Research On Non-iteration Feature Preserving Denoising Of3D Point Clouds From Large Forgings

Large forgings is the basis of Manufacture major equipment, the accuratemeasurement of the forging size plays an important role in forging raw materialutilization, better product quality and increase of the rate of qualified products.Because of strong earthquakes, dust, forgings Self-luminous, self-heating and otherfactors, it makes the measured three-dimensional data point often contain serious noisewhen use the existing measurements to do three-dimensional measurement of largeforgings. Study of3D point-clouds, eliminate of large forgings three-dimensionalcloud noise and while maintaining the original structural features have greatsignificance to improve the accuracy of three-dimensional measurement. For thecontradiction between the existing large forgings point cloud de-noising method ofnoise cancellation and feature maintaining, we carry out the study of large forgingspoint cloud de-noising based on three-dimensional point cloud features maintainingtheory, specific work as follows:First of all, based on the implicit surfaces orthogonal projection theory, to carryout study of large-scale forgings three-dimensional point cloud of non-iterativedenoising. By moving least squares fitting of point cloud smooth surface, using therelationship of projection point between sampling point and neighborhood point on thelocal surface to realize the constraint of Point cloud surface deviate, combinedwith the distance weight factor and curvature Weight Factors for characterize thecurvature rights factor to reflect the local structural characteristics of the samplingpoints. On this basis, we put forward large-scale forgings three-dimensional pointcloud denoising algorithm based onconstraints of the orthogonal projection to realize of large forgings three-dimensionalpoint cloud denoising.Secondly, based on bilateral filtering theory, we carry out the study of the largeforgings three-dimensional point cloud non-iterative feature to keep noise cancellation.According to Surface variational curve estimation process to find the optimal noise reduction to each sampling point neighborhood, which will helpful to estimate of thelocal area’s geometric properties of the fairing sampling point; And propose animproved vertex position estimation method. Laid a solid foundation for theimprovement of the efficiency of point cloud denoising algorithm. On this basis,we put forward the large forgings three-dimensional point cloud characteristics ofnon-iterative ensure noise cancellation algorithm, to realize the effective maintenanceof the structural features in the process of the large forgings three-dimensional pointcloud noise cancellation.Finally, we carry out experimental research on the the large-scale the forgingsline three-dimensional laser scanning measurement platform which was developed bythe research group, according to the processing of on the measured of large forgingsmodel three-dimensional point cloud data, to verify the performance of largeforgings3D point cloud denoising method in this paper.