Point Cloud Transmission Using Octree-Based LOD Generation

The point cloud data is divided into multiple Levels of Detail (LoDs), where each LoD represents a different level of spatial resolution. This segmentation allows the receiver to process and display data at coarser levels first (reducing latency) while optionally refining to finer levels later, thereby reducing both latency and processing complexity through hierarchical organization.

The patent introduces a hierarchical dimension to the point cloud data by organizing points into multiple LoDs with different spatial resolutions. This dimensional organization transforms the flat data structure into a multi-level hierarchy, enabling progressive transmission and decoding where coarser levels are processed first, thus reducing latency and complexity.

The patent applies different attribute compression techniques to different spatial regions based on their local characteristics. For regions with asymmetric point distributions, specific prediction and compression methods are applied locally rather than uniformly across the entire point cloud. This local adaptation improves compression efficiency for asymmetric regions while keeping the overall system manageable by only applying complex methods where needed.

The patent dynamically adjusts compression parameters based on the local point distribution characteristics. By analyzing the asymmetry of point distributions in different regions, the system changes compression parameters (such as prediction modes, quantization levels, and transformation types) to optimize compression efficiency for each local region, thereby improving overall compression while avoiding uniform application of complex algorithms.