Point cloud geometric lossy compression method based on voxel convolution

Abstract

The invention discloses a point cloud geometric lossy compression method based on voxel convolution. Compression and decompression are carried out by carrying out convolution and deconvolution on voxels of a point cloud through a training model. Firstly, voxelization is carried out on point cloud data, a certain grid size is selected for voxels of point cloud to carry out 3D convolution operationto obtain feature data with smaller shapes and sizes, quantization processing is carried out on the feature data after convolution, uniform quantization noise is added during model training to improvethe generalization of the model, and the quantized data is compressed. During decompression, deconvolution is performed on the quantized feature data to obtain feature data of which the size is consistent with the shape and size of the initial point cloud voxel, normalizing the feature data, and judging whether each voxel unit is empty or not through a threshold value to obtain decompressed pointcloud data. During model training, focus loss is used as distortion loss to reduce the influence of too many voxel hollow samples on the model. According to the method, geometric compression can be efficiently carried out on the point cloud data, and the distortion rate after restoration is reduced.

Description

technical field [0001] The invention relates to the technical field of point cloud compression, in particular to a point cloud geometric lossy compression method based on voxel convolution. Background technique [0002] With the development of 3D scanning technology, 3D information is widely used in the real world, and the digitization of 3D information is becoming more and more important. Point cloud is an important way to express 3D scenes and 3D surfaces of objects. Point clouds are discretely distributed in three-dimensional space in the form of points, and each point contains information such as geometric position, color, and texture, allowing users to experience viewing from any angle. In practical applications, the point cloud is mainly obtained by sampling the scene or the surface of the object by the 3D scanning device. There are many points and a large amount of information, so the data volume of the point cloud is often very large. Considering the problem of stora...

AI Technical Summary

Problems solved by technology

The point cloud data collected by 3D scanning equipment is usually a scattered point cloud. There is no corresponding topological relationship between the points of the scattered point cloud, so it is difficult to process. The existing traditional compression methods include uniform grid algorithm, bounding box algorithm, and average point distance. Algorithms, etc., the compression rate is low and the compression effect is not ideal, so an efficient compression method is needed to solve this problem

Images