为了解决精密工件电镀过程中因电流波动导致的镀层不均匀的问题,在保留原电镀电源控制方式不变的基础上,完成了对其控制系统的升级改造.改造后的系统采用西门子WinCC开发人机交互界面,以PLC为控制核心,通过实时在线调整PID参数控制变频器输出合适的频率以改变电镀电源电机的转速,提高输出电流的调节精度,以满足高精度镀件产品的需求.实验证明,改造后的电源控制系统在不同负载下均能保证电流的调节品质并且具有较强的抑制扰动能力,镀层质量得到显著改善.

Pose-based action recognition has always been an important research field in computer vision. However, most existing pose-based methods are built upon human skeleton data, which cannot be used to exploit the feature of the motion-related object, i.e., a crucial clue of discriminating human actions. To address this issue, we propose a novel pose-flow relational model, which can benefit from both pose dynamics and optical flow. First, we introduce a pose estimation module to extract the skeleton data of the key person from the raw video. Second, a hierarchical pose-based network is proposed to effectively explore the rich spatial–temporal features of human skeleton positions. Third, we embed an inflated 3D network to capture the subtle cues of the motion-related object from optical flow. Additionally, we evaluate our model on four popular action recognition benchmarks (HMDB-51, JHMDB, sub-JHMDB, and SYSU 3D). Experimental results demonstrate that the proposed model outperforms the existing pose-based methods in human action recognition.

语音端点检测(Voice Activity Detection,VAD),是指在给定语音信号帧中判别语音是否存在,鲁棒的VAD有助于提高语音应用的自动化效率,例如语音增强、说话人识别、助听器等.为了提高低信噪比下语音端点检测的精度以及效率,提出了一种新的语音特征—低频消噪能量(Low Frequency De-noising Energy,LFDE),将其应用于VAD中,并利用LFDE与现有的声学特征(梅尔频率倒谱参数、共振峰频率)结合训练极限学习机(Extreme Learning Machine,ELM)分类器.仿真实验发现,端点检测的精度与效率都有提高.

Human action recognition in videos is still an important while challenging task. Existing methods based on RGB image or optical flow are easily affected by clutters and ambiguous backgrounds. In this paper, we propose a novel Pose-Guided Inflated 3D ConvNet framework (PI3D) to address this issue. First, we design a spatial–temporal pose module, which provides essential clues for the Inflated 3D ConvNet (I3D). The pose module consists of pose estimation and pose-based action recognition. Second, for multi-person estimation task, the introduced pose estimation network can determine the action most relevant to the action category. Third, we propose a hierarchical pose-based network to learn the spatial–temporal features of human pose. Moreover, the pose-based network and I3D network are fused at the last convolutional layer without loss of performance. Finally, the experimental results on four data sets (HMDB-51, SYSU 3D, JHMDB and Sub-JHMDB) demonstrate that the proposed PI3D framework outperforms the existing methods on human action recognition. This work also shows that posture cues significantly improve the performance of I3D.

A fractional order moments-based detector is proposed for the detection of weak signals in additive impulsive noise environment assumed as generalized Gaussian distribution with properly selected parameter values. The asymptotic detection performance is derived and compared with some traditional detectors optimized for operations in Gaussian noise with Nakagami fading communication channels. The analytical and computer simulation results of the fractional order moment-based detector are shown for signal detection with fading channels in the impulsive noise.

Pose-based action recognition has aroused increasing attention for its broad application prospects and excellent performance. Though the pose-based action recognition methods have been significantly advanced, pose-based action recognition remains a challenging task for various human action categories and subtle changes in human poses. To solve those problems, we propose pose-based multisource networks. First, human pose features are extracted from the raw video, followed by a filtration. Then, using a convolutional neural network (CNN) and long short-term memory (LSTM), the extracted pose sequence is fed into the proposed multisource networks. Subsequently, the CNN-based spatial model processes the relative position in each frame, and the LSTM-based temporal model is built to learn the temporal correlation of pose sequence. Afterward, the temporal model contains three sublevels to fully exploit the subtle information in the temporal domain. Finally, the experimental results verify the effectiveness of the proposed approach on SUBJHMDB, MPII Cooking Activities, SYSU 3D Human-Object Interaction, and NTU RGB+D. (C) 2019 SPIE and IS&T

Differential microphone arrays have been widely used in hands-free communication systems because of their frequency-invariant beampatterns, high directivity factors and small apertures. Considering the position of acoustic source always moving within a certain range in real application, this letter proposes an approach to construct the steerable first-order differential beampattern by using four omnidirectional microphones arranged in a non-orthogonal circular geometry. The theoretical analysis and simulation results show beampattern constructed via this method achieves the same direction factor (DF) as traditional DMAs and higher white noise gain (WNG) within a certain angular range. The simulation results also show the proposed method applies to processing speech signal. In experiments, we show the effectiveness and small computation amount of the proposed method.

“数字系统集成化设计”是高等院校电类专业本科生必修课程之一,在整个课程体系中不可或缺.本文针对该课程实验教学,从现有问题出发,对实验教学内容、教学方法和考核方式进行了深入研究,借助于“口袋实验室”系统提高学生数字系统开发设计能力,从而进一步培养学生的创造能力,以达到培养电子专业卓越工程师的要求.

α stable distribution model is used to describe the non-Gaussian noise of the cognitive communication system based on the theoretical study of the spectrum sensing problem of the primary user in the case of non-Gaussian noise.A perceptual method based on fractional lower order covariance is given.Fractional lower order covariance spectrum is used to estimate the primary user signals under α stable distributed noise,which solves the problem of failure of the traditional power spectrum estimation performance under non-Gaussian noise.On the basis of the characteristics of the FPGA,the algorithm is further optimized,and the sensing system based on the algorithm is designed and implemented on the FPGA.The system uses FPGA to generate QPSK signal with center frequency of 25 MHz,bandwidth of 12.5 MHz and α stable distributed noise with characteristic index of 1 as the primary user signal.The corresponding digital signal processing module is designed and verified in this system.The sensing method based on the fractional lower order covariance can effectively detect the presence of the primary signal from the α stable distributed noise.The system is stable and portable,and it is suitable for different primary user spectrum detection schemes to be implemented and verified on this system.

In view of the fact that the alpha distribution does not possess two order moment and power spectrum, the detection performance of the traditional detectors, such as energy detector (ED) and power spectral density (PSD) detector, will be degraded or even failed when the background noise be modeled as alpha-stable distribution in CR system. This paper presents a novel spectrum-sensing scheme based on fractional lower order statistics power spectral density (FPSD). The proposed algorithm, combining pseudo-PSD and Fourier transform (FT), calculates the FPSD of the received signal to determine whether primary user (PU) is present or absent. Via the numerous simulations, the performance of the FPSD versus the characteristic exponents \( \alpha \), the moment \( p \), and generalized signal-to-noise (GSNR) of the noise has been studied. Simulations show that the proposed FPSD detector has greater performance than ED and PSD in alpha-stable noise environment. In addition, the new detector, as a blind detector, has high probability detection without the prior knowledge of PU signal and noise.

In view of the problem that the performance of the traditional spectrum sensing algorithm of the of the maximum minimum eigenvalues (MME) (noise environment is assumed to be a Gaussian distribution) may degrade severely due to the heavy tail characteristics of the probability density function (PDF) of the non-Gaussian noise in the actual environment.To this end,an improved the fractional lower order moment sampling covariance MME that does not require any a priori knowledge about the signal,channels and noise is presented in this paper.The algorithm use fractional lower order moment of observation data preprocessing,scoring low moments of covariance matrix,and maximum ratio of the minimum eigenvalue of matrix as a statistic.This paper adopted the Alpha and Laplace distribution fitting non-Gaussian noise environment,Monte Carlo simulation results show that the performance of the fractional lower order moment of covariance MME is superior to MME in non-Gaussian situation.

本文根据电子信息工程专业创新应用型人才的培养目标,介绍了该专业实践教学体系的建设与实施情况,重点讲述了课程设计环节的设置思路、内容、方法,实施过程及效果.

《微机系统与接口技术》是高等院校理工科非计算机专业学生必修的一门计算机基础教育课程,也是高等教育电类本科生的专业基础平台课,在整个课程体系中起着承上启下的关键作用.从现有的实验教学存在的问题出发,对实验教学内容和教学方法两方面进行深入研究,借助于虚拟实验系统使学生具有灵活的微机系统开发设计能力,从而进一步培养学生的创造能力,以达到新时期对电子专业卓越工程师的培养要求.

当BD2/GPS卫星信号受到多径效应干扰时,接收机的定位精度将会严重下降.针对这一问题提出了一种抑制多径的BD2/GPS双模自适应扩展卡尔曼滤波算法.通过观测误差协方差估计和粗差检测来调整卫星参与定位的受信任程度和个数,分析了多径效应对伪距残差和多普勒残差的影响,同时对比了原始EKF算法和AREKF算法在多径干扰下复杂动态路况的定位性能.实验结果表明,AREKF算法能够有效抑制多径信号对定位效果的干扰,明显提高了定位精度和可靠性.

In cognitive radio systems, noise samples are often assumed to be independent Gaussian in order to simplify the spectrum sensing problem. However, due to the high frequency of sampling, a certain level of correlation exists among the noise samples. Furthermore, non-Gaussian noise often has a negative effect on the signals which the secondary users finally receive. Spectrum sensing methods based on the independent Gaussian noise assumption may not achieve satisfying detection performance when noise samples are correlated and non-Gaussian distributed. A novel signal detection method based on pth order moments (POM) in a multi-user cooperative scheme is proposed to address spectrum sensing issue for both independent and weakly correlated Laplace noise. Different from other detectors, our detector does not require a priori knowledge of PU, noise and communication channels. Theoretical performance measures are derived and verified for both independent and weakly correlated Laplace noise. Moreover, the detection performances versus signal-to-noise ratio SNR, order p, scale parameter b and correlation coefficient T of the background noise are investigated by computer simulation. It is shown that, for both independent and weakly correlated Laplace noises, the POM-based detector outperforms energy detector (ED) and polarity-coincidence-array (PCA) detector when p < 2. (C) 2017 Elsevier B.V. All rights reserved.

Under the background of Gaussian noise,the performance of spectrum sensing based on energy detection is optimal.But under the background of non-Gaussian noise,the performance of spectrum sensing drops greatly.In order to deal with the problem,a method of cancellation processing based on energy detection method is put forward,which is signal data minus no signal noise data.Compared with the method of energy detection,the way of cancellation processing improves detection probability by increasing the signal-to-noise ratios,so as to improve the performance of spectrum sensing.The effective spectrum sensing platform is designed and implemented based on USRP,GNURADIO and MATLAB,which verifies the feasibility of the algorithm.

It is hard for the traditional robotic portraying technology to represent the extracted nose contour, thus influencing the encoding and vectorization for the nose. Hence, a nose contour extraction algorithm has been presented based on the least square method. Firstly, the nose tip and nostrils are positioned after the nose has been detected. Then the key feature points are positioned, and the contour reference points are searched by constraining and controlling the key feature points. At last, the nose contour curve is fit based on the least square method, thus obtaining the nose contour curve. Experiments show that our method can ef-fectively extract the features of nose contour and also facilitate the encoding and vectorization.

Projector calibration is one of the key technologies in computer vision.This paper presented a calibration method of projector based on the positional relationship between the camera and projector. The method established the relation which was one to one correspondence of pixels between the image of the camera and projector image.The projector calibration was completed when getting homography between the projected image and the camera image by a series of operations to the captured image such as background subtraction,feature points extraction and so on.The method was simple and the experimental results indicated the method was practical,since the back projection error was between 0.3to 1pixel units.

The design of the high performance DSP circuit based on EDA technology is studied. The process that the full-custom sequence circuit based on EDA technology is used to realize the automatic extraction of the function model is given,and the automatic extraction algorithm of the sequence circuit function model is studied. An automatic extraction tool AutoExtra of the function model was designed and implemented. The simulation results show that the AutoExtra is effective,and can satisfy the design requirements of the high performance and module level DSP circuit based on EDA technology.

为加快空调生产线上检测工位的生产速率,提高整条生产线的生产效率.以组态王与空调检测的生产模式为研究对象,提出了一种更加智能化,人性化的检测系统.该系统在不改变原生产工艺与要求的前提下,通过优化上位机自动下发与存储功能,并增加自动诊断报警功能,将检测工位的空调平均生产耗时减少到10s,从而提高了整条检测线的智能化水平,继而向“智能工厂”的目标迈出了重要一步.