摘要
针对高速公路施工期间人工监管违法用地难度大,而现有的图像识别与对比分析技术对数据要求高且误差显著的问题,提出一种基于数字图像处理的违法用地检测方法。首先对无人机正射影像进行颜色空间转换,分离背景区域与施工裸土区域;继而对征地红线二值影像取反,运用形态学连通域检测对取反掩膜图像分类,结合像素占比分析识别合法区域,并通过差值分析提取违法图斑。进一步,基于像素统计计算违法图斑面积,辅以用地要素相关数据的叠置分析,实现违法占地类型与行政区的判别;最后依据像素点饱和度与亮度的计算,评估违法严重程度。与深度学习、多序影像对比分析等传统方法相比,该方法漏检率为0,误报率低,不依赖海量训练样本,具有轻量化、智能化程度高的特点,能有效提高违法用地监管水平。
To address the challenges of manual supervision of illegal land use during the construction of highways,where existing image recognition and comparative analysis technologies demand high data quality and yet exhibit significant errors,this paper proposed an illegal land use detection method based on digital image processing.First,color space conversion was performed on the orthophoto image of the unmanned aerial vehicle to separate the background areas from the construction-exposed soil regions.Then,the binary images of the land acquisition red line were inverted.The inverted mask images were classified using morphological connected domain detection.The legal areas were identified by combining pixel proportion analysis,and the illegal patches were extracted through difference analysis.Furthermore,the area of illegal patches was calculated based on pixel statistics,supplemented by the superimposed analysis of relevant data of land use elements,to achieve the discrimination of illegal land occupation types and administrative regions.Finally,the severity of the violation was evaluated based on the calculation of the saturation and brightness of the pixel points.Compared with traditional methods such as deep learning and multi-order image comparison analysis,this method has a zero missed detection rate,a low false alarm rate,does not rely on a large number of training samples,and has the characteristics of lightweight and high intelligence.It can effectively improve the supervision level of illegal land use.
作者
熊燕文
卢照孔
XIONG Yanwen;LU Zhaokong(Guangzhou Congpu Expressway Co.,Ltd.,Guangdong University,Guangzhou 510000,China;Guangzhou Tianqin Digital Technology Co.,Ltd.,Guangdong University,Guangzhou 510000,China)
出处
《国外电子测量技术》
2025年第4期43-52,共10页
Foreign Electronic Measurement Technology
基金
2019年度交通运输行业重点科技项目(2019-ZD5-028)。
作者简介
熊燕文,硕士,高级工程师;通信作者:卢照孔,本科,工程师。E-mail:guwoai2025@163.com。
