Automatic recognition of defects in plasma-facing material using image processing technology

Observing and analyzing surface images is critical for studying the interaction between plasma and irradiated plasma-facing materials.This paper presents a method for the automatic recognition of bubbles in transmission electron microscope(TEM)images of W nanofibers using image processing techniques and convolutional neural network(CNN).We employ a three-stage approach consisting of Otsu,local-threshold,and watershed segmentation to extract bubbles from noisy images.To address over-segmentation,we propose a combination of area factor and radial pixel intensity scanning.A CNN is used to recognize bubbles,outperforming traditional neural network models such as Alex Net and Google Net with an accuracy of 97.1%and recall of 98.6%.Our method is tested on both clear and blurred TEM images,and demonstrates humanlike performance in recognizing bubbles.This work contributes to the development of quantitative image analysis in the field of plasma-material interactions,offering a scalable solution for analyzing material defects.Overall,this study's findings establish the potential for automatic defect recognition and its applications in the assessment of plasma-material interactions.This method can be employed in a variety of specialties,including plasma physics and materials science.

Semi-supervised learning based probabilistic latent semantic analysis for automatic image annotation

In recent years,multimedia annotation problem has been attracting significant research attention in multimedia and computer vision areas,especially for automatic image annotation,whose purpose is to provide an efficient and effective searching environment for users to query their images more easily. In this paper,a semi-supervised learning based probabilistic latent semantic analysis( PLSA) model for automatic image annotation is presenred. Since it's often hard to obtain or create labeled images in large quantities while unlabeled ones are easier to collect,a transductive support vector machine( TSVM) is exploited to enhance the quality of the training image data. Then,different image features with different magnitudes will result in different performance for automatic image annotation. To this end,a Gaussian normalization method is utilized to normalize different features extracted from effective image regions segmented by the normalized cuts algorithm so as to reserve the intrinsic content of images as complete as possible. Finally,a PLSA model with asymmetric modalities is constructed based on the expectation maximization( EM) algorithm to predict a candidate set of annotations with confidence scores. Extensive experiments on the general-purpose Corel5k dataset demonstrate that the proposed model can significantly improve performance of traditional PLSA for the task of automatic image annotation.

In-situ strain measurement and error analysis of arc welding with 2D digital image correlation

The deformation and residual stress generated by the welding process can seriously affect the use of components.As a result,it is very important to understand the evolution of stress and strain during the welding process.The strain measurement method based on digital image correlation(DIC)is an excellent method to detect welding strain and residual stress.The out-of-plane translation and out-of-plane rotation introduce errors to the two-dimensional DIC.In this paper,the causes of errors are analyzed theoretically,and the formulas of errors caused by the out-of-plane displacement and the out-of-plane rotation are derived.The out-of-plane translation experiment and the out-of-plane rotation experiment were carried out to verify the theory,and the experimental results are consistent with the theoretical analysis results.The error caused by the out-of-plane translation can be reduced by increasing the object distance;the error caused by the out-of-plane rotation is greatly affected by the rotation angle.

Automatic Computer Analysis of Digital Images of Triple-Antibody-Stained Prostate Biopsies

Background: Worldwide, prostatic adenocarcinoma is the most common tumour type among men. Aim: The aim of the present investigation was to develop a computer program to identify normal prostate biopsies and distinguish them from biopsies showing premalignant alterations (LGPIN, HGPIN) and adenocarcinoma. Method: Prostate biopsies (n = 2094) taken from 191 consecutive men during 2016 were stained with triple immunehistochemisty (antibodies to AMACRA, p63 and CK 5). Digital images of the biopsies were obtained with a scanning microscope and used to develop an automatic computer program (CelldaTM), intended to identify the morphological alterations. Visual microscopic finding was used as a reference. Result: Of the 191 men, 121 (63.4%) were diagnosed as having prostate adenocarcinoma and 70 (36.6%) as having no malignancy on the basis of the visual microscopy. In comparison, computer analysis identified 134 (70.2%) men with malignant disease and 57 (29.8%) with non-malignant disease after exclusion of artifacts, which constituted 10.4% of areas (indicated as malignant disease). Discrepant results were recorded in 15 (7.9%) men, and in 14 of these cases, HGPIN and areas suggestive of early invasion were common. Thus, it was uncertain whether these cases should be regarded as malignant or not. The agreement between the visual examination and the computer analysis was 92.1% (kappa value 0.823, sensitivity 99.2 and specificity was 0.80). Conclusion: It seems that computer analysis could serve as an adjunct to simplify and shorten the diagnostic procedure, first of all by ensuring that normal prostate biopsies are sorted out from those sent for visual microscopic evaluation.

Evaluation of Characteristics of Non-Metallic Inclusions in P/M Ni-Base Superalloy by Automatic Image Analysis

Non-metallic inclusions,especially the large ones,within P/M Ni-base superalloy have a major influence on fatigue characteristics,but are not directly measurable by routine inspection.In this paper,a method,automatic image analysis,is proposed for estimation of the content,size and amount of non-metallic inclusions in superalloy.The methodology for the practical application of this method is described and the factors affecting the precision of the estimation are discussed.In the experiment,the characteristics of the non-metallic inclusions in Ni-base P/M superalloy are analyzed.

Image-based Water Level Measurement Method under Stained Ruler

This paper propoes the water level measuring method based on the image, while the ruler used to indicate the water level is stained. The contamination of the ruler weakens or eliminates many features which are required for the image processing. However, the feature of the color difference between the ruler and the water surface are firmer on the environmental change compare to the other features. As the color differeaces are embossed, only the region of the ruler is limited to eliminate the noise, and the average image is produced by using several continuous frames. A histogram is then produced on the height axis of the produced intensity average image. Local peaks and local valleys are detected, and the section between the peak and valley which have the greatest change is looked for. The valley point at this very moment is used to detect the water level. The detected water level is then converted to the actual water level by using the mapping table. The proposed method is compared to the ultrasonic based method to evaluate its accuracy and efficiency on the various contaminated environments.

Fusing PLSA model and Markov random fields for automatic image annotation

A novel image auto-annotation method is presented based on probabilistic latent semantic analysis(PLSA) model and multiple Markov random fields(MRF).A PLSA model with asymmetric modalities is first constructed to estimate the joint probability between images and semantic concepts,then a subgraph is extracted served as the corresponding structure of Markov random fields and inference over it is performed by the iterative conditional modes so as to capture the final annotation for the image.The novelty of our method mainly lies in two aspects:exploiting PLSA to estimate the joint probability between images and semantic concepts as well as multiple MRF to further explore the semantic context among keywords for accurate image annotation.To demonstrate the effectiveness of this approach,an experiment on the Corel5 k dataset is conducted and its results are compared favorably with the current state-of-the-art approaches.

Digital Image Correlation Using Specific Shape Function for Stress Intensity Factor Measurement

The stress intensity factor (SIF) is a critical parameter associated with the fracture behaviour of materials. In this paper, we select the displacement function around a crack tip as the shape function of the digital image correlation (DIC), which makes it possible to directly calculate the SIF by the correlation scheme. Moreover, we use a non-rectangular subset, which can reduce the influence of plastic deformation and crack width on the DIC measurement accuracy. We measured the SIF of a mode I crack in a super-hard aluminium alloy specimen to verify the performance of the proposed method. Our experimental results show that a DIC with a specific shape function can be used to accurately and efficiently calculate the SIF. Furthermore, we also present a practical application of our proposed method for determining the SIF, crack propagation angle and crack tip displacement. © 2017, Tianjin University and Springer-Verlag Berlin Heidelberg.

Evaluation of Flight Trajectory and Unsteady Fluid Forces on Kicked Non-Spinning Soccer Ball by Digital Image Analysis

This paper describes the experimental method for evaluating the flight trajectory and the aerodynamic performance of a kicked non-spinning soccer ball. The flight trajectory measurement is carried out using the digital image analysis. A centroid method and a template matching method are tested for the flight trajectory analysis using the artificial images generated by the data of a free-fall experiment. The drag coefficient obtained by the centroid method is better suited for the sports ball experiment than that by the template matching method, which is due to the robustness of the centroid method to the non-uniform illumination. Then, the flight trajectory analysis is introduced to a kicked experiment for a non-spinning soccer ball. The experimental result obtained from the stereo observation indicates that the S-shaped variation is found in the three-dimensional flight trajectory and in the side force coefficient during the flight of the non-spinning soccer ball.

“To measure is to know”: how advances in image analysis are supporting neural repair strategies

Neuronal regeneration in the peripheral nervous system arises via a synergistic interplay of neurotrophic factors,integrins,cytoskeletal proteins,mechanical cues,cytokines,stem cells,glial cells and astrocytes.

Support vector machine with mixture of kernels for automatic image annotation

Automatic image annotation(AIA)has become an important and challenging problem in computer vision due to the existence of semantic gap.In this paper,a novel support vector machine with mixture of kernels(SVM-MK)for automatic image annotation is proposed.On one hand,the combined global and local block-based image features are extracted in order to reflect the intrinsic content of images as complete as possible.On the other hand,SVM-MK is constructed to shoot for better annotating performance.Experimental results on Corel dataset show that the proposed image feature representation method as well as automatic image annotation classifier,SVM-MK,can achieve higher annotating accuracy than SVM with any single kernel and mi-SVM for semantic image annotation.

基于合成数据集的多目标识别与6-DoF位姿估计

多目标识别及六自由度(6-DoF)位姿估计是实现物料无序堆放状态下机器人自动分拣的关键。近年来,基于深度神经网络的方法在目标识别及位姿估计领域受到广泛关注,但此类方法依赖大量训练样本,而样本的采集及标注费时费力,限制了其实用性。其次,当成像条件差、目标相互遮挡时,现有位姿估计方法无法保证结果的可靠性,进而导致抓取失败。为此,文中提出了一种基于合成数据样本的目标识别、分割及位姿估计方法。首先,以目标对象的3维(3D)几何模型为基础,利用3D图形编程工具生成虚拟场景的多视角RGB-D合成图像,并对生成的RGB图像及深度图像分别进行风格迁移和噪声增强,从而提高合成数据的真实感,以适应真实场景的检测需要;接着,利用合成数据集训练YOLOv7-mask实例分割模型,运用真实数据进行测试,结果验证了该方法的有效性;然后,以分割结果为基础,基于ES6D目标位姿估计模型,提出了一种在线姿态评估方法,以自动滤除严重失真的估计结果;最后,采用基于主动视觉的位姿估计校正策略,引导机械臂运动到新的视角重新检测,以解决因遮挡而导致位姿估计偏差的问题。在自行搭建的6自由度工业机器人视觉分拣系统上进行了实验,结果表明,文中提出的方法能较好地适应复杂环境下工件的识别与6-DoF姿态估计要求。

人工智能在视网膜图像自动分割和疾病诊断中的应用指南(2024)

人工智能技术的快速发展推动了医学的智能化进程。近年来,随着机器学习和深度学习等技术的不断提高,人工智能技术在眼底疾病诊疗中得到了快速发展和应用。眼底疾病主要包括视网膜血管病、黄斑疾病、视网膜脱离、视网膜色素变性等,早期诊断及治疗对改善眼底疾病的预后具有重大意义。本文就人工智能在视网膜图像自动分割和疾病诊断中的应用形成指南,为人工智能在该领域中的进一步研究和应用提供参考。

深度学习对膝骨关节炎MRI图像智能分割和测量分析的作用及意义

背景:MRI对诊断早期膝骨关节炎有重要意义。利用深度学习方法进行膝骨关节炎的MRI图像识别和智能分割,是目前人工智能在影像诊断方面的研究热点。目的:通过对膝骨关节炎病例MRI图像的深度学习,能够全自动分割膝关节的股骨、胫骨、髌骨、软骨、半月板、韧带、肌肉及关节积液,并测量膝关节积液体积和肌肉含量。方法:筛选出100个正常膝关节和100个膝骨关节炎患者数据,按照8︰1︰1的比例随机分为训练集(traindataset,n=160)、调优集(validation dataset,n=20)和测试集(test dataset,n=20)。采用Coarse-to-Fine序贯训练的方法训练3D-UNET网络深度学习模型,先训练一个膝关节矢状面MRI粗略分割模型,将得到的粗略分割结果作为掩膜(mask),再训练精细分割模型。输入膝关节矢状面T1WI、T2WI图像和各结构的标注文件,运用DeepLab v3,分割骨、软骨、韧带、半月板、肌肉、积液,最后显示三维重建,显示自动测量结果(肌肉的含量、积液的体积),完成深度学习的应用程序。再筛选出26例正常人和38例膝骨关节炎患者的膝关节MRI数据进行测试验证。结果与结论:①26例正常人中女13例,男13例,平均年龄(34.88±11.75)岁,膝关节中肌肉含量平均值(1051322.94±2007249.00)mL,均值中位数为631165.21 mL;积液的体积平均值(291.85±559.59)mL,均值中位数0 mL。②38例膝骨关节炎患者中女30例,男8例,平均年龄(68.53±9.87)岁;肌肉含量平均值(782409.18±331392.56)mL,均值中位数689105.66 mL;积液的体积平均值(1625.23±5014.03)mL,均值中位数178.72 mL。③正常人的肌肉含量与膝骨关节炎患者的相差不大,差异无显著性意义;而膝骨关节炎患者积液的体积高于正常人,差异有显著性意义(P<0.05)。④提示通过深度学习对膝骨关节炎MRI图像进行智能分割,可摒弃以往手工分割的缺陷;对膝骨关节炎的评估需要更加精细化,将图像分割处理得更加精细,以提高结果的精度。

基于神经网络深度学习模型的踝关节X线片标志点自动定位研究

目的探索基于神经网络深度学习模型的踝关节X线片标志点自动定位方法及其应用价值。方法选取陕西省人民医院2019年1月至2022年11月间行X线检查的360例成年人正常左踝关节正、侧位片影像资料为研究对象,将其随机分配至训练集(210例)、验证集(90例)和测试集(60例)。以人工标注作为参考,对图像预处理后分别建立基于神经网络Unet架构的踝关节X线片标志点预测模型,生成对应的热力图,并用测试集数据进行验证。结果在踝关节X线正位片6个标志点的预测中,2 mm阈值的平均正确估计比例(Percentage of Correct Keypoints,PCK)可达99.7%,总体平均径向误差(Mean Radial Errors,MRE)为0.411,总体标准差(Standard Deviation,SD)为0.290。距骨顶端内点的预测准确度最高,1 mm阈值时的PCK可达100%,同时其MRE及SD在正位片6个点中最小,分别为0.290和0.178。在踝关节X线侧位片9个标志点的预测中,2 mm阈值的平均PCK达到95.0%,总体MRE为0.669,总体SD为0.710。胫骨下段最前点的预测准确度最高,1 mm阈值时的PCK可达100%,同时其MRE及SD在侧位片9个点中最小,分别为0.334和0.173。正位片和侧位片所有标志点的预测位置坐标与对应参考标准标志点位置坐标差异均无统计学意义(P>0.05)。结论基于神经网络深度学习模型能够实现对踝关节X线片标志点的有效自动定位,对辅助踝关节X线片形态学自动测量和疾病诊疗具有应用价值。

地面目标的中波红外高光谱成像特性测量

利用Telops公司设计的Hyper-Cam中波高光谱成像仪对水泥地上某坦克模型进行3~5μm波段的红外高光谱成像测量。对实验数据进行了逐像素点的大气校正。基于矫正后的实验数据,分析了仪器误差、随机误差、大气校正传递误差等对测试不确定度的影响。分析了该坦克模型在不同谱段上的亮度分布和模型上不同部位的光谱分布特性。结果表明:在波数为2000~3000 cm-1波段,其相对不确定度一直稳定在10%以内,但是在大于3000 cm-1的波段,其误差迅速上升,主要因为常温物体的中波辐射低,在该波段目标辐射与大气路程辐射接近,而引起的噪声增大,造成的测试信噪比降低。该坦克模型光谱辐射亮度测试数据的平均相对不确定度在20%以内,整体误差较低。光谱分布上,坦克各部分特征点和水泥地表的光谱辐射亮度在波长为4.2~5μm波段时的差异性比波长为3.0~4.2μm波段时的差异性更大,且探测器接收到的4.2~5μm波段的光谱辐射亮度要大于3.0~4.2μm的光谱辐射亮度。由于4.3μm波段处于大气吸收带,大气透射率几乎为0,以至于无法获得目标的光谱辐射亮度的真实测量值。空间分布上,模型的辐射亮度主要集中在侧边涂层上,炮口处和履带处的辐射亮度较小。目标各部位辐射亮度分布与周边水泥地表的辐射亮度有着明显差异性。在3~5μm的辐射亮度中,4.4~4.8μm波段的辐射亮度占比较高,该波段成像效果最好。3.2~3.8μm波段的辐射亮度占比较低,该波段成像效果较差。处于大气吸收带的4.2~4.4μm波段,由于大气透射率几乎为0,无法获得其准确的空间辐射亮度分布。研究表明该拓展中波高光谱成像仪在地面目标的中波红外高光谱成像特性研究上具有目标区分精确、误差小、“图像光谱合一”的特点,测试数据可以应用于目标的红外高光谱隐身设计、目标的分类与识别等领域。

岩体暴露面图像中裂隙全自动提取方法研究

裂隙显著影响岩体物理与力学性质,针对从岩体暴露面图像中自动提取裂隙存在的提取结果不完整、信噪比低,以及提取方法经验参数多、鲁棒性差等问题,提出一种全自动裂隙提取优化程序。首先,通过引入广义伽马校正初步提升裂隙与岩壁面对比度;其次,考虑裂隙路径像素间相互影响,设计灰度传递算法提升裂隙连续性;最后,通过改进Frangi滤波器提取裂隙并充分抑制噪音像素响应。结果表明:该程序充分融合了裂隙低灰度和高线型两大特征,图像增强与裂隙提取两阶段协同运作,显著改善了裂隙对比度不均问题,在较完整提取裂隙的同时还能够有效防止噪音与伪裂隙的产生,对多种岩体暴露面图像均具有很高的鲁棒性。将该方法与常用的裂隙识别算法进行比较,展现了该算法流程在处理低对比度、高噪声岩体裂隙图像方面的显著优势。

基于自动脑分割技术对海马硬化型颞叶内侧癫痫颞叶白质体积的分析及其应用价值

目的:利用自动脑分割技术(FreeSurfer)分析海马硬化型颞叶内侧癫痫(MTLE-HS)患者颞叶亚区白质体积的变化,探讨该技术在MTLE-HS中的临床应用价值。方法:搜集2021年1月至2023年9月经术后病理或MRI诊断为MTLE-HS的患者53例,其中左侧MTLE-HS(LMTLE-HS)30例,右侧MTLE-HS(RMTLE-HS)23例;同期招募性别、年龄相匹配的健康对照者43例。所有受试者均在3.0T磁共振上行T 1加权三维磁化强度预备梯度回波序列(3D-T 1WI-MPRAGE)扫描。使用FreeSurfer软件对T 1-MARAGE图像进行全脑分割,搜集颞叶亚区(颞上回后部、颞极、颞横回、颞上回、颞中回、颞下回、梭状回、海马旁回)的白质体积数据,采用配对t检验比较对照组左、右两侧颞叶各亚区白质体积的差异,采用独立样本t检验分别比较对照组与LMTLE-HS组、RMTLE-HS组患侧、对侧之间颞叶各亚区白质体积的差异。结果:对照组左、右两侧颞横回、颞上回、颞中回、颞下回、梭状回的白质体积差异有统计学意义(P<0.05)。对照组左、右两侧分别与LMTLE-HS组、RMTLE-HS组患侧、对侧进行比较,结果显示对照组左侧与LMTLE-HS组患侧颞极、颞横回、颞上回、颞中回、颞下回、梭状回、海马旁回的白质体积差异有统计学意义(P<0.05),对照组右侧与LMTLE-HS组对侧颞上回后部、颞上回、颞中回、颞下回、梭状回、海马旁回及RMTLE-HS组患侧颞上回后部、颞中回、颞下回、梭状回、海马旁回的白质体积差异有统计学意义(P<0.05),其余亚区体积差异无统计学意义(P>0.05)。结论:基于自动脑分割技术的MTLE-HS颞区白质体积定量分析对颞叶癫痫术前评估及手术方式的选择具有一定意义。

小波分析的自适应卡尔曼滤波模型在地铁隧道变形监测中的应用

为了加强地铁安全保护机制,基于测量机器人智能化、自动化的特点,对某试验区的地铁隧道进行变形监测,选择Trimble S9 HP测量机器人进行数据采集,通过云平台进行数据处理及变形分析,最后利用小波分析的自适应卡尔曼滤波模型对后期形变量进行预测。结果表明,自动测量机器人的测量精度满足隧道监测要求,完成了地铁隧道变形监测的预设目标,分析隧道结构的变形特征并通过小波分析的自适应卡尔曼滤波模型进行变形预测,所得预测数据精度较高,可以为今后工程建设和地铁维护提供参考。

基于频谱分析模块的机载雷达信号参数流程化自动测量方法

针对部队外场飞机雷达维护工作需求,本文提出了一种基于频谱分析模块的机载雷达信号参数流程化自动测量方法。文中分析了典型雷达信号时域和频域的特征,研究了多项参数自动测量的原理和步骤,并开发了相应的测试软件用以执行相关算法和实现流程化测量。此外,还设计了4种信号仿真实验来验证本文所提方法的有效性,实验结果表明,该方法无需借助传统通用仪器,仅基于频谱分析模块在不同工作模式下提供的数据序列,即可实现典型雷达信号多项参数一键式流程化测量,且测量结果精确有效,符合部队外场维护保障条件,因此具有较强的工程应用价值。