基于人工智能的路面抛洒事件检测及应用研究

Research on Detection and Application of Road Sprinkling Events Based on Artificial Intelligence

摘要: 公路路面抛洒事件与交通安全密切相关，本文通过路侧布设的固定视频图像，基于高斯混合模型(GMM)中的HSV颜色空间(色调、饱和度、明度)，构建路面抛洒事件智能检测算法，突破复杂交通环境下的公路视频图像样本采集难题，实现路面抛洒事件的智能识别及主动预警，研究结果表明检测准确率达到90.42%，漏检率低于10%。

Abstract: Road surface sprinkling events are closely related to traffic safety. This paper uses fixed video images deployed on the roadside to construct an intelligent detection algorithm for road surface sprinkling events through the HSV color space (hue, saturation, and brightness) of the Gaussian Mixture Model (GMM). The proposed method overcomes the difficulty of collecting road video image samples in complex traffic environments, and achieves intelligent recognition and active warning of road surface sprinkling events. The results indicate that the detection accuracy reaches 90.42%, and the missed detection rate is less than 10%.