基于机器视觉的隔水管法兰端面位姿检测

为提高隔水管抗压检验效率,引导完成待测隔水管在隔水管检验检测装置上的在线自动对接,提出基于机器视觉的隔水管法兰端面位姿检测方法。首先通过基于形状的模板匹配算法定位出待测隔水管法兰端面的定位坐标和定位角度,同时使用图像金字塔对算法进行加速。然后使用单目三维重建技术将像素坐标系下的模板坐标及定位坐标转换到世界坐标系中,并计算出引导待测隔水管精确对接到隔水管检验检测装置上所需的位移量和旋转量。针对隔水管与注水端进行了仿真试验检测和现场试验检测,试验结果表明,该方法实现了快速精确的隔水管法兰端面位姿检测,在工程实践中具有很强的实用性。

Riser flange flat pose detection based on the machine vision

In this paper,the method for riser flange flat pose detection based on the machine vision is proposed,with a view to the efficient test of the riser’s compressive strength and the online automatic docking the riser with the detection and testing devices,Firstly,thanks to the algorithm of shape-based template matching,the positioning coordinates and angle of riser flange flat are located,and the image pyramid is applied to accelerate the algorithm.Secondly,with the aid of monocular three-dimensional reconstruction,the template and positioning coordinates in the pixel coordinate system are converted into the world coordinate system,and the displacement and rotation values required for the docking tasks are calculated.Both the simulation test and field test are carried out for the riser and the water-injection end.The results show that this method realizes the riser flange flat pose detection,which is rapid and accurate,and has a potentially wide application in those engineering undertakings.