【作者】 彭静； 陈芒； 程治国； 刘允才；

【Author】 Peng Jing, Chen Mang, Cheng Zhi-guo, Liu Yun-cai(Institute of Image Processing and Pattern Recognition, Shanghai Jiaotong University, Shanghai 200030)

【机构】 上海交通大学图像处理与模式识别所；

【摘要】 本文提出了一种航空图像中道路线特征的提取算法,适用于复杂背景下城区中多等级道路的特征识别。在本文提出的多等级城市道路模型的基础上,算法利用道路线状结构的尺度空间特性对道路特征进行描述、分离与定位,最后,结合道路特征的图形和图像特征利用道路线状特征的方向尺度不变性,通过对道路特征图像局部方向能量的判断提取有效的线状道路特征,形成道路特征的目标中心链。与目前大量采用的基于边缘检测的道路提取算法,本算法复杂性大大降低,试验结果表明算法具有良好的抗噪性和普适性。更多还原

【Abstract】 This paper proposed an algorithm to extraction hierarchical road features from high-resolution aerial images of urban areas. Firstly, a hierarchical road model is presented. Then, road features are characterized, separated and located using the behaviors of line structure in scale-space. Finally, using the orientation invariance of road texture in scale-space, linear road features are extracted in both image and graph geometry using the orientation of feature’s local energy, and chains of object centers of road features are generated with good smoothness. Compared with most published edge-based algorithms, experiments show that this algorithm is more applicable in complex urban areas of various conditions.更多还原