Point cloud compression method and device and computer readable storage medium
A compression method and compression device technology, applied in the network field, can solve problems such as easy loss of three-dimensional geometric features of point clouds, failure to automatically optimize compression rate and distortion rate indicators, etc., and achieve the effect of high-efficiency point cloud lossless compression
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0050] This embodiment provides a point cloud compression method, such as figure 1 As shown, the method includes:
[0051] Step S102: According to the geometric structure of the target point cloud, adaptively divide the target point cloud into a plurality of voxel blocks of different sizes, and mark the plurality of voxel blocks using an octree.
[0052] In this embodiment, firstly, according to the geometric structure of the target point cloud, the target point cloud is adaptively divided into multiple d×d×d voxel blocks v of different sizes i (eg d={128,64,32,16,8}). It should be noted that if the division of voxel blocks is too sparse or dense, it will affect the efficiency of the network model, so adaptive division is required. Recursive method can be used for adaptive division and a threshold can be set (assuming that the number of points allowed for each voxel block does not exceed 5), that is, first divide the target point cloud into 8 equal voxel blocks, if the first...
Embodiment 2
[0077] like Figure 4 As shown, the present embodiment provides a point cloud compression device, including:
[0078] The point cloud division module 12 is used to adaptively divide the target point cloud into a plurality of voxel blocks of different sizes according to the geometric structure of the target point cloud, and use an octree to mark the plurality of voxel blocks;
[0079] Encoding and compression module 14, connected with described point cloud division module 12, for encoding and compressing the described plurality of voxel blocks after marking based on the mask 3D convolutional neural network model trained and adaptive arithmetic coder, obtain Encoding the compressed target point cloud; wherein, the trained mask 3D convolutional neural network model is used to obtain the distribution probability of the plurality of voxel blocks;
[0080] The decompression and reconstruction module 16 is connected with the encoding and compression module 14, and is used for decomp...
Embodiment 3
[0092] refer to Figure 5 , this embodiment provides a point cloud compression device, including a memory 22 and a processor 24, a computer program is stored in the memory 22, and the processor 24 is configured to run the computer program to perform the point cloud compression method in Embodiment 1 .
[0093] Wherein, the memory 22 is connected with the processor 24, and the memory 22 may adopt flash memory or read-only memory or other memory, and the processor 24 may adopt a central processing unit or a single-chip microcomputer.
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More