Steel rail profile matching algorithm based on computer vision

The invention discloses a steel rail profile matching algorithm based on computer vision, which realizes computer simulation of an artificial matching result through convolutional neural network learning or a Transform algorithm based on multi-head attention so as to carry out profile matching. Comp...

Full description

Saved in:
Bibliographic Details
Main Authors WANG ZHONGYUAN, YANG LIANG, YANG DONGSHENG, SI DAOLIN, ZHAO ZHENHUA, WANG SHUGUO, GE JING, WANG KUNQI, WANG PU, WANG MENG, GAO YUAN
Format Patent
LanguageChinese
English
Published 12.08.2022
Subjects
CALCULATING
COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
COMPUTING
COUNTING
HANDLING RECORD CARRIERS
PHYSICS
PRESENTATION OF DATA
RECOGNITION OF DATA
RECORD CARRIERS
Online AccessGet full text

Cover

More Information
Summary:The invention discloses a steel rail profile matching algorithm based on computer vision, which realizes computer simulation of an artificial matching result through convolutional neural network learning or a Transform algorithm based on multi-head attention so as to carry out profile matching. Compared with an existing matching method, the matching algorithm has higher advantages in the aspects of matching accuracy, matching precision and calculation efficiency; more importantly, artificial profile matching experience can be introduced into the algorithm, the problem of complex and changeable railway field conditions is solved, and the algorithm is a fundamental way for solving the profile matching problem; the method is universal for rail pieces in a turnout zone besides a standard profile; the problem of manual matching efficiency is solved, different profile matching experiences and criteria caused by human factors are avoided, and a profile matching standard is formed through the constructed steel rail p
Bibliography:Application Number: CN202210091144