Research On Three-dimensional Imaging For Low-temperature Adiabatic Cylinder Appearance Defect Detection

The cylinder inspection institutions have formed the marketization with low-temperature adiabatic cylinder applied more and more widely.Traditional appearance defect detection has been difficult to meet the required precision and efficiency,so 3D visualization detection technology is widely used in testing field due to the characteristics of non-contact,no damage,moderate speed and accuracy of measurement.The research background of this thesis is based on Changzhou cryogenic insulation cylinder three-dimensional imaging detection system.And the research status and development trend of 3d detection technology at home and abroad is analyzed firstly.Secondly,complete structure of the light strip center extraction by the process such as distortion correction,Gaussian filtering,threshold segmentation,connected domain extraction,refining,deburring in view of the current visual images of the sensors.Then,a 3D reconstruction of cylinder surface shape data is realized through the camera model.And the iterative circle fitting,deformation more outline and Statistical Outlier Removal filter denoising effect are studied respectively in view of point cloud data of reconstruction and the reason that the formation of the noise points.Thirdly,Statistical Outlier Removal filter is used for preprocess the point cloud data due to the iterative circle fitting,and the outline polygon can to remove part of the cylinder surface point cloud data.Finally,a sag detection method for the point cloud data after pretreatment is proposed according to the current method to detect cylinders sag of artificial.And to solve the problem of the lack of point cloud data after pretreatment and select the corresponding filling algorithm,the cubic spline and BP neural network to the filling effect of missing data and the prediction precision are compared.The experiments verify the method designed in this thesis and the experiments indicate that the application of this method given in this thesis can effectively remove noise from the group to the cylinder of the point cloud data,and realize the hollow part of the cylinder extraction and fill the missing data at the same time of when process gas cylinder point cloud data.The cryogenic insulation cylinder sag detection method is feasible and has certain application value and development significance.