3D Point Cloud Encoding Using Dynamic Bit Depth Adaptation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for encoding and decoding 3D point cloud data suffer from inefficient bit depth processing and lack of error correction for attribute information, leading to errors in reconstructed data due to near-lossless/lossy compression.
Innovation Solution
A method and apparatus that signal bit depth information in a sequence parameter set, perform bit depth transformation, and generate an error correction table to accurately reconstruct attribute information, improving encoding/decoding efficiency and correcting errors in attribute data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If fixed bit depth processing (e.g., 16-bit) is performed on attribute information, then encoding/decoding is simplified, but encoding/decoding performance deteriorates when attribute values require less than 16 bits or need compression
Solution Approach 1:
The patent introduces dynamic bit depth adaptation where the system determines the actual bit depth required for attribute information based on the data characteristics, rather than using a fixed 16-bit depth. This allows the encoding/decoding process to adapt to the actual needs of the data, improving performance when fewer bits are sufficient while maintaining simplicity when full precision is required.
Solution Approach 2:
The patent changes the bit depth parameter dynamically based on the attribute information characteristics. By signaling bit depth information in the sequence parameter set and transforming bit depth for each attribute, the system optimizes the balance between processing simplicity and encoding/decoding performance according to the actual data requirements.
2Productivity
If near-lossless/lossy compression is applied to attribute information through quantization, then compression ratio is improved, but errors are introduced in the reconstructed attribute information
Solution Approach 1:
The patent performs preliminary error correction by generating an error correction table before final reconstruction. This table is constructed using bit-depth-changed attribute values and is used to correct errors in the decoded attribute information, proactively addressing potential reconstruction errors before they affect the final output quality.
Solution Approach 2:
The patent implements a feedback mechanism where error correction information is derived from the compressed data itself. The error correction table is generated based on the actual compressed and decompressed attribute values, creating a feedback loop that identifies and corrects errors introduced during quantization and compression, thereby improving reconstruction accuracy.
Data Source
AI summary
There is provided a method of decoding point cloud data, the method comprising: decoding a bitstream to obtain attribute information of point cloud data, acquiring attributes of the point cloud data using the decoded attribute information and reconstructing the point cloud data using the obtained attributes, wherein the attribute information contains bit depth information of each attribute of the point cloud data.