复杂背景下电力线自动提取算法

为从具有复杂背景的无人机航拍图像中准确完整地提取电力线,提出了一种结合改进Ratio算子和改进Hough变换的电力线提取新方法,命名为基于比率算子聚类和霍夫变换编组的方法(ratio-basedclusteringand hough-basedgrouping,RCHG)。首先采用Ratio算子进行电力线边缘检测,在综合考虑电力线连续性和噪声抑制能力的理念下,给出了线特征检测阈值的参考范围。然后,对边缘图像进行四连通聚类分析,消除大部分背景噪声。最后,采用Hough变换提取电力线,进一步设计直线段聚类算法,对提取结果进行直线编组并进行最小二乘拟合处理,以解决电力线断裂和重叠的问题。实验结果表明,相比传统Ratio算子结合Hough变换的方法及Line Segment Detector(LSD)算法,所提方法具有更好的抗噪性能和更高的电力线提取精度,能从具有复杂背景的无人机航拍图像中准确完整地提取出电力线,有较高的工程应用价值。

Automatic Extraction Algorithm of Power Line in Complex Background

In order to extract power lines from unmanned aerial vehicle(UAV) aerial images with complex background accurately and completely, we proposed a new power line extraction method combined an improved Ratio operator with improved Hough transform, named ratio-based clustering and hough-based grouping(RCHG). Firstly, the Ratio operator was used to detect the edge of the power line. Taking the compromise between the continuity of the power line and the ability of noise suppression into consideration, we put forward a reference range of the line feature detection threshold. Then, the edge image was analyzed by 4-connected clustering to eliminate most of background noise. Finally, the Hough transform was used to extract the power line. In order to solve the problem of power line rupture and overlap, we designed the straight line clustering algorithm to organize the lines into group and fit them by the least square fit method. As experimental results shown, compared with the traditional Ratio combined Hough transform method and the Line Segment Detector(LSD) method, the proposed method has better anti-noise performance and higher power line extraction accuracy, which can extract power lines accurately and completely from the UAV aerial images in complex background, and it has high engineering application value.