Spinning LiDAR Point Cloud Entropy Coding for Low-Latency Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing point cloud codecs fail to combine simplicity, low latency, and high compression performance, particularly in encoding and decoding sparse geometry data captured by spinning Lidar sensors for applications like autonomous driving.
Innovation Solution
A method and apparatus for encoding and decoding point clouds that orders points based on azimuthal angles and sensor indices, using context-adaptive entropy coding for order index differences and residuals, allowing efficient compression and low latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If existing point cloud codecs are used, then compression performance can be achieved, but encoding complexity and latency increase
Solution Approach 1:
The patent segments the point cloud encoding process into distinct stages: organizing points by azimuthal angle and sensor index, computing order index differences, and applying context-adaptive entropy coding. This segmentation allows each component to be optimized independently, reducing overall complexity while maintaining compression efficiency.
Solution Approach 2:
The patent performs preliminary organization of points by azimuthal angle and sensor index before encoding. This preliminary sorting action creates a structured representation that simplifies the subsequent entropy coding process and reduces latency by preparing data in advance.
2Loss of energy
If existing point cloud codecs are used, then compression can be achieved, but encoding latency increases
Solution Approach 1:
The patent performs preliminary organization of points by azimuthal angle and sensor index before encoding. This preliminary sorting action creates a structured representation that simplifies the subsequent entropy coding process and reduces latency by preparing data in advance.
Solution Approach 2:
The patent extracts and encodes only the order index differences rather than encoding all point coordinates directly. This extraction of differential information reduces the amount of data to process and enables more efficient, lower-latency encoding while maintaining compression performance.
3Loss of time
If simple encoding methods are used, then latency is reduced, but compression performance deteriorates
Solution Approach 1:
The patent applies context-adaptive entropy coding that adjusts compression strength based on local characteristics of the point cloud data. Different contexts are assigned different coding parameters, allowing simple encoding operations to achieve high compression performance when data patterns permit, without increasing latency.
Solution Approach 2:
The patent changes the encoding parameters dynamically based on context, using context-adaptive entropy coding. This allows the system to switch between more compressive and more efficient coding modes depending on the local data characteristics, achieving both low latency and high compression performance.
4Loss of energy
If context-adaptive entropy coding is applied, then compression performance improves, but device complexity increases
Solution Approach 1:
The patent segments the point cloud encoding process into distinct stages: organizing points by azimuthal angle and sensor index, computing order index differences, and applying context-adaptive entropy coding. This segmentation allows each component to be optimized independently, reducing overall complexity while maintaining compression efficiency.
Data Source
Figure 1~3
Figure 4~5
Figure 6
AI summary
There is provided methods and apparatus of encoding/decoding a point cloud representing a physical object. Points of the point cloud are ordered based on azimuthal angles representing capture angles of sensors and sensor indices associated with sensors. The encoding method comprises encoding, into a bitstream, at least one order index difference representing a difference between order indices of two consecutive ordered points by: obtaining at least one binary data by binarizing the at least one order index difference; and for each binary data, selecting a context based on a distance between an azimuthal angle associated with the binary data and an azimuthal angle of an already encoded point, and context-based entropy coding the binary data in the bitstream, based on the selected context.