AR Point Cloud Filtering for Low-Accuracy GNSS Zones
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Linking virtual reality (VR) and augmented reality (AR) devices to high-accuracy global navigation satellite system (GNSS) data has proven difficult.
Innovation Solution
A method for capturing point clouds using an AR device that involves determining a GNSS point, projecting depth images into 3D space, and displaying accurate points within a defined zone based on GNSS accuracy, with options to store and transform point clouds using AR and geospatial reference frames.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If depth images are projected into 3D space to create point clouds, then the visualization capability is improved, but the accuracy of point cloud data deteriorates when GNSS accuracy is low
Solution Approach 1:
The system applies different quality standards to different regions of the point cloud data. High-accuracy points within the zone are preserved and displayed with full detail, while low-accuracy points outside the zone are filtered out. This creates a spatially varying quality profile where only regions meeting accuracy thresholds contribute to the final visualization, resolving the contradiction between comprehensive visualization and data accuracy.
Solution Approach 2:
The system dynamically changes the accuracy threshold parameter based on the determined zone. By adjusting the minimum accuracy requirement for point inclusion based on the GNSS accuracy at each location, the system adapts the data quality standards to local conditions. This allows the point cloud to maintain high overall accuracy while still capturing useful data from areas with variable GNSS reception.
2Quantity of substance
If all captured points are displayed, then the completeness of the point cloud is improved, but the reliability of the AR system deteriorates due to inclusion of low-accuracy points
Solution Approach 1:
The system creates a virtual copy of the physical space through point cloud representation, but applies selective copying based on accuracy. Only points that meet the accuracy threshold within the determined zone are copied into the final point cloud dataset. This selective copying mechanism ensures that the virtual representation maintains high reliability by excluding low-accuracy measurements while still capturing the essential structure of the environment.
3Measurement precision
If points outside the zone are removed, then the accuracy of the displayed point cloud is improved, but the coverage area deteriorates
Solution Approach 1:
The system performs preliminary determination of the accuracy zone before finalizing the point cloud display. By pre-calculating which regions meet the accuracy threshold and defining the zone in advance, the system can efficiently filter points without requiring post-processing of the entire point cloud. This preliminary action allows the system to maintain high accuracy within the zone while minimizing the loss of coverage area by including all valid points within the determined boundaries.
Data Source
AI summary
Techniques for capturing and displaying point clouds using an AR device are described. A GNSS point of the AR device is determined based on received satellite signals. A depth image is captured while the AR device is positioned at the GNSS point. The depth image is projected into 3D space to obtain a point cloud having a set of points. An accuracy of the GNSS point is determined. In response to determining that the accuracy of the GNSS point is below a threshold, the GNSS point is determined to be a low-accuracy GNSS point, points from the set of points that are outside of a zone surrounding a previously determined high-accuracy GNSS point are removed, and remaining points from the set of points that are inside the zone are displayed on a display of the AR device.