激光视觉传感的焊缝成形测量

利用线激光视觉传感技术测量焊缝成形尺寸,关键在于准确提取焊缝特征点。针对常用的斜率解析法与斜率截距法在提取复杂焊缝特征点时定位精度及稳定性较低的问题,在常规斜率截距法的基础上,提出一种改进的焊缝特征提取算法实现复杂焊缝特征点的识别。根据焊缝中心条纹寻找焊缝轮廓最高点,利用常规截距法提取焊缝候选特征点,根据焊缝最高点与候选特征点的几何关系确定焊缝特征点。试验结果表明,对于复杂焊缝改进后的算法提取的焊缝特征点精度及可靠性高于一般焊缝特征检测法,能有效提高焊缝特征点的检测精度及可靠性。     

基于结构光视觉的激光拼焊焊缝背面质量检测方法研究

在激光拼焊中,背面焊缝成形质量直接影响焊件的机械性能.文中在分析结构光视觉检测原理的基础上,建立了基于结构光视觉传感的焊后焊缝背面质量检测系统并对焊缝背面质量检测系统图像处理方法进行了深入研究.在图像处理过程中,首先通过开窗处理来获得兴趣区域,采用中值滤波去除图像噪声;其次使用迭代自动阈值法分割出结构光;在结构光条纹中心线提取过程中,提出了一种新的模板法获得了条纹的边界并用几何中心法提取了条纹中心线;然后,将斜率分析法引入到条纹中心线特征点检测中并获得了中心线上的一系列特征点.最后通过背面焊缝的图像序列,计算获得了背面焊缝不同位置处的几何参数及缺陷值.试验表明,该焊缝背面质量检测系统图像处理方法可靠性高、运算速度快、抗干扰能力强,具有较高实用价值.     

基于结构光视觉的激光拼焊焊缝质量检测方法研究

在基于结构光视觉的激光拼焊焊缝表面质量检测系统中,需要准确获得焊缝结构光条纹中心线和特征点的位置.由于焊接过程中的各种噪声干扰,使得焊缝图像复杂多变,因此选择合理高效的图像处理算法是非常重要的.针对这一问题,对结构光视觉焊缝质量检测系统图像处理方法进行了深入研究.在图像预处理中,首先通过开窗处理来获得兴趣区域,采用中值滤波去除图像噪声.在结构光条纹中心线提取过程中,提出了一种新的模板法获得了条纹的边界并用几何中心法提取了条纹中心线;最后,使用斜率分析法提取了特征点.试验表明,该方法具有较高的特征点检测可靠性,并且运算速度快、抗干扰能力强,具有较高实用价值.     

基于结构光视觉的激光拼焊焊缝质量检测方法研究

焊缝质量自动检测是实现焊接自动化的关键技术。文中在分析结构光视觉检测原理的基础上,建立了基于结构光视觉的激光拼焊焊缝质量检测系统并对检测系统的图像处理方法进行了研究。对原始图像进行了开窗处理和中值滤波,并使用迭代阈值法获得了结构光光纹。提出了一种简化的模板获得了光纹的边界点并使用几何中心法提取了光纹中心线。使用平均斜率法识别出光纹中心线的特征点。分析了激光拼焊焊缝截面轮廓的特点并建立了截面轮廓几何参数计算方法。实验结果表明该检测系统可以完成焊缝截面轮廓相关参数的计算。     

基于局部阈值的激光拼焊焊缝条纹中心线快速检测方法研究

基于视觉的激光拼焊焊缝表面质量在线实时检测系统中,结构光条纹中心线能否快速、准确提取是影响检测系统性能的关键因素之一。传统的高斯拟合法和Hessian矩阵法虽然具有较高的亚像素提取精度,但其计算量非常大,无法满足实时性的要求。文中在分析激光拼焊焊缝质量检测系统中结构光条纹图像特点的基础上,将传统的几何中心法引入到结构光条纹中心线提取中,提出了一种精度介于像素级到亚像素级之间的局部阈值几何中心法。实验表明:该算法具有较高的精度和较强的抗干扰能力,实现了视觉检测系统中结构光条纹中心线的快速提取,为激光拼焊焊缝质量视觉检测系统在线实时检测奠定了基础。     

基于视觉特征的厚板机器人焊接焊缝轮廓的有效提取

基于结构光的主动视觉传感技术是用于厚板机器人自动焊接中检测焊缝信息的一种有效途径。针对焊接过程中采集的焊缝图像包含或不包含电弧区域的两种情形,提出了一种基于视觉特征的适用于典型接头的激光条纹提取的通用算法。该算法利用Gabor滤波生成方向特征图,阈值分割后利用最近邻聚类算法对激光条纹和干扰数据进行聚类,根据聚类结果获得各类数据点在图像中的空间跨度,且只保留跨度值大于平均跨度的类,实现激光条纹的粗略提取;在从粗略提取结果中获取首个激光条纹分段的前提下,利用后续待识别的类与前一次已提取的激光条纹分段的欧式距离和类的自身跨度,设计了一个能提取激光条纹后续分段的竞争策略,从而实现激光条纹主体轮廓的准确提取;根据激光条纹视觉投射特点,设计了一个提取主体轮廓各分段之间的激光条纹细节的有效规则,实现激光条纹细节的提取。利用厚板T形接头和对接接头焊接中采集的焊缝图像验证了文中激光条纹提取算法的有效性,为厚板机器人自主焊接智能系统面向多种接头焊接应用的集成提供了可能。     

角点检测与光流跟踪的焊缝特征提取与定位研究

针对V型坡口中厚板对接焊焊缝特征点检测精度不高的问题,研究了一种基于激光视觉传感的角点检测与光流(LK)跟踪的焊缝特征快速提取与定位方法。根据三角测量原理,设计了能够实时检测焊缝特征图像的激光视觉传感器,并建立了由激光条纹特征点像素坐标到焊缝特征点三维坐标的数学模型;对焊缝图像进行了预处理,采用Shi-Tomasi角点检测提取了焊缝特征;最后使用光流法为后续帧匹配特征角点,实时计算出了图像中焊缝特征点的亚像素位置。研究结果表明:基于角点检测与光流法跟踪的焊缝特征提取与定位方法,其特征点检测精度较高,平均误差在±0.13 mm以内,可以实时、准确地识别焊缝特征。     

斜率表征与卡尔曼滤波的焊缝跟踪方法研究

针对V型坡口中厚板对接焊焊接过程中存在的系统噪声和测量噪声的问题,采用了斜率分析与卡尔曼滤波相结合的方法进行了焊缝跟踪。使用激光视觉传感系统采集了V型坡口的条纹图像,获取了测量系统内外标定参数,通过锯齿靶标建立了焊缝特征点坐标转换模型;提出了一种基于极值法的边界约束灰度平方加权重心法,提取出了激光条纹线的中心点,采用斜率分析法寻找到了激光条纹中心线拐点,对每一段条纹中心线通过最小二乘法拟合求交,提取出了焊缝特征点;利用卡尔曼滤波方法对系统噪声和测量噪声进行了优化;对V型坡口进行了厚板对接焊试验,焊炬能始终对准焊缝中心。研究结果表明:基于斜率分析法与卡尔曼滤波的焊缝跟踪方法能够减小系统噪声干扰的影响,实现了焊缝特征点的准确提取,提高了焊缝跟踪精度。     

基于分段线性拟合的典型焊缝特征提取

针对角接、搭接、V型坡口对接和窄对接等四种典型焊缝特征的提取问题,搭建了一套基于激光视觉的焊缝图像采集系统,通过图像预处理和焊缝激光条纹中心线提取,建立了基于分段线性拟合的焊缝特征提取方法.该方法通过计算斜率划分中心线像素坐标点集,然后采用最小二乘法对各分段点集进行直线拟合来实现焊缝特征点的有效判别.结果 表明:特征点的平均提取时间为9ms,平均提取误差为0.035 mm,为基于线激光视觉图像的焊缝跟踪系统提供判据.     

基于RANSAC算法的焊缝余高识别与检测

开发了基于激光结构光视觉的焊缝余高检测系统,并提出了一种基于RANSAC算法的焊缝余高识别算法,它通过利用RANSAC算法"局外点"的特性可对焊缝特征点进行精确提取,并实现对焊缝余高的检测。在实验平台上对该算法进行了验证,实验结果表明该算法可行,精度较高,速度快,具有一定的实用价值。     

基于激光三角法和嵌入式的微位移实时检测系统

基于激光三角法和嵌入式的微位移实时检测技术,采用激光三角法测距和灰度重心法提取激光条纹中心的原理,使线结构光、透镜、CCD位置信息满足Scheimpflug共轭清晰成像条件。通过图像处理获取像面坐标系位移信息,得到待测物体物面坐标系的位移,利用灰度重心法实时提取结构光条纹中心的特点,结合嵌入式系统尺寸小、稳定、便携的特性,实现对待测物体进行实时、非接触和微位移测量。     

利用轮廓线多边形表示实时提取光带中心线

提出了一种亚像素光带中心线实时提取方法.首先,通过跟踪光带轮廓线并进行多边形表示将光带分段;然后,利用轮廓线的逆时针排序特性以及分段光带方向和横截面方向的正交关系,采用序号表示方法计算光带横截面扫描线方向;最后,在扫描线上设计了自适应光带横截面宽度变化的灰度重心法计算亚像素坐标.针对6传感器线激光人脚扫描仪采集的分辨率为640 pixel×480 pixel的典型位置光带位图进行了实验,结果显示,采用本方法提取光带中心线用时不超过3.1 ms.精度达到亚像素级,能够满足多传感器线激光扫描视觉系统对光带中心线提取的实时性和高精度的要求.     

Evaluation of system models for an endoscopic fringe projection system

To be able to perform inline inspection of complex geometries, which exhibit for example undercuts or internal structures, a new endoscopic micro fringe projection system has been developed. It is designed to perform areal measurements for tool inspection inside the limited space of metal forming presses by employing flexible image fibers to couple the measurement system’s camera and projector to a compact sensor head. The projector features a laser light source and a digital micro-mirror device to generate high-contrast fringe patterns. To increase the depth of field of the sensor heads, custom gradient-index lenses have been designed as an approximation to the Scheimpflug principle. Challenges arise for both calibration and phase measuring algorithms from the optics, as well as from the reduction in resolution introduced by the fiber bundles. This paper presents an evaluation of two different system models for the endoscopic fringe projection system, which are based on the pinhole camera model and a black box model. An automated calibration process, which gathers the calibration data for two calibration algorithms that are robust to artifacts introduced by the optical path, is demonstrated. Based on a comparison of measurements, differences between the two modeling approaches are discussed. Finally, results of measurements of a demonstrational metal forming tool are shown as an application example.     

Development of a real-time laser-based machine vision system to monitor and control welding processes

In this study, a laser-based machine vision system is developed and implemented to monitor and control welding processes. The system consists of three main modules: a laser-based vision sensor module, an image processing module, and a multi-axis motion control module. The laser-based vision sensor is designed and fabricated based on the principle of laser triangulation. By developing and implementing a new image processing algorithm on the platform of LabVIEW, the image processing module is capable of processing the images captured by the vision sensor, identifying the different types of weld joints, and detecting the feature points. Based on the detected feature points, the position information and geometrical features of the weld joint such as its depth, width, plates mismatch, and cross-sectional area can be obtained and monitored in real time. Meanwhile, by feeding these data into the multi-axis motion control module, a non-contact seam tracking is achieved by adaptively adjusting the position of the welding torch with respect to the depth and width variations of the weld joint. A 3D profile of the weld joint is also obtained in real time for the purposes of in-process weld joint monitoring and post-weld quality inspection. The results indicate that the developed laser-based machine vision system can be well suited for the measurement of weld joint geometrical features, seam tracking, and 3D profiling.     

Measurements of fillet weld by 3D laser scanning system

The external geometry of a fillet weld affects not only the quality but also the safety of a structure. Two parameters that influence the geometric quality of a fillet weld are the weld size and the weld surface profile. A 3D laser scanning measurement system integrating the techniques of reverse engineering is adopted in this study to provide a more accurate measurement of the weld profile. The section profiles of the external weld geometry along the longitudinal axis of the weld are generated by the measured data set through CAD software. A computer programme based on the AWS Structural Welding Code is used to judge the acceptability of a fillet weld based on the CAD section profile of the weld. The proposed 3D laser scanning system with integrated CAD software can provide a more accurate and efficient method to estimate the geometric quality of a fillet weld .     

面向重建的结构光中心快速提取方法

在逆向工程的非接触光学测量系统中,准确、快速提取光条中心是保证获取实物表面形状数据精度和算法实时性的关键因素。根据激光强度符合高斯分布和光条边界的连续性特点,对传统的光条中心提取方法进行改进。考虑光条在图像中的形状和分布特征,通过从图像的不同位置和方向扫描得到光条中心,并比较了不同扫描方法所用的时间。利用本文方法重建了动车轮毂的三维轮廓。实验结果表明,本文方法比传统方法有明显提高,具有较强的实用性。     

基于图像识别的齿轮尺寸测量研究

为解决传统工业测量中,齿轮参数尺寸测量方法过于繁琐的问题,利用数字图像处理技术对其进行检测,提出了一种基于圆投影分割的齿轮参数尺寸检测方法。该方法首先通过图像预处理得到细化的齿轮轮廓,使用改进随机采样算法找到齿轮中心;然后利用轮廓点和图像中心点连线与所建立的图像坐标系正向X轴所成角度和两点距离的映射关系,建立新的极坐标系,并对齿轮进行圆分割,记录新坐标系中峰值点个数和峰值点纵坐标,这2个参数分别对应齿轮齿数和齿顶圆半径;最后利用齿顶圆半径与齿数、模数之间的关系式,求出模数,从而求出齿轮其他尺寸。实验结果表明,用该方法检测齿轮参数尺寸,检测速度快、精确度高,具有较高的实用价值。     

激光冲击软模大面积微弯曲成形方法

为了实现金属箔板大面积微弯曲成形,本文结合激光冲击微弯曲成形技术与软模成形技术的优点,提出了激光冲击软模大面积微弯曲成形方法。 该方法是在脉冲激光冲击波压力下,将软模作为柔性冲头作用于金属箔板来实现工件成形的。实验中使用了Innolas Gmbit公司生产的Spitlight 2000 THG脉冲激光器,将250 μm厚的聚氨酯橡胶薄膜作为软模,采用德国LPKF-ProtoMat-C60型雕刻机在印刷电路板上加工出深度为120 μm的U型多槽模具,实现了在厚度为30 μm的铜箔板上一次性对3个U型凹槽冲击成形。用KEYENCE VHX-1000C超景深三维显微系统进行工件观测,结果显示工件上的微成形槽具有良好的轮廓质量。以ANSYS/LS-DYNA为平台,使用有限元建模（FEM）方法对微弯曲过程进行了数值模拟。实验和模拟结果均表明,加载软模的工件与模具的U型凹槽特征在形状上更加接近,成形工件更加均匀,而且具有较好的表面质量,其最大平均成形深度可达110 μm,大于激光直接冲击成形的最大深度（88 μm）,说明使用软模提高了充型能力。     

Measurement point prediction of flatness geometric tolerance by using grey theory

The objective of this study is to improve the efficiency in achieving accurate and speedy inspection when coordinate measuring machine (CMM) is used to measure geometric tolerance. The grey prediction in grey theory was applied in developing the algorithm for predicting the number of measuring points required for measuring geometric tolerance in this article. The said algorithm was used to plan the number of measuring points of the next workpiece and to predict the geometric tolerance dimension of the next workpiece. This step provided better foundation for on line inspection to determine the number of measuring points required for measurement inspection of the next workpiece. Besides, it can predict whether the geometric tolerance of the next workpiece conform with the geometric tolerance dimension marked on the design drawing. This algorithm is used to determine whether the manufacturing process requires modification, in an attempt to save human and material resources and improve failure rate.In this paper, the said algorithm refers to the flatness measurement point algorithm in a plane of the workpiece. In planning for the increase of measuring points, this algorithm is bound by the principle of keeping the measurement scope a rectangular shape by adding a row of measuring points. When the prediction value exceeds the designation dimensions range, immediate inspection of the manufacturing process is needed. The concept of on line inspection of the addition of measuring points is also incorporated together with the grey prediction model and concept of standard deviation model in an effort to develop the said algorithm.Experimental data were also introduced to verify the flatness geometric tolerance measuring points algorithm for the flatness geometric tolerance in a workpiece plane. The objective is to reach a balance between the smallest number of measuring points possible required for inspection and inspection accuracy. Consequently, one can avoid using too many measuring points, which increases the inspection time, while achieving the required measurement accuracy.