Spinning LiDAR Point Cloud Geometry Coding With Order Index Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing point cloud codecs fail to combine simplicity, low latency, and high compression performance for sparse geometry data captured by spinning Lidar sensors, particularly in applications like autonomous driving, where real-time processing and efficient data transmission are crucial.
Innovation Solution
A method and apparatus for encoding and decoding point cloud data that orders points based on azimuthal angles and sensor indices, using a lexicographic order and encoding order index differences, along with optional quantization and entropy coding of radius and residual coordinates, to simplify encoding and reduce latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If existing point cloud codecs are used to compress sparse geometry data from spinning Lidar sensors, then compression is performed, but the codecs fail to achieve a combination of simplicity, low latency, and high compression performance
Solution Approach 1:
The patent segments the point cloud data processing by separating geometry data and attribute data into independent encoding streams. The geometry data is encoded using a dedicated geometry encoder that exploits the specific structure of spinning Lidar data, while attribute data is encoded separately using an attribute encoder. This segmentation allows each encoder to be optimized for its specific data type, achieving high compression performance without requiring a complex unified encoding system.
Solution Approach 2:
The patent transforms the 3D point cloud data into a 2D representation by unfolding the spherical coordinate system into a 2D plane. Points are mapped from 3D spherical coordinates (radius, azimuthal angle, polar angle) to 2D array positions based on their angular coordinates. This dimensionality change enables the application of 2D compression techniques to 3D data, achieving high compression performance while maintaining encoding simplicity.
2Productivity
If existing point cloud codecs are used to compress sparse geometry data, then compression is performed, but latency is high due to complex encoding processes
Solution Approach 1:
The patent performs preliminary organization of points based on their angular coordinates (azimuthal and polar angles) before encoding. Points are sorted and arranged in a predetermined order that exploits the sequential nature of spinning Lidar data acquisition. This preliminary action reduces the computational complexity during the actual encoding phase, enabling low-latency real-time processing while maintaining high compression performance.
Solution Approach 2:
The patent changes the parameter representation by using angular coordinates (azimuthal and polar angles) as the primary sorting and encoding parameters instead of traditional 3D Cartesian coordinates. This parameter change aligns with the natural acquisition sequence of spinning Lidar sensors, reducing the computational overhead during encoding and enabling low-latency real-time processing.
3Device complexity
If simple encoding methods are used for point cloud data, then encoding complexity is reduced, but compression performance deteriorates
Solution Approach 1:
The patent applies 2D compression techniques to the angular dimension of 3D point cloud data by mapping points to a 2D array based on their azimuthal and polar angles. This allows the use of simple and efficient 2D compression algorithms while achieving high compression performance on 3D data, resolving the contradiction between encoding simplicity and compression ratio.
Solution Approach 2:
The patent creates a universal encoding framework that can handle both geometry data and attribute data using similar processing principles. The geometry encoder and attribute encoder follow analogous processing steps (organizing, encoding, and transmitting data), allowing the system to achieve high compression performance through a simple and unified approach rather than complex specialized encoders.
4Productivity
If complex encoding algorithms are used to achieve high compression performance, then compression ratio improves, but processing speed decreases
Solution Approach 1:
The patent performs preliminary organization of points based on angular coordinates and pre-computes the order in which points will be encoded. This preliminary action reduces the computational work required during actual encoding, enabling the use of effective compression algorithms without sacrificing processing speed. The pre-organization exploits the sequential acquisition pattern of spinning Lidar data to minimize runtime computations.
Data Source
Figure 1~2
Figure 3
Figure 4~5
AI summary
There is provided methods and apparatus of encoding/decoding a point cloud representing a physical object. Points are captured by a spinning sensors head and are represented by sensor indices associated with sensors that captured the points, azimuthal angles representing capture angles of said sensors, and radius values of spherical coordinates of the point. Points are ordered based order indices obtained from the azimuthal angles and the sensor indices. Order index differences are encoded. An order index difference represents a difference between order indices associated with two consecutive ordered points. Optionally, the method encodes radius values, residual azimuthal angles associated with ordered points and residuals of three-dimensional cartesian coordinates of ordered points based on their three-dimensional cartesian coordinates, decoded azimuthal angles based on azimuthal angles, decoded radius values and sensor indices.