基于改进Sobel算子的遥感图像道路边缘检测方法  

Road edge detection from remote sensing image based on improved Sobel operator

作　　者：谭媛[1] 黄辉先[1] 徐建闽[2] 陈任[1] 

机构地区：[1]湘潭大学信息工程学院,湘潭411105 [2]华南理工大学土木与交通学院,广州510640

出　　处：《国土资源遥感》2016年第3期7-11,共5页Remote Sensing for Land & Resources

基　　金：国家自然科学基金面上项目"基于群体动力学的交叉口群协调控制理论与方法研究"(编号:61174184);湖南省教育厅重点项目(编号:12A136)共同资助

摘　　要：从遥感图像中提取道路边缘可以大量简化道路网的测绘与规划工作。传统边缘检测算子由于方向和模板尺寸的局限性,易造成检测结果中边缘点散乱、不连续或过多边缘点误判。基于道路边缘完整且连续的特点,针对传统检测效果并不理想的问题,提出了一种改进的Sobel算子,即5×5的8方向模板。从Sobel算子的基本原理出发,根据Pascal三角形理论推导出各方向的最优模板。研究表明,该算子不仅能较好地检测出更多方向上的边缘,而且能有效减少误判点,检测出的边缘线条更加平滑、完整,轮廓清晰且连续性好,尤其在弯曲道路检测中表现得更为突出,优于其他算子的检测效果。Extracting road edge from remote sensing image can extremely simplify the land survey workload for planning traffic networks. Because of the direction and size limitations of the template, the edge detection result derived by traditional algorithms shows high rate of false positive points and discontinuity, which is the reason why traditional edge detection algorithms can seldom achieve an ideal result in detecting continuous and integral road edge. To deal with this problem, this study proposes an improved Sobel operator which is based on an 8 directional 5 × 5 template. The optimal settings of each direction in the template are derived by the Pascal＇ s triangle theory. The improved operator not only achieves a better performance of edge detection in different directions but reduces the false positive point effectively as well. In comparison with other operators, the improved Sobel operator proposed by this study has better integrity and continuity in road edge extraction results, especially in road curve detection.

关 键 词：遥感图像 道路边缘 SOBEL算子 8方向 Pascal三角形 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]