Train Wagon Number Recognition And Application Based On Image Processing

As a result of widely applications of computer in field of engineering, the technique of classical digital image processing is not satisfied. Based on an overview of the latest development on transportation management information system on international, a representing method of the vehicle number recognition based on digital image processing technology is proposed in this paper. This paper describes an effective method of segmenting and recognizing the characters on freight train. This method has been used in practical system and obtained satisfying results.At first, the methodology of the train wagon number locating is presented in this paper based on color space and grayscale morphology. This methodology need not determine the threshold which is the most unstable factor in image binarization for the technology of the train wagon number locating. Next, the method makes real-time segmentation of the train wagon number image in background of vehicle image, and then extract train plate detail character feature. Finally, an approach is proposed to get the train wagon number by image matching. On basis modeling after in projection, be resolve it's lower problem of reliability, be tied in wedlock some characteristic of wagon number character, have carried out certain improvement on the basis projection law, have improved it's success rate carving up aspect in character thereby. In this paper, the BP neural network has been applied in train wagon number recognition, the results obtained by simulation show that the method is an efficient way for number recognition.All routines in the paper were implemented in Delphi, and were debugged on computer. Finally, the examples are given step by step to demonstrate the feasibility of the approach suggested.