一种基于SVM的无人机影像中单个建筑物的角点检测方法
发布时间：2018-07-16 17:31
【摘要】：针对目前无人机影像中单个建筑物角点的检测现状,提出了一种基于支持向量机(SVM)的无人机影像中建筑物的角点检测方法。首先对4个波段的无人机影像进行多尺度分割,计算影像的NDVI,通过植被与非植被区域的波谱差异剔除植被的影响。其次,用面向对象分类法将"建筑物块"从影像中提取出来,对"建筑物块"区域用Harris算子进行边缘检测,形成建筑物边缘点集数据。随后通过设计高斯径向基将边缘样本点映射到高维特征空间,构建特征向量,采用边缘点集训练SVM分类模型,最终通过SVM分类模型从粗提取的边缘点集中检测出正确的建筑物角点,实现了单个建筑物的角点提取。
[Abstract]:According to the current situation of single building corner detection in UAV image, a new method of building corner detection in UAV image based on support vector machine (SVM) is proposed. Firstly, the multi-scale segmentation of UAV images in four bands was carried out, the NDVI of the images was calculated, and the influence of vegetation was eliminated by the spectral differences between vegetation and non-vegetation regions. Secondly, the "building block" is extracted from the image by the object-oriented classification method, and the edge of the "building block" area is detected by Harris operator to form the building edge point set data. Then, the edge sample points are mapped to the high dimensional feature space by designing Gao Si radial basis function, and the feature vectors are constructed, and the classification model is trained by edge point set. Finally, the correct corner points of buildings are detected from rough edge points by SVM classification model, and the corner points of a single building are extracted.
【作者单位】： 桂林理工大学测绘地理信息学院;广西空间信息与测绘重点实验室;南宁市勘察测绘地理信息院;
【基金】：国家自然科学基金(41161073) 广西自然科学基金(2016GXNSFAA380013;2014GXNSFDA118038) 桂林市科学研究与技术开发计划(2016012601) 重庆基础科学与前沿技术研究项目(cstc2015jcyj B028)
【分类号】：P237
本文编号：2127107
[Abstract]:According to the current situation of single building corner detection in UAV image, a new method of building corner detection in UAV image based on support vector machine (SVM) is proposed. Firstly, the multi-scale segmentation of UAV images in four bands was carried out, the NDVI of the images was calculated, and the influence of vegetation was eliminated by the spectral differences between vegetation and non-vegetation regions. Secondly, the "building block" is extracted from the image by the object-oriented classification method, and the edge of the "building block" area is detected by Harris operator to form the building edge point set data. Then, the edge sample points are mapped to the high dimensional feature space by designing Gao Si radial basis function, and the feature vectors are constructed, and the classification model is trained by edge point set. Finally, the correct corner points of buildings are detected from rough edge points by SVM classification model, and the corner points of a single building are extracted.
【作者单位】： 桂林理工大学测绘地理信息学院;广西空间信息与测绘重点实验室;南宁市勘察测绘地理信息院;
【基金】：国家自然科学基金(41161073) 广西自然科学基金(2016GXNSFAA380013;2014GXNSFDA118038) 桂林市科学研究与技术开发计划(2016012601) 重庆基础科学与前沿技术研究项目(cstc2015jcyj B028)
【分类号】：P237
【相似文献】
相关期刊论文 前1条
1 闫利,林宗坚,于子凡;基于结构分析的线划角点检测[J];武汉测绘科技大学学报;1996年03期
相关会议论文 前1条
1 张睿;陈雪;马建文;;基于递归SVM的高光谱数据特征选择算法研究[A];第八届成像光谱技术与应用研讨会暨交叉学科论坛文集[C];2010年
相关硕士学位论文 前1条
1 张汉女;基于SVM的海岸线提取方法研究[D];东北师范大学;2010年
，本文编号：2127107
本文链接：https://www.wllwen.com/kejilunwen/dizhicehuilunwen/2127107.html
