主办：陕西省汽车工程学会
ISSN 1671-7988  CN 61-1394/TH
创刊：1976年
汽车实用技术

汽车实用技术 ›› 2025, Vol. 50 ›› Issue (8): 22-28.DOI: 10.16638/j.cnki.1671-7988.2025.008.005

• 智能网联汽车 • 上一篇    

融合插值测量和单目测距的盲区安全预警

张云飞 1,2，李斌 1,2，杨良义 1,2，韩中海 1,2   

  1. 1.智能汽车安全技术全国重点实验室； 2.中国汽车工程研究院股份有限公司
  • 发布日期:2025-04-25
  • 通讯作者: 张云飞
  • 作者简介:张云飞（1993－），男，助理工程师，研究方向为智能网联汽车驾驶辅助系统算法及测试评价技术
  • 基金资助:
    中国汽车工程研究院股份有限公司科研项目（0002KTCP20230880）：智能驾驶系统实车测试技术能力提升

Safety Warning of Blind Spots by Integrating Interpolation Measurement and Monocular Ranging

ZHANG Yunfei1,2 , LI Bin1,2 , YANG Liangyi1,2, HAN Zhonghai1,2   

  1. 1.National Key Laboratory of Intelligent Vehicle Safety Technology; 2.China Automotive Engineering Research Institute Company Limited
  • Published:2025-04-25
  • Contact: ZHANG Yunfei

PDF

摘要： 提出一种融合插值测量和单目测距的盲区安全预警方法，旨在为驾驶员提供足够的反 应时间，以有效规避潜在的交通事故风险。通过 YOLOv3 算法对视野盲区障碍物进行实时、 高精度检测；采用单目测距技术，结合摄像机的内外参数以及障碍物的图像特征，初步估算 出视野盲区障碍物与车辆之间的距离；引入插值测量技术，结合预先在已知距离下采集的精 确测量点，动态地校正和优化测距结果；综合考虑驾驶员在应对突发状况时的反应时间、车 速以及摩擦系数等因素，计算出视野盲区安全预警距离，并精确判断何时发出安全预警信号。 实验结果表明，该方法能准确检测出车辆周围的盲区障碍物，实时测量出其与车辆之间的距 离，并提前发出预警信号，从而显著提升了车辆的行驶安全性。

关键词: 插值测量；单目测距；YOLOv3 算法；安全预警

Abstract: A safety warning method for blind field of vision combining interpolation measurement and monocular distance measurement is proposed to provide sufficient response time for drivers and effectively avoid potential traffic accident risks. Real time and high-precision detection of blind spots and obstacles vision using YOLOv3 algorithm; Using monocular ranging technology, utilizing the internal and external parameters of the camera and the image features of obstacles, the distance between blind spots and obstacles in the field of view and vehicles is preliminarily estimated; Further introduce interpolation measurement technology to dynamically correct and optimize distance measurement results by collecting precise measurement points at known distances in advance;Taking into account the reaction time, vehicle speed, and friction coefficient of vehicles when responding to emergencies, calculate the safety warning distance of blind spots field of view, and accurately determine when to issue a safety warning. The experimental results show that this method can accurately detect blind spots and obstacles around vehicles, measure the distance between them in real time, and issue warning signals in advance, driving safety of the vehicle can be significantly improved.

Key words: interpolation measurement; monocular distance measurement; YOLOv3 algorithm; safety warning