3D Feature Clustering for Compact Camera Positioning Databases
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing camera-based positioning methods require large 3D maps with significant computational and data storage demands, making them inefficient for use in shaded areas with weak GPS signals.
Innovation Solution
A method and apparatus for constructing a positioning DB using clustering of local features, where local features are grouped into clusters based on 3D keypoints and descriptors, storing only representative keypoints and descriptors for each cluster, reducing the data capacity required.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If camera-based positioning is used to achieve high positioning accuracy, then positioning accuracy is improved, but data storage capacity and computational load increase significantly
Solution Approach 1:
The patent segments the large 3D map data into multiple keyframes, and further segments each keyframe into multiple clusters of local features. By organizing features into clusters with representative positions and descriptors, the system reduces the overall data storage requirement while maintaining positioning accuracy through selective feature matching.
Solution Approach 2:
The patent extracts only the essential representative information (representative 3D keypoint and representative descriptor) from each cluster of local features, storing only this condensed representation in the positioning database. This extraction approach eliminates redundant data while preserving the critical information needed for accurate positioning.
2Measurement precision
If a detailed 3D map is constructed to improve positioning accuracy, then positioning accuracy is improved, but computational load increases
Solution Approach 1:
The patent divides the computational task of positioning into smaller segments by organizing features into clusters. Instead of processing all individual local features across the entire 3D map, the system processes cluster representatives, significantly reducing computational complexity while maintaining positioning precision through the preserved geometric and descriptive information.
Solution Approach 2:
The patent extracts and stores only the representative 3D keypoint and descriptor for each cluster, eliminating the need to process redundant individual feature data during positioning operations. This extraction reduces computational load by focusing processing on essential representative features rather than all原始 features.
3Measurement precision
If all local features are stored individually in the positioning DB, then positioning accuracy is maintained, but database size increases
Solution Approach 1:
The patent merges multiple similar local features into single clusters, combining their information into a representative 3D keypoint and descriptor. This merging process reduces database size by eliminating redundant individual feature entries while preserving the essential geometric and descriptive characteristics needed for accurate positioning through cluster-based matching.
Solution Approach 2:
The patent extracts only the essential representative information from each group of local features, storing the representative 3D keypoint and representative descriptor in the database. This selective extraction maintains positioning accuracy by preserving critical feature characteristics while dramatically reducing the volume of stored data by eliminating redundant individual feature records.
Data Source
AI summary
A method of constructing a positioning DB performed by an apparatus for constructing the positioning DB, may comprise: extracting a plurality of local features from a plurality of keyframes capturing a predetermined region; determining an individual 3D keypoint including information on a 3-dimensional position of each of the plurality of local features; clustering the plurality of local features into a plurality of clusters based on the individual 3D keypoint; determining a representative position information representatively indicating a position of each of the plurality of clusters by using the individual 3D keypoint of the local feature included in each of the plurality of clusters; and storing, for each of the plurality of keyframes, an cluster identification for identifying each of the plurality of clusters and the representative position information of each of the plurality of clusters in the positioning DB.