基于图像处理的输电线路线下树木检测算法研究

Study on Transmission Line Tree Detection Algorithm Based on Image Processing

针对电力系统日益突出的输电线路沿线树线矛盾问题,提出一种基于图像处理技术的输电线路线下树木自动检测算法,以输电线路图像/视频监测装置采集的输电线路现场图像为对象,当在输电导线下检测到树木时发送报警信息。首先,对原始图像进行图像灰度化、图像平滑、图像增强等预处理;其次,通过Hough变换检测到导线位置,通过导线定位感兴趣区域,鉴于树木所具有的特殊纹理,在该区域内通过纹理分析进行树木检测;最后,通过纹理图像分割及后处理,结合Sobel算子进行树木区域边缘检测,标识出树木边界轮廓并发送报警信息。通过实例分析证明该方法易于实施,直观有效,可实时检测出接近输电导线的树木。

Aiming at increasingly prominent contradiction problem between transmission lines and trees along the line in power system, this paper proposes a new algorithm for automatic tree detection based on image processing technology which takes field image of transmission lines collected by transmission line image and video monitoring devices as object and sends warning messages when detecting trees close to transmission lines. It firstly preprocesses the original images for image gra- ying, image smoothing and image intensification. Secondly, by Hough transform, it is able to detect position of the conduc- tor and locate the region of interest according to conductor positioning. In consideration of special texture of trees, it is able to process tree detection in this region based on texture analysis. At last, by texture image segmentation and postprocessing, it is able to process tree boundary contours detection combining with Soble operators and identify tree boundary contours and send out warning messages. On the basis of practical analysis, it verifies this method is liable to be implemented which is in-tuitive and effective for real-time detection of trees close to transmission lines.