Visual Identification Method and Implementation of Impurities in Food Grain

A method to check impurities in food grain based on digital image processing is proposed in this study. According to the feature of the impurity, one image with twice processing is put forward based on blob analysis, which eliminates impurities that were brighter or darker than normal particles. With the image processing system developed by Matrox Imaging Library, black and white blobs are extracted for twice with one function, and then the size and the position of impurities are calculated quickly through mathematical morphologic and blob analysis. The method can be extended and applied in other fields due to its advantages of quick speed and high accuracy.

Illumination Adaptive Identification Algorithm of a Reconfigurable Modular Robot

Reconfigurable modular robots feature high mobility due to their unconstrained connection manners.Inspired by the snake multi-joint crawling principle,a chain-type reconfigurable modular robot(CRMR)is designed,which could reassemble into various configurations through the compound joint motion.Moreover,an illumination adaptive modular robot identification(IAMRI)algorithm is proposed for CRMR.At first,an adaptive threshold is applied to detect oriented FAST features in the robot image.Then,the effective detection of features in non-uniform illumination areas is achieved through an optimized quadtree decomposition method.After matching features,an improved random sample consensus algorithm is employed to eliminate the mismatched features.Finally,the reconfigurable robot module is identified effectively through the perspective transformation.Compared with ORB,MA,Y-ORB,and S-ORB algorithms,the IAMRI algorithm has an improvement of over 11.6%in feature uniformity,and 13.7%in the comprehensive indicator,respectively.The IAMRI algorithm limits the relative error within 2.5 pixels,efficiently completing the CRMR identification under complex environmental changes.

SVM-based Identification and Un-calibrated Visual Servoing for Micro-manipulation

This paper presents an improved support vector machine （SVM） algorithm, which employs invariant moments-based edge extraction to obtain feature attribute. A heuristic attribute reduction algorithm based on rough set＇s discernible matrix is proposed to identify and classify micro-targets. To avoid the complicated calibration for intrinsic parameters of camera, an improved Broyden＇s method is proposed to estimate the image Jacobian matrix which employs Chebyshev polynomial to construct a cost function to approximate the optimization value. Finally, a visual controller is designed for a robotic micromanipulation system. The experiment results of micro-parts assembly show that the proposed methods and algorithms are effective and feasible.

Visual identification of silver ornament by the structural color based on Mie scattering of ZnO spheres

Structural colors originated from Mie scattering of dielectric spheres can be regulated by the coupling effect between them and substrates.Here a rapid visual identification method of silver ornaments was proposed by the coupling effect of Zn O spheres with them.Both simulation and experimental results proved that,by coupling with different metal substrates,the Mie resonance scattering peaks of ZnO spheres with dimeter of 700 nm showed different degrees of redshift,which lead to different structural color appeared when ZnO spheres deposited on different metal surfaces with a similar appearance.A red structural color was displayed on the surface of the real silver ornament and a yellow-green structural color was shown on the surface of the cupronickel ornament.This method is quite simple and low-cost because it only needs to spray the dispersion of ZnO spheres on the ornament surface.Due to the mild chemical properties of the ZnO,covering and erasing ZnO spheres on the surface of silver would not corrode the silver ornament.Finally,an atomizer method was used for portable and daily testing.This work opens new perspectives on the visual identification of silver.

Visual Modflow模型在马城铁矿床地下水数值模拟中的应用

通过马城铁矿床水文地质条件概化,用勘探时抽水试验测得的水位变化资料对Visual Modflow三维地下水模型进行模拟与识别,校准后的模型真实可靠,可用于马城铁矿床地下水涌水量的预测,考虑上下两个阶段采场同时开采,按不同开采阶段组合模式计算了地下水涌水量和动态流场,为排水系统设计提供可靠依据。

Physical modeling and visualization of soil liquefaction under high confining stress

The mechanisms of seismically-induced liquefaction of granular soils underhigh confining stresses are still not fully understood.Evaluation of these mechanisms is generallybased on extrapolation of observed behavior at shallow depths.Three centrifuge model tests wereconducted at RPI's experimental facility to investigate the effects of confining stresses on thedynamic response of a deep horizontal deposit of saturated sand.Liquefaction was observed at highconfining stresses in each of the tests.A system identification procedure was used to estimate theassociated shear strain and stress time histories.These histories revealed a response marked byshear strength degradation and dilative patterns.The recorded accelerations and pore pressures wereemployed to generate visual animations of the models.These visualizations revealed a liquefactionfront traveling downward and leading to large shear strains and isolation of upper soil layers.

Computer Identification of Multispectral Satellite Cloud Imagery

A dynamic clustering method based on multispectral satellite imagery to identify the different features is described. The channel combinations selected are for the different purposes in classification. Several cases are presented using the polar-orbiting satellite imageries.

Dynamic Visualization Approach of the Multiphase Flow Using Electrical Capacitance Tomography

Identifying the flow patterns is vital for understanding the complicated physical mechanisms in multiphase flows.For this purpose,electrical capacitance tomography(ECT) technique is considered as a promising visualization method for the flow pattern identification,in which image reconstruction algorithms play an important role.In this paper,a generalized dynamic reconstruction model,which integrates ECT measurement information and physical evolution information of the objects of interest,was presented.A generalized objective functional that simultaneously considers the spatial constraints,temporal constraints and dynamic evolution information of the objects of interest was proposed.Numerical simulations and experiments were implemented to evaluate the feasibility and efficiency of the proposed algorithm.For the cases considered in this paper,the proposed algorithm can well reconstruct the flow patterns,and the quality of the reconstructed images is improved,which indicates that the proposed algorithm is competent to reconstruct the flow patterns in the visualization of multiphase flows.

基于Visual Basic的木材识别系统

基于Visual Basic编程语言与Microsoft Office Access数据库,研发了一套木材识别系统,包含200余个常见商品材树种数据,具有提取木材微观图像中的木材参量、根据木材特征自动匹配相似木材、储存与管理检测木材样本信息、检索与浏览木材数据库等功能,可辅助木材树种准确识别,并提高木材鉴定速度。

内河航道船舶大气污染物扩散模拟研究

为探究内河航道船舶大气污染物在沿岸建筑物群的扩散规律,基于船舶自动识别系统数据构建南京市内河某航道段船舶大气污染物排放计算模型,采用“自下而上”方法计算得到船舶SO_(2)、NO_(2)、PM_(10)、PM_(2.5)、CO和挥发性有机物6种大气污染物排放量。通过Revit软件与Fluent软件建立船舶大气污染物在沿江居民区的扩散模型并进行可视化模拟分析。结果表明:在常风向全年平均风速条件下,污染物扩散与建筑群布局密切相关,在密集建筑物群背风面易出现负压区,该处污染物浓度较低,但在整体建筑物群背风面稍远处,出现气流旋涡区,导致较高浓度的污染物聚集,短期内产生持续污染影响。研究思路与结果可为航道沿岸居民区规划建设,降低船舶大气污染物影响提供理论依据。

基于视觉图像与激光点云融合的交通标志快速识别方法

交通标志对车辆交通起到重要作用和意义,而智能交通中交通标志识别由于标志特征提取效果差,导致识别率低、识别时间长,因此,提出一种新的基于视觉图像与激光点云融合的交通标志快速识别方法。采用双边滤波方法预处理原始激光点云数据;通过归一化处理得到视觉图像激光点云融合的目标空间激光点云位置测距数值;通过测距值获取目标图像位置,归一化处理交通标志视觉图像,引入k均值聚类算法二聚类处理图像,采用制作的切割模板切割图像感兴趣区域,提取交通标志图像的深度特征,结合卷积神经网络二次过滤特征,重新标定二次过滤后的特征,最终利用卷积神经网络模型实现交通标志快速识别;经实验对比证明,采用所提方法提取各个类型交通标志特征的提取效果较好,并且识别率达到89.74%,识别时间仅为13.1 s,干扰下识别时间最高仅为15.1 s,验证了该方法可以快速且准确识别各个类型的交通标志。

荧光探针合成与可视化检测三硝基苯酚综合实验开发

根据待测物2,4,6-三硝基苯酚(TNP)的性质,以香豆素为荧光团、噻二唑为识别基团,调整共轭体系和电子流向,设计具有绿色荧光的有机小分子探针,实现特异性检测与可视化识别。引导学生深入探索化学学科的内在联系,掌握基础知识的同时了解学科领域发展。设置智能手机检测模块,将科学研究与日常生活紧密联接,拉近学生与科学研究的距离,激发学生探索创新的意识。

基于孪生神经网络的楔形环连接结构预紧状态辨识

楔形环连接结构由于其连接简单可靠、同时兼具节省空间及减重的优势,常被应用于鱼雷、航天飞行器等武器装备。针对楔形环连接结构预紧状态辨识方面存在的机理模型复杂、样本量小且类别不平衡的问题,提出了一种基于孪生神经网络模型的预紧状态辨识方法。为提高模型训练效率和效果,首先利用时频处理技术进行孪生神经网络模型特征增强,基于增强特征建立了3层孪生神经网络分类模型,实现楔形环预紧状态宏观分类。同时,为指导楔形环精密装配,通过特征可视化技术,深入分析了孪生神经网络训练过程特征聚类效果,并基于二维特征建立了预紧状态定量表征模型,引入目标状态聚类中心与接受域参量,用于实现楔形环连接结构预紧状态定量评估。通过试验验证了所提方法的有效性,该方法可为楔形环连接结构定量辨识提供新的技术途径和思路,具有一定工程应用价值。

多特征融合的无人艇视觉目标长时相关鲁棒跟踪

[目的]针对显著海浪遮挡、相机剧烈晃动引起的无人艇视觉目标跟踪脱靶问题,提出一种基于多特征融合的长时相关鲁棒跟踪算法。[方法]首先,采用多特征融合技术,增强目标特征表达,提高目标模型鲁棒性;其次,利用高维特征降维和响应图子网格插值,提高目标跟踪的效率与精度;然后,设计水面目标重识别机制,解决目标完全脱离视野时的稳定跟踪问题;最后,采用多个代表性视频数据集进行验证和比较分析。[结果]实验结果表明,相较于传统的长时相关跟踪算法,平均成功率提升15.7%,平均距离精度指标提升30.3%,F-Score指标提升7.0%。[结论]所提算法能够处理恶劣海况下的目标脱靶问题,对于提升无人船艇及海洋机器人智能感知能力,具有重要技术支撑意义。

基于行车风险场的快速路短交织区车辆交互风险识别

为更加准确和直观地识别快速路短交织区行驶车辆之间的交互风险,解决传统风险识别模型无法连续识别换道时车辆间交互风险的问题,本文利用无人机采集快速路短交织区内车辆轨迹数据,并利用Tracker进行读取,筛选跟驰车辆对和换道车辆对,在现有行车风险场理论的基础上,综合考虑车辆面积、车辆前进时前后方风险差异、转弯时左右方风险差异以及车辆之间横向距离,适应性改进现有行车风险场模型DRF(Driving Safety Field),利用遗传算法和波良可夫模型标定模型参数。对比改进后行车风险场模型、车头时距倒数(THWI)、碰撞时间倒数(TTCI)对车辆间跟驰交互风险与换道交互风险的识别情况,验证该模型的有效性。结果表明,本文模型较THWI和TTCI更符合驾驶员的驾驶心理,且感知风险变化先于驾驶员,对车辆换道交互风险的识别率较THWI提升52.45%,较TTCI提升83.66%,本模型在换道交互风险的识别性能上表现的更加优越。基于本文模型对短交织区区域内多车共同作用产生的风险进行可视化,可辅助交通管理部门确定需要精细化组织的关键区域,也可作为评价管理措施改善效果的可视化手段。

视觉形象的“他者化”与“自我他者化”探微——从中外视觉艺术中的“眯眯眼”谈起

艺术中的“他者”形象有时不是直接依据客观真实进行塑造的,其间既有带着偏见的丑化的形象,也有出于认同的幻想的形象。“他者”形象的塑造受到权力关系的制约,对“他者”的误读、扭曲、美化往往与本土艺术创作者所处时代的地缘政治、话语权力分布及文化认同程度关系密切。我们应当尊重真实、客观地反映生活的作品,反对其他族群视觉艺术中隐含着的对中国人、中国文化的歪曲表现,同时应能判析本国视觉艺术中是否存在“自我他者化”的倾向,并展现出真正的原创力,创作出切实促进文明交流互鉴的作品。

基于Faster-RCNN网络的接箍自动识别方法

在可视化测井中,深度对于判断油管缺陷位置、射孔位置至关重要,而现有的测深系统具有一定的深度误差。在观测井下视频人工查找接箍,存在耗时、检测速度慢等问题。先对VideoLog油气井可视化测井技术采集到的井下视频进行图像增强,引入Faster-RCNN模型,采用ResNet50网络作为特征提取网络提取接箍特征,最终通过兴趣区域池化网络和全连接层完成接箍的识别定位。该模型有着0.99的平均精度,在实验中,视频中的接箍均可被准确识别,具有识别速度快和准确率高等优点。

遥感影像水体识别研究进展及热点分析

洪涝灾害是严重危害人类社会发展的自然灾害之一,从遥感影像中准确识别水体信息对提升突发洪涝灾害事件的应急处理、灾害预警具有重要意义,一直受到专家学者的关注。针对国内外专家学者利用遥感影像进行水体识别的相关研究成果,文章从传统方法以及深度学习法两大方面,对前人提出的水体识别理论和方法进行分析研究;在此基础上,利用文献计量的方法对水体识别当前的研究热点进行梳理分析,针对现存问题和当前热点对基于遥感影像的水体识别未来发展趋势进行了展望。

面向目视解译的全极化SAR船只精细化特征表征方法

随着卫星技术的发展,极化合成孔径雷达(PolSAR)数据的分辨率和数据质量得到大幅提升,为人造目标的精细化目视解译提供了良好的数据条件。目前主要采用多分量分解的方法,但是易造成像素错分问题,为此,该文结合Yamaguchi极化分解和极化熵提出了一种非固定阈值划分的方法用于实现全极化SAR图像船只结构精细化特征表征。Yamaguchi极化分解能够识别基本散射机制,其修正后的体散射模型更符合实测数据,可有效对人造目标进行表征。极化熵H在弱去极化状态下可以看成某一指定等效点的目标散射机制,能够有效突出船只主散射特征。因此,该文通过将Yamaguchi极化分解算法的非固定三分量与极化熵的低中高熵内嵌,将其分为非固定阈值的九分类成分,从而降低硬阈值处理在阈值边界处受噪声影响产生的类别随机性。并且将二次散射和单次散射均显著的区域称为混合散射(MSM),以更好匹配实验中船只典型结构的散射类型。在此基础上,利用广义相似性参数进一步缩短类内距离,采用改进后的GSP-Wishart分类器进行迭代聚类,旨在通过提高二次散射和混合散射机制以提高不同类型船只可区分度。最后,该文采用中国上海某港口的高分三号全极化SAR数据进行实验,为了验证每艘船只特征表征正确性,通过船舶自动识别系统(AIS)收集并筛选了该港口船只信息及光学数据,并与极化SAR数据中每艘船只进行匹配。实验结果表明该方法可有效区分散货船、集装箱船和油轮3种类型船只。

融合密集卷积网络和注意力机制的拱桥损伤识别

针对传统深度学习方法缺乏对网络特征的差异化利用且损伤识别精度易受环境因素影响的问题,提出了一种融合密集卷积网络(DenseNet121)和混合注意力机制(convolutional block attention module,CBAM)的拱桥损伤识别方法。首先,获取拱桥加速度响应数据,利用连续小波变换将其转换成时频图,形成拱桥损伤识别数据集;其次,将CBAM嵌入DenseNet121模型,加强断层特征的传播和特征的差异化利用,经训练得到拱桥损伤识别模型;然后,基于测试集评估损伤识别模型的性能,引入t分布随机邻域嵌入非线性降维技术对特征进行可视化分析;最后,通过数值案例验证了该方法的可行性和鲁棒性,并应用于劲性骨架拱肋的损伤识别。结果表明:所提方法可增强有用信息的权重,实现网络特征的差异化利用;与传统方法相比,该方法在单损伤和多损伤识别中准确率分别达到了91.67%和92.78%,准确率更高,且具有较强的鲁棒性和实用价值。