Three-dimensional point cloud coding and decoding method, compression method and device based on graph dictionary learning

Abstract

The invention provides a three-dimensional point cloud image dictionary learning method, an encoding and decoding method based on image dictionary learning, and a compression method and device. The method comprises the steps of obtaining N pieces of training set point cloud data; performing voxelization processing on the point cloud data to obtain voxelized training set point cloud data; carrying out voxel block division on the training set point cloud data, selecting a plurality of voxel blocks as a training set, and constructing a graph dictionary learning model according to the training set; and performing iterative optimization on the objective function of graph dictionary learning to obtain a graph dictionary for encoding and decoding of the three-dimensional point cloud signal. According to the method, the spatial correlation between the point cloud signals is effectively utilized, and the redundancy between the point cloud signals is gradually and optimally removed; according to the method, predictive coding is carried out from the low level to the top by using the data correlation between signals of different levels, so that the compression efficiency of the 3D point cloud attribute signals is effectively improved, the coding overhead is effectively reduced, the requirements for decoded signals of different qualities in actual requirements are flexibly met, and the method has scalability.

Description

technical field [0001] The present invention relates to a solution in the technical field of 3D point cloud data compression, in particular to a three-dimensional point cloud encoding and decoding method, compression method and device based on graph dictionary learning. Background technique [0002] In recent years, with the rapid development of 3D acquisition and perception devices, irregular data such as 3D point clouds have been widely used in many emerging technologies such as autonomous driving, virtual reality, augmented reality, 3D long-range video communication, and relic reconstruction. A 3D point cloud image usually contains millions of points containing geometric information and attribute information, so its data scale is often huge, which requires a lot of computing resources and storage space. Restricted by the actual network bandwidth and limited storage space, the existing storage and transmission requirements in this way have far exceeded the resource constra...

AI Technical Summary

Problems solved by technology

However, V-PCC will introduce inevitable distortion during the projection process, while the attribute space transformation scheme of G-PCC only relies on geometric information, and does not consider the statistical characteristics of the attribute signal itself, thus reducing the compression quality of the attribute signal

Images