Banknote magnetic anti-counterfeiting feature authentication method based on magnetic spectrogram ViT classification model

Abstract

The invention discloses a banknote magnetic anti-counterfeiting feature authentication method based on a magnetic spectrogram ViT classification model. The banknote magnetic anti-counterfeiting feature authentication method comprises the following steps: collecting banknote magnetic signals; reading a magnetic spectrogram ViT classification model, wherein the model has a preset intensity upper limit, a preset intensity lower limit, a preset frequency upper limit, a preset signal length and preset speed-frequency ratio data of the magnetic characteristic signal; effective magnetic characteristic signals in the magnetic signals are detected and intercepted; according to the speed-frequency ratio of the intercepted effective magnetic characteristic signal, the signal length is adjusted to be a preset signal length; performing magnetic spectrum transformation on the magnetic characteristic signal according to the speed-frequency ratio of the magnetic characteristic signal, and drawing a magnetic spectrum by adopting a preset intensity upper limit, a preset intensity lower limit and a preset frequency upper limit of the magnetic characteristic signal; and inputting the drawn magnetic spectrogram into a magnetic spectrogram ViT classification model for identification, analyzing an identification result, and completing paper money magnetic anti-counterfeiting feature identification according to the identification result.

Description

Technical Field

[0001] The present invention relates to the technical field of banknote counterfeiting identification, and in particular to a banknote magnetic anti-counterfeiting feature identification method based on a magnetic spectrum ViT classification model. Background Art

[0002] With the continuous development of magnetic sensor technology, magnetic image sensors have been successfully developed and have begun to be used in the field of banknote counterfeiting. Through magnetic image sensors, the complex magnetic distribution on banknotes can be detected in conjunction with a uniform paper feed device and converted into a magnetic image, thereby providing the possibility for quantitative counterfeiting of banknote magnetic characteristics. Therefore, quantitative analysis and counterfeiting of the magnetic distribution on banknotes through magnetic images has important research value and has great application potential in the field of counterfeiting. Summary of the invention

[0003] In order to solve the above technical problems, the purpose of the present invention is to provide a banknote magnetic anti-counterfeiting feature authentication method based on the magnetic spectrum ViT classification model.

[0004] The purpose of the present invention is achieved through the following technical solutions:

[0005] A banknote magnetic anti-counterfeiting feature identification method based on a magnetic spectrum ViT classification model, comprising:

[0006] A collects magnetic signals of banknotes;

[0007] B. reading a magnetic spectrum ViT classification model, wherein the model has a predetermined intensity upper limit, a predetermined intensity lower limit, a predetermined frequency upper limit, a predetermined signal length, and a predetermined speed frequency ratio data of a magnetic characteristic signal; detecting and intercepting a valid magnetic characteristic signal in the magnetic signal;

[0008] C. adjusting the signal length to a predetermined signal length according to the speed-frequency ratio of the intercepted effective magnetic characteristic signal;

[0009] D. performing magnetic spectrum transformation on the magnetic characteristic signal according to the speed-frequency ratio of the magnetic characteristic signal, and drawing a magnetic spectrum diagram using a predetermined upper intensity limit, a predetermined lower intensity limit and a predetermined upper frequency limit of the magnetic characteristic signal;

[0010] E inputs the drawn magnetic spectrum into the magnetic spectrum ViT classification model for recognition, analyzes the recognition results, and completes the banknote magnetic anti-counterfeiting feature authentication based on the recognition results.

[0011] Compared with the prior art, one or more embodiments of the present invention may have the following advantages:

[0012] This method pre-processes the signal according to the unstable scanning speed and frequency in the actual scanning process, and adjusts the signal length to a predetermined length according to the speed-frequency ratio of the magnetic characteristic signal to solve the problem of the consistency of the horizontal axis of the magnetic spectrum. By adopting the speed-frequency ratio of each scan, the consistency of the vertical axis of the magnetic spectrum can be ensured, and the fixed upper and lower limits of the intensity can ensure the consistency of the intensity axis (color) of the magnetic spectrum. Figure 1 Consistency ensures recognition accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 This is a flow chart of the banknote magnetic anti-counterfeiting feature authentication method based on the magnetic spectrum ViT classification model. DETAILED DESCRIPTION

[0014] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be further described in detail below in conjunction with embodiments and drawings.

[0015] like Figure 1 As shown in FIG. 1 , the method flow of banknote magnetic anti-counterfeiting feature identification based on the magnetic spectrum ViT classification model includes:

[0016] A collects magnetic signals of banknotes;

[0017] Multi-channel magnetic sensors are used to collect magnetic signals at key locations along the short side of the banknote. Key locations on the banknote with magnetic anti-counterfeiting features include: security thread, two-color special-shaped horizontal numbers, and main background pattern.

[0018] B. taking a magnetic spectrum ViT classification model, the model having a predetermined intensity upper limit, a predetermined intensity lower limit, a predetermined frequency upper limit, a predetermined signal length and a predetermined speed frequency ratio data of a magnetic characteristic signal; detecting and intercepting a valid magnetic characteristic signal in the magnetic signal;

[0019] The effective magnetic feature signals in the magnetic signal can be detected by learning with Transformer or LSTM deep learning models, or by detecting based on signal strength.

[0020] For detection based on signal strength, the convolution operator 1×n dimensional I is used det =[1,1,1,....,1,]; let the magnetic signal be S mag The upper limit of strength is p signal_upper， Then judge that

[0021] Magnetic characteristic signal S mag_on

[0022] S mag_on ={x|x∈S mag *I det , x>0.05×p signal_upper}

[0023] C. adjusting the signal length to a predetermined signal length according to the speed-frequency ratio of the intercepted effective magnetic characteristic signal;

[0024] In this scan, the average scanning speed v signal , the magnetic sensor sampling frequency is f signal , the length of the banknote is L cash , that is, the velocity frequency ratio of the magnetic characteristic signal can be obtained as r signal , signal length l signal :

[0025] r signal =f signal / v signal

[0026] l signal =L cash ×f signal / v signal

[0027] For a predetermined signal length l signal_set , predetermined speed frequency ratio r signal_set .

[0028] Then the normalized predetermined signal length l can be obtained signal_set_norm for

[0029] l signal_set_norm = l signal_set × signal ÷r signal_set

[0030] Then the process of adjusting the signal length to the predetermined length is

[0031]

[0032] It can solve the problem of horizontal axis consistency in drawing magnetic spectrum.

[0033] D. performing magnetic spectrum transformation on the magnetic characteristic signal according to the speed-frequency ratio of the magnetic characteristic signal, and drawing a magnetic spectrum diagram using a predetermined upper intensity limit, a predetermined lower intensity limit, and a predetermined upper frequency limit of the magnetic characteristic signal;

[0034] The velocity frequency ratio of the magnetic characteristic signal is r signal , the upper limit of the intensity of the predetermined magnetic characteristic signal is p upper The lower limit of strength is p lower , the upper frequency limit is f upper Drawing a magnetic spectrum includes:

[0035] Draw the magnetic spectrum according to the upper limit of intensity, the lower limit of intensity and the upper limit of frequency: it means that the horizontal axis starting point of the magnetic spectrum is: 0, and the horizontal axis end point is: signal length l signal;

[0036] The starting point of the vertical axis of the magnetic spectrum is: 0, and the end point of the horizontal axis is: the upper limit of the frequency value f upper ;

[0037] The starting point of the intensity color axis of the magnetic spectrum is: the lower limit of the intensity of the magnetic sensor signal p down , the end point of the intensity color axis is: the upper limit of the intensity of the magnetic sensor signal p upper .

[0038] By adopting the frequency ratio of each scanning speed, the consistency of the vertical axis of the magnetic spectrum can be ensured, and the fixed upper and lower limits of intensity can ensure the consistency of the intensity axis of the magnetic spectrum (color). Figure 1 Consistency ensures recognition accuracy.

[0039] E inputs the drawn magnetic spectrum into the magnetic spectrum ViT classification model for recognition, analyzes the recognition results, and completes the banknote magnetic anti-counterfeiting feature authentication based on the recognition results.

[0040] The identification types include: suspicious, security line, two-color special-shaped horizontal number, main background pattern and other self-set magnetic features; the identification result is analyzed and determined according to the identification type. The analysis method of the identification result is shown in Table 1:

[0041] Table 1

[0042]

[0043] If the identification result is suspicious or the confidence level is lower than 90% or the identification magnetic feature results are inconsistent, the individual item is set as suspicious and the overall conclusion is suspicious. Only when all scanning positions meet the normal requirements, the overall conclusion is normal.

[0044] Although the embodiments disclosed in the present invention are as above, the above contents are only embodiments adopted for facilitating the understanding of the present invention and are not intended to limit the present invention. Any technician in the technical field to which the present invention belongs can make any modifications and changes in the form and details of the implementation without departing from the spirit and scope disclosed in the present invention, but the patent protection scope of the present invention shall still be subject to the scope defined in the attached claims.

Claims

1. A method for identifying banknotes' magnetic anti-counterfeiting features based on a magnetic spectrum ViT classification model, characterized in that: include: A collects magnetic signals of banknotes; B. reading a magnetic spectrum ViT classification model, wherein the model has a predetermined intensity upper limit, a predetermined intensity lower limit, a predetermined frequency upper limit, a predetermined signal length, and a predetermined speed frequency ratio data of a magnetic characteristic signal; detecting and intercepting a valid magnetic characteristic signal in the magnetic signal; C. adjusting the signal length to a predetermined signal length according to the speed-frequency ratio of the intercepted effective magnetic characteristic signal; D. performing magnetic spectrum transformation on the magnetic characteristic signal according to the speed-frequency ratio of the magnetic characteristic signal, and drawing a magnetic spectrum diagram using a predetermined upper intensity limit, a predetermined lower intensity limit, and a predetermined upper frequency limit of the magnetic characteristic signal; E inputs the drawn magnetic spectrum into the magnetic spectrum ViT classification model for recognition, analyzes the recognition results, and completes the banknote magnetic anti-counterfeiting feature authentication based on the recognition results.

2. The banknote magnetic anti-counterfeiting feature identification method based on the magnetic spectrum ViT classification model according to claim 1 is characterized in that: The A includes: using a multi-channel magnetic sensor to collect magnetic signals at specified key positions along the short side of the banknote; the key positions are the security line, two-color special-shaped horizontal numbers and the main background pattern.

3. The banknote magnetic anti-counterfeiting feature identification method based on the magnetic spectrum ViT classification model according to claim 1 is characterized in that: The effective magnetic characteristic signal in the magnetic signal in B is detected after learning through a Transformer or LSTM deep learning model.

4. The banknote magnetic anti-counterfeiting feature identification method based on the magnetic spectrum ViT classification model according to claim 1 is characterized in that: The effective magnetic characteristic signal in the magnetic signal in B is detected according to the signal strength, including: Using the convolution operator 1×n dimension I det =[1,1,1,....,1,]; let the magnetic signal be S mag The upper limit of strength is p signal_upper， Then the magnetic characteristic signal S mag_on The calculation formula is: S mag_on ={x|x∈S mag *I det ,x>0.05×p signal_upper }。 5. The banknote magnetic anti-counterfeiting feature identification method based on the magnetic spectrum ViT classification model according to claim 1 is characterized in that: The C includes: During scanning, let the average scanning speed be v signal , the magnetic sensor sampling frequency is f signal 、The length of the banknote is L cash , then the velocity frequency ratio r of the magnetic characteristic signal signal With the signal length l signal The calculation formula is: r signal =f signal / v signal l signal =L cash ×f signal / v signal The predetermined signal length is l signal_set , the predetermined speed frequency ratio is r signal_set , by pre-determining the signal length to be l signal_set , the predetermined speed frequency ratio is r signal_set and the velocity sampling ratio r of the magnetic characteristic signal signal Get the normalized predetermined signal length l signal_set_norm : l signal_set_norm ＝l signal_set ×r signal ÷r signal_set Then the signal length is adjusted to the predetermined length:

6. The banknote magnetic anti-counterfeiting feature identification method based on the magnetic spectrum ViT classification model according to claim 1 is characterized in that: The velocity frequency ratio of the magnetic characteristic signal in D is r signal , the upper limit of the predetermined intensity of the magnetic characteristic signal is p upper , the predetermined lower limit of intensity is p lower , the predetermined upper frequency limit is f upper ; The speed frequency ratio r of the magnetic sensor signal signal , the upper limit of the predetermined intensity of the magnetic characteristic signal p upper , the predetermined lower strength limit p lower With the predetermined upper frequency limit f upper Drawing a magnetic spectrum includes: Draw the magnetic spectrum according to the upper limit of intensity, the lower limit of intensity and the upper limit of frequency: it means that the horizontal axis starting point of the magnetic spectrum is: 0, and the horizontal axis end point is: signal length l signal ; The starting point of the vertical axis of the magnetic spectrum is: 0, and the end point of the horizontal axis is: the upper limit of the frequency value f upper ; The starting point of the intensity color axis of the magnetic spectrum is: the lower limit of the intensity of the magnetic sensor signal p down , the end point of the intensity color axis is: the upper limit of the intensity of the magnetic sensor signal p upper .

7. The banknote magnetic anti-counterfeiting feature identification method based on the magnetic spectrum ViT classification model according to claim 1 is characterized in that: The identification types include: suspicious, security line, two-color special-shaped horizontal number and main background pattern; the identification result is analyzed and determined according to the identification type.

8. The banknote magnetic anti-counterfeiting feature identification method based on the magnetic spectrum ViT classification model according to claim 7 is characterized in that: If the recognition result in the recognition type is suspicious or any one of the security line, two-color special-shaped horizontal number, main background pattern and other self-set magnetic features has a confidence level lower than 90%, the recognition result is set as suspicious.

Images