Research On Technology Of Discerning RMB Through Blind Photograghing

RMB is the legal currency in China, plays a very important role in the daily consumption activities. Although the RMB feature is rich, the main character is also so easy for us normal people to capture, but for the loss of visual ability of the blind, they can only feel touch. Although blind markers are increased in the human- based design of RMB, these markers only in new currencies are sensitive. With circulating in the market, many currencies are repeatedly worn, which brings judging value by touching the braille marks great obstacles. Certainly, six banknotes have their own dimensions, but inevitably, technological restrictions generally result to the deviation of size. only by size to judge, it is difficult to ensure the accuracy of the judgement.For this intractable problem concerning the blind,this paper proposes general Smartphone Application Design, and the realize a complete set of algorithm for this application. This algorithm is divided into two parts: the value recognition and RMB serial number recognition. Value recognition, is the main part of the algorithm. The serial number of banknotes is composed by the crown code and numbers. Serial number to banknotes as identity card is to a person, it is unique. Automatic identification of serial number of banknotes for achieving the effective management and reducing the false banknotes identification is very pretty important.The image processing algorithm mainly includes the following contents:Wave filtering of Image-- Identification need locating the boundaries of RMB. Because the texture information appears too rich, there’s need to remove the noise(i.e. those texture and miscellaneous lines) to make image fuzzy easy to boundary judgment. At the same time, filtering is also in order to acquire the national emblem symbol and prepares for judging a banknote Orientation and value.Banknote edge detection-- after study of frequently-used algorithm on edge detection, the final selection is the Canny algorithm. The image filtered which is processed by the Canny algorithm, clearly appears a few miscellaneous lines except for notes boundaries.Note the Banknote boundry location-- The particularity of the scene in this design(banknote in the relaxed state), causes the banknote boundaries only show the characteristics of linear class. Therefore, noting the warping boundaries becomes difficult. There is need to explore a detection algorithm for twisting line. In this paper, I propose a detection algorithm on straight lines based on K mean clustering. The new method is efficient and accurate, basically meet the design requirements.Banknote face recognition-- The traditional recognition is by means on texture information of RMB. But the faces of banknote in a little rolled- up condition cannot be recognized effectively. In this paper, I capture emblem symbol locating on banknote positive, and extract the outer contour of the emblem, via shape matching to identify orientation. After verification, this method is proven to be more reliable.Banknote value recognition-- Six denomination banknote based on color difference, I categorize the six kinds of colors by neural network algorithm. After the test, the recognition rate is pretty high.Sequence number positioning, segmentation and recognition-- The emblem has been identified, serial number is on the opposite of the national emblem of which position will be easily fixed. Specific locking character position needs binarization of Wide Lines OSTU. I acquire single characters by the horizontal and vertical projection. This article achieves high accuracy through the method of template matching to recognize the character.