Mobile Edge Image Matching for GPS-Denied Positioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing GPS positioning systems are unreliable in scenarios such as under bridges, main and auxiliary roads, indoor environments, and tall-building dense commercial areas due to satellite and network conditions, necessitating a more reliable positioning method.
Innovation Solution
A visual positioning method utilizing mobile edge computing (MEC) nodes that receive environmental images from devices, match them with pre-stored candidate images, and calculate position and pose information using feature point pairs and least square methods to determine accurate positioning without relying on GPS.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If GPS positioning is used, then positioning can be obtained in open areas, but positioning reliability deteriorates in scenarios such as under bridges, main and auxiliary roads, indoor environments, and tall-building dense commercial areas
Solution Approach 1:
The system divides the positioning task into two segments: GPS positioning for open areas and visual positioning for GPS-denied environments. This segmentation allows each method to operate in its optimal scenario, maintaining high reliability where applicable while providing fallback capability where needed.
Solution Approach 2:
The patent introduces visual positioning as an intermediary method that bridges the gap when GPS fails. By using image matching between captured environment images and pre-stored candidate images, the system provides continuous positioning capability across all scenarios, including under bridges, in tunnels, and dense urban areas where GPS signals are blocked.
2Reliability
If visual positioning with image matching is implemented, then positioning reliability improves in GPS-denied scenarios, but calculation complexity increases
Solution Approach 1:
The system performs preliminary action by pre-storing candidate images and their corresponding position information before actual positioning is needed. When positioning is required, the system only needs to match the captured image against these pre-prepared candidates, significantly reducing real-time calculation complexity compared to processing from scratch.
Solution Approach 2:
The patent uses copying by storing pre-captured images as reference candidates. Instead of performing complex feature extraction and matching algorithms on all possible images, the system copies and compares against a curated set of candidate images that have already been processed and stored with their position metadata, simplifying the matching process.
3Measurement precision
If image matching and feature point calculation are performed, then positioning precision improves, but processing time increases
Solution Approach 1:
The system applies partial action by selecting only the most relevant feature points for matching rather than processing all image features. By identifying and matching only key feature points between the captured image and candidate images, the system achieves sufficient positioning precision while reducing the computational burden and processing time.
Solution Approach 2:
The patent replaces complex mechanical calculation systems with optimized image processing algorithms. By using efficient feature point extraction and matching techniques, along with direct comparison against pre-stored candidate images, the system achieves high positioning precision with reduced computational time compared to traditional coordinate transformation methods.
Data Source
Figure 1a~1b
Figure 2a~2b
Figure 3~4a
AI summary
An embodiment of the present disclosure discloses a method and apparatus for visual positioning based on mobile edge computing, and relates to the technical field of computer visual image, in which image data is input to a trained model for image processing. Specifically, the mobile edge computing node receives an environment image captured by a to-be-positioned device in an area covered by the mobile edge computing node; the mobile edge computing node determines a target image matching the environment image from multiple candidate images, and calculates the position and pose information of the to-be-positioned device based on the position and pose information of the device for capturing the target image; and the mobile edge computing node sends the position and pose information of the to-be-positioned device to the to-be-positioned device, so that the to-be-positioned device determines the positioning information in the electronic map according to the position and pose information. The present embodiment adopts a positioning method based on computer vision, and is particularly suitable for positioning in a complex scenario.