Research On Line Laser Measurement System And Registration For Point Clouds

As the products needing to be measured are becoming more and more complex and people’s requirement of the measurement accuracy is becoming more and more demanding, vision measurement with traditional machines was unable to meet the requirements of accuracy because of its two-dimensional limitation. At the same time, the laser measurement technology is getting more and more popular with its strong robustness and high accuracy. According to the actual needs of the project, the present paper aims at building and designing the line laser measurement system. It mainly discusses how to design the hardware structure and build the software frame. The procedure design flow is also analyzed.Because there will be a bind spot in the line laser equipment when it scans an object, it is difficult to get all 3D data from an object whose shape is complicated at one time. Sometimes the object needs to be scanned repeatedly from different angle of view. Therefore, the 3D point clouds registration is needed to get the whole model of the object. The present paper analyzes the registration algorithm for 3D point clouds.First, in order to extract the feature of the point cloud, the point cloud data can be mapped to a two-dimensional image according to the structure of line laser scanning point cloud data. Then, the potential feature of point cloud data can be estimated by using the method of image processing and improving the Canny algorithm. At last, the final feature based on curvature information can be extracted. Compared with the traditional feature extraction algorithm of 3D point cloud data, the proposed algorithm has higher efficiency.As for the description of the point cloud’s feature, although the high dimensional feature description algorithm can directly solve the registration problem for different 3D point clouds, its calculation is complex. Besides, for a large number of point cloud data, the matching efficiency will be greatly reduced. For the consideration of efficiency, the paper proposes a description method based on the information of neighboring curvature. By replacing the high dimensional descriptor with the low dimensional one, the calculating time can be reduced.In order to match corresponding points between different point cloud data, the paper construct an evaluation function to find potential corresponding points for each feature point according to the feature description algorithm. Because the low dimensional feature description could not provide enough information, each feature point will be matched to a plurality of corresponding points, which is necessary for further screening. Based on the principle of rigid transformation, the paper proposes a regional voting system to match the corresponding point for each feature point accurately.At the end of the paper, the experiment on registration algorithm is carried out and the laser measurement system is applied for flatness detection. First, the measuring machine scans the object from different angles of view to get different point cloud models. The features of the point cloud models are analyzed based on the feature extraction algorithm, feature description algorithm and corresponding points matching algorithm proposed by the paper. Moreover, the paper bases the registration for the point cloud models on ICP algorithm to test the accuracy and efficiency of the registration algorithm. In practical application, the experimental analysis about the calibration of line laser measuring machine is conducted. Also, the author uses the flatness detection function of the line laser measuring machine to measure some products which require high flatness of the surface. Finally the results of the measurement and the 3D scan charts of the products are showed to validate the function of the measuring machine.