【作者】 刘鹏；

【导师】 候华毅；

【作者基本信息】 武汉工程大学 ， 电子信息， 2024， 硕士

【摘要】 随着自动化技术的快速发展,激光加工技术因其低材料浪费、显著的规模化生产成本效益以及高度的材料适应性,在军事、工业、医疗等众多领域得到广泛应用。激光加工作为一种非接触式加工方式,通过高能激光束的灵活调控,适应多样化的加工需求,极大地提高了加工效率和精度,推动了相关领域的技术创新和发展。同时,机器视觉检测技术通过捕捉图像并将其转化为数字信号进行系统分析,提取目标特征,并控制现场设备动作,进一步提升了激光加工的生产效率和自动化水平,促进了工业制造水平的显著进步。这种两种技术的结合不仅大幅优化了生产流程,还增强了产品质量和生产安全性。本课题来源于某公司“基于视觉引导的激光自动化焊接机”项目,设计和实现了一款基于视觉引导的工件定位系统。本文的主要研究内容和成果如下:1.对视觉引导的工件定位系统进行了硬件设计分析了工件定位的上料需求,设计了系统整体性的方案,完成了实验平台的硬件选型。整个硬件模块主要由加工平台、机器视觉模块和运动控制模块组成。加工平台主要提供稳定而可靠的工作环境;机器视觉模块主要捕获高质量的图像数据;运动控制模块负责实现精确的工件定位和激光振镜的移动控制。2.对视觉引导的工件定位系统进行了算法设计针对加工工件的表面细节特征不一致,对比基于特征点的模板匹配和基于灰度信息的模板匹配方法,选择更适合加工工件的基于灰度信息的模板匹配方法。提出了一种适用于模板匹配的旋转匹配方法,针对原始模板进行一系列的角度旋转,生成多个旋转模板以覆盖可能的目标方向。这些模板被用于在目标图像中进行滑动窗口搜索,每个窗口的匹配程度通过计算如相关系数的相似度度量进行评分。从所有的匹配结果中选择评分最高的作为最优匹配点。将该匹配点的信息传递给运动控制系统,用于引导PLC自动化设备进行精准的定位和操控。实验结果表明,该方法的匹配角度的标准差为0.510°,匹配坐标的标准差为2.5像素。满足实际工业生产中工件定位的精度和通用性需求。提出了一种结合ROI区域和图像金字塔算法的优化方法,在完整的图像上直接标定ROI,通过自动化算法识别和裁剪出这些区域。由ROI区域应用图像金字塔算法,生成多个分辨率层级的图像,以便在不同层级上进行详细的分析和处理。同时通过限制最顶层图像的大小,确保金字塔各层之间的尺寸比例得当,从而优化后续匹配算法的效率与稳定性。实验结果表明,此方法对比未使用优化方法的匹配算法的效率提升了8倍。针对于工件种类的多样性和加工环境下光照情况变换,可能限制基于灰度的模板匹配方法,引入了基于卡尺工具的边缘拟合方法。该方法采用Canny边缘检测算法以识别工件的显著边缘,随后提取出一维的边缘点。为了增强结果的准确性和鲁棒性,进一步使用RANSAC算法剔除数据中的异常边缘点,从而仅保留与主要边缘对应的点。最后,通过基于最小二乘法拟合精确地定义出工件的几何边界。3.对视觉引导的工件定位系统开发了软件软件设计主要包括可视化操作模块、图像采集模块、模板制作模块、图像处理模块以及通讯模块。基于多线程的设计思路,使用QT平台开发了工件定位系统的可操作界面模块和模板制作模块,结合海康机器人的相机SDK完成了图像采集模块,并使用双边滤波、CLAHE算法、模板匹配和卡尺拟合算法,共同实现了图像处理模块。最后,通讯模块使用TCP/IP协议,实现与运动控制板卡的实时通讯。4.对视觉引导的工件定位系统进行了整体的实验与测试为了确定设计的系统性能与精度指标达到了设计要求,对硬件系统稳定性、软件和算法稳定性和系统整体稳定性进行了性能测试。实验结果表明:本文所设计并研制的基于视觉引导的工件定位系统具备设计的合理性和测试的准确性,通过一系列激光加工实验验证,证明了该系统能够实现对工件的精确定位。本系统不仅能够适应多种工件的加工需求,为工业生产领域提供了一种可靠的定位引导解决方案。同时,该研究还为类似的工业应用场景提供了研究思路,具有一定的现实参考价值。更多还原

【Abstract】 With the rapid development of automation technology,laser processing technology has been widely used in many fields such as military,industry,and medical due to its low material waste,significant cost-effectiveness of large-scale production,and high material adaptability.As a non-contact processing method,laser processing adapts to diverse processing needs through the flexible regulation of high-energy laser beams,greatly improves processing efficiency and accuracy,and promotes technological innovation and development in related fields.At the same time,machine vision inspection technology further improves the production efficiency and automation level of laser processing by capturing images and converting them into digital signals for systematic analysis,extracting target features,and controlling on-site equipment movements,and promotes significant progress in industrial manufacturing levels..This combination of the two technologies not only significantly optimizes the production process,but also enhances product quality and production safety.This topic comes from a company’s "laser automated welding machine based on vision guidance" project,and a workpiece positioning system based on vision guidance is designed and implemented.The main research contents and results of this article are as follows:1.Conducted hardware design of vision-guided workpiece positioning systemIn view of the inconsistent surface details of the processed workpiece,compare the template matching method based on feature points and the template matching method based on grayscale information,and choose the template matching method based on grayscale information that is more suitable for the processed workpiece.A rotation matching method suitable for template matching is proposed.A series of angle rotations are performed on the original template to generate multiple rotation templates to cover possible target directions.These templates are used to perform a sliding window search in the target image,and the degree of match for each window is scored by calculating a similarity measure such as the correlation coefficient.Select the one with the highest score from all matching results as the optimal matching point.The information of the matching point is transmitted to the motion control system to guide the PLC automation equipment to perform precise positioning and control.Experimental results show that the standard deviation of the matching angle of this method is 0.510°,and the standard deviation of the matching coordinates is 2.5 pixels.Meet the accuracy and versatility requirements of workpiece positioning in actual industrial production.An optimization method is proposed that combines the ROI area and the image pyramid algorithm to directly calibrate the ROI on the complete image,and identify and crop these areas through an automated algorithm.The image pyramid algorithm is applied to the ROI area to generate images at multiple resolution levels for detailed analysis and processing at different levels.At the same time,by limiting the size of the topmost image,the appropriate size ratio between each layer of the pyramid is ensured,thereby optimizing the efficiency and stability of the subsequent matching algorithm.Experimental results show that this method is 8 times more efficient than the matching algorithm that does not use optimization methods.In view of the diversity of workpiece types and changes in lighting conditions in the processing environment,which may limit the grayscale-based template matching method,an edge fitting method based on the caliper tool is introduced.This method uses the Canny edge detection algorithm to identify significant edges of the workpiece,and then extracts one-dimensional edge points.In order to enhance the accuracy and robustness of the results,the RANSAC algorithm is further used to eliminate abnormal edge points in the data,thereby retaining only the points corresponding to the main edges.Finally,the geometric boundaries of the workpiece are accurately defined by fitting based on the least squares method.2.Developed software for vision-guided workpiece positioning systemThe software design mainly includes visual operation module,image acquisition module,template production module,image processing module and communication module.Based on the multi-threaded design idea,the QT platform was used to develop the operable interface module and template production module of the workpiece positioning system.The image acquisition module was completed with the camera SDK of Hikvision Robot,and bilateral filtering,CLAHE algorithm,template matching and calipers were used.Fitting algorithm,jointly implements the image processing module.Finally,the communication module uses the TCP/IP protocol to achieve real-time communication with the motion control board.3.Conducted overall experiments and tests on the vision-guided workpiece positioning systemIn order to ensure that the designed system performance and accuracy indicators meet the design requirements,performance tests were conducted on hardware system stability,software and algorithm stability,and overall system stability.The experimental results show that the workpiece positioning system based on visual guidance designed and developed in this article has the rationality of design and accuracy of testing.It has been verified through a series of laser processing experiments that the system can achieve precise positioning of the workpiece.This system can not only adapt to the processing needs of a variety of workpieces,but also provides a reliable positioning and guidance solution for the industrial production field.At the same time,this study also provides research ideas for similar industrial application scenarios,which has certain practical reference value.更多还原