KMV-Cast scheme. In this paper, an adaptive Wiener filtering denoising algorithm is proposed to reduce such noise at the receiver in order to maximize the signal-to-noise ratio (SNR) of the reconstructed video frame. The simulation results show that the proposed Wiener filter algorithm is superior to other schemes without the Wiener filter under different sources and channel qualities. At lower-SNR channels (i.e., −5dB), the proposed algorithm achieves 2dB gains in terms of peak signal-to-noise ratio (PSNR), while at higher-SNR channels (i.e., 10dB), the proposed algorithm achieves 3dB gains in terms of PSNR.

State of charge (SoC) imbalance and dc bus voltage deviations are significant issues for distributed battery energy storage systems in autonomous dc microgrid applications. Accordingly, a high-pass filter (HPF) based SoC balancing method is proposed to achieve SoC balance by considering different SoCs and capacities; A band-pass filter (BPF) based power droop control is used to accomplish power-sharing and voltage compensation. Through this approach, battery storage units (BSUs) with higher SoC and capacity deliver more power during discharging mode than those with lower SoC and capacity. During charging mode, those BSUs with higher SoC and lower capacity are controlled to absorb less power than those with higher capacity and lower SoC. Therefore, SoC balance can be achieved between distributed BSUs. Further, the dc bus voltage is maintained within the desired range by adopting the proposed method. In addition, the control method employed also considers the impact of line resistance. The proposed control strategy is implemented in a fully decentralized way which does not require any communication link, while maintaining system stability. MATLAB/Simulink and processor-in-the loop (PIL) simulation results verify that the proposed control strategy is effective and feasible.

In this study, a precision bandgap reference with a v-curve correction (VCC) circuit is presented. The proposed VCC circuit generates a correction voltage to reduce the temperature drift of the reference voltage and achieves a low temperature coefficient (TC) in a wide temperature range. The proposed bandgap reference was designed and fabricated using a standard TSMC 0.18-$\mu \text{m}$ 1P6M CMOS technology with an active area of 0.0139 mm2. The measured results show that the proposed bandgap reference achieves a TC of 1.9-5.28 ppm/°C over a temperature range of -40°C to 140 °C at a supply voltage of 1.8 V. In addition, the circuit demonstrated a line regulation of 0.033 %/V for supply voltages of 1.2 - 1.8 V at room temperature.

Image deformation has ubiquitous usage in multimedia applications. It morphs one image into another through a seamless transition. Existing techniques either mainly focus on the correspondence mapping of interior features of the objects in two images, without considering object contours, or sketch contours manually, resulting in tedious work for users. Thus, we propose a 2D image deformation method, which extracts object contours automatically, considers both inner features and contours as constraints and preserves image features in terms of visual importance. Our method first automatically extracts the object contours in the source and target images and then allows users to sketch some interior features in both the images. Then, our method tessellates two images to generate two triangular meshes and builds a guaranteed bijective mesh mapping between them. We also prove the bijectivity of our mesh mapping and discuss its other desirable properties. Then, our method generates the intermediate images between the source and target images by calculating the intermediate meshes and pixels of each intermediate image. Our method realizes automatic contour extraction, provides an intuitive user interface and utilizes harmonic maps to establish a bijective mesh mapping. Therefore, it preserves more significant features with less distortion and works well for many image deformation cases in real time.

To mitigate the contradiction between scarcity spectrum resources and heavily wireless services, cognitive radio (CR) has been proposed to improve spectrum utilization through allowing CR users to access licensed channels opportunistically. In large-scale networks, spectrum status is not the same at different locations due to the heterogeneity of CR network (CRN). To cope with such heterogeneity, some noncooperative spectrum sensing and distributed cooperative sensing algorithms were proposed. Such spectrum decision results will be exploited in this paper, in order to draw the spectrum map of the entire largescale CRN. The proposed spectrum mapping scheme contains three processing steps, and a boundary CR users searching algorithm using kennel function based supportive vector machine is adopted to improve the performance of the proposed scheme. The simulation results show that radial basis function kennel performs the best in the proposed scheme, and when accuracy threshold θα = 0.95 and filtration threshold θf = 0.02, the proposed scheme can draw a spectrum map with the accuracy 99.3% using only about 28% CR users performing spectrum sensing. Furthermore, the number of CR users and energy detection threshold have little effects on the performance of the proposed scheme.

Using the array technique, the multi-channel synthetic aperture radar (SAR) has the capability of high-resolution imaging, detection, and location of the ground moving targets, which is a powerful tool during the remote sensing of smart city. Recently, the theory of compressive sensing has been applied to radar imaging with data of compressive sampling, which can effectively reduce the burden of current radar system. In this paper, we focus on SAR moving targets imaging from sparse aperture (SA) data with accurate target motion compensation. The procedure of motion compensation is decomposed into two steps: the SPECAN processing in the range frequency and azimuth time domain is first applied and then followed by the residue component correction embedded into sparse imaging. To overcome the SA and target motion, a novel parametric sparse imaging approach is proposed by addressing the problems of Doppler ambiguity and phase errors. In the scheme, a parametric and dynamic dictionary is used to include these two important issues. Then, a modified orthogonal matching pursuit method is presented for high-quality imaging with Doppler ambiguity number estimation and phase error correction, which can deal with the case of multiple moving targets. Finally, experimental analysis is performed to confirm the effectiveness of the proposed algorithm.

Under the case of exponentially growth of wireless services and the scarcity of spectrum resources, cognitive radio (CR) has been proposed to access licensed channels opportunistically, and thus improve spectrum utilization. In CR devices, accurate spectrum sensing is the prerequisite for opportunistic access. The current cooperative spectrum sensing still cannot effectively exploit the temporal correlations among sensing data, especially the correlations between the current sensing data and the historical data. This paper uses sticky hierarchical Dirichlet process-hidden Markov model to exploit the historical sensing data of multiple users, and classifies the historical sensing data into groups according to their latent spectrum states. The proposed spectrum sensing algorithm can fuse the historical sensing data into prior knowledge, which can be used to improve the accuracy in spectrum decision. Furthermore, a rejection process is proposed to filter out some sensing data with high uncertainty in classification, which guarantees the effectiveness of historical sensing data. The simulation results show that the proposed algorithm performs the best, compared with other three typical cooperative spectrum sensing algorithms, in terms of detection probability and false alarm probability. Specifically, when the false alarm probability is 0.2, the proposed algorithm has more than 10% and 60% detection probability improvement under channel signal-to-noise ratio as 0 and -5 dB, respectively.