Point Cloud Attribute Encoding with Slice-Based Reference Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for encoding 3D data, such as point clouds, fail to restrict search targets for nearby points, leading to potential decoding failures when node groups are not decoded, which can result in unreliable attribute data decoding.
Innovation Solution
Implement a nearby point search that restricts reference points to nodes decoded earlier than the processing target node within a tree structure, and encode the difference value between attribute data and prediction values using these points, ensuring independent encoding and decoding of slices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If nearby point search is performed without restricting search targets, then more points can be considered as reference points, but decoding reliability deteriorates because nodes from undecoded groups may be incorrectly referenced
Solution Approach 1:
The patent divides the point cloud into multiple node groups (slices) that can be decoded independently. Each node group is processed separately, and the nearby point search is restricted to only include points within the current node group being decoded. This segmentation prevents references to points in undecoded groups, ensuring decoding reliability while maintaining the ability to process large point clouds through parallel independent decoding of each segment.
2Productivity
If node groups are formed for independent decoding, then decoding efficiency improves, but decoding reliability deteriorates when unrestricted nearby point search includes nodes from undecoded groups
Solution Approach 1:
The patent implements segmentation by dividing the point cloud into node groups (slices) that can be decoded independently and in parallel. Each node group is processed as a separate unit, improving decoding efficiency. The nearby point search is then restricted to only include points within the current node group, preventing references to undecoded groups and maintaining decoding reliability.
Solution Approach 2:
The patent applies local quality by making the nearby point search criteria different for different node groups. Specifically, the search is restricted to points within the current node group being decoded, rather than allowing global search across all points. This local restriction ensures that each node group can be decoded reliably using only locally available decoded points.
Data Source
AI summary
There is provided an information processing apparatus and method that enable attribute data to be decoded more reliably. Nearby point search for setting a reference point to be referred to when deriving a prediction value of attribute data of a processing target node is executed only for a node that is decoded earlier than the processing target node at the time of decoding among nodes of a tree structure in which attribute data of each point of a point cloud expressing an object having a three-dimensional shape as a set of points is a node and slices serving as node groups that can be encoded independently of each other are formed. The present disclosure can be applied to, for example, an information processing apparatus, an encoding apparatus, a decoding apparatus, an electronic device, an information processing method, or a program.