【作者】 王邦国；

【导师】 贾振元；

【作者基本信息】 大连理工大学 ， 机械电子工程， 2011， 博士

【摘要】 大型构件是电力、船舶、航空航天等重大工程核心装备中的关键结构件,其特点是大重量和大尺寸,多用于承载力大、性能要求高的场合,通常先由锻造方法将坯料锻成所需形状、尺寸的锻件,再经切削加工制造而成。加热后的钢锭在锻造过程中需要对其尺寸进行测量,以保证锻件的锻后尺寸,由于锻造条件极为恶劣,锻造温度通常高达一千余摄氏度,导致在线测量大型锻件的热态几何参数十分困难。视觉测量具有非接触、速度快、精度高的特点,研究如何将视觉测量技术应用于大型锻件热态几何参数测量对提高锻件材料利用率、降低生产成本具有十分重要的意义。本文在相关项目的资助下,较系统的研究了大型锻件视觉测量的方法,主要针对以下几个关键问题开展了研究：(1)针对热态锻件图像采集的难题,在分析锻件热辐射性质和反射性质的基础上提出了热态锻件光谱选择性图像采集方法,该方法借助照明光源照射热态锻件,并用黑白CCD摄像机接收锻件反射的照明光线成像,光谱范围由照明光源辐射出射度与热态锻件辐射出射度差值最大的波长位置确定；该方法可有效消除高温锻件辐射光线的影响,为测量大型锻件热态尺寸提供了基础。(2)根据大型锻件尺寸测量现场条件,以热态锻件光谱选择性图像采集方法为基础,构建了基于双目视觉的大型锻件热态尺寸测量系统,系统中使用照明设备投射结构光到锻件表面,以增加锻件表面特征信息；分析了系统结构参数及结构光条中心线提取精度对系统测量精度的影响规律,根据分析结果并结合测量现场情况对两摄像机的间距、摄像机焦距以及测量系统到被测锻件距离等参数进行了调整。(3)针对锻件加工车间复杂环境下测量系统难以标定的问题,研究了大型锻件热态尺寸测量系统的标定方法；该方法以主动视觉标定技术为基础,通过对正交运动前后所拍摄图像中投影标识的识别及对应点提取计算摄像机内参数；标定摄像机外参数时,先由基本矩阵求解相差一个尺度的旋转和平移矩阵,再根据内参数及旋转、平移矩阵三维重建两特定标识点,用双经纬仪系统测量两标识点间的实际间距,最后由两标识点测量间距与重建间距的比值得到尺度因子,从而实现外参数标定。(4)针对大型锻件热态尺寸在线测量的难题,提出了由投影特征线所反映的锻件形状特征求锻件热态尺寸的方法。测量圆柱形及方形锻件热态尺寸时分别向锻件投影编码特征线和等间距特征线；对于圆柱形锻件,使用椭圆拟合在线测量其尺寸,并通过圆柱拟合检测其锻后尺寸；测量方形锻件时,根据重建特征线拟合曲线,搜索曲率大于阈值的位置,进而可求锻件长度。在锻件加工厂进行了大型锻件热态尺寸测量实验,并对该检测方法的测量精度进行了检验,实验结果表明该方法可在线测量锻件热态尺寸。本文中热态锻件光谱选择性图像采集方法、通过重建的投影特征线反映锻件几何形状并测量其尺寸方法及研制的大型热态锻件几何参数测量系统等研究工作为大型锻件热态几何参数提供了一种有效的测量手段。更多还原

【Abstract】 Large components are the key parts of important equipment which are widely used in the field of power, ship and aerospace. In order to let the large components can be employed in extreme condition, the blanks which are used to machine the large components are manufactured through forging. The large size forgings have to be measured in the forming process. However, the high temperature of hot large forgings and the large size make it hard to be measured accurately. Machine vision technology has been used as a high-speed, non-contact method in dimension measurement. In order to reduce the waste and production cost, the dimension measurement method with the machine vision technology is researched. This paper is mainly focuses on some key technical as follows:(1) To solve the problem of acquiring the image of hot parts, the spectrum selective method is proposed. By analyzing the self-emitted radiation character and reflection character of hot large forging, the illumination lamp is used to acquiring image of hot large forging. When the external light source projects light toward the high temperature objects, the external light strikes the hot objects and is reflected toward CCD cameras along with the self-emitted radiation. One component of the reflected light is selectively detected by the CCD cameras, and the image of hot objects can be acquired. The range of wavelength is obtained according to the wavelength when difference between the radiant exitance of illumination lamp and the radiant exitance of hot forging is the maximum value. This method presents an important base for measuring the dimension of hot forging.(2) Considering to the measurement condition in the forging plant, a stereo vision system which using the spectrum selective method is developed for mesuring the dimension of hot large forging. Structure light is projected toward the surface of hot forging in the dimension measurement process for increasing characteristics on it. The effect of the structural parameters of the stereo vision system on the error of the measurement is analysised. Based on the analysis results, the distance of two cameras, the focus length of cameras and the distance between the stereo vision system and the hot forging are adjusted accordingly.(3) The camera calibration technique that can be used to calibrate the dimension measurement system in the forging plant is proposed. The intrinsic parameters of camera are calibrated according to the active vision. By recognizing the corresponding points in the images acquired after each orthogonal translation, the intrinsic parameters can be calculated. The extrinsic parameters of camera are determined up to indeterminate scale by the essential matrix. Then, scale is calculated by dividing the distance of two reconstructed feature points in 3D space by the length of the two feature points which is measured by the double-theodolite system.(4) In order to measure the dimension of hot forging in the forming process, the method based on projected feature lines which can obtain the shape of forgings is presented. When measuring the dimension of round forging, a coded single pattern is projected toward hot forging to label characteristic lines and simplify the process of finding correspondences between image pairs. If the feature line is vertically to the axis of forging, the diameter of round forging can be estimated by fitting an ellipse using the reconstructed feature line in 3D space. Otherwise, the diameter of round forgings are acquired by fitting a cylinder with the reconstructed feature lines in 3D space. The lengths of square forgings are measured by finding the position of the curvature which is greater than the threshold.The spectrum selective method for acquiring the image of hot part, the dimension measurement method by reconstructing the featured lines that are projected on the surface of hot forging, the measurment system presented in this paper can be used to measure the geometric parameters of hot large forging effeictively.更多还原