3D Point Cloud Positioning Model Pruning for Mobile AR
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image-based positioning methods rely on high computing power and network connections, limiting their application on mobile devices due to efficiency and bandwidth constraints, and are not suitable for real-time Augmented Reality (AR) applications.
Innovation Solution
A method to optimize a positioning model by calculating the significance of each 3D point in a point cloud and retaining only points with a significance greater than a threshold, reducing the number of points and descriptors, enabling efficient positioning on mobile devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a complete positioning model with all 3D points and descriptors is used, then positioning accuracy is maintained, but positioning speed and processing efficiency deteriorate
Solution Approach 1:
The patent extracts and removes redundant 3D points and descriptors from the positioning model based on significance calculations. By calculating the significance of each 3D point and selectively retaining only those above a threshold, the system eliminates unnecessary data elements that contribute to processing overhead without compromising positioning accuracy.
Solution Approach 2:
The patent changes the parameter of the positioning model by adjusting the number of 3D points and descriptors through significance-based filtering. By dynamically selecting which points to retain based on calculated significance values, the system optimizes the balance between model completeness and processing efficiency.
2Measurement precision
If high computing power is used to process all 3D points and descriptors, then positioning accuracy is improved, but device complexity and resource requirements worsen
Solution Approach 1:
The patent extracts only the essential 3D points and descriptors needed for accurate positioning by calculating significance values. This extraction process removes redundant computational elements, thereby reducing the computing power and device complexity required while maintaining positioning accuracy.
Solution Approach 2:
The patent applies partial action by processing and retaining only a subset of 3D points that meet the significance threshold. Instead of processing all points excessively, the system performs selective processing on relevant points only, reducing computational burden while achieving sufficient positioning accuracy.
3Measurement precision
If network connections with high bandwidth are used for server-based positioning, then positioning accuracy is improved, but adaptability to mobile devices deteriorates
Solution Approach 1:
The patent extracts the essential positioning information into a optimized model that can be stored locally on mobile devices. By removing redundant 3D points and descriptors, the system creates a compact positioning model that fits within mobile device storage and processing capabilities, enabling offline positioning without high-bandwidth network connections.
Solution Approach 2:
The patent enables mobile devices to perform positioning independently using the optimized positioning model stored locally. By reducing the model size through significance-based filtering, the system allows mobile devices to self-service positioning tasks without relying on server connections or high bandwidth, improving adaptability to mobile environments.
Data Source
AI summary
A positioning model optimization method, a positioning method, and a positioning device are provided. The positioning model optimization method includes: inputting a positioning model for a scene, the positioning model including a three-dimensional (3D) point cloud and a plurality of descriptors corresponding to each 3D point in the 3D point cloud; calculating a significance of each 3D point in the 3D point cloud, and if the significance is greater than a predetermined threshold, outputting the 3D point and the plurality of descriptors corresponding to the 3D point to an optimized positioning model for the scene; and outputting the optimized positioning model for the scene.