Multi-color space SLAM (Simultaneous Localization and Mapping) method for enhancing low-illumination image

By integrating the low-light enhancement module of multi-color space on the front end of the SLAM, processing the brightness and color information of low-light images, the problem of poor image quality in the SLAM system in low-light environment is solved, and positioning accuracy and robustness are improved.

CN119693266APending Publication Date: 2025-03-25TONGXIANG GENERAL ARTIFICIAL INTELLIGENCE RESEARCH INSTITUTE +1
0 Cites 0 Cited by

Patent Information

Application Number
CN202411774750.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-12-05
Publication Date
2025-03-25

AI Technical Summary

Technical Problem

In low-light environments, the image quality of the SLAM system is poor, resulting in difficulty in extracting and matching feature points, affecting positioning accuracy and stability.

Method used

The low-light enhancement module integrating a multi-color space in the front end of the SLAM, by processing images in the LAB and YUV color spaces, the multi-head self-attention module and Fourier brightness processing module are used to finely restore the brightness details, and reduce noise through the channel denoising module. Finally, the brightness and color information are fused through the multi-stage compression excitation fusion module to optimize the modeling of lighting and color characteristics.

Benefits of technology

It significantly improves the performance of SLAM system in low-light environments, improves positioning accuracy and robustness, and enhances the extraction and matching effect of feature points.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention provides a multi-color space SLAM (Simultaneous Localization and Mapping) method for enhancing a low-illumination image. According to the method, a multi-color space low-illumination image enhancement module is integrated at the front end of an SLAM system, and the image enhancement module adopts an LAB color space and a YUV color space to respectively process brightness and color information of an image. The system specifically comprises a multi-head self-attention module and a Fourier brightness processing module which are used for finely recovering brightness details; and the channel denoising module is combined with a four-scale U-shaped network and a self-attention mechanism to reduce image noise and maintain color accuracy. And feature extraction and matching are carried out on the enhanced image through an ORB-SLAM2 framework, so that efficient pose estimation and map construction are realized. In order to optimize the image quality, a comprehensive loss function is adopted, and the brightness, the contrast ratio and the color fidelity of the image are ensured in combination with pixel-level and perception-level loss. In addition, the method has the characteristics of high real-time performance and excellent calculation efficiency, and is suitable for the fields of robot navigation, unmanned driving, augmented reality and the like.
Need to check novelty before this filing date? Find Prior Art