基于无人机技术的交通路标图像的识别方法研究

Research on Recognition Method of Traffic Sign Images Based on UAV Technology

摘要: 针对传统道路养护需要人工巡查的弊端，文中提出利用无人机技术，将其获取的图像作为道路标线识别与检测的一种技术手段。通过对路面颜色和梯度变化特征的分析，将图像中的路面区域划分出来，并精准识别出这些区域内的连续部分。结合Faster R-CNN算法与连通区域的颜色面积特征，剔除非标线区域。识别并保留了所有标线区域，实现公路标线的自动识别和准确提取。通过实验验证了本文提出方法具有较高的适应性和灵活性，即使在复杂的公路环境下也能保持高准确率。

Abstract: In response to the drawbacks of traditional road maintenance that requires manual inspection, this article proposes using drone technology to obtain images as a technical means for road marking recognition and detection. By analyzing the characteristics of road color and gradient changes, the road areas in the image are divided and the continuous parts within these areas are accurately identified. By combining the Faster R-CNN algorithm with the color area features of connected regions, non-marking areas are removed. Identified and retained all marked areas, achieving automatic recognition and accurate extraction of highway markings. The proposed method has been experimentally verified to have high adaptability and flexibility and can maintain high accuracy even in complex highway environments.