AR Geolocation via Filtered Image Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining the geolocation of user devices in augmented reality (AR) systems lack precision, leading to inaccurate placement of AR objects in real-world environments, and require substantial processing power, which is inefficient for dynamic user movements.
Innovation Solution
A system that uses image matching techniques by comparing live image content from a user device with a filtered group of stored images, selected based on condition attributes such as geolocation, tilt, heading, altitude, and environmental conditions, to determine a high-precision geolocation, reducing the number of images to compare and minimizing processing requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image matching is performed against all stored images to achieve high-precision geolocation, then measurement precision is improved, but processing power requirements increase substantially
Solution Approach 1:
The patent segments the image matching process into two distinct phases: (1) a filtering phase that divides stored images into groups based on condition attributes like geolocation, time, and environmental factors, and (2) a matching phase that compares live images only against the filtered subset. This segmentation reduces the number of images requiring full comparison while maintaining matching accuracy, thereby reducing processing power consumption.
Solution Approach 2:
The patent applies preliminary action by pre-filtering stored images based on condition attributes (geolocation, time, environmental conditions) before performing the actual image matching. This preliminary filtering step prepares a reduced subset of candidate images that are more likely to match the live image, avoiding unnecessary comparisons with irrelevant images and thus reducing processing requirements.
2Measurement precision
If a large number of stored images are compared to determine high-precision geolocation, then measurement precision is improved, but loss of time increases due to extensive processing
Solution Approach 1:
The patent segments the stored image collection into multiple groups based on condition attributes such as geolocation, time of day, and environmental conditions. This segmentation allows the system to quickly identify and compare images only from the relevant group, dramatically reducing the time required for image matching while maintaining precision by ensuring the compared images are contextually appropriate.
Solution Approach 2:
The system performs preliminary filtering of stored images based on condition attributes before the actual matching process. This preliminary action pre-sorts images into relevant groups, so when a live image needs matching, the system only needs to search through the pre-filtered subset rather than all stored images, significantly reducing processing time.
3Manufacturing precision
If arbitrary placement of AR objects is used in existing systems, then device complexity is reduced, but manufacturing precision of AR presentation deteriorates
Solution Approach 1:
The patent replaces the simple but imprecise arbitrary placement mechanism with an image-matching-based positioning system. Instead of placing AR objects at arbitrary coordinates, the system uses visual comparison between live and stored images to determine precise real-world locations, substituting a mechanical coordinate system with a visual recognition system that achieves higher placement precision.
Solution Approach 2:
The system changes the parameters used for AR object placement from arbitrary screen coordinates to precise real-world geolocation data derived from image matching. By transforming the placement parameters from abstract screen positions to actual spatial coordinates, the system achieves higher manufacturing precision for AR presentations.
Data Source
AI summary
Disclosed is a system for determining high-precision geolocation. The system includes a user device having a camera, a non-transitory memory containing computer readable instructions, and a processor. The processor is configured to process the instructions receive live image content from the camera, receive information indicative of one or more device condition attribute that is associated with the user device, select a group of stored images based on the one or more stored image condition attribute, compare the live image content to the group of stored images, based on the comparing, select, from the group of stored images, one or more matching stored image that matches the live image content, and determine a high-precision geolocation of the user device based on the matching one or more stored image.