基于HSV空间距离模型和超像素分割的公路护栏图像提取  

Highway guardrail extraction based on HSV spatial distance model and superpixel segmentation

作　　者：靳霄微 高明星[1] 方淑艳 范井丽 `JIN Xiaowei;GAO Mingxing;FANG Shuyan;FAN Jingli(College of Energy and Transportation Engineering,Inner Mongolia Agricultural University,Hohhot,Inner Mongolia 010018,China)

机构地区：[1]内蒙古农业大学能源与交通工程学院,内蒙古呼和浩特010018

出　　处：《公路交通科技》2025年第4期24-32,共9页Journal of Highway and Transportation Research and Development

基　　金：内蒙古自治区重点科技攻关项目(2021GG0178);内蒙古自治区高校重点科研项目(NJZZ21013);中央引导地方科技发展项目(2024ZY0042,2024ZY0111)。

摘　　要：【目标】针对复杂背景下公路波形护栏分割精度较低、背景噪声干扰较强的问题,提出一种结合HSV颜色空间和超像素分割的公路护栏分割算法,以提高护栏分割的准确性和鲁棒性。【方法】首先,基于改进的HSV颜色空间距离模型自适应提取目标区域的HSV分量,生成ROI区域的二值图像,初步定位护栏的位置;其次,通过形态学重建和连通域分割去除背景中小面积强干扰噪声,使护栏轮廓更加清晰;最后,融合Gabor滤波器和超像素分割技术,结合纹理特征解决相似语义的干扰问题,Gabor滤波器提取图像的纹理特征,超像素分割则将图像划分为具有相似特征的区域,得到ROI区域的二值化图像。【结果】与传统的大津法和Sobel算法相比,该算法在复杂背景下的公路波形护栏分割中表现优异,能够清晰地提取护栏的结构和轮廓,分割准确率达到90%以上,同时,在面对复杂背景和噪声干扰时,准确率和鲁棒性上均得到显著提升。【结论】基于HSV颜色空间和超像素分割的公路护栏分割算法,能够有效展现护栏的结构特征和形状特征,为后续的护栏缺陷检测奠定坚实的基础,具有较高的应用价值和广阔的应用前景。[Objective]In view of the problems of low segmentation accuracy and strong background noise interference of highway wave-shaped guardrail in complex background,a highway guardrail segmentation algorithm combining HSV colour space and superpixel segmentation was proposed to improve the guardrail segmentation accuracy and robustness.[Method]First,the HSV components in target region were adaptively extracted based on the improved HSV colour spatial distance model.A binary image in ROI region was generated to initially locate the guardrail position.Second,the small-area interference noise in the background was removed by morphological reconstruction and connectivity domain segmentation,so as to make the guardrail outline clearer.Finally,the Gabor filter and superpixel segmentation technology were integrated.The texture features were combined to solve the interference problem of similar semantics.Gabor filters extracted the texture feature of images;while the super pixel segmentation divided the images into regions with similar features.The binary images in ROI region were obtained.[Result]Compared with the traditional Otsu method and Sobel algorithm,the proposed algorithm performs well in the highway wave-shaped guardrail segmentation under complex background.It can clearly extract the structure and outline of guardrail,and the segmentation accuracy reaches over 90%.Meanwhile,when facing the complex background and noise interference,the accuracy and robustness are significantly improved.[Conclusion]The highway guardrail segmentation algorithm,based on HSV colour space and superpixel segmentation,can effectively display the structural and shape features of guardrail.It lays a solid foundation for the subsequent guardrail defect detection,and has high application value and broad application prospects.

关 键 词：智能交通 图像分割 HSV空间距离 波形护栏 超像素分割 Gabor过滤器 

分 类 号：U417.12[交通运输工程—道路与铁道工程] TP391.41[自动化与计算机技术—计算机应用技术]