Three-dimensional lane line detection method based on monocular camera

Through a collaborative detection method based on a monocular camera and an inertial measurement unit, the hardware dependence and environmental robustness problems of existing three-dimensional lane line detection methods are solved, and low-cost and efficient three-dimensional lane line reconstruction is achieved, which is suitable for a variety of road environments.

CN120689830APending Publication Date: 2025-09-23城市之光(深圳)无人驾驶有限公司
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202510786143.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-06-12
Publication Date
2025-09-23

AI Technical Summary

Technical Problem

Existing 3D lane detection methods suffer from strong hardware dependence, uncontrollable costs, complex algorithm structures, difficulty in achieving both real-time and accuracy, poor environmental robustness, and low generalization capabilities. In particular, when relying on lidar or deep learning 3D reconstruction networks, it is difficult to achieve the requirements of low cost, high accuracy, and strong generalization.

Method used

A three-dimensional lane line detection method based on a monocular camera is adopted. Through the steps of image acquisition, two-dimensional lane line detection, coordinate transformation, curve fitting and result output, the monocular camera and inertial measurement unit are used to collaboratively complete the three-dimensional lane line reconstruction, abandoning the dependence on lidar and complex deep neural networks, and adopting the back-projection geometric mapping algorithm and polynomial fitting technology.

Benefits of technology

It achieves low-cost, high-efficiency, and highly robust three-dimensional lane line detection, reduces the system's computational complexity, improves detection accuracy and environmental adaptability, is suitable for a variety of road environments, and has good engineering promotion value.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120689830A_ABST
    Figure CN120689830A_ABST
Patent Text Reader

Abstract

The invention discloses a three-dimensional lane line detection method based on a monocular camera. The method comprises the following steps: acquiring an RGB image of a front road scene by using the monocular camera installed on a vehicle; inputting the image into a two-dimensional lane line detection model, and extracting pixel coordinates of a lane line in an image coordinate system; converting the two-dimensional pixel points into three-dimensional coordinate points under an IMU coordinate system through a back projection method based on the internal reference and external reference matrix of the camera; performing curvature analysis and fitting on the obtained three-dimensional point cloud data to generate a smooth and continuous three-dimensional lane line curve equation; and applying the curve to path planning, navigation control or visual output. Compared with the prior art depending on laser radar or deep neural network reasoning, the method has the advantages of low hardware cost, small computing resource consumption, simple system structure, high environment adaptability and the like, and is particularly suitable for three-dimensional lane line modeling and sensing tasks under a resource-limited platform; and the method has good engineering popularization value and commercialization potential.
Need to check novelty before this filing date? Find Prior Art