Dynamic analysis of memristive neural system with unbounded time-varying delays

Memristive models are currently the dominant technique for neuromorphic VLSI design. Critical dynamics research of memristive systems is very meaningful in both theoretical importance and practical significance. In this paper, by using nonsmooth analysis and control theory, nonlinear dynamics on global asymptotic stability and global exponential stability of memristive neural system with unbounded time-varying delays are explored. The theoretical analysis may yield a deep insight into the electric characteristics of memristor devices.     

A promising route to neuromorphic vision

The human vision system has a very strong capability in perceiving surround-ings.Such visual perception starts in the retina,which receives and preprocesses the...     

An integrated actuating and sensing system for light-addressable potentiometric sensor (LAPS) and light-actuated AC electroosmosis (LACE) operation

To develop a lab on a chip (LOC) integrated with both sensor and actuator functions, a novel two-in-one system based on optical-driven manipulation and sensing in a microfluidics setup based on a hydrogenated amorphous silicon (a-Si:H) layer on an indium tin oxide/glass is first realized. A high-intensity discharge xenon lamp functioned as the light source, a chopper functioned as the modulated illumination for a certain frequency, and a self-designed optical path projected on the digital micromirror device controlled by the digital light processing module was established as the illumination input signal with the ability of dynamic movement of projected patterns. For light-addressable potentiometric sensor (LAPS) operation, alternating current (AC)-modulated illumination with a frequency of 800 Hz can be generated by the rotation speed of the chopper for photocurrent vs bias voltage characterization. The pH sensitivity, drift coefficient, and hysteresis width of the Si₃N₄ LAPS are 52.8 mV/pH, −3.2 mV/h, and 10.5 mV, respectively, which are comparable to the results from the conventional setup. With an identical two-in-one system, direct current illumination without chopper rotation and an AC bias voltage can be provided to an a-Si:H chip with a manipulation speed of 20 μm/s for magnetic beads with a diameter of 1 μm. The collection of magnetic beads by this light-actuated AC electroosmosis (LACE) operation at a frequency of 10 kHz can be easily realized. A fully customized design of an illumination path with less decay can be suggested to obtain a high efficiency of manipulation and a high signal-to-noise ratio of sensing. With this proposed setup, a potential LOC system based on LACE and LAPS is verified with the integration of a sensor and an actuator in a microfluidics setup for future point-of-care testing applications.     

基于视觉传感与图像处理技术的焊接过程控制技术

随着视觉传感、图像处理和人工智能等技术的发展,计算机视觉以其信息量大、精度高、通用性好、检测范围大等特点,在焊接领域得到了广泛的应用。本文简要地介绍了目前的计算机视觉技术和方法,综述了该技术在焊接过程控制中的典型应用。     

一种基于CIEDE2000的烟叶颜色分级方法

烟叶人工分级易受主观因素影响,分级结果存在主观差异性,需要建立与人眼视觉感受相吻合的客观分级方法。本文针对烟叶颜色属性进行分级判断,给出一种基于CIEDE2000的均匀色差颜色分级方法。该方法包含了亮度、色调和饱和度综合色差视觉感受的图像评价标准,将待测烟叶与标准烟叶库中所有标准烟叶分别做色差计算,根据距离准则,寻找与当前待测烟叶色差最小的标准烟叶,同时对这个最小距离进行阈值判定,来确定当前待测烟叶的等级。实验结果表明根据CIEDE2000色差公式的颜色分级方法优于RGB颜色分级方法,与人眼视觉感受更为吻合,为建立烟叶自动分级系统提供了新的依据。     

浅谈基于稀疏表示的图像融合算法

图像融合已成为图像处理和计算机视觉领域的一项重要技术。随着压缩感知理论的发展,稀疏表示成为了压缩感知中的一个重要的研究内容。文章系统地阐述了当前图像融合技术的研究现状,重点介绍了稀疏表示在图像融合中的应用,并对目前图像融合算法存在的问题以及未来的发展进行了展望。     

Robust visual servoing control for quadrotors landing on a moving target

In this paper, a robust visual servoing control approach is proposed to address the landing problem for quadrotors on a moving platform. A vision system is implemented to estimate the position and velocity of the quadrotor. A robust cascade controller is proposed by following backstepping-like fundamentals and robust compensating theory. The effects of time-varying uncertainties, including parameter uncertainties and external disturbances, and time-varying delays resulted from image acquisition, image processing, and sensor measurement delays can be restrained. Experimental results using a quadrotor to land on a ground moving target illustrate the effectiveness of the proposed approach.     

模极大值小波算法在红外图像点目标检测中的应用研究

在总结以往红外项目实现的基础上,着重学习研究了小波变换点目标检测技术,并就其工程实现进行了论证,尝试将模极大值小波算法应用到红外点目标检测项目中。     

基于国产GF-1遥感影像的山区细小水体提取方法研究

以国产"高分一号"16m遥感图像为数据源,在特克斯县选取两个研究区域,针对山区细小线状河流提取难度较大的问题,使用基于规则的面向对象的方法实现了对山区细小水体的精确化提取。首先,在总结前人选择最优分割尺度的基础上,考虑了各层权重信息,针对某一特定地物,提出了指示最优分割尺度的指标——改进的与邻域绝对均值差分方差比(MRMAS),并由此获取了影像上细小水体的最优分割尺度。其次,为区分水体和山体阴影,构建阴影水体指数SWI=B1+B2-B4,成功剔除了绝大部分阴影信息。最后,利用形态学膨胀滤波及Pavlidis异步细化算法对提取的细小水体进行后处理,最终得到细小河流的矢量化水系图。实验结果表明,该方法可以完整、快速地提取出山区细小线状河流信息,总体精度在90%以上,Kappa系数在85%以上,有效排除阴影等暗色地物的干扰,基本消除椒盐噪声。该研究成果或对国产高分影像处理系统的研发与应用提供一定的科学参考。     

数据感知技术在电力物资供应链数据质量管理中的应用

目前在电力企业资源管理系统、电子商务平台、电工装备智慧物联平台、电力物流服务平台和供应链运营中心("五E一中心")建设中,存在电力物资供应链数据分布相对分散、规范化和结构化程度不一等问题,影响供应链数据挖掘和能源大数据分析的准确性和有效性.从时空数据质量感知、异常数据感知、供应商信息系统标识感知等维度剖析数据感知技术在电力物资供应链数据质量管理中的应用,构建包括数据资源层、数据清洗层、数据校正层、数据分析层和可视化交互层的全面数据质量管理体系,通过基于时空相关性的多源感知数据清洗技术、基于隐藏规则挖掘的异构异常数据修复技术和基于供应商信息系统标识的全息画像认知技术,实现数据治理、问题数据自动纠错;并基于电力物资供应链生态圈全寿命周期管理发展需求,提出数据感知技术在电力物资智慧供应链延伸及拓展应用方向:从上下游对接、全链业务融合贯通及画像需求出发构建生态圈,基于多源感知能力实现信息全面互联共享,实现整个智能制造业价值链的智能应用与创新服务.     

基于稀疏表示的人脸识别算法

近年来,人脸识别一直是计算机视觉和模式识别领域一个很热门研究课题。随着压缩感知理论的发展,稀疏表示已经成功应用于人脸识别。文章系统地综述了人脸识别的研究现状、系统,并重点介绍了基于稀疏表示的人脸识别算法及求解,最后对人脸识别算法存在的问题以及发展进行了展望。     

MEMSWear-biomonitoring system for remote vital signs monitoring

This paper proposes a remote vital signs monitoring system, which integrates wireless body area network (WBAN) and personal digital assistant (PDA) phone technology. Four different physiological signs, e.g., ECG, SpO₂, temperature and blood pressure, can be continuously acquired or derived from two wireless sensor nodes—ECG sensor and integrated SpO₂/temperature sensor. Once sentinel events happened or the request to real-time display vital signs is confirmed, all physiological signs and critical indices will be immediately transmitted to patient's PDA phone through Bluetooth and further relayed to doctor's PDA phone through global system for mobile communication (GSM) technology. A prototype of such system has been successfully developed and implemented, which will offer high standard of healthcare with a major reduction in cost for our society.     

Informed Patch Enhanced HyperGCN for skeleton-based action recognition

Human skeleton, as a compact representation of action, has attracted numerous research attentions in recent years. However, skeletal data is too sparse to fully characterize fine-grained human motions, especially for hand/finger motions with subtle local movements. Besides, without containing any information of interacted objects, skeleton is hard to identify human–object interaction actions accurately. Hence, many action recognition approaches that purely rely on skeletal data have met a bottleneck in identifying such kind of actions. In this paper, we propose an Informed Patch Enhanced HyperGraph Convolutional Network that jointly employs human pose skeleton and informed visual patches for multi-modal feature learning. Specifically, we extract five informed visual patches around head, left hand, right hand, left foot and right foot joints as the complementary visual graph vertices. These patches often exhibit many action-related semantic information, like facial expressions, hand gestures, and interacted objects with hands or feet, which can compensate the deficiency of skeletal data. This hybrid scheme can boost the performance while keeping the computation and memory load low since only five extra vertices are appended to the original graph. Evaluation on two widely used large-scale datasets for skeleton-based action recognition demonstrates the effectiveness of the proposed method compared to the state-of-the-art methods. Significant accuracy improvements are reported using X-Sub protocol on NTU RGB+D 120 dataset.     

Non-contact real-time detection of trace nitro-explosives by MOF composites visible-light chemiresistor

To create an artificial structure to remarkably surpass the sensitivity, selectivity and speed of the olfaction system of animals is still a daunting challenge. Herein, we propose a core-sheath pillar (CSP) architecture with a perfect synergistic interface that effectively integrates the advantages of metal–organic frameworks and metal oxides to tackle the above-mentioned challenge. The sheath material, NH₂-MIL-125, can concentrate target analyte, nitro-explosives, by 10¹² times from its vapour. The perfect band-matched synergistic interface enables the TiO₂ core to effectively harvest and utilize visible light. At room temperature and under visible light, CSP (TiO₂, NH₂-MIL-125) shows an unexpected self-promoting analyte-sensing behaviour. Its experimentally reached limit of detection (∼0.8 ppq, hexogeon) is 10³ times lower than the lowest one achieved by a sniffer dog or all sensing techniques without analyte pre-concentration. Moreover, the sensor exhibits excellent selectivity against commonly existing interferences, with a short response time of 0.14 min.     

一种基于多通道分解的图像测试信号设计方法

HVS中多通道分解模型用于度量误差图像空间尺寸的视觉感受,为了降低主观评价实验方法的条件要求,建立了一种基于多通道分解的图像测试信号设计方法。该方法通过控制白噪声在频域存在的区间,即白噪声经正交变换、掩膜量化矩阵,产生信噪比相同且空间尺寸不同的测试图像信号。分析和实验表明,多通道分解测试图像信号产生一对与高频和低频对应的误差图像,其信噪比基本相等但其空间尺寸的视觉感受不同,有利于考察与视觉感知相吻合的图像评价方法。     

Library on-shelf book segmentation and recognition based on deep visual features

On-shelf book segmentation and recognition are crucial steps in library inventory management and daily operation. In this paper, a detailed investigation of related work is conducted. RFID and barcode-based solutions suffer from expensive hardware facilities and long-term maintenance. Digital Image processing and OCR techniques are flawed due to a lack of accuracy and robustness. On this basis, we propose a visual and non-character system utilizing deep learning methods to accomplish on-shelf book segmentation and recognition tasks. Firstly, book spine masks are extracted from the image of on-shelf books by instance segmentation model, followed by affine transformation to rectangle images. Secondly, a spine feature encoder is trained to learn the deep visual features of spine images. Finally, the book inventory search space is constructed and the similarity metric between spine visual representations is calculated to recognize the target book identity. To train the models we collect high-resolution datasets of 10k-level and develop a data annotation software accordingly. For validation, we design simulated scenarios of recognizing 3.6k IDs from 5.6k book spines and achieve a best top1 accuracy of 99.18% and top5 accuracy of 99.91%. Furthermore, we develop a prototype of a mobile library management robot with embedded edge intelligence. It can automatically perform on-shelf book image capturing, spine segmentation and recognition, and target book grasping workflow.     

数字曲线的多边形逼近方法研究进展

在模型简化、图像处理和分析、模式识别以及计算机视觉等应用中常常涉及到数字曲线的多边形逼近问题。对该问题的国内外研究成果进行了较为系统地介绍,对一些典型算法的优缺点进行了分析,对相关技术的发展进行了展望。     

Human-swarm collaboration with coverage control under nonidentical and limited sensory ranges

Human intelligence plays a significant role in the operation of a multi-agent system. This study proposes a control framework that allows a human operator to collaboratively interact with a swarm robot to accomplish environmental exploration, detection, and coverage. A r_i-limited Voronoi partition is proposed herein for improving the all-territory sensing range for coverage control. Subsequently, an interactive control framework and control algorithms are presented for an abstract task function that allows a human operator to control the movement of a swarm robot in a working environment. Environmental information is fed back to the master devices so that the human operator can realize the swarm robots coverage control situation. Stability and position tracking with static coverage control and input-to-state stability with dynamic coverage control of the human-swarm system are investigated. The efficiency and efficacy of the proposed system are validated via numerical examples and experiments.