Method and system for vehicle positioning from camera images

By employing semantic segmentation and 3D map matching, the problem of feature extraction difficulties caused by environmental changes in camera image localization was solved, achieving efficient and accurate vehicle localization and improving the robustness and accuracy of localization.

CN115035338BActive Publication Date: 2026-07-03VOLVO CAR CORP
View PDF 4 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
VOLVO CAR CORP
Filing Date
2017-11-27
Publication Date
2026-07-03

AI Technical Summary

Technical Problem

Existing technologies for vehicle localization using camera images face challenges such as difficulty in feature extraction due to changes in ambient light, and insufficient accuracy or low efficiency when the environment changes.

Method used

By using a semantic segmentation-based approach, two-dimensional images captured by a camera are classified into pixels. Combined with a three-dimensional map of a predetermined classification group, a learning algorithm and a Bayesian filter are used to match vehicle locations, reducing feature dependence and improving positioning accuracy and efficiency.

Benefits of technology

It achieves efficient and accurate vehicle positioning in the face of environmental changes, reduces the storage requirements for specific identifiable targets, and improves the robustness and accuracy of positioning.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115035338B_ABST
    Figure CN115035338B_ABST
Patent Text Reader

Abstract

A method for determining vehicle location is provided, comprising acquiring a 2D image showing the vehicle's surrounding environment using a vehicle-based camera; classifying pixels in the image such that each classified pixel belongs to a category of a predetermined classification group, thereby forming a classified 2D image. Classification is performed using a learning algorithm. The method further includes determining an initial estimated vehicle location and defining possible vehicle locations based on the initial location. Next, the method includes matching the classified 2D image with a 3D map including a plurality of geometric objects, each belonging to a category of the predetermined classification group, by comparing the classification of geometric objects for possible vehicle locations in a 3D map with the classification of at least one corresponding pixel in the classified 2D image; determining a matching score for at least one possible vehicle location based on the matching pixels in the 2D image; and determining the vehicle location based on the score of the possible vehicle location.
Need to check novelty before this filing date? Find Prior Art