An image registration technology for fault detection of EMU underbody

Abstract

In order to efficiently match and use the monitoring information obtained by the EMU fault detection robot, image registration technology can play a role. Facing the operation site of EMU fault detection robot, based on the original image data information of vehicle bottom fault collected by EMU fault detection robot, this paper carries out gamma transform and windowing preprocessing on the vehicle bottom image of EMU, puts forward the image registration algorithm of vehicle bottom fault detection of EMU based on Fourier transform, and gives the specific operation process of this algorithm. The experimental verification and comparative analysis show that compared with the image registration technology of EMU underbody fault detection based on Fourier transform, the image registration technology of EMU underbody fault detection based on conventional affine transform has obvious advantages, high effectiveness, strong robustness, accuracy and reliability, and can complete the high-precision, stable and reliable matching of images in a reasonable time, which meets the requirements of field practical application. The research of this paper provides theoretical support and technical basis for the intelligent identification of vehicle bottom fault defects of EMU, and has a good application prospect.