Study of image processing for V-shape groove and robotic weld seam tracking based on laser vision

Single-stripe laser was applied to acquire V-shape groove contour information. Most of arc light and splash noise was removed and stripe laser image was kept by wavelet transform. Half-threshold algorithm was used for image segmentation and stripe laser image was gotten after refining. Weld seam center position was identified and extracted by extreme curvature corner detection method. The location of torch was detected to accord the frequency of computer program with robot program and serial communication program. The tracking experiments of sidelong, reflex and curve weld line show that the system can meet the demand of the tracking precision under normal welding conditions.

Study on the image processing of laser vision seam tracking system

Seam image processing is the basis of the realization of automatic laser vision seam tracking system, and it has become one of the important research directions. Adding windows processing, gray processing, fast median filtering, binary processing and image edge extraction are used to pretreat the seam image. In the post-processing of seam image, the feature points of the target image are succesfully detected by using center line extraction and feature points detection algorithm based on slope analysis. The whole processing time is less than 150 ms, and the real-time processing of seam image can be implemented.

Detection and processing of molten-pool image of electron-beam deep-penetration welding based on visual sensing

A visual sensing system was developed. The system is suitable for titanium-alloy electron-beam welding, and senses and detects molten-pool dynamic processes. A suite of processing programs for colored molten-pool images in titanium-alloy electron-beam welding was developed using Matlab software; molten-pool edge images are completely obtained using the program. The Matlab software was used to write a program which could extract the molten-pool width. The functional relationship between the molten-pool width and penetration under the experimental conditions was obtained by a curve-fitting method, and provided the theoretical basis for further penetration control.

Vision Navigation Based PID Control for Line Tracking Robot

In a controlled indoor environment,line tracking has become the most practical and reliable navigation strategy for autonomous mobile robots.A line tracking robot is a self-mobile machine that can recognize and track a painted line on thefloor.In general,the path is set and can be visible,such as a black line on a white surface with high contrasting colors.The robot’s path is marked by a distinct line or track,which the robot follows to move.Several scientific contributions from the disciplines of vision and control have been made to mobile robot vision-based navigation.Localization,automated map generation,autonomous navigation and path tracking is all becoming more frequent in vision applications.A visual navigation line tracking robot should detect the line with a camera using an image processing technique.The paper focuses on combining computer vision techniques with a proportional-integral-derivative(PID)control-ler for automatic steering and speed control.A prototype line tracking robot is used to evaluate the proposed control strategy.

A machine vision approach to seam tracking in real-time in PAW of large-diameter stainless steel tube

Manual monitoring and seam tracking through watching weld pool images in real-time, by naked eyes or by industrial TV, are experience-depended, subjective, labor intensive, and sometimes biased. So it is necessary to realize the automation of computer-aided seam tracking. A PAW (plasma arc welding) seam tracking system was developed, which senses the molten pool and the seam in one frame by a vision sensor, and then detects the seam deviation to adjust the work piece motion adaptively to the seam position sensed by vision sensor. A novel molten pool area image-processing algorithm based on machine vision was proposed. The algorithm processes each image at the speed of 20 frames/second in real-time to extract three feature variables to get the seam deviation. It is proved experimentally that the algorithm is very fast and effective. Issues related to the algorithm are also discussed.

Weld pool boundary and weld bead shape reconstruction based on passive vision in P-GMAW

A passive visual sensing system is established in this research, and clear weld pool images in pulsed gas metal arc welding （ P-GMA W） can be captured with this system. The three-dimensional weld pool geometry, especially the weld height, is not only a crucial factor in determining workpiece mechanical properties, but also an important parameter for reflecting the penetration. A new three-dimensional （3D） model is established to describe the weld pool geometry in P-GMAW. Then, a series of algorithms are developed to extract the model geometrical parameters from the weld pool images. Furthermore, the method to reconstruct the 3 D shape of weld pool boundary and weld bead from the two-dimensional images is investigated.

Development of an automatic post-weld inspection system based on laser vision

In order to overcome the limitations of manual post-weld visual inspection approach, an automated inspection system is developed which uses three-dimensioual laser vision system based on the principle of optical triangulation. The system hardware consists of a modular development kit （MDK）, a computer, an actuating mechanism and so on. In image processing algorithms, extraction accuracy of centric line of laser stripe is the critical factor that determines the system performance. So according to the features of laser stripe image, a novel algorithm is developed to detect the central line of laser stripe fast and accurately. Experiments have demonstrated that this system can be used in various weld features inspection of both butt and fillet types of weld. Compared with traditional manual inspection method, this method has obvious dominance. The three-dimensional reconstruction result shows that this system has high accuracy and reliability.

An adaptive thresholding for weld seam image recognition in real time

Image prooessing of wehl seam in real time is an importunity to make welding rohot be able to track weld seam. The algorithm described in this paper combines some image technologies, such as modified Sobel edge detector and Hough transformation function, and especially the thresholds for image processing are ore aled adaptively by Ineans of a neural network. aests proved that this algorithm has a high reliability and rapidity in distinguishing the position of weld seam even with noises. The algorithm can be used ac the basic program .for robot to track welding seam and furthermore for calculating 3 dimensional information plan robot movement automatically.

Multiple Object Tracking through Background Learning

This paper discusses about the new approach of multiple object track-ing relative to background information.The concept of multiple object tracking through background learning is based upon the theory of relativity,that involves a frame of reference in spatial domain to localize and/or track any object.Thefield of multiple object tracking has seen a lot of research,but researchers have considered the background as redundant.However,in object tracking,the back-ground plays a vital role and leads to definite improvement in the overall process of tracking.In the present work an algorithm is proposed for the multiple object tracking through background learning.The learning framework is based on graph embedding approach for localizing multiple objects.The graph utilizes the inher-ent capabilities of depth modelling that assist in prior to track occlusion avoidance among multiple objects.The proposed algorithm has been compared with the recent work available in literature on numerous performance evaluation measures.It is observed that our proposed algorithm gives better performance.

基于电流多特征融合的窄间隙P-GMAW摆动电弧传感焊缝跟踪方法

电弧传感是实现窄间隙焊接焊缝跟踪的主要方式之一。针对电弧传感稳定性差、可靠性低的问题,提出融合摆动周期内电流信息的多个统计学特征进行偏差提取的方法,克服单一数据特征容易受到电弧稳定性影响、导致传感精度下降的问题。首先,提取电流信号的多个时域特征,计算特征矩阵与偏差矢量相关性;然后,取相关性高的特征用主成分分析的方法进行融合,取前两个主成分作为观测数据;最后,基于多分类的支持向量机模型对其进行分类试验。试验结果表明最大误差为0.2 mm,0.1 mm以内的误差占93.75%。该方法对比传统方法精度有所提升,对比神经网络方法,所用训练样本少,训练过程更加简单。

基于单目视觉的石化作业区外因明火感知与测量方法

针对目前石化作业区外因火灾监测方法大都没有火源定位功能的现状,提出基于单目视觉的外因明火感知与测量方法。首先改进YOLO v7深度学习网络的SiLU激活函数,并引入注意力机制CBAM,使感知网络获得更好的准确性、互适性;随后基于相机成像原理建立物距与像素、分辨率等因子的多元关系测量模型进行距离预测。实验表明:优化后的网络mAP_0.5值提升了2.2%,Precision值提升了5.0%,预测距离误差率绝对值小于3.1%。

Web-Based Object Tracking Using Collaborated Camera Network

The paper presents a web based vision system using a networked IP camera for tracking objects of interest. Three critical issues are addressed in this paper. First is the detection of moving objects in the foreground;second is the control of pan-tilt-zoom (PTZ) IP cameras based on object location;and third is the collaboration of multiple cameras over the network to track objects of interests independently. The developed system utilized a network of PTZ cameras along with a number of software tools for this implementation. The system was able to track a single and multiple objects successfully. The difficulties in the detection of moving objects are also analyzed while multiple cameras are collaborating over a network utilizing PTZ cameras.

基于归一化注意力机制的特征自适应融合目标跟踪算法

针对快速运动目标跟踪时图像的形变和低分辨率等问题,基于当前的孪生网络,提出一种基于归一化注意力机制的特征自适应融合目标跟踪算法。首先,通过轻量级的注意力机制抑制不太明显的权重,对注意力模块施加权重稀疏惩罚,并对主干网络最后4个特征层进行路径增强;其次,为捕捉在线跟踪过程中目标的外观变化,提升算法鲁棒性,提出了一种插件式的模板在线更新方法;最后,利用回归增强分类的方法完成对目标的跟踪。实验结果表明:该算法在OTB100,UAV123两个挑战性数据集上分别取得了63.3%和59.5%的较高成功率;同时,在外界光照变化、图像背景复杂、目标平面内旋转时,算法具有较强的鲁棒性。

基于机器学习的不锈钢薄板MIG焊焊穿缺陷识别

随着焊接工艺在工业环境中的广泛应用,对焊接缺陷的研究受到越来越多的关注,焊穿缺陷作为焊接过程中最严重的缺陷之一,必须采取相应的措施对其进行监测以及控制。提出了一种采用被动视觉传感技术,利用工业CCD相机结合减光、滤光系统消除焊接过程中强弧光干扰,获得MIG焊熔池图像的方法。设计正交试验并从图像的多维特征出发,提取熔池图像4个种类对应的10个维度特征作为机器学习的输入。在进行焊穿缺陷的标签标定时,基于所提出的特征提取对正交试验中获得的熔池图像数据进行预处理,通过反向搜索实现对无效数据的剔除和有效数据的正确归类,同时发现了数据在空间分布的规律。最后探究了不同维度输入特征下三种分类器的识别效果,结果表明:所选择的特征在三种分类器上均获得了不低于97%的分类精度,为后续开发基于视觉在线焊穿缺陷检测系统提供了技术支撑和理论依据。

基于改进型U-net神经网络的焊缝图像处理

对于基于线结构光的视觉焊缝跟踪系统,焊缝信息提取的精度和速度是焊缝图像处理算法的两个关键指标.对于含有强噪声的焊缝图像,传统的图像处理算法很难达到较高的提取精度;而基于传统全卷积神经网络的图像处理算法则可以有效地提取焊缝信息,但该网络仅对深层的抽象特征进行连续的上采样,忽略了焊缝像素点间空间位置的关系,从而割裂了空间的一致性,降低了焊缝信息提取的精度.针对以上问题,提出基于改进型U-net全卷积神经网络的焊缝图像处理模型以实现焊缝信息的逐级恢复.该模型通过引入双U型结构,将下采样倍率从16倍降为4倍,更多地保留了下采样时的空间信息;通过桥接的方式将第1次下采样时的焊缝特征传入第2次上采样阶段,为抽象的特征信息融入更多的空间信息,提高了焊缝信息提取的精度;将带泄漏的修正线性单元作为神经网络的激活函数,有效避免了原生U-net网络神经元坏死的现象.网络训练结果表明,使用相同数据集训练时,与传统FCN-32s网络和原生U-net网络相比,该模型的像素精度、平均像素精度和平均交集对联合均为最高.实验结果表明:该模型的焊缝位置提取平均偏差为1.64 mm,单帧焊缝图像处理时间为6.4 ms;该模型对含强噪声图像的焊缝信息提取精度和速度均优于Sterger算法和传统FCN-32s网络.

焊接轨迹跟踪控制中的深度视觉研究进展

建立深度视觉的焊缝跟踪控制系统,能准确快速识别焊缝轨迹,实现焊缝实时跟踪,维持焊炬在整个焊接过程中始终与焊缝精确实时对中,在自动焊接领域应用潜力巨大。总结了传统图像焊缝识别方法和深度图像焊缝识别方法的发展现状,归纳分析了传统图像识别方法、焊接面积预测方法、焊缝图像识别改进算法、多样本训练学习法、神经网络法等方法的关键技术及优缺点。阐述了视觉焊缝跟踪技术的进展,经过图像采集和深度学习处理,实现了机器人运动轨迹的实时偏差修正,可以有效地实时提取初始焊缝,提高了自动焊接质量与效率。建立适用场合广、鲁棒性好、精度高、效率好、安全机制完善的深度视觉焊缝跟踪系统是未来进一步研究的方向。

基于空间光传感信息的运动姿态实时跟踪分析与研究

为了排除运动动态性干扰,探讨人体运动下空间光传感信息可作为视觉跟踪特征的可行性,建立起空间三维点与运动姿态对应点的关系,通过计算运动姿态目标点的投影,对摄像机进行标定,根据人体运动动作点的标定,采集到人体运动姿态空间光传感特征,通过提取运动姿态特征,根据人体运动姿态光传感样本点的逆映射原理,重构运动姿态图像,实现运动姿态的实时跟踪。实验结果表明,基于空间光传感信息的运动姿态实时跟踪的横向偏移量还是跟踪准确率方面,都不会受到运动姿态的动态性影响,跟踪方法的准确率可达到1.0,跟踪效率高达0.1,具有更好的稳定性。

基于图像处理与数值仿真辅助的电弧增材成形研究

为解决Fronius CMT TPS3200焊接电源提供的焊接模式(CMT或CMT+Pulse)与增材策略(单向或往复路径)的选择问题,在相同线能量密度条件下,基于不同模式的单道单层优质焊道成形,探索了单道多层的层宽、层高变化规律,并结合图像处理与数值仿真评价单道10层的增材试验。研究发现,针对焊道成形,CMT+Pulse模式与CMT模式分别从第4层和第3层开始趋于稳定。为优选扫描策略,通过顶层焊道成形和数值仿真结果对比,确定了往复路径优于单向路径。在往复路径下,CMT模式的焊道成形均优于CMT+Pulse模式。研究结果表明,CMT模式率先达到散热稳定条件,结合优选往复路径,在焊道成形方面具有优势,为后续电弧增材奠定了基础。

运用形态学运算的焊接前离线路径规划

焊接特征点的自动提取是机器人自主规划焊接路径的关键。传统的焊接通常需要示教焊接中间点,当工件更改或装配出现误差时,通常需要重新进行示教,导致焊接效率较低。文章提出了一种基于单目相机的焊接特征点自动化提取的方法,首先采用双边滤波平滑图像;随后运用Sobel算子计算像素的梯度,以图像的平均灰度值为阈值进行二值化;接着通过闭运算填充细小孔洞并连接间断的细小轮廓线;使用改进的轮廓跟踪算子消除图像上的残留噪声;采用平均法从图像上提取特征点;然后通过相机标定将特征点从像素坐标系转换到用户坐标系。采取不同形状的平面薄板对接焊缝工件进行特征点提取实验,通过对比特征点与激光传感器采集得到的焊缝坡口中心点在用户坐标系y轴方向上的差值,获得混合型焊缝、直线型焊缝、曲线型焊缝的平均误差分别为0.45mm、0.23mm、0.42mm,表明该方法能够从图像中自动识别焊缝并提取焊接特征点,免除焊接系统的示教环节。

基于改进核相关算法的钢管塔焊缝特征跟踪研究

针对输电钢管塔纵向焊缝焊接过程中因弧光、飞溅等干扰导致焊缝偏差识别不准确的现状,提出一种基于改进核相关算法的钢管塔焊缝特征跟踪方法。通过图像处理技术得到初始帧焊缝特征点,将最小值图像融合和核相关算法相结合,并采用自适应更新策略,解决了长时间强噪声下核相关滤波模型漂移问题。试验数据表明:该方法能够在强弧光、飞溅干扰下稳定、准确地识别焊缝点,识别焊缝误差均在0.15 mm之内,满足焊接生产实际要求,提高了钢管塔焊接的效率和质量。