Study Of Surface Measurement Based On Multilinear Structured Light

As a kind of reverse engineering technology,the structured light-based surface measurement technology has the advantages of non-contact,high accuracy and high speed compared with the traditional 3D measurement methods,and can obtain the geometric contour information of the object surface without physical contact.After obtaining the image information of the intersection of the structured light and the measured point through image processing,the 3D spatial coordinates of the measured point can be calculated by using the model parameters of the system,so that the 3D contour data of the measured object can be obtained quickly,which can better meet the demand for real-time 3D measurement technology in many occasions.In this thesis,based on the line structured light3 D measurement technology,a multi-line structured light surface measurement system composed of multiple line lasers,cameras,computers and motion platforms is realized,and the key technologies are studied and analyzed,and the main research contents are:(1)Combined with the measurement principle of laser triangulation,a measurement model based on multilinear structured light can be established,which can obtain the conversion relationship between the two-dimensional image points in the image and the corresponding points in the three-dimensional space.The overall structure of the multilinear structured light surface measurement system is also constructed,which includes a software part and a hardware part.(2)In order to improve the fitting accuracy of light plane parameters in light plane calibration and reduce the influence of error points on the estimation of light plane parameters,a light plane calibration algorithm based on ROI and LMed S is proposed.The ROI extraction is used to extract the feature points after segmenting the light stripe regions on different light planes,and the LMed S algorithm is used to eliminate the anomalies in the feature points and fit the remaining feature points to obtain the light plane parameters,which improves the light plane calibration parameters and lays the foundation for the subsequent object surface measurement.(3)In order to better extract the light stripe center,a line structured light stripe center extraction algorithm based on the improved UNet network is proposed.The CNN units in the UNet network are replaced by the improved RCNN units,and the attention mechanism is introduced to enhance the feature extraction ability of the light stripe image.At the same time,experimental comparison and analysis with the traditional light stripe center extraction algorithm are carried out to realize the effective extraction of the light stripe center.(4)The hardware selection and experimental platform in the measurement system were completed,and objects to be measured with different shapes and different surface materials were selected for experiments.Different parts of different objects to be measured were selected for multiple measurements,and the data obtained from the experiments were analyzed for errors,and the causes of errors in the system were also analyzed.