Salient-point-sampling-based point cloud data classification and segmentation method

Abstract

The invention discloses a salient-point-sampling-based point cloud data classification and segmentation method, and provides a salient-point-sampling-based point cloud data classification and segmentation network, two network backbone parts are composed of a new salient point sampling algorithm (SPS) and a multi-scale local salient feature extraction network (MS-LSFE), and the SPS and the MS-LSFEcan be flexibly inserted into other networks. And sampling or feature extraction is assisted. In order to facilitate calling, SPS and MS-LSFE are packaged to form a multi-scale significant feature extraction module (MS-SFE), and balance between performance and parameter quantity is achieved. According to the method, a classification test is carried out on a standard public data set ModelNet40, andthe classification precision reaches 92.42% while the number of parameters is only 0.3 * 10<6>; the segmentation experiments are carried out on standard public data sets ShapeNet, S3DIS and Semantic3D, and the segmentation precision of 85.1%, 61.8% and 65.8% can be achieved respectively. The classification and segmentation results are in the leading or equivalent level in the same kind of work, and the feasibility and effectiveness of the method are verified.

Description

technical field [0001] The invention relates to the technical fields of computer graphics, computer vision and intelligent recognition, in particular to a point cloud data classification and segmentation method based on salient point sampling. Background technique [0002] With the popularity of 3D point cloud acquisition equipment such as lidar and stereo cameras, and the development of related fields such as autonomous robots and autonomous driving, there is an increasing demand for direct processing of point clouds in order to avoid costly mesh reconstruction. As a result, large-scale repositories of 3D point clouds are beginning to emerge, and convolutional neural networks (CNNs) are becoming one of the most important techniques to greatly improve point cloud processing capabilities. [0003] The application scenarios of the point cloud model, such as unmanned driving, mostly require real-time acquisition of point cloud data to complete the preliminary segmentation and i...

AI Technical Summary

Problems solved by technology

[0006] Qi et al. (C.R.Qi, H.Su, K.Mo, and L.J. Guibas. Pointnet: deep learning on pointsets for 3d classification and segmentation[C] / / Proceedings of the IEEEConference on Computer Vision and Pattern Recognition, pages 652–660, 2017.) The PointNet network proposed for the characteristics of point cloud data first applies deep learning to point cloud classification tasks. PointNet uses T-Net to achieve effective alignment of data and features, and uses Maxpooling symmetric function to extract order-independent global features. ModelNet40 A classification accuracy of 89.20% has been achieved on the above. However, PointNet cannot extract the local features of point cloud data, and a series of work has been designed on how to effectively extract the local features of point cloud data.

Images