Banknotes Number Recognition System Design And Realization

The main purpose of this dissertation is to study the automatic recognition technology of bill serial number based on image processing, The relative theories and algorithms are analyzed and researched systematically. Then the effective image preprocessing and recognition algorithms are proposed through testing many kinds of them. Taking these algorithms as kernel, an automatic recognition system for bill serial number is developed. At first, image is gathered through CCD camera, then the serial number is outputted through process of image processing, image segmentation, and image recognition. Experiment result reveals that this system can deals with many problems happened in the practical environment, also it satisfies demands of high recognition rate and high speed.In the process of system recognizing, because the serial number area is only small part as to the whole image, so in order to decrease the time cost of recognition, number area orientation must be done firstly. In this paper, a new way of serial number area orientation using gray image is proposed. This method can extract serial number area quickly.Serial number deformation is a common situation in image sampling process. It makes number recognition more difficult than ever. By analyzing features of those deformed serial number, algorithm to serial number correction is presented, which have been proved effective in practice.Through analyzing template matching and the recognition method based on characteristics, it reveals that recognition rate of any one of these methods is not very high. So a new method which combined these two ways is proposed in this paper. At first, it classifies the sample image into different subgroups using some simple characteristics, and then recognizes them with template matching in the smaller subgroup. This combined multiple recognition method not only has high recognition rate but also has better whole performance than any other single method.In a word, all these mended technical measures are proposed in this paper can greatly improve the system's adaptability in different environments, and it has been verified by lots of experiments. This technology must have a good prospect in the future.