对比Harris算子和Susan算子的建筑物边界规则化方法

A building boundary regularization method by contrasting Harris operator and Susan operator

建筑物顶部边界的精确提取在建立数字城市等方面发挥着非常重要的作用。本文针对从遥感影像中粗提取的建筑物边界不规则的问题,结合Harris算子和Susan算子,提出了一种对粗提取后的建筑物边界进行规则化拟合的处理方法。首先对粗提取的建筑物顶部边界进行预处理以剔除噪声影响;然后分别使用Harris算法和Susan算法对预处理后的建筑物边界进行角点检测提取;最后对检测提取的边界角点进行点号排序和规则化拟合连接得到规则的建筑物边界。试验结果表明,通过该方法处理后的建筑物边界平滑且与其实际边界基本一致。

The precise extraction of the top boundary of a building plays a very important role in establishing a digital city and so on. In order to solve the problem of irregular boundary of buildings extracted from remote sensing images,combined with Harris operator and Susan operator,a method for regular fitting of rough extracted building boundaries is proposed. Firstly,pre-process the rough-extracted top boundary of the building to eliminate noise effects. Then the Harris algorithm and Susan algorithm are used to detect the corners of the pre-processed building boundary respectively. Finally,the corner points of the extracted boundary corner points are sorted and regular fitting is connected to obtain the regular building boundary. The experimental results show that the boundary of the building treated by this method is smooth and basically consistent with the actual boundary of the building.