轨道供电变电站分合板图像识别方法研究

Research on image recognition method of rail power supply station

轨道供电变电站分合板图像识别方法基于传统图像算法对分合板状态进行实时监测。文章通过变电站固定位置的球机拍摄变电站内的分合板图像,并在图像中设定一个矩形的分合识别区域,对分合板在分合识别区域中的像素面积要求大于50%;通过HSV颜色识别方法找出分合板位置;通过遍历像素的方式寻找分合板端点;通过仿射变换提取和矫正分合板图像;通过几何方法或霍夫直线变换方法找到分界线位置,将分合板切割为“分”“合”两部分;通过HSV颜色识别分别检测“合”部分与“分”部分图像中的黄色箭头,并转为二值图;通过计算两二值图中白色像素点个数,最终判断分合状态。

The image recognition method of rail power supply substation is based on traditional image algorithm to monitor the status of the split plate in real time. In this paper, we use a ball machine at a fixed position in the substation to capture the image of the split and composite board in the substation, and set a rectangular split and composite recognition area in the image. The pixel area of the split and composite board in the split and composite recognition area is required to be greater than 50%. It finds out the position of the split plate by HSV color identification method. The endpoint of the splitter board is found by traversing pixels, extracting and correcting the image of the parting board by affine transformation. The location of the dividing line can be found by geometric method or Hough line transformation method, and the dividing plate can be cut into two parts: “opening” and “closing”. It detects the yellow arrows in the images of the “closing” part and the “opening” part respectively through HSV color recognition, and turns them into binary images. By calculating the number of white pixels in the binary image, the final judgment of the opening and closing state is made.