The invention discloses a double-
block type sleeper laying fine adjustment
system and method based on
machine vision, and relates to the field of
rail traffic engineering.The method comprises the steps that firstly, embedded
metal measuring nails are arranged along a rail to serve as physical reference marks, and a stable three-dimensional coordinate
system is constructed through back measurement, least square optimization and dynamic remeasurement; then,
machine vision is used for collecting sleeper height, transverse center line position and
pitch angle data in real time, and the corresponding relation between adjusting operation and parameters is synchronously recorded; kalman filtering
noise reduction is carried out on the collected data, a normalized deviation cumulant is calculated based on a sliding window, and an error accumulation trend is identified through a continuous increasing situation; and finally, extracting high-frequency operation in an error accumulation period, calculating a
unit operation quantity deviation change rate, and executing a differential adjustment strategy, including
independent parameter adjustment, reverse compensation operation or pause operation, according to the
unit operation quantity deviation change rate.