Research On The Method And Tool For Automatic Recognition Of DMI Information Of CTCS-3Onboard Equipment

CTCS-3train control system is an advanced train control system which is based on radio communication. It has been widely applied in high speed railways with the speed above300km/h. Before a railway line is put into operational service, a series of test shall be carried out for the train control system, which include laboratory testing, field testing, integrated testing and commissioning and interoperability testing.. The black-box testing is a common method for the lab testing of the on-board equipment. The testing facility usually consists of the equipment under test, the test script generating module, the test process controlling module, the test results recording and analyzing module and so on. Before the testing, some test sequences shall be prepared according to the test cases and test script shall be generated according to the track data of the test line. During the testing process, the state of each interface (e.g. DMI and wireless interface) shall be observed and recorded. After the testing, the test data will be analyzed manually. The whole process is time consuming and with heavy workload. To realize the automatic testing of on-board equipment, several key techniques shall be addressed, which include the automatic generation of test scripts, the automatic execution of the test process and the automatic analysis and validation of the test results.This paper mainly studies the method of automatic recognition of DMI information, which involves the positioning of the DMI display area and the segmentation of DMI information. The main works of this paper include:1. The CTCS-3on-board device DMI interface is analyzed in detail, and the features of the DMI interface are summarized according to the practical needs of the testing of the on-board equipment, and the solution for the automatic identification of the DMI information is proposed.2. The locating technique of DMI display area is studied. The method for DMI border tracking based on Hough transform and the method for DMI image tilt correction based on perspective transform are proposed, with the aim to solving the locating problem caused by different viewing angle, viewing direction, stochastic vision and camera shaking, so as to achieve the automatic locating of DMI information.3. Different methods for image segmentation, feature extraction and pattern recognition are adopted for different types of images of the DMI interface information. Decision tree is used for classifying icons and SVM for numbers, letters and Chinese characters. 4. An on-board DMI interface information recognizing tool is developed by using C++developing language and OpenCV library under Visual Studio2012, which can recognize the DMI interface information and manage the recognition results.