Research And Realization Of Driver Machine Interface Recognition Technology For Train Control System

In the interoperability testing platform of on-board equipment for CTCS-3train control system in high-speed rail, Driver Machine Interface is a window not only for testers to understand the running status of platform, but also for data input and instructions issued. Traditional information collection method takes up human resources and low-level efficiency. To operate DMI by establishing machine vision system can significantly improve the automation level of the interoperability test platform. Therefore, to research and realize methods for DMI identification is of important practical significance.DMI identification has the feature of various kinds of information and unstable illumination. The key of DMI identification technology of consists of three parts:DMI information location, DMI information segmentation, DMI information identification. After analyzing and comparing various algorithms for image location comprehensively, Dynamic edge extraction is proposed to locate information. This method combines the advantages of edge detection and dynamic thresholding segmentation, being able to locate information automatically in inhomogeneous illumination and complex background.Automatic threshold method is applied to extract icons regions and dynamic threshold segmentation method is used to extract character regions, overcoming the effects of changing illumination. As Chinese characters are of complex structure, character segmentation algorithm based on the maximum broadband regression is put forward to solve the disconnect problem of Chinese characters better, getting the full character.Finally, Methods based on template matching and region features are applied to classify icons according to the feature of icon after analyzing various kinds of image recognition technologies, improving the classification rate significantly. Packet-based neural network approach is applied to train and recognize character, reducing network complexity and improving character recognition rate.In this paper, Visual Studio2008is adopted as software development platform, programming languages C#and visual tools HALCON are employed to integrate algorithms to build vision system is combined with the manipulator platform to operate DMI automatically, being reliable and stable in DMI operation in laboratory.