Train positioning method based on catenary strut number image recognition

The invention discloses a train positioning method based on catenary strut number image recognition. The method comprises the following steps: S1, acquiring a high-definition video of catenary equipment in real time by using a catenary suspension state detection device; S2, key frames in the inspection video are extracted, and a high-definition image containing all complete overhead line system supporting columns along the way is obtained; S3, preprocessing the high-definition image, and determining an overhead line system strut area; S4, adopting an HOG feature description method and SVM classification to locate a rectangular area of the pole number of the overhead line system strut in the image, and carrying out binarization processing on the image after gray scale transformation; S5, onthe basis of a horizontal projection method, cutting the binarized rectangular area of the strut number t, and segmenting characters; S6, normalizing the character material, and zooming the originalimage; S7, directly carryi