EMU Electrical Cabinet Wiring Recognition System

At present, the EMU connection troubleshooter still rely on manual inspection. Manual inspection investigation work efficiency is very low, and the result have greater impact by human factors. To shorten the connection troubleshooter test cycle, and improve the investigation accuracy, it is necessary to develop automatic recognition system. In recent years, as the image processing is widely used in all aspects of our lives, such as fingerprint identification, traffic regulation, OCR and other application aspects is also mature.More and more areas began to use computer vision to detect by image processing. In domestic and foreign,the development based on image processing technology to solve practical problems has entered a mature stage.The system judges its electrical module connection by identifying EMU the cabinet connection number.Since the connection status is very complicate,the study of this paper is only about the situation that module electrical cables are intact, the system can recognize cable number by determining the cord location whether proper.Determining the appropriate cord location is the most important part of this system.Because each cable has a unique number,the result of connection can be judged by the unique cable.Acquiring target images through a hand held device,this images are transmitted to a computer for image processing. Due to the complicated wiring in the cabinet, and label area is relatively small, so,positioning accurately the connection area is the most difficulty part.This system takes the minimum bounding rectangle localization algorithm to achieve the objectives targeting,and stores the inclination angle during positioning,finally output the rotation correction destination rectangle.After identifying number by BP neural network,character are identified and the last result are displayed.Comparison correct identify results to the stored number, recognizing and judging the EMU distribution cabinet connection state based on opencv.