3D Point Cloud Encoding Using Dynamic N-ary Tree Occupancy Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
There is a demand for a three-dimensional data decoding device that can decode a bitstream appropriately, but existing methods struggle with efficient encoding and decoding of three-dimensional data, particularly in point cloud compression.
Innovation Solution
The proposed solution involves a three-dimensional data encoding method that converts geometry information of three-dimensional points into occupancy maps using an N-ary tree, and then encodes these maps into a bitstream using a table that indicates the correspondence between occupancy maps and indexes. This method also determines whether to initialize the table based on the lead point of the next data unit, allowing for efficient encoding and decoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a table indicating correspondence between occupancy maps and indexes is used for encoding, then coding efficiency is improved, but device complexity increases due to table management and initialization requirements
Solution Approach 1:
The table initialization is made dynamic based on the lead point position. The decoder determines whether to initialize the table by checking if the lead point of the next data unit is the same as or different from the lead point of the current data unit. This dynamic approach allows the system to adapt table management to the actual data characteristics, improving coding efficiency while avoiding unnecessary initialization operations that would increase complexity.
Solution Approach 2:
The system uses the lead point information already present in the bitstream to automatically determine table initialization requirements. By embedding the lead point position in the encoded data and using it to control table initialization, the system makes autonomous decisions without requiring external control signals, thereby improving efficiency while keeping the control mechanism simple.
2Measurement precision
If the table is initialized for each data unit, then decoding accuracy is improved, but processing time increases
Solution Approach 1:
The table initialization is dynamically controlled based on lead point comparison. When the lead point of the next data unit differs from the current one, the table is initialized to ensure decoding accuracy. When the lead points are the same, the table is reused without initialization, reducing processing time. This dynamic strategy balances accuracy requirements with time efficiency.
Solution Approach 2:
The system uses feedback from the lead point position information to control table initialization. By comparing the lead point of the next data unit with the current one, the system determines whether initialization is necessary, thereby maintaining decoding accuracy only when needed and avoiding unnecessary time consumption.
3Loss of time
If the table is reused across data units, then processing time is reduced, but decoding reliability decreases
Solution Approach 1:
The table reuse strategy is dynamically adjusted based on lead point comparison. The system reuses the table when lead points are identical (reducing processing time) but initializes the table when lead points differ (maintaining decoding reliability). This dynamic approach optimizes the balance between speed and reliability based on actual data characteristics.
Solution Approach 2:
Lead point information serves as feedback to control table reuse. The decoder compares the lead point of the next data unit with the current one and uses this feedback to decide whether to reuse or reinitialize the table, thereby maintaining reliability when necessary while minimizing processing time.
Data Source
AI summary
A three-dimensional data encoding method includes: obtaining data units each including three-dimensional points; encoding the three-dimensional points included in each of the data units; and generating a bitstream including encoded data obtained by encoding the three-dimensional points. In the encoding: geometry information of three-dimensional points included in a data unit to be encoded is converted into occupancy maps using an N-ary tree; each of the occupancy maps is converted into an index using a table indicating a correspondence relation between occupancy maps and indexes, and the encoded data is generated by encoding the index; the table is updated according to the index resulting from the conversion, and is stored into the memory; and when it is determined that the table stored in the memory is not to be initialized, encoding of a next data unit is started using the table stored in the memory.