基于梯度对称性的无人机影像电力线自动提取方法

针对航空影像电力线提取过程中,背景线特征(电力杆塔、道路和植被等)干扰问题和电力线提取准确性问题,提出一种基于梯度对称性的电力线特征判别方法。采用Line Segment Detector(LSD)算法,对航空影像进行线特征提取;基于电力线附近梯度的对称性特征,实现从大量线段中准确筛选电力线段;采用缓冲区序列化线段连接算法,实现了电力线的自动连接。通过实验证明,提出的方法能够排除背景干扰并准确、完整地提取整条电力线。

Automatic Extraction of Power Lines Based on Gradient Symmetry from UAV Image

To get rid of the noise of background line features(such as the edges of electricity towers, roads and plants) and improve the extraction accuracy of power lines, this paper presents a gradient symmetry based power line extraction method. Firstly, power line segment candidates are extracted from aerial images by Line Segment Detector(LSD). Then the true power line segments are selected from the large number of power line segment candidates based on gradient symmetry. At last, the selected power line segments are connected into complete power lines by a buffered line segements sorting algorithm. Experiments showed that the proposed method can extract the power lines entirely and accurately.