Transform method, inverse transform method, encoder, decoder and storage medium

Patent metadata

Field	Value
Publication number	US-2022207781-A1
Application number	US-202217655409-A
Country	US
Kind code	A1
Filing date	Mar 18, 2022
Priority date	Sep 30, 2019
Publication date	Jun 30, 2022
Grant date	—

Abstract

Provided in the embodiments of the present application are a transform method, inverse transform method, encoder, decoder and storage medium. The method comprises: determining a two-dimensional Morton code of an encoding point in a preset space of an encoding point cloud, wherein the two-dimensional Morton code is a Morton code corresponding to two-dimensional coordinates of the encoding point obtained by removing one of three-dimensional coordinates of the encoding point; performing right shift of the two-dimensional Morton code by K binary digits to determine point sets respectively corresponding to three-dimensional coordinate components; determining, from the point sets, point quantities of maximum included encoding points respectively corresponding to the three-dimensional coordinate components; determining, based on the point quantities, a transform order corresponding to the preset space; performing RAHT based on the transform order; and signaling the transform order and an attribute encoding result obtained after RAHT into a bitstream.

First claim

1 . A transform method, applied to an encoder, comprising:
determining a two-dimensional Morton code of an encoding point in a preset space of an encoding point cloud, wherein the two-dimensional Morton code of the encoding point is a Morton code corresponding to two-dimensional coordinates of the encoding point obtained by removing one of three-dimensional coordinate components of the encoding point; performing right shift of the two-dimensional Morton code by K binary digits to determine point sets respectively corresponding to three-dimensional coordinate components, wherein K is a positive integer greater than 1; determining, from the point sets, point quantities of maximum comprised encoding points respectively corresponding to the three-dimensional coordinate components; determining, based on the point quantities respectively corresponding to the three-dimensional coordinate components, a transform order corresponding to the preset space; performing Region Adaptive Hierarchical Transform (RAHT) based on the transform order corresponding to the preset space; and signaling the transform order and an attribute encoding result obtained after the RAHT is performed into a bitstream.

2 . The method of claim 1 , wherein, before determining the two-dimensional Morton code of the encoding point, in the preset space of the encoding point cloud, the method further comprises:
partitioning a point cloud space where the encoding point cloud is located into slices to obtain N slices; wherein the preset space is each slice of the N slices; and N is a positive integer greater than or equal to 1.

3 . The method of claim 1 , wherein, after determining the two-dimensional Morton code of the encoding point in the preset space of the encoding point cloud, and before performing right shift of the two-dimensional Morton code by K binary digits to determine point sets respectively corresponding to three-dimensional coordinate components, the method further comprises:
performing iteration and traversal based on the two-dimensional Morton code to determine the K.

4 . The method of claim 1 , wherein, before performing right shift of the two-dimensional Morton code by K binary digits to determine point sets respectively corresponding to three-dimensional coordinate components, the method further comprises:
acquiring a maximum side length of a point cloud space where the encoding point cloud is located; determining a geometric accuracy of the point cloud space based on the maximum side length; wherein, the geometric accuracy is a number of binary digits corresponding to the maximum side length represented by binary; and determining the K based on the geometric accuracy.

5 . The method of claim 1 , wherein, after determining the two-dimensional Morton code of the encoding point in the preset space of the encoding point cloud, and before performing right shift of the two-dimensional Morton code by K binary digits to determine point sets respectively corresponding to three-dimensional coordinate components, the method further comprises:
determining a difference value between a maximum value and a minimum value in the two-dimensional Morton code; determining a quotient of the difference value and a point quantity of encoding points; determining a number of binary digits by which the quotient is shifted to equal to zero; and determining the K based on the number of binary digits.

6 . The method of claim 1 , wherein the performing RAHT based on the transform order corresponding to the preset space comprises:
determining, based on the transform order corresponding to the preset space, transform directions and order corresponding to the preset space and forming preset angles with respective coordinate components; acquiring a three-dimensional Morton code of the encoding point; and performing the RAHT based on the transform directions and order and the three-dimensional Morton code.

7 . The method of claim 6 , wherein the acquiring the three-dimensional Morton code of the encoding point comprises:
determining respective side lengths of three-dimensional coordinate components in a point cloud space where the encoding point cloud is located; determining, based on the respective side lengths of the three-dimensional coordinate components, weight values respectively corresponding to the three-dimensional coordinate components; and coding the encoding point in the encoding point cloud based on the weight values to obtain the three-dimensional Morton code.

8 . The method of claim 1 , wherein, before determining the two-dimensional Morton code of the encoding point in the preset space of the encoding point cloud, the method further comprises:
selecting part of a point cloud space where the encoding point cloud is located as the preset space.

9 . The method of claim 8 , wherein the performing RAHT based on the transform order corresponding to the preset space comprises:
performing RAHT based on the transform order corresponding to the preset space and a preset transform order corresponding to other spaces; wherein, the other spaces are spaces excluding the preset space in the point cloud space.

10 . The method of claim 1 , wherein after determining the two-dimensional Morton code of the encoding point in the preset space of the encoding point cloud, and before performing Region Adaptive Hierarchical Transform (RAHT) based on the transform order corresponding to the preset space, the method further comprises:
respectively performing right shift of the two-dimensional Morton code of the encoding point to determine Ks respectively corresponding to the three-dimensional coordinate components of which point sets with maximum quantities of encoding points meet a preset point quantity threshold range; and determining, according to the Ks respectively corresponding to the three-dimensional coordinate components, the transform order corresponding to the preset space.

11 . The method of claim 1 , wherein the determining, based on the point quantities respectively corresponding to the three-dimensional coordinate components, the transform order corresponding to the preset space comprises:
determining the transform order to be zyx when the point quantities respectively corresponding to the three-dimensional coordinate components are NumX>NumY>NumZ; determining the transform order to be yzx when the point quantities respectively corresponding to the three-dimensional coordinate components are NumX>NumZ>NumY; determining the transform order to be zxy when the point quantities respectively corresponding to the three-dimensional coordinate components are NumY>NumX>NumZ; determining the transform order to be xzy when the point quantities respectively corresponding to the three-dimensional coordinate components are NumY>NumZ>NumX; determining the transform order to be yxz when the point quantities respectively corresponding to the three-dimensional coordinate components are NumZ>NumX>NumY; and determining the transform order to be xyz when the point quantities respectively corresponding to the three-dimensional coordinate components are NumZ>NumY>NumX

12 . The method of claim 10 , wherein the determining, according to the Ks respectively corresponding to the three-dimensional coordinate components, the transform order corresponding to the preset space comprises:
determining the transform order to be zyx when Ks respectively corresponding to the three-dimensional coordinate components are K(X)<K(Y)<K(Z); determining the transform order to be yzx when Ks respectively corresponding to the three-dimensional coordinate components are K(X)<K(Z)<K(Y); determining the transform order to be

Assignees

• Guangdong Oppo Mobile Telecommunications Corp Ltd

Inventors

• Zhang Wei
• Yang Fuzheng
• Wan Shuai
• Ma Yanzhuo
• Huo Junyan
• Dai Na
• Sun Zexing
• Yang Lihui

Classifications

• G06T9/007

  Transform coding, e.g. discrete cosine transform · CPC title

• H04N19/597

  specially adapted for multi-view video sequence encoding · CPC title

• G06T9/001 Primary

  Model-based coding, e.g. wire frame · CPC title

• H04N19/85

  using pre-processing or post-processing specially adapted for video compression · CPC title

• G06F5/01

  for shifting, e.g. justifying, scaling, normalising {(digital stores in which the information is moved stepwise, e.g. shift-registers G11C19/00; digital stores in which the information circulates G11C21/00)} · CPC title