Abstract: To implement the standardization of data format for heterogeneous station signal plan drawings, this paper proposed an image identification method based on OpenCV by analyzing the morphological characteristics and logical relationships of each module in the station signal plan (referred to as the plan) to identify the main station signal equipment in the plan. The paper converted CAD floor plans into PNG format images of the same resolution and size, and used grayscale and edge detection algorithms for image preprocessing to obtain the contour of the station, analyzed the structure of the floor plan and use Hough circle detection and line detection to extract features of the floor plan elements, and based on the identified circles and straight line segments, combined with the attribute characteristics of each station yard signal equipment and the logical relationship of the physical connections between attribute characteristics to design identification methods separately. The experimental results show that the OpenCV based image identification method can accurately and completely identify various types of station yard signal equipment, and obtain station yard data related to each equipment.