Research on Point Cloud Matching of Casting Risers Based on Fusion of PCA-SIFT Features
The current development of automated casting riser cutting robot system requires precision in i-dentifying and positioning the riser,and faces problems such as complex scene environment,large amount of data collected by the camera,low processing efficiency,and unstable algorithms leading to large errors.Aiming at the above problems,this paper proposes a PCA-SIFTS4 point cloud alignment method,firstly,the collected field point cloud data are filtered by multi-stage filtering and clustering segmentation techniques to extract high-quality riser point clouds from complex scenes;secondly,the feature attributes of riser point clouds are extracted based on the fusion of PCA and 3D-SIFT features,which reduces the searching com-plexity of the Super4PCS coarse alignment and Finally,the point-to-face ICP algorithm is used for fine a-lignment.In this paper,the proposed method is applied to the typical shape of the riser point cloud for the a-lignment experiment,compared with the traditional alignment algorithm,the alignment time and root mean square error are reduced by 78.53%and 64.93%on average,which shows that the PCA-SIFTS4 algorithm has better real-time and accuracy for the riser point cloud with complex environment,large data volume and inconspicuous features.
NETL
NSTL
万方数据
