Image-Based Map Layer Fusion for GPS-Denied Parking Localization

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The reliability of location information determined by GPS signals in parking lot environments is low, leading to inaccurate parking lot localization maps due to GPS signal interference.

Innovation Solution

A method involving feature point matching and map layer fusion based on image coordinates and motion parameters, eliminating the need for GPS-dependent location information, and utilizing projection error functions to determine displacement relationships between map layers.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If GPS signal is used to determine location information for building parking lot localization maps, then the process is simple and can be implemented with existing infrastructure, but the reliability and accuracy of location information deteriorates due to GPS signal interference in parking lot environments

Engineering Contradiction:
Improvereliability of location informationVSAvoidcomplexity of map building process
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent replaces the GPS-based mechanical positioning system with a visual-inertial navigation system using cameras and IMU sensors. Instead of relying on satellite signals, the system uses image capture components to obtain images of the parking lot environment, extracts feature points from these images, and determines displacement relationships between map layers through feature point matching and projection error minimization. This substitution eliminates GPS signal interference issues while maintaining positioning capability.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Manufacturing precision

If GPS-based location information is used to establish displacement relationships between feature points, then the implementation is straightforward, but the manufacturing precision and accuracy of the localization map deteriorates

Engineering Contradiction:
Improveaccuracy of parking lot localization mapVSAvoidcomplexity of feature point matching process
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent creates a visual copy of the parking lot environment through image capture and feature point extraction. Instead of directly using GPS coordinates, the system captures images of the environment, extracts key feature points from these images, and uses these visual copies to establish displacement relationships. The feature points serve as visual markers that replicate the spatial structure of the parking lot, enabling accurate map construction through image-based geometric relationships rather than GPS data.

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent transforms the problem from GPS coordinate-based positioning to image coordinate-based positioning. By changing the parameter space from absolute GPS coordinates to relative image coordinates and feature point descriptors, the system achieves higher precision in parking lot environments. The displacement relationships are expressed in terms of image coordinate transformations and feature point matching scores rather than GPS latitudes and longitudes, improving accuracy while accounting for the specific characteristics of indoor/covered parking environments.

Inventive Principle:
Principle #35Parameter changes

3Measurement precision

If traditional GPS-dependent methods are used for map building, then the system architecture is simple, but the measurement precision of location information deteriorates due to signal interference

Engineering Contradiction:
Improveprecision of location informationVSAvoidcomplexity of image-based map building system
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent divides the map building process into distinct segments: image capture, feature point extraction, feature point matching between map layers, displacement relationship determination, and map layer fusion. Each segment handles a specific aspect of the positioning task, allowing for specialized optimization at each stage. The feature point matching segment specifically addresses precision by using descriptor comparison and geometric verification, while the fusion segment integrates multiple map layers with calculated transformation relationships, achieving high measurement precision through modular processing.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP3876142B1Map building method, apparatus and system, and storage medium
Publication Date: 2026.02.11 YINWANG INTELLIGENT TECHNOLOGIES CO LTD
  • EP3876142B1 patent drawingFigure 1~2
  • EP3876142B1 patent drawingFigure 3
  • EP3876142B1 patent drawingFigure 4~6

AI summary

This application discloses a map building method, apparatus, and system, and a storage medium, and relates to the field of data processing technologies. The method may include: performing feature point matching on feature points at every two map layers of a plurality of map layers used to build a map, to obtain a plurality of feature point pairs, wherein each map layer includes feature points in a plurality of frames of images, and each frame of image carries image information of a target environment; obtaining a target displacement relationship between the two map layers based on map layer space coordinates, at a corresponding map layer, of each feature point in each feature point pair, wherein the map layer space coordinates are obtained based on image coordinates of the feature point in a target image and a relative displacement between capture locations of every two frames of target images, and the relative displacement is obtained based on a motion parameter of an image capture component at the image capture locations; and performing map layer fusion on the every two map layers based on the target displacement relationship between the every two map layers, to obtain a map of the target environment. This application effectively improves accuracy of the map obtained through map layer fusion.