JP3408075B2 - Banknote identification method - Google Patents

Description

Description: BACKGROUND OF THE INVENTION [0001] 1. Field of the Invention [0002] The present invention relates to a method for identifying banknotes, and more particularly, to suppressing the influence of various types of dirt on the identified banknotes to the identification accuracy, and achieving high accuracy and high speed. It relates to an identification method that can be determined. 2. Description of the Related Art A prior art of the present invention is disclosed in
No. 215293. In this publication, a bill is divided into a plurality of zones, detection data for each zone is compared with reference data predetermined for each zone, and the bill is identified based on a comparison result in each zone. In the bill discriminating method, a plurality of bills are set in accordance with the front and back, the orientation, and the displacement at the time of discrimination, and the data of each zone are totaled for one bill, and a standard value is determined by a ratio value to the total value. It is stored as pattern data, a total value of the detection data is obtained, and a ratio value to the total value is calculated as detection pattern data, and whether or not the detection pattern data is within an allowable value range of the reference pattern data is determined. Is determined, and the absolute value of the difference between the reference pattern data and the detected pattern data is calculated for each of the zones, and the total is calculated. The total value banknote identification method and performing to the bill validator determines whether less than the allowable value is disclosed. [0003] In the above-mentioned prior art, however, discrepancies in identification due to dirt, distortion, and other reasons are tolerated. There is a problem that the identification accuracy is reduced. SUMMARY OF THE INVENTION The present invention has been made to solve the problems of the conventional method, and is capable of accurately and quickly identifying paper sheets without being affected by dirt, distortion and the like. The aim is to provide a way to do so. According to the method of the present invention, image data of a plurality of genuine reference banknotes is obtained, and the obtained image data is subjected to cluster analysis and classified into several data patterns to obtain a plurality of data. of creating the reference image data, the difference was calculated respectively between the input image data with the plurality of reference image data of the identified bill input, extracts dirt, the first irregular components such as distortion, said first The plurality of reference images based on one irregular component
A second irregular component corresponding to each reference image data of the data is predicted, and a sum of squares of differences between the second irregular component and the first irregular component is calculated. This is a method of selecting reference image data having a minimum value, and determining a genuine bill or a fake bill by comparing the value of the sum of squared differences corresponding to the selected reference image data with a predetermined threshold. . An embodiment of the method for identifying bills according to the present invention will be described below in detail with reference to the accompanying drawings. The method of the present invention is broadly divided into pre-processing and bill insertion processing. FIGS. 1 and 2 are flowcharts showing basic algorithms of the pre-processing and bill insertion processing, respectively. First, in the preprocessing, when a program is started by inserting a plurality of genuine bills prepared in advance (step S1), multi-value density data by an optical sensor of the bills is fetched (step S2). The plotted gray-scale data is shown in FIG. 3 by plotting the gradation value on the vertical axis and the point on the horizontal axis. Cluster analysis is performed using the plurality of waveform diagrams, and waveform diagrams that are very similar to each other are classified as one unit (cluster) (Step S).
3). Then, an average waveform diagram is created for each of the obtained clusters, and a plurality of reference image data (waveform diagrams) are created (step S4). The method of cluster analysis is shown in the flowchart of FIG. When the program is started in step S31, first, in step S32, the number of banknotes to be cluster-sorted as initialization is set to N. Next, the distance (corresponding to the similarity) between the bills in the image space is calculated (step S33). In step S34, the number K of clusters is calculated as K = N
Set to. Then, using the distance data between the banknotes previously obtained in step S33, a cluster set having the closest distance is selected (step S35). When a set of clusters is selected, the set of target clusters is integrated into one cluster (step S
36). Therefore, in step S37, K is changed to K-1. In step S38, it is determined whether or not the result K of step S37 has become 1. If K is 2 or more, the nearest cluster set including the newly integrated cluster is searched and selected again (step S39). The similar cluster integration process described above is repeated until K becomes 1 in step S38. When K = 1, the process ends, and a tree diagram is created in step S40 (see FIG. 5). Based on the tree diagram thus obtained, an appropriate number of a plurality of reference waveforms are created, and the processing shifts to a bill insertion process. For example, in the example of FIG. 5, the reference waveform obtained when the distance A is a threshold is one, the reference waveform is B when the distance is B, and the reference waveform is C.
Then there are three. FIG. 3 shows an example in which the threshold value is C and three reference waveforms are created. In the processing at the time of bill insertion, when the program is started in step S11, image data of a bill to be identified is first taken in (step S12). This image data is formed in a waveform diagram as in the preprocessing. Next, after carrying out a conveyance deviation correcting process, a level adjusting process, etc., in step S13, a difference from the reference waveform data created in the pre-process is obtained, and irregularities such as dirt and distortion on the bill are obtained. A component (first irregular component) extraction process is performed. Using the obtained irregular component, AR
(Auto regression) Irregular component (second irregularity)
The prediction process (step S14) for the rule component is performed. In step S15, a sum of squared differences is calculated using the extracted irregular component and the irregular component corresponding to each predicted reference waveform, and a prediction error is calculated. Then, the reference image waveform data having the smallest value of the obtained sum of squared differences is selected (step S16), and the value of the sum of squared differences between the selected reference image waveform data and the input image waveform data is determined in advance. By the comparison with the threshold value thus determined, the authenticity is determined (step S17). According to the present invention, when the authenticity is determined by comparing the captured image data of the banknote to be identified with the reference image data as described above, the reference image data is classified according to the similarity. By comparing the representative data with the image data of the banknote to be identified by dividing the set into several sets, the authenticity can be determined at high speed without significantly lowering the accuracy.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a flowchart showing pre-processing in an embodiment of a bill discriminating method of the present invention. FIG. 2 is a flowchart showing a bill insertion process in a bill discriminating method according to an embodiment of the present invention. FIG. 3 is a conceptual diagram of one embodiment of a bill discriminating method of the present invention. FIG. 4 is a flowchart illustrating a reference image data generation method using a cluster analysis method. FIG. 5 is a diagram showing an example of a dendrogram.

────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-5-182063 (JP, A) JP-A-2-108187 (JP, A) JP-A-3-210692 (JP, A) JP-A-6-106 44437 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G07D ^7/ 00-7/20

Claims (1)

(57) [Claims 1] A plurality of reference images are obtained by obtaining image data of a plurality of genuine reference banknotes, classifying the obtained image data into clusters and classifying them into several data patterns. create a data difference and the respective calculation of the inputted plurality of reference image data and the input image data of the identified bill was, dirt, and extracting a first random components such as distortion, said first non Each reference image data of the plurality of reference image data is
Predicting a corresponding second irregular component, calculating a sum of squared differences between the second irregular component and the first irregular component,
The calculated difference sum of squares selects the reference image data having the minimum value, and the true and false bills are obtained by comparing the value of the difference square sum corresponding to the selected reference image data with a predetermined threshold. A bill discriminating method characterized by performing the following judgment.