Encoding / decoding positions of points of a point cloud contained in a cuboid volume

By quantizing vertex positions and residual values, the TriSoup encoding scheme was improved, which enhanced the efficiency and visual quality of point cloud encoding, solved the lossy compression problem of dense point clouds, and achieved better encoding efficiency and reconstruction results.

CN118985133BActive Publication Date: 2026-07-10BEIJING XIAOMI MOBILE SOFTWARE CO LTD
View PDF 1 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
BEIJING XIAOMI MOBILE SOFTWARE CO LTD
Filing Date
2023-02-21
Publication Date
2026-07-10

AI Technical Summary

Technical Problem

Existing TriSoup encoding schemes are insufficient in terms of visual quality and objective metrics (such as PSNR) of point cloud reconstruction, making it difficult to achieve efficient lossy compression on dense point clouds.

Method used

By quantizing vertex positions and residual values, vertex information is encoded using quantization parameters and dequantized during decoding to construct triangles to reconstruct point cloud positions, thereby improving encoding efficiency and visual quality.

Benefits of technology

It improves the trade-off between bit rate and distortion in point cloud encoding, enhances the visual quality and objective metrics (such as PSNR) of point cloud reconstruction, and is suitable for efficient lossy compression of dense point clouds.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118985133B_ABST
    Figure CN118985133B_ABST
Patent Text Reader

Abstract

There is provided a method and apparatus for encoding / decoding positions of points of a point cloud contained in a cuboid volume, said positions being represented in a three-dimensional coordinate system, at least three vertices being located on edges of the cuboid volume, one vertex per edge. The encoding method comprises: - obtaining (110) quantized positions (p Q ) of the at least three vertices by quantizing positions of the at least three vertices based on a vertex quantization step (AV), the vertex quantization step (AV) depending on at least one quantization parameter; - encoding (120) vertex information defining the quantized positions (p Q ) of the at least three vertices into a bitstream; and - encoding (130) the at least one quantization parameter into the bitstream.
Need to check novelty before this filing date? Find Prior Art