An Optimized Transformerless Photovoltaic Grid-Connected Inverter
无源毫米波成像改进POCS超分辨率算法

取 有 效 后 处理 措 施 增 强 分 辨 率 。提 出 了一 种 改进 的 P C O S超 分 辨 率 算 法 , 算 法结 合 了 Wi e 滤 波 器 复 原 该 e r n 算 法和 凸 集投 影 ( OC ) 法 的优 点 , 用 W i e 滤 波 复 原 算 法 恢 复 图像 通 带 内的 低 频 分 量 , 用 P C P S算 使 e r n 运 O S算
A b t a t The sr c : pr e oblm of oo r s l in a q r d m a i t pa sv m ilm e e w a e m a ng p r e o uto of c uie i ge n he s ie li t r v i gi
( c o l f Elcr ncE g n e ig, n v r i f E eto i S in ea d S h o e to i n i ern U i e s y o lcr nc ce c n o t
T c n l g f C ia,C e g u 6 1 3 ,C ia eh o o yo h n hn d 1 7 1 hn )
g rt m m p o e h o v r e tr t n s c mp t t n l c r fii n h n P oi h i r v st ec n eg n aea d i o u a i a l mu h mo ee fce tt a OCS ag r h o y l o i m.Th t e ag rt m se sl mp e n e n r a i o a sv li t rwa e i g n . lo i h i a i i l me t d i e lt y me f rp s ie mi me e v ma i g l Ke r s p s i e mi i t r wa e i g n ; s p r r s l t n;POCS ag rt m ; i g e t r to y wo d : a sv l me e v ma i g l u e—eo ui o lo i h ma e r so a i n;
优美斯(Optimax Systems)的相位平移干扰光学测量方法白皮书说明书

The Effect Of Phase Distortion On InterferometricMeasurements Of Thin Film Coated Optical SurfacesJon Watson, Daniel SavageOptimax Systems, 6367 Dean Parkway, Ontario, NY USA*********************©Copyright Optimax Systems, Inc. 2010This paper discusses difficulty in accurately interpreting surface form data from a phase shifting interferometer measurement of a thin film interference coated surfaces.PHASE-SHIFTING INTERFEROMETRYPhase-shifting interferometry is a metrology tool widely used in optical manufacturing to determine form errors of an optical surface. The surface under test generates a reflected wavefront that interferes with the reference wavefront produced by the interferometer 1. A phase-shifting interferometer modulates phase by slightly moving the reference wavefront with respect to the reflected test wavefront 2 . The phase information collected is converted into the height data which comprises the surface under test3.Visibility of fringes in an interferometer is a function of intensity mismatch between the test and reference beams. Most commercially available interferometers are designed to optimize fringe contrast based on a 4% reflected beam intensity. If the surface under test is coated for minimum reflection near or at the test wavelength of the interferometer, the visibility of the fringe pattern can be too low to accurately measure.OPTICAL THIN-FILM INTERFERENCE COATINGSOptical thin-film interference coatings are structures composed of one or more thin layers (typically multiples of a quarter-wave optical thickness) of materials deposited on the surface of an optical substrate.The goal of interference coatings is to create a multilayer film structure where interference effects within the structure achieve a desired percent intensity transmission or reflection over a given wavelength range.The purpose of the coating defines the design of the multilayer structure. Basic design variables include:• Number of layers• Thickness of each layer• Material of each layerThe most common types of multilayer films are high reflector (HR) and anti-reflection (AR) coatings. HR coatings function by constructively interfering reflected light, while AR coatings function by destructively interfering reflected light. These coatings are designed to operate over a specific wavelength range distributed around a particular design wavelength.To produce the desired interference effects, thin-film structures are designed to modulate the phase of the reflected or transmitted wavefront. The nature of the interference effect depends precisely on the thickness of each layer in the coating as well as the refractive index of each layer. If the thickness and index of each layer is uniform across the coated surface, the reflected wavefront will have a constant phase offset across the surface. However, if layer thicknesses or index vary across the coated surface, then the phase of thereflected wavefront will also vary. Depending on the design of the coating and the severity of the thickness or index non-uniformity, the distortion of the phase of the reflected wavefront can be severe. 4Layer thickness non-uniformity is inherent in the coating process and is exaggerated by increasing radius of curvature of the coated surface.5 All industry-standard directed source deposition processes (thermal evaporation, sputtering, etc) result in some degree of layer thickness non-uniformity.5 Even processes developed to minimize layer non-uniformity, such as those used at Optimax, will still result in slight layer non-uniformity (within design tolerance).TESTING COATED OPTICS INTERFEROMETRICALLYPhase-shifting interferometers use phase information to determine the height map of the surface under test. However, surfaces coated with a thin-film interference coating can have severe phase distortion in the reflected wavefront due to slight layer thickness non-uniformities and refractive index inhomogeneity. Therefore, the measured irregularity of a coated surface measured on a phase shifting interferometer at a wavelength other than the design wavelength, may not represent the actual irregularity of the surface. Even using a phase shifting interferometer at the coating design wavelength does not guarantee accurate surface irregularity measurements. If a coating has very low reflectance over any given wavelength range (such as in the case of an AR coating), the phase shift on reflection with wavelength will vary significantly in that range.7 Figure 1 shows an example of how the phase can vary with coating thickness variations.Figure 1In this particular case, if a point at the lens edge has the nominal coating thickness and the coating at lens center is 2% thicker, expect ~38° phase difference in the measurement (~0.1 waves). This will erroneous be seen as height by the interferometer, despite the actual height change in this case being less than 7nm (~0.01 waves). Also, depending on coating design, low fringe visibility may inhibit measurements.There is an extreme method to determine the irregularity of a thin-film interference coated surface by flash coating it with a bare metal mirror coating. A metal mirror coating is not a thin-film interference coating, and the surface of the mirror represents the true surface, This relatively expensive process requires extra time, handling, and potential damage during the metal coating chemical strip process.CONCLUSIONS•There can be practical limitations to getting accurate surface form data on coated optical surfaces due to issues with phase distortion and fringe visibility.•The issues are a function of thin film coating design particulars and the actual deposition processes.1 R.E. Fischer, B. Tadic-Galeb, P. Yoder, Optical System Design, Pg 340, McGraw Hill, New York City, 20082 H.H. Karow, Fabrication Methods For Precision Optics, Pg 656, John Wiley & Sons, New York City, 19933 MetroPro Reference Guide OMP-0347J, Page 7-1, Zygo Corporation, Middlefield, Connecticut, 20044 H.A. Macleod, Thin Film Optical Filters, Chapter 11: Layer uniformity and thickness monitoring, The Institute of Physics Publishing, 2001.5 R.E. Fischer, B. Tadic-Galeb, P. Yoder, Optical System Design, Pg 581, McGraw Hill, New York City, 2008。
基于太赫兹波谱图像识别无缺陷水果

基于太赫兹波谱图像识别无缺陷水果近年来,基于太赫兹波谱图像的无缺陷水果识别技术逐渐受到人们的关注。
太赫兹波谱图像识别是一种基于太赫兹波谱技术的无损检测方法,通过分析水果的太赫兹波谱图像,可以准确判断水果的品质和完整性,为消费者提供高质量的水果选择。
太赫兹波(terahertz wave)是介于微波与红外线之间的电磁波,其频率范围约为100GHz至10THz。
太赫兹波谱图像由太赫兹波谱仪获取,其中记录了水果对太赫兹波的反射、透射和散射等信息。
通过对这些信息的分析,可以得出水果的成分、结构和缺陷等有用的信息。
无缺陷水果识别是太赫兹波谱图像识别的重要应用之一。
随着人们对食品安全和质量的要求越来越高,无缺陷水果的识别成为了一个重要的问题。
传统的水果检测方法,如视觉检测和机械检测,通常需要繁琐的操作和专业知识,且容易出现误判的情况。
而基于太赫兹波谱图像的无缺陷水果识别技术能够准确、快速地判断水果是否有缺陷,提高水果的品质和安全。
在太赫兹波谱图像识别无缺陷水果方面,有许多研究工作已经取得了令人瞩目的成果。
其中,最常用的方法是基于机器学习和图像处理技术。
首先,需要建立一个太赫兹波谱图像数据库,其中包含了大量的无缺陷水果样本。
然后,利用机器学习算法对这些样本进行训练,建立一个识别模型。
最后,通过对新的太赫兹波谱图像进行处理和分析,将其与识别模型进行比对,从而得出水果是否有缺陷的结果。
在机器学习算法方面,支持向量机(SVM)和随机森林(Random Forest)是常用的分类算法。
SVM是一种广泛应用于模式识别和分类问题的机器学习算法,通过在高维空间中寻找一个最优超平面对样本进行分类。
而随机森林是一种集成学习方法,通过构建多个决策树,并对结果进行投票,来得出最终的分类结果。
这些算法都能够有效地对太赫兹波谱图像进行分类,准确判断水果是否有缺陷。
除了机器学习算法,图像处理技术在太赫兹波谱图像识别中也起到了重要的作用。
基于小波变换和去噪模型的光照不变人脸识别

中图 分 类 号 : P 9 . 1 T 3 14 3
文 献标 志码 : A
Il i a i n i a i n a e r c g ii n ba e n lum n to nv r a t f c e o n to s d o wa e e r n f r n e ii o l v l t t a s o m a d d no sng m de
Ab ta t h e o n t n o r na a i p e r n e w t l mi ain i a d f c l ts o a e r c g i o . I h s s r c :T e r c g i o ffo tlf ca a p aa c i i u n t s i u t a k fr fc e o t n n t i i l h l o i n i p p r o e l mi ai n i v r n xr cin me h d w s p o o e o d a t h l mi ai n p o lm a e n w v lt a e ,a n v li u n t n ai te ta t t o a r p s d t el w h t e i u n t r be b s d o a ee o a o i l o t n f r a d d n iig mo e .T e i u i ai n iv ra tW Se ta td i v l td man b sn a ee — a e e o s g r s m n e osn d 1 h l m n t n a in a x rc e n wa ee o i y u ig w v ltb d d n ii a o l o s n tc n q e .T r u h ma i ua ig te h g e u n y wa ee o f ce tc mbn d w t e osn d e ,te e g e t r so e h i u s h o g np lt h ih f q e c v ltc e in o i e i d n iig mo l h d e f au e f n r i h te i u n t n iv r n sw r n a c d a d mo e u e u no ma in w sr soe n i u n t n i v ra t,w ih c ud la h l mi ai n a i t e e e h n e n r s f l f r t a e t rd i lmi ai ai n s h c o l e d o a i o l o n t n e c l n a e r c g i o e fr n e o a x el tfc e o nt n p r ma c .T e e p r n a e ut n Y l a e d tb s n e i o h x ei me t lrs l o a e fc aa a e B a d CMU I a e d tb s s P E f c aa a e s o h ts t fc oy r c g i o a e c n b c iv d b h r p s d me h d h w t a ai a tr e o nt n rt a e a h e e y t e p o o e t o . s i Ke r s l mia in i v r n ;f c e o i o ;w v ltt n fr y wo d :i u n t n a i t a e rc g t n l o a n i a ee r so m;d n ii g mo e a e o sn d l
基于深度学习的低光照图像增强技术研究

基于深度学习的低光照图像增强技术研究深度学习是当前人工智能技术中的热门研究方向之一,它已经被应用于许多领域,如语音识别、图像识别、自然语言处理等。
在图像处理领域中,深度学习也有着广泛的应用,其中之一就是低光照图像增强。
低光照图像增强是指对光线不足或光线环境恶劣的图像进行处理,使其变得更加清晰明亮、细节更加丰富。
这是一个非常具有挑战性的问题,因为低光照图像通常由于光线不足导致图像信息缺失、噪点增多、色彩失真等现象,传统的图像处理方法难以有效处理。
而深度学习基于卷积神经网络的特征学习和表示能力,能够有效地处理低光照图像增强问题。
要实现低光照图像增强,需要解决以下几个问题:一、建立适合于低光照图像增强的深度学习模型传统的图像增强方法大多建立在颜色空间变换或梯度域变换技术之上,但是这些方法并不能很好地捕捉到图像的高级特征和语义信息,也不能很好地利用复杂的马尔可夫随机场模型来进行处理。
而基于深度学习的方法可以学习到更高级别的特征,通过模型的层次化特性来逐步提取图像中的语义信息,使得低光照图像增强更加准确和精细。
在建立深度学习模型时,需要对训练数据进行合理的选择和处理,以保证模型的泛化能力和鲁棒性。
同时,需要针对不同程度的低光照图像进行训练,以增强模型的适应性。
二、应用适当的损失函数损失函数是深度学习中的关键组成部分之一。
在低光照图像增强问题中,传统的损失函数往往只能通过像素级比较误差来进行刻画,不能很好地利用图像整体的特征和语义信息。
而基于深度学习的方法能够利用更丰富的先验知识,选择适当的损失函数来确保输出结果的质量。
针对低光照图像增强问题,一些研究者提出了不同的损失函数,例如平均绝对误差、结构相似性算法等。
这些损失函数可以提高图像增强效果和图像质量,提高模型的稳定性和鲁棒性。
三、提高模型的效率和速度在低光照图像增强过程中,需要处理大量的图像数据,如果深度学习模型的效率和速度不高,会导致图像增强的过程无法实时进行,大大降低用户的体验。
智能反射面增强的多无人机辅助语义通信资源优化

doi:10.3969/j.issn.1003-3114.2024.02.018引用格式:王浩博,吴伟,周福辉,等.智能反射面增强的多无人机辅助语义通信资源优化[J].无线电通信技术,2024,50(2): 366-372.[WANG Haobo,WU Wei,ZHOU Fuhui,et al.Optimization of Resource Allocation for Intelligent Reflecting Surface-enhanced Multi-UAV Assisted Semantic Communication[J].Radio Communications Technology,2024,50(2):366-372.]智能反射面增强的多无人机辅助语义通信资源优化王浩博1,吴㊀伟1,2∗,周福辉2,胡㊀冰3,田㊀峰1(1.南京邮电大学通信与信息工程学院,江苏南京210003;2.南京航空航天大学电子信息工程学院,江苏南京211106;3.南京邮电大学现代邮政学院,江苏南京210003)摘㊀要:无人机(Unmanned Aerial Vehicle,UAV)为无线通信系统提供了具有高成本效益的解决方案㊂进一步地,提出了一种新颖的智能反射面(Intelligent Reflecting Surface,IRS)增强多UAV语义通信系统㊂该系统包括配备IRS的UAV㊁移动边缘计算(Mobile Edge Computing,MEC)服务器和具有数据收集与局部语义特征提取功能的UAV㊂通过IRS 优化信号反射显著改善了UAV与MEC服务器的通信质量㊂所构建的问题涉及多UAV轨迹㊁IRS反射系数和语义符号数量联合优化,以最大限度地减少传输延迟㊂为解决该非凸优化问题,本文引入了深度强化学习(Deep Reinforce Learn-ing,DRL)算法,包括对偶双深度Q网络(Dueling Double Deep Q Network,D3QN)用于解决离散动作空间问题,如UAV轨迹优化和语义符号数量优化;深度确定性策略梯度(Deep Deterministic Policy Gradient,DDPG)用于解决连续动作空间问题,如IRS反射系数优化,以实现高效决策㊂仿真结果表明,与各个基准方案相比,提出的智能优化方案性能均有所提升,特别是在发射功率较小的情况下,且对于功率的变化,所提出的智能优化方案展示了良好的稳定性㊂关键词:无人机网络;智能反射面;语义通信;资源分配中图分类号:TN925㊀㊀㊀文献标志码:A㊀㊀㊀开放科学(资源服务)标识码(OSID):文章编号:1003-3114(2024)02-0366-07Optimization of Resource Allocation for Intelligent ReflectingSurface-enhanced Multi-UAV Assisted Semantic CommunicationWANG Haobo1,WU Wei1,2∗,ZHOU Fuhui2,HU Bing3,TIAN Feng1(1.School of Communications and Information Engineering,Nanjing University of Posts and Telecommunications,Nanjing210003,China;2.College of Electronic and Information Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing211106,China;3.School of Modern Posts,Nanjing University of Posts and Telecommunications,Nanjing210003,China)Abstract:Unmanned Aerial Vehicles(UAV)present a cost-effective solution for wireless communication systems.This article introduces a novel Intelligent Reflecting Surface(IRS)to augment the semantic communication system among multiple UAVs.The system encompasses UAV equipped with IRS,Mobile Edge Computing(MEC)servers,and UAV featuring data collection and local semantic feature extraction functions.Optimizing signal reflection through IRS significantly enhances communication quality between drones and MEC servers.The formulated problem entails joint optimization of multiple drone trajectories,IRS reflection coefficients,and the number of semantic symbols to minimize transmission delays.To address this non-convex optimization problem,this paper introduces a Deep收稿日期:2023-12-31基金项目:国家重点研发计划(2020YFB1807602);国家自然科学基金(62271267);广东省促进经济发展专项资金(粤自然资合[2023]24号);国家自然科学基金(青年项目)(62302237)Foundation Item:National K&D Program of China(2020YFB1807602);National Natural Science Foundation of China(62271267);Key Program of Marine Economy Development Special Foundation of Department of Natural Resources of Guangdong Province(GDNRC[2023]24);National Natural Sci-ence Foundation of China(Young Scientists Fund)(62302237)ReinforcementLearning(DRL)algorithm.Specifically,theDuelingDoubleDeepQNetwork(D3QN)isemployedtoaddressdiscreteactionspaceproblemssuchasdronetrajectoryandsemanticsymbolquantityoptimization.Additionally,DeepDeterministicPolicyGra dient(DDPG)algorithmisutilizedtosolvecontinuousactionspaceproblems,suchasIRSreflectioncoefficientoptimization,enablingefficientdecision making.Simulationresultsdemonstratethattheproposedintelligentoptimizationschemeoutperformsvariousbenchmarkschemes,particularlyinscenarioswithlowtransmissionpower.Furthermore,theintelligentoptimizationschemeproposedinthispaperexhibitsrobuststabilityinresponsetopowerchanges.Keywords:UAVnetwork;IRS;semanticcommunication;resourceallocation0 引言当前技术飞速发展的背景下,无人机(UnmannedAerialVehicle,UAV)已经成为无线通信系统中一种重要的技术[1]。
低光照图像增强算法综述
低光照图像增强算法综述一、本文概述随着计算机视觉技术的快速发展,图像增强技术成为了研究的重要领域之一。
其中,低光照图像增强算法是处理低质量、低亮度图像的关键技术,对于提高图像质量、增强图像细节、提升图像识别精度等方面具有重要的应用价值。
本文旨在对低光照图像增强算法进行全面的综述,介绍其研究背景、发展历程、主要算法及其优缺点,并探讨未来的发展趋势。
本文将对低光照图像增强的研究背景进行介绍,阐述低光照图像增强技术在视频监控、医学影像分析、军事侦察、航空航天等领域的应用需求。
本文将回顾低光照图像增强技术的发展历程,分析不同算法在不同历史阶段的发展特点和主要贡献。
接着,本文将重点介绍当前主流的低光照图像增强算法,包括基于直方图均衡化的算法、基于Retinex理论的算法、基于深度学习的算法等,并详细阐述其原理、实现方法、优缺点等。
本文将展望低光照图像增强技术的未来发展趋势,探讨新技术、新算法在提升图像质量、提高识别精度等方面的潜在应用。
通过本文的综述,读者可以全面了解低光照图像增强算法的研究现状和发展趋势,为相关领域的研究和实践提供有益的参考和借鉴。
二、低光照图像增强的基本原理低光照图像增强算法的核心目标是在保持图像细节和色彩信息的提高图像的亮度和对比度,从而改善图像的视觉效果。
这通常涉及到对图像像素值的调整,以及对图像局部或全局特性的分析和优化。
基本的低光照图像增强算法可以分为两类:直方图均衡化和伽马校正。
直方图均衡化是一种通过拉伸像素强度分布来增强图像对比度的方法。
这种方法假设图像的可用数据跨度大,即图像包含从暗到亮的所有像素值。
然而,对于低光照图像,由于大部分像素值集中在较低的亮度范围内,直方图均衡化可能会过度增强噪声,导致图像质量下降。
伽马校正则是一种更为柔和的增强方法,它通过调整图像的伽马曲线来改变图像的亮度。
伽马曲线描述了输入像素值与输出像素值之间的关系,通过调整这个关系,可以改变图像的亮度分布。
一种改进的高斯频率域压缩感知稀疏反演方法(英文)
AbstractCompressive sensing and sparse inversion methods have gained a significant amount of attention in recent years due to their capability to accurately reconstruct signals from measurements with significantly less data than previously possible. In this paper, a modified Gaussian frequency domain compressive sensing and sparse inversion method is proposed, which leverages the proven strengths of the traditional method to enhance its accuracy and performance. Simulation results demonstrate that the proposed method can achieve a higher signal-to- noise ratio and a better reconstruction quality than its traditional counterpart, while also reducing the computational complexity of the inversion procedure.IntroductionCompressive sensing (CS) is an emerging field that has garnered significant interest in recent years because it leverages the sparsity of signals to reduce the number of measurements required to accurately reconstruct the signal. This has many advantages over traditional signal processing methods, including faster data acquisition times, reduced power consumption, and lower data storage requirements. CS has been successfully applied to a wide range of fields, including medical imaging, wireless communications, and surveillance.One of the most commonly used methods in compressive sensing is the Gaussian frequency domain compressive sensing and sparse inversion (GFD-CS) method. In this method, compressive measurements are acquired by multiplying the original signal with a randomly generated sensing matrix. The measurements are then transformed into the frequency domain using the Fourier transform, and the sparse signal is reconstructed using a sparsity promoting algorithm.In recent years, researchers have made numerous improvementsto the GFD-CS method, with the goal of improving its reconstruction accuracy, reducing its computational complexity, and enhancing its robustness to noise. In this paper, we propose a modified GFD-CS method that combines several techniques to achieve these objectives.Proposed MethodThe proposed method builds upon the well-established GFD-CS method, with several key modifications. The first modification is the use of a hierarchical sparsity-promoting algorithm, which promotes sparsity at both the signal level and the transform level. This is achieved by applying the hierarchical thresholding technique to the coefficients corresponding to the higher frequency components of the transformed signal.The second modification is the use of a novel error feedback mechanism, which reduces the impact of measurement noise on the reconstructed signal. Specifically, the proposed method utilizes an iterative algorithm that updates the measurement error based on the difference between the reconstructed signal and the measured signal. This feedback mechanism effectively increases the signal-to-noise ratio of the reconstructed signal, improving its accuracy and robustness to noise.The third modification is the use of a low-rank approximation method, which reduces the computational complexity of the inversion algorithm while maintaining reconstruction accuracy. This is achieved by decomposing the sensing matrix into a product of two lower dimensional matrices, which can be subsequently inverted using a more efficient algorithm.Simulation ResultsTo evaluate the effectiveness of the proposed method, we conducted simulations using synthetic data sets. Three different signal types were considered: a sinusoidal signal, a pulse signal, and an image signal. The results of the simulations were compared to those obtained using the traditional GFD-CS method.The simulation results demonstrate that the proposed method outperforms the traditional GFD-CS method in terms of signal-to-noise ratio and reconstruction quality. Specifically, the proposed method achieves a higher signal-to-noise ratio and lower mean squared error for all three types of signals considered. Furthermore, the proposed method achieves these results with a reduced computational complexity compared to the traditional method.ConclusionThe results of our simulations demonstrate the effectiveness of the proposed method in enhancing the accuracy and performance of the GFD-CS method. The combination of sparsity promotion, error feedback, and low-rank approximation techniques significantly improves the signal-to-noise ratio and reconstruction quality, while reducing thecomputational complexity of the inversion procedure. Our proposed method has potential applications in a wide range of fields, including medical imaging, wireless communications, and surveillance.。
杜克大学推出AI图像生成器 模糊图像五秒可变清晰
杜克大学推出AI图像生成器模糊图像五秒可变清晰作者:来源:《中国计算机报》2020年第25期近日,美国杜克大学的研究团队研发了一个AI图像生成模型PULSE。
PULSE可以在5秒钟内将低分辨率的人像转换成清晰、逼真的人像。
据了解,PULSE所做的工作并不是把输入的低分辨率人像变成一张高分辨率的人像,而是“一对多”地输出许多张面部细节各不相同的高分辨率人像。
比如,用户输入一张16×16分辨率的图像,PULSE可输出一组1024×1024分辨率的图像。
上述研究已在计算机视觉与模式识别会议CVPR 2020上发表,论文标题为《PULSE:通过对生成模型的潜在空間探索实现自监督照片上采样》。
研究人员用高分辨人脸数据集CelebA HQ评估PULSE的性能。
为了进行对比,研究人员利用CelebA HQ数据集训练了监督模型BICBIC、FSRNET和FSRGAN。
所有模型均以16×16分辨率的图像作为输入,BICBIC、FSRNET和FSRGAN模型以128×128分辨率图像作为输出,PULSE模型以128×128分辨率图像和1024×1024分辨率图像作为输出。
评估结果显示,图像质量方面,PULSE模型在生成眼睛、嘴唇等图像细节方面的能力优于其他模型。
此外,研究人员还利用平均意见分数(MOS)测试来定量评估模型的分辨率。
研究人员邀请40位评估者对6个模型的输出结果进行打分。
结果显示,PULSE的MOS分数最高。
研究人员称,未来,PULSE或可被用于天文学、医学等领域。
比如,一位天文学研究人员输入一张模糊的黑洞图像,就可以获得许多张可能的天体照片。
纹理物体缺陷的视觉检测算法研究--优秀毕业论文
摘 要
在竞争激烈的工业自动化生产过程中,机器视觉对产品质量的把关起着举足 轻重的作用,机器视觉在缺陷检测技术方面的应用也逐渐普遍起来。与常规的检 测技术相比,自动化的视觉检测系统更加经济、快捷、高效与 安全。纹理物体在 工业生产中广泛存在,像用于半导体装配和封装底板和发光二极管,现代 化电子 系统中的印制电路板,以及纺织行业中的布匹和织物等都可认为是含有纹理特征 的物体。本论文主要致力于纹理物体的缺陷检测技术研究,为纹理物体的自动化 检测提供高效而可靠的检测算法。 纹理是描述图像内容的重要特征,纹理分析也已经被成功的应用与纹理分割 和纹理分类当中。本研究提出了一种基于纹理分析技术和参考比较方式的缺陷检 测算法。这种算法能容忍物体变形引起的图像配准误差,对纹理的影响也具有鲁 棒性。本算法旨在为检测出的缺陷区域提供丰富而重要的物理意义,如缺陷区域 的大小、形状、亮度对比度及空间分布等。同时,在参考图像可行的情况下,本 算法可用于同质纹理物体和非同质纹理物体的检测,对非纹理物体 的检测也可取 得不错的效果。 在整个检测过程中,我们采用了可调控金字塔的纹理分析和重构技术。与传 统的小波纹理分析技术不同,我们在小波域中加入处理物体变形和纹理影响的容 忍度控制算法,来实现容忍物体变形和对纹理影响鲁棒的目的。最后可调控金字 塔的重构保证了缺陷区域物理意义恢复的准确性。实验阶段,我们检测了一系列 具有实际应用价值的图像。实验结果表明 本文提出的纹理物体缺陷检测算法具有 高效性和易于实现性。 关键字: 缺陷检测;纹理;物体变形;可调控金字塔;重构
Keywords: defect detection, texture, object distortion, steerable pyramid, reconstruction
II
- 1、下载文档前请自行甄别文档内容的完整性,平台不提供额外的编辑、内容补充、找答案等附加服务。
- 2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
- 3、如文档侵犯您的权益,请联系客服反馈,我们会尽快为您处理(人工客服工作时间:9:00-18:30)。
An Optimized Transformerless Photovoltaic Grid-Connected InverterHuafeng Xiao,Student Member,IEEE,Shaojun Xie,Member,IEEE,Yang Chen,and Ruhai HuangAbstract—Unipolar sinusoidal pulsewidth modulation(SPWM) full-bridge inverter brings high-frequency common-mode voltage, which restricts its application in transformerless photovoltaic grid-connected inverters.In order to solve this problem,an op-timized full-bridge structure with two additional switches and a capacitor divider is proposed in this paper,which guarantees that a freewheeling path is clamped to half input voltage in the freewheeling period.Sequentially,the high-frequency common-mode voltage has been avoided in the unipolar SPWM full-bridge inverter,and the output currentflows through only three switches in the power processing period.In addition,a clamping branch makes the voltage stress of the added switches be equal to half input voltage.The operation and clamping modes are analyzed, and the total losses of power device of several existing topologies and proposed topology are fairly calculated.Finally,the common-mode performance of these topologies is compared by a universal prototype inverter rated at1kW.Index Terms—Clamping,common-mode voltage,full-bridge inverter,unipolar sinusoidal pulsewidth modulation(SPWM).I.I NTRODUCTIONT RANSFORMERLESS grid-connected inverters have a lot of advantages such as high efficiency,small size,light weight,low cost,etc.[1]–[6].However,there is a galvanic connection between power grid and solar cell array.Depending on the inverter topology,this may causefluctuation of the potential between the solar cell array and the ground,and these fluctuations may have a square wave at switching frequency. When energized by afluctuating potential,the stray capacitance to ground formed by the surface of the photovoltaic(PV) array may lead to the occurrence of ground currents.A person, connected to the ground and touching the PV array,may con-duct the capacitive current to the ground,causing an electrical hazard[4].At the same time that the conducted interference and radiated interference will be brought in by the ground current, the grid current harmonics and losses will also increase[5]. The unipolar sinusoidal pulsewidth modulation(SPWM) full-bridge inverter has received extensive attentions,owing to its excellent differential mode characteristics such as higher dc voltage utilization,smaller current ripple in thefilter inductor, and higher processing efficiency.However,the switching fre-Manuscript received November7,2009;revised January28,2010, March17,2010,and May1,2010;accepted June4,2010.Date of publication June28,2010;date of current version April13,2011.This work was supported by the National Natural Science Foundations of China under Award51077070. The authors are with the Department of Electrical Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing210016,China(e-mail: xiaohf@).Color versions of one or more of thefigures in this paper are available online at .Digital Object Identifier10.1109/TIE.2010.2054056quency time-varying common-mode voltage(whose amplitude is equal to a dc input voltage)is brought in.Therefore,a transformer(low frequency or high frequency)is needed to isolate the solar cell array from the grid in grid-connected ap-plications,and at the same time,the high-frequency common-mode voltage endangers the insulation layer of the transformers [7],which increases its manufacturing cost.In order to remove this transformer from the unipolar SPWM full-bridge grid-connected inverter,a lot of in-depth researches,where new freewheeling paths are constructed to separate the PV array from the grid in the freewheeling period,have been done[6], [8]–[11].A pair of switches between the two midpoints of the bridge leg[ac side,shown in Fig.1(a)]has been added in[8]to construct a new freewheeling path in the freewheeling period. In[9],Gonzalez et al.bring a double clamping branch to the solar cell array side[shown in Fig.1(b)],and the potential can be clamped in the freewheeling period by a capacitor divider in the input side.Only one additional high-frequency switch is brought to the positive terminal of the PV array[shown in Fig.1(c)]to achieve the disconnection with the grid in the free-wheeling period in[10].Based on the high-frequency common-mode equivalent model of the full-bridge circuit derived by Gonzalez et al.[11]and Gubia et al.[12],it is necessary that the potential of the freewheeling path is clamped to half input voltage in the freewheeling period instead of disconnecting the PV array from the grid simply,and by which,the high-frequency common-mode voltage can be completely avoided in the unipolar SPWM full-bridge inverter.In[8]and[10], the potential of the freewheeling path cannot be clamped in the freewheeling period,and its level depends on the parasitic parameters of the path and the grid voltage amplitude.The clamping branch proposed by Gonzalez et al.[9]guarantees that the freewheeling path is clamped to half input voltage in the freewheeling period,but the output currentflows through four switches in the power processing period,which increases the conduction losses.Thin-film panels have a lot of advantages such as low cost and are suitable for building-integrated PV[13],[14].However, its power density is lower than the conventional crystalline silicon module(which means that its conversion efficiency is lower).Thus,the stray capacitor of unit power module to the ground increases from50–150nF/kW for crystalline silicon module up to1µF/kW for thin-film module[4].Unfortunately, the transformerless grid-connected inverters make the ground current suppression become much more challenging in applica-tions of thin-film panels.Considering both of the advantages and disadvantages of the existing topologies mentioned earlier,an optimized0278-0046/$26.00©2010IEEEXIAO et al.:OPTIMIZED TRANSFORMERLESS PHOTOVOLTAIC GRID-CONNECTED INVERTER1889Fig.3.Equivalent circuits of working mode.(a)Power processing mode and(b)freewheeling mode in the positive half period of the grid current.(c)Power processing mode and(d)freewheeling mode in the negative half period of the grid current.1890IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS,VOL.58,NO.5,MAY2011Fig. 6.Equivalent circuits in the clamped mode.(a)Potential down.(b)Potential up.(c)Potential fluctuates with grid voltage in the positive half period of the grid current.This deviation can be suppressed by some means such as adding a resistor divider or an active voltage balancing circuit.In this paper,a simple resistor divider is used to balance the capacitor voltage.B.Operation Principle of Clamping BranchThe equivalent circuit of the converter in the clamping period is shown in Fig.6.It can be seen that,regardless of the grid current direction,if the freewheeling path potential falls,the current flows through the antiparallel diode (or body diode)of the clamp switch S 2to step up this freewheeling path potential to (1/2)V pv ,as shown in Fig.6(a);if the potential rises,the clamp switch S 2will be on so that the potential falls back to (1/2)V pv ,as shown in Fig.6(b).However,during dead time be-tween the switches S 1and S 2,the potential of thefreewheeling2+V CE I L t b 2+V CE I RR t b 3(1)where V CE is IGBT’s blocking voltage across the collector andemitter,I L is the filter inductor current,and I RR is the reverse recovery current.The diode turn-off loss can now be computed as [21],[22]W Diode ,turn -off=V F (I L +I RR )(t r +t a )2+(V D +V F )I RR t b 6(2)where V F is the ON -state voltage of the diode and V D is the diode’s blocking voltage across the cathode and anode.XIAO et al.:OPTIMIZED TRANSFORMERLESS PHOTOVOLTAIC GRID-CONNECTED INVERTER1891B.Losses for IGBT Turn-off and Diode Turn-onThe IGBT’s turn-off and the diode’s turn-on behavior shownin Fig.7are also characterized by the IGBT’s turn-off delaytime t d,fall time t f,and tail time t tail.The turn-off loss ofIGBT is calculated as[18],[21]W IGBT,turn-off=V CE I L t d2+11·V CE I L t f20+V CE I L t tail20.(3)The diode turn-on loss can now be computed asW Diode,turn-on=9·V F I L t f20+19·V F I L t tail20.(4)C.On-State Losses for IGBT and DiodeThe conduction losses of IGBT and diode can be cal-culated asW IGBT,on-state=V CE(on)·I L·[dT S−(t r+t a+t b)](5) W Diode,on-state=V F·I L·[(1−d)T S−(t d+t f+t tail)](6) where V CE(on)is the ON-state voltage of IGBT,d is IGBT’s duty cycle,and T S is the switching period.D.Calculation ResultsThe device loss power can be derived by integral in a grid periodP IGBT,loss=1T gNi=1(W IGBT,turn-on(i)+W IGBT,turn-off(i)+W IGBT,on-state(i))(7)P Diode,loss=1T gNi=1(W Diode,turn-on(i)+W Diode,turn-−off(i)+W Diode,on-state(i))(8)where T g is the grid period,i represents one switching process, and N is the total switching time in a grid period.Table I shows the voltage rate and distribution of the de-vice’s number in these topologies.International Rectifier’s IRG4PSC71UD(600V/60A)IGBT with ultrafast soft recovery diode was chosen for the switches rated at600V.The1200-V IGBT used was IRG4PSH71UD(1200V/50A),which is of the same family as the IRG4PSC71UD.The total device losses in different switching frequencies are listed in Table II under selected devices and shown as histogram in Fig.8with each component’s percent.It can be seen that Heric[8]is with the least device loss and H5[10]is the pared with H5,the oH5proposed in this paper reduces the device loss significantly.In particular,the advantage of the efficiency of the optimized topology oH5becomes more and more obvious as the switching frequency increases,and it is gradually closeTABLE IA NALYSIS OF D EVICE O PERATION IN S EVERAL TOPOLOGIESTABLE IIT OTAL S EMICONDUCTOR L OSSES OF S EVERAL T OPOLOGIES R ATED AT 5kW U NDER D IFFERENT S WITCHING FREQUENCIESFig.8.Total device loss distribution for a5-kW rate.to the Heric topology.The calculation results are in agreement with the theoretical estimation.The comparison of total device losses is helpful for the selection of the high-efficiency topology in practice.V.E XPERIMENTAL R ESULTSIn order to verify the operation principle and performance comparison,a universal prototype inverter has been built in our laboratory,as shown in Fig.9.The specifications of the converter are listed in Table III.In Fig.9,modules“Leg1U,”“Leg1D,”“Leg2U,”and “Leg2D”are leg switches of the conventional full-bridge in-verter.Modules“DC Bypass1,”“DC Bypass2,”and“Clamping Branch”are partially selected in H5,H6,and oH5inverters according to the topology structure.Module“AC Bypass”is enabled in the Heric inverter.L1,L2,and C1make up the1892IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS,VOL.58,NO.5,MAY2011XIAO et al.:OPTIMIZED TRANSFORMERLESS PHOTOVOLTAIC GRID-CONNECTED INVERTER1893mon-mode voltage and ground current waveforms in Heric topology.(a)Common-mode voltage(v g:400V/div,i g:6.7A/div,v3N and v4N: 200V/div,v CM:200V/div,and time:4ms/div).(b)Ground current(v g:400V/div,i g:6.7A/div,i Ground:80mA/div,time:4ms/div,and M:4mA/div, 5kHz/div).mon-mode voltage and ground current waveforms in H6topology.(a)Common-mode voltage.(b)Ground current.mon-mode voltage and ground current waveforms in H5topology.(a)Common-mode voltage.(b)Ground current.mon-mode voltage and ground current waveforms in oH5topology.(a)Common-mode voltage.(b)Ground current.1894IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS,VOL.58,NO.5,MAY2011Fig.14.Detailed Common-mode voltage waveforms in oH5topology.(a)At the peak of grid current (v g :400V/div,i D2:40mA/div,v 3N and v 4N :200V/div,v CM :200V/div,and time:4ms/div).(b)At the vale of gridcurrent.Fig.15.Differential-mode voltage waveform in oH5topology at V pv =400V (v g :200V/div,i g :6.7A/div,v 34:200V/div,and time:4ms/div).VI.C ONCLUSIONAn optimized transformerless grid-connected PV inverter has been proposed in this paper,which has the following advantages.1)The common-mode voltage is clamped to a constant level,so the ground current can be suppressed well.2)The good differential-mode characteristic can be achieved like the unipolar SPWM full-bridgegrid-Fig.16.Experimental waveform of power device in oH5topology at V pv =400V.(a)S 1(v GS1:20V/div and v DS1:200V/div).(b)S 2(i D2:40mA/div,v GS2:20V/div,and v DS2:100V/div).Fig.17.Experimental waveform of capacitor divider in oH5topology at V pv =400V (v g :200V/div,i g :6.7A/div,v pv and v 2N :100V/div,and time:4ms/div).connected inverter with galvanic isolation,but with higher efficiency.3)The blocking voltage of the added switches is only half of the input voltage.These merits are verified and compared by a universal proto-type rated at 240V/50Hz,1kW.It can be concluded that the proposed inverter is extremely suitable for high-power single-phase grid-connected systems with thin-film solar cell.XIAO et al.:OPTIMIZED TRANSFORMERLESS PHOTOVOLTAIC GRID-CONNECTED INVERTER1895R EFERENCES[1]J.M.Carrasco,L.G.Franquelo,J.T.Bialasiewicz, E.Galvan,R.C.P.Guisado,M.Prats,J.I.Leon,and N.Moreno-Alfonso,“Power-electronic systems for the grid integration of renewable energy sources:A survey,”IEEE Trans.Ind.Electron.,vol.53,no.4,pp.1002–1016,Jun.2006.[2]S.B.Kjaer,J.K.Pedersen,and F.Blaabjerg,“A review of single-phasegrid-connected inverters for photovoltaic modules,”IEEE Trans.Ind.Appl.,vol.41,no.5,pp.1292–1306,Sep./Oct.2005.[3]B.Sahan,A.N.Vergara,N.Henze,A.Engler,and P.Zacharias,“A single-stage PV module integrated converter based on a low-power current-source inverter,”IEEE Trans.Ind.Electron.,vol.55,no.7,pp.2602–2609, Jul.2008.[4]J.M.A.Myrzik and M.Calais,“String and module integrated inverters forsingle-phase grid connected photovoltaic systems—A review,”in Proc.IEEE Bologna Power Tech Conf.,Bologna,Italy,2003,pp.430–437. [5]M.Calais and V.G.Agelidis,“Multilevel converters for single-phase gridconnected photovoltaic systems—An overview,”in Proc.IEEE Int.Symp.Ind.Electron.,Pretoria,South Africa,1998,pp.224–229.[6]T.Kerekes,R.Teodorescu,and U.Borup,“Transformerless photovoltaicinverters connected to the grid,”in Proc.IEEE Appl.Power Electron.Conf.Expo.,Anaheim,CA,2007,pp.1733–1737.[7]C.X.Mao,W.B.Li,L.Jiming,and F.Shu,“Study of the common-modevoltage in a high-voltage ASD’s system,”Proc.CSEE,vol.23,no.9, pp.57–62,2003.[8]S.Heribert,S.Christoph,and K.Jurgen,“Inverter for transforming a DCvoltage into an AC current or an AC voltage,”Europe Patent1369985 (A2),May13,2003.[9]R.Gonzalez,J.Lopez,P.Sanchis,and L.Marroyo,“Transformer-less inverter for single-phase photovoltaic systems,”IEEE Trans.Power Electron.,vol.22,no.2,pp.693–697,Mar.2007.[10]M.Victor,F.Greizer,S.Bremicker,and U.Hübler,“Method of convertinga direct current voltage from a source of direct current voltage,morespecifically from a photovoltaic source of direct current voltage,into a alternating current voltage,”U.S.Patent7411802,Aug.12,2008. [11]R.Gonzalez,E.Gubia,J.Lopez,and L.Marroyo,“Transformerlesssingle-phase multilevel-based photovoltaic inverter,”IEEE Trans.Ind.Electron.,vol.55,no.7,pp.2694–2702,Jul.2008.[12]E.Gubia,P.Sanchis,and A.Ursua,“Ground currents in single-phasetransformerless photovoltaic systems,”Prog.Photovolt.,vol.15,no.7, pp.629–650,May2007.[13]R.Ruther,A.J.G.daSilva,A.A.Montenegro,and I.T.Salamoni,“Assessment of thin-film technologies most suited for BIPV applications in Brazil:The PETROBRAS44kWp project,”in Proc.3rd World Conf.Photovoltaic Energy Convers.,Osaka,Japan,2003,pp.2294–2297. [14]X.L.Li,“Application of CIS thinfilm solar cells in the BIPV systemwith large commercial project,”New Energy Environ.,vol.37,pp.46–48, 2008.[15]H.du Toit Mouton,“Natural balancing of three-level neutral-point-clamped PWM inverters,”IEEE Trans.Ind.Electron.,vol.49,no.5, pp.1017–1025,Oct.2002.[16]R.Stala,S.Pirog,M.Baszynski,A.Mondzik,A.Penczek,J.Czekonski,and S.Gasiorek,“Results of investigation of multicell converters with balancing circuit—Part I,”IEEE Trans.Ind.Electron.,vol.56,no.7, pp.2610–2619,Jul.2009.[17]P.N.Tekwani,R.S.Kanchan,and K.Gopakumar,“A dualfive-levelinverter-fed induction motor drive with common-mode voltage elimina-tion and DC-link capacitor voltage balancing using only the switching-state redundancy—Part II,”IEEE Trans.Ind.Electron.,vol.54,no.5, pp.2609–2617,Oct.2007.[18]A.D.Rajapakse,A.M.Gole,and P.L.Wilson,“Electromagnetic tran-sients simulation models for accurate representation of switching losses and thermal performance in power electronic systems,”IEEE Trans.Power Electron.,vol.20,no.1,pp.319–327,Jan.2005.[19]T.Shimizu and S.Iyasu,“A practical iron loss calculation for ACfilterinductors used in pwm inverter,”IEEE Trans.Ind.Electron.,vol.56,no.7, pp.2600–2609,Jul.2009.[20]Y.L.Xiong,S.Sun,H.W.Jia,P.Shea,and Z.J.Shen,“New physi-cal insights on power MOSFET switching losses,”IEEE Trans.Power Electron.,vol.24,no.2,pp.525–531,Feb.2009.[21]F.Hong,R.Z.Shan,H.Z.Wang,and Y.Yangon,“Analysis and calcu-lation of inverter power loss,”Proc.CSEE,vol.28,no.15,pp.72–78, May2008.[22]N.Shammas,D.Chamund,and M.Calais,“Forward and reverse recoverybehavior of diodes in power converter applications,”in Proc.Int.Conf.Microelectron.,Montenegro,Serbia,2004.Huafeng Xiao(S’10)was born in Hubei,China,in1982.He received the B.S.and M.S.degreesin electrical engineering from Nanjing Universityof Aeronautics and Astronautics,Nanjing,China,in2004and2007,respectively,where he is cur-rently working toward the Ph.D.degree in electricalengineering.His main research interests are high-frequencysoft-switching conversion and photovoltaicapplications.Shaojun Xie(M’05)was born in Hubei,China,in1968.He received the B.S.,M.S.,and Ph.D.degreesin electrical engineering from Nanjing Universityof Aeronautics and Astronautics(NUAA),Nanjing,China,in1989,1992,and1995,respectively.In1992,he joined the Faculty of Electrical En-gineering,Teaching and Research Division,NUAA,where he is currently a Professor with the Collegeof Automation Engineering.He has authored over50technical papers in journals and conference pro-ceedings.His main research interests include avia-tion electrical power supply systems and power electronicconversion.Yang Chen was born in Shanxi,China,in1985.He received the B.S.degree in electrical engineer-ing from Nanjing University of Aeronautics andAstronautics,Nanjing,China,in2008,where heis currently working toward the M.S.degree inelectrical engineering in the College of AutomationEngineering.He mainly focuses his research on high-performance dc–dc converters for photovoltaicapplications.Ruhai Huang was born in Jiangsu,China,in1987.He received the B.S.degree from Nanjing Universityof Aeronautics and Astronautics,Nanjing,China,in2009,where he is currently working toward the M.S.degree in electrical engineering.His research interests include grid-connected con-verters and parallel technology of inverters.。