The Design And Implementation Of Train Ticket ManageMent System Based On Android

In recent years, traditional management of tickets is becoming more and more cumbersome,especially when it comes to identification of a sleeper ticket. The main reason lies in the absence of a proper hardware client and the relevant management software.However, with the development of embedded technol-ogy, the performance of mobile phones has also experienced an explosive en-hancement. Mobile phones are capable enough to store and process the infor-mation of tickets. In addition, the versatility of the mobile phones has greatly simplified the demands for image recognition.Compared with the existing client hardware equipped in the train, a client based on the cellphone is quite more convenient and functional, while the cost of massive applications is really cheap. This project has developed a cellphone APP of train ticket management system based on the Android platform. The system adopts the architecture of Client/Server. As a client, a Cellphone is the hardware environment which running the Android operating system and pro-gramed with Java/C++ programming language.The client focuses on recognizing the necessary information of a ticket im-age captured by some Android device, generating a record based on the rec-ognized result, and adding the record to the APP’s database. The server is a fully functional Web Server and it contacts the client using the normal HTTP protocol which following the REST style, take the management of basic data and transports them to the client in the format of JSON.When fetching the required information from the captured ticket, the sys-tem will recognize the QR code and the text on the ticket. To make a faster and more accurate operation, a pre-processing method will be applied to that image. The procedure can be described as below: First, block-based adaptive thresholding and morphological filtering can generate the mask of the ticket, and then we extract the outline of the mask by morphological gradient method. Then, we use the Hough Transform or the text flow check to get the ticket boundary lines describing by the equations, after that, the corners can be calcu-lated, and be used to rectify and crop the original image eliminate the inference caused by background and the perspective distortion.The recognition result will be sent to a trained Tesseract OCR engine or QR code processing element, as a result, the system will match the character recognition result.With the overall system testing, this system works quite well on recogniz-ing all the information of a ticket. There is no memory leak during running the system and the processing is really fast, while the average recognition ac-curance is in the tolerable range.