用近红外光提升视觉定位精度，解决光照干扰问题
Research on visual simultaneous localization and mapping technology based on near infrared light
- 采用近红外光作为光源，规避环境光干扰
- 提出光照参数估计与辐射校正方法，提升DSO定位精度
- 在四个大场景验证可行性，适用于自动驾驶与机器人
针对视觉同步定位与地图构建（VSLAM）易受环境光干扰及亮度不一致的问题，本文提出基于近红外感知的视觉同步定位与地图构建技术（NIR-VSLAM）。通过创新性选用近红外光作为光源，避免环境光影响。提出误差能量函数中的亮度参数估计、眩光因子与曝光时间建模方法，以及光源辐照度校正策略，显著提升直接稀疏里程计（DSO）的定位精度。在四个大型场景中验证了该方法的可行性，为自动驾驶与移动机器人的视觉定位提供了有效参考。
原文摘要 · Abstract (English)
In view of the problems that visual simultaneous localization and mapping (VSLAM) are susceptible to environmental light interference and luminosity inconsistency, the visual simultaneous localization and mapping technology based on near infrared perception (NIR-VSLAM) is proposed. In order to avoid ambient light interference, the near infrared light is innovatively selected as the light source. The luminosity parameter estimation of error energy function, halo factor and exposure time and the light source irradiance correction method are proposed in this paper, which greatly improves the positioning accuracy of Direct Sparse Odometry (DSO). The feasibility of the proposed method in four large scenes is verified, which provides the reference for visual positioning in automatic driving and mobile robot.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。