Magnetic security thread identification method, paper resource identification method and device

Abstract

The application relates to a magnetic security thread identification method and device and a paper resource identification method and device, which can be used in the field of financial technology or other related fields. The magnetic security thread identification method comprises the following steps: acquiring a waveform magnetic signal corresponding to a to-be-identified magnetic security thread, and acquiring a magnetic signal change speed and a magnetic signal direction corresponding to a plurality of preset time points of the waveform magnetic signal; obtaining a path feature of the waveform magnetic signal based on the waveform magnetic signal; inputting the path feature into a first branch of a pre-trained magnetic security thread identification model, acquiring an identification feature of the waveform magnetic signal through the first branch, inputting the magnetic signal change speed and the magnetic signal direction into a second branch of the magnetic security thread identification model, and acquiring a calibration feature of the waveform magnetic signal through the second branch; and fusing the identification feature and the calibration feature, and obtaining an identification result corresponding to the to-be-identified magnetic security thread according to the fused feature. The method can accurately identify the magnetic security thread.

Description

Technical Field

[0001] This application relates to the field of financial technology, and in particular to a method and apparatus for identifying magnetic security threads and paper resources. Background Technology

[0002] With the development of the financial technology field, magnetic security thread identification technology for financial paper resources has emerged. This technology identifies magnetic security threads by acquiring the peak of the magnetic wave signal generated by the magnetic security thread.

[0003] However, the above technical solution only considers the local features of the security thread waveform, namely the peak, and ignores some important features, which makes the accuracy of magnetic security thread identification low, and consequently the accuracy of financial paper resource identification low. Summary of the Invention

[0004] Therefore, it is necessary to provide a magnetic security thread identification method, paper resource identification method, device, paper resource identification equipment, computer-readable storage medium, and computer program product that can accurately identify magnetic security threads in response to the above-mentioned technical problems.

[0005] Firstly, this application provides a method for identifying magnetic security threads. The method includes:

[0006] Acquire the waveform magnetic signal corresponding to the magnetic security line to be identified, and acquire the magnetic signal change rate and magnetic signal direction corresponding to multiple preset time points of the waveform magnetic signal;

[0007] Based on the waveform magnetic signal, the path characteristics of the waveform magnetic signal are obtained;

[0008] The path features are input into the first branch of the pre-trained magnetic security line recognition model to obtain the recognition features of the waveform magnetic signal. The magnetic signal change rate and magnetic signal direction are input into the second branch of the magnetic security line recognition model to obtain the calibration features of the waveform magnetic signal. The recognition features are used to identify the magnetic security line, and the calibration features are used to calibrate the recognition features.

[0009] The identification features are fused with the calibration features, and the identification result corresponding to the magnetic security line to be identified is obtained based on the fused features.

[0010] In one embodiment, the step of inputting the path features into a first branch of a pre-trained magnetic security line recognition model and obtaining the recognition features of the waveform magnetic signal through the first branch includes: inputting the path features into the first branch and extracting features from the magnetic signal attributes to obtain the extracted path features corresponding to the path features; obtaining the context information of the extracted path features and obtaining the recognition features of the waveform magnetic signal based on the context information.

[0011] In one embodiment, obtaining the context information of the extracted path features includes: obtaining sub-features corresponding to multiple time points of the extracted path features, and obtaining the current sub-feature corresponding to the current time point; obtaining the weights corresponding to each of the remaining sub-features other than the current sub-feature, and obtaining the first branch state corresponding to the current time point based on the remaining sub-features and their respective weights; and obtaining the context information of the extracted path features based on the first branch states corresponding to each of the time points.

[0012] In one embodiment, obtaining the path characteristics of the waveform magnetic signal based on the waveform magnetic signal includes: dividing the waveform magnetic signal into multiple signal segments with the same time using a pre-obtained preset time window; obtaining the sub-path characteristics corresponding to each signal segment and the order of each signal segment; and obtaining the path characteristics of the waveform magnetic signal based on each sub-path characteristic and the order of each signal segment.

[0013] In one embodiment, obtaining the sub-path features corresponding to each signal segment includes:

[0014] Obtain multiple time points corresponding to the current signal segment, and multiple magnetic signal values ​​corresponding to each of the multiple time points; based on the multiple time points and the multiple magnetic signal values, obtain multiple iterative integration elements corresponding to the current signal segment; based on the multiple iterative integration elements, obtain the sub-path features corresponding to the current signal segment.

[0015] In one embodiment, obtaining the magnetic signal change rate and magnetic signal direction corresponding to multiple preset time points of the waveform magnetic signal includes: obtaining the current magnetic signal value of the current waveform magnetic signal at the current preset time point, and obtaining the next magnetic signal value of the current waveform magnetic signal at the next time point of the current preset time point; using the difference between the next time point and the current time point, and the difference between the next magnetic signal value and the current magnetic signal value, obtaining the current magnetic signal change rate of the current waveform magnetic signal at the current time point; and obtaining the current magnetic signal direction of the current waveform magnetic signal at the current time point based on the current magnetic signal change rate.

[0016] In one embodiment, obtaining the identification result corresponding to the magnetic security line to be identified based on the fusion features includes: obtaining the output result corresponding to the magnetic security line based on the fusion features; comparing the output result with a preset benchmark result to obtain the identification result corresponding to the magnetic security line to be identified.

[0017] Secondly, this application provides a method for identifying paper resources. The method includes:

[0018] Obtain the identification result corresponding to the magnetic security thread carried on the paper resource to be identified; the identification result is obtained according to the magnetic security thread identification method described in the first aspect;

[0019] The identification results are used to identify the paper resources to be identified.

[0020] Thirdly, this application also provides a magnetic security thread identification device. The device includes:

[0021] The signal attribute acquisition module is used to acquire the waveform magnetic signal corresponding to the magnetic security line to be identified, and to acquire the magnetic signal change rate and magnetic signal direction corresponding to multiple preset time points of the waveform magnetic signal.

[0022] The path feature acquisition module is used to obtain the path features of the waveform magnetic signal based on the waveform magnetic signal;

[0023] The identification feature acquisition module is used to input the path features into a first branch of a pre-trained magnetic security line identification model, and to acquire the identification features of the waveform magnetic signal through the first branch; and to input the magnetic signal change rate and the magnetic signal direction into a second branch of the magnetic security line identification model, and to acquire the calibration features of the waveform magnetic signal through the second branch; the identification features are used to identify the magnetic security line, and the calibration features are used to calibrate the identification features;

[0024] The identification result acquisition module is used to fuse the identification features with the calibration features, and obtain the identification result corresponding to the magnetic security line to be identified based on the fused features.

[0025] Fourthly, this application also provides a paper resource identification device. The device includes:

[0026] The security thread identification result acquisition module is used to acquire the identification result corresponding to the magnetic security thread carried on the paper resource to be identified; the identification result is obtained according to the magnetic security thread identification method described in the embodiments of the first aspect.

[0027] The paper resource identification module uses the identification results to identify the paper resource to be identified.

[0028] Fifthly, this application also provides a paper resource identification device. The paper resource identification device includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0029] Acquire the waveform magnetic signal corresponding to the magnetic security line to be identified, and acquire the magnetic signal change rate and magnetic signal direction corresponding to multiple preset time points of the waveform magnetic signal;

[0030] Based on the waveform magnetic signal, the path characteristics of the waveform magnetic signal are obtained;

[0031] The path features are input into the first branch of the pre-trained magnetic security line recognition model to obtain the recognition features of the waveform magnetic signal. The magnetic signal change rate and magnetic signal direction are input into the second branch of the magnetic security line recognition model to obtain the calibration features of the waveform magnetic signal. The recognition features are used to identify the magnetic security line, and the calibration features are used to calibrate the recognition features.

[0032] The identification features are fused with the calibration features, and the identification result corresponding to the magnetic security line to be identified is obtained based on the fused features.

[0033] When the processor executes the computer program, it may also perform the following steps:

[0034] Obtain the identification result corresponding to the magnetic security thread carried on the paper resource to be identified; the identification result is obtained according to the magnetic security thread identification method described in the embodiments of the first aspect;

[0035] The identification results are used to identify the paper resources to be identified.

[0036] Sixthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, performs the following steps:

[0037] Acquire the waveform magnetic signal corresponding to the magnetic security line to be identified, and acquire the magnetic signal change rate and magnetic signal direction corresponding to multiple preset time points of the waveform magnetic signal;

[0038] Based on the waveform magnetic signal, the path characteristics of the waveform magnetic signal are obtained;

[0039] The path features are input into the first branch of the pre-trained magnetic security line recognition model to obtain the recognition features of the waveform magnetic signal. The magnetic signal change rate and magnetic signal direction are input into the second branch of the magnetic security line recognition model to obtain the calibration features of the waveform magnetic signal. The recognition features are used to identify the magnetic security line, and the calibration features are used to calibrate the recognition features.

[0040] The identification features are fused with the calibration features, and the identification result corresponding to the magnetic security line to be identified is obtained based on the fused features.

[0041] When the processor executes the computer program, it may also perform the following steps:

[0042] Obtain the identification result corresponding to the magnetic security thread carried on the paper resource to be identified; the identification result is obtained according to the magnetic security thread identification method described in the embodiments of the first aspect;

[0043] The identification results are used to identify the paper resources to be identified.

[0044] Seventhly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, performs the following steps:

[0045] Acquire the waveform magnetic signal corresponding to the magnetic security line to be identified, and acquire the magnetic signal change rate and magnetic signal direction corresponding to multiple preset time points of the waveform magnetic signal;

[0046] Based on the waveform magnetic signal, the path characteristics of the waveform magnetic signal are obtained;

[0047] The path features are input into the first branch of the pre-trained magnetic security line recognition model to obtain the recognition features of the waveform magnetic signal. The magnetic signal change rate and magnetic signal direction are input into the second branch of the magnetic security line recognition model to obtain the calibration features of the waveform magnetic signal. The recognition features are used to identify the magnetic security line, and the calibration features are used to calibrate the recognition features.

[0048] The identification features are fused with the calibration features, and the identification result corresponding to the magnetic security line to be identified is obtained based on the fused features.

[0049] When the processor executes the computer program, it may also perform the following steps:

[0050] Obtain the identification result corresponding to the magnetic security thread carried on the paper resource to be identified; the identification result is obtained according to the magnetic security thread identification method described in the embodiments of the first aspect;

[0051] The identification results are used to identify the paper resources to be identified.

[0052] The aforementioned magnetic security thread identification method, paper resource identification method, device, paper resource identification equipment, storage medium, and computer program product acquire the waveform magnetic signal corresponding to the magnetic security thread to be identified, as well as the magnetic signal change rate and magnetic signal direction corresponding to multiple preset time points of the waveform magnetic signal; based on the waveform magnetic signal, obtain the path features of the waveform magnetic signal; input the path features into the first branch of a pre-trained magnetic security thread identification model, and obtain the identification features of the waveform magnetic signal through the first branch; input the magnetic signal change rate and magnetic signal direction into the second branch of the magnetic security thread identification model, and obtain the calibration features of the waveform magnetic signal through the second branch; the identification features are used to identify the magnetic security thread, and the calibration features are used to calibrate the identification features; the identification features and calibration features are fused, and the identification result corresponding to the magnetic security thread to be identified is obtained based on the fused features. This application obtains the path characteristics of the waveform magnetic signal corresponding to the magnetic security line to be identified, as well as the magnetic signal change rate and magnetic signal direction of the waveform magnetic signal. Then, through a pre-trained magnetic security line identification model, it obtains identification features for identifying the magnetic security line based on the path characteristics, and obtains calibration features for calibrating the identification features based on the magnetic signal change rate and magnetic signal direction. Finally, based on the above identification features and the above calibration features, the magnetic security line can be accurately identified. Attached Figure Description

[0053] Figure 1 This is a flowchart illustrating a magnetic security thread identification method in one embodiment;

[0054] Figure 2 This is a waveform magnetic signal diagram of a magnetic security wire in one embodiment;

[0055] Figure 3 This is a schematic diagram of the process for obtaining identification features in one embodiment;

[0056] Figure 4 This is a schematic diagram of the structure of a magnetic security thread identification model in one embodiment;

[0057] Figure 5 This is a schematic diagram of the process for obtaining context information in one embodiment;

[0058] Figure 6 This is a flowchart illustrating the process of obtaining path features in one embodiment;

[0059] Figure 7 This is a flowchart illustrating a paper resource identification method in one embodiment;

[0060] Figure 8 This is a structural block diagram of a magnetic security thread identification device in one embodiment;

[0061] Figure 9 This is a structural block diagram of a paper resource identification device in one embodiment;

[0062] Figure 10 This is an internal structural diagram of a paper resource identification device in one embodiment. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0064] It should be noted that the terms "first" and "second" used in the embodiments of the present invention are merely to distinguish similar objects and do not represent a specific ordering of objects. It is understood that "first" and "second" can be interchanged in a specific order or sequence where permissible. It should be understood that the objects distinguished by "first" and "second" can be interchanged where appropriate so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein.

[0065] In one embodiment, such as Figure 1 As shown, a method for identifying magnetic security threads is provided. This embodiment illustrates the application of this method to a terminal. It is understood that this method can also be applied to a server, and further to a system including both a terminal and a server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0066] Step S101: Obtain the waveform magnetic signal corresponding to the magnetic security line to be identified, and obtain the magnetic signal change rate and magnetic signal direction corresponding to multiple preset time points of the waveform magnetic signal.

[0067] The magnetic security thread can be a magnetic security thread embedded in paper resources. This thread consists of evenly spaced lines woven segment by segment into the paper resource, containing embedded magnetic signals. The waveform magnetic signal is the magnetic signal emitted by this magnetic security thread, such as... Figure 2 As shown in the waveform magnetic signal diagram, the horizontal axis represents time t, and the vertical axis represents the magnetic signal value. The preset time point refers to a preset moment within the duration of the waveform magnetic signal. The magnetic signal change rate is the rate of change of the magnetic signal value at the current moment, while the magnetic signal direction is the tangent direction of the waveform magnetic signal at the current moment. Both the magnetic signal change rate and the magnetic signal direction are attributes of the aforementioned waveform magnetic signal.

[0068] Specifically, the waveform magnetic signal corresponding to the magnetic safety line to be identified is acquired by a magnetic sensor. Then, a preset time point and the magnetic signal value corresponding to the preset time point are acquired. The next moment after the preset time point and the next magnetic signal value corresponding to the next moment are acquired. Based on the preset time point, the magnetic signal value, the next moment after the preset time point and the next magnetic signal value, the rate of change of the magnetic signal value corresponding to the preset time point is obtained. Finally, based on the rate of change of the magnetic signal value and the inverse trigonometric function formula, the direction of the magnetic signal corresponding to the preset time point is obtained.

[0069] Step S102: Based on the waveform magnetic signal, obtain the path characteristics of the waveform magnetic signal.

[0070] Among them, the path feature is the path integral feature of the above-mentioned waveform magnetic signal, which can be used to characterize the geometric features of the above-mentioned waveform magnetic signal.

[0071] Specifically, each preset time point of the above-mentioned waveform magnetic signal and each magnetic signal value corresponding to each preset time point are obtained. Based on each magnetic signal value and each preset time point, the path characteristics of the waveform magnetic signal can be calculated.

[0072] Step S103: Input the path features into the first branch of the pre-trained magnetic security line recognition model, obtain the recognition features of the waveform magnetic signal through the first branch, and input the magnetic signal change speed and magnetic signal direction into the second branch of the magnetic security line recognition model, obtain the calibration features of the waveform magnetic signal through the second branch; the recognition features are used to identify the magnetic security line, and the calibration features are used to calibrate the recognition features.

[0073] The pre-trained magnetic security thread recognition model is a neural network model used to identify magnetic security threads. This model includes two branches, namely the first branch and the second branch. The first and second branches have the same structure but different parameters. The structure of the first and second branches includes an input layer, a one-dimensional convolutional layer, a fully connected layer, a bidirectional long short-term memory layer, and a self-attention layer. Then, the recognition features are the main features for identifying magnetic security threads, obtained based on path features, while the calibration features are the features used to assist in calibrating the recognition features, obtained based on the rate of change and direction of the magnetic signal.

[0074] Specifically, the path features are input into the first branch of the pre-trained magnetic security line recognition model. The path features are extracted through an input layer, a one-dimensional convolutional layer, and a fully connected layer. Then, the bidirectional long short-term memory layer is used to further obtain the contextual information of the path features based on the time series. Finally, the recognition features of the waveform magnetic signal are obtained based on the contextual information. Afterward, the magnetic signal change rate and magnetic signal direction are input into the second branch of the magnetic security line recognition model. The magnetic signal change rate and magnetic signal direction are extracted through an input layer, a one-dimensional convolutional layer, and a fully connected layer. Then, the bidirectional long short-term memory layer is used to further obtain the contextual information of the magnetic signal change rate and magnetic signal direction based on the time series. Finally, the calibration features of the waveform magnetic signal are obtained based on the contextual information.

[0075] Step S104: The identification features and calibration features are fused together, and the identification result corresponding to the magnetic security line to be identified is obtained based on the fused features.

[0076] Among them, the fusion feature is the feature obtained by fusing the identification feature and the calibration feature, while the identification result is the identification result of the paper resource corresponding to the magnetic security thread to be identified, including the face value and authenticity.

[0077] Specifically, the identification features and calibration features are input into the connection layer for processing and summarization to obtain the aforementioned data fusion features. These fusion features are then input into a fully connected layer containing multiple preset identification units. Each preset identification unit contains a baseline result. After obtaining the output result based on the fusion features, the output result is compared with the baseline result to finally obtain the corresponding paper resource identification result. For example, it can identify genuine banknotes with denominations of 100, 50, 20, and 10 yuan, as well as unidentifiable counterfeit banknotes.

[0078] In the aforementioned magnetic security wire identification method, the method acquires the waveform magnetic signal corresponding to the magnetic security wire to be identified, as well as the magnetic signal change rate and magnetic signal direction corresponding to multiple preset time points of the waveform magnetic signal. Based on the waveform magnetic signal, the path features of the waveform magnetic signal are obtained. The path features are input into the first branch of a pre-trained magnetic security wire identification model to obtain the identification features of the waveform magnetic signal. The magnetic signal change rate and magnetic signal direction are input into the second branch of the magnetic security wire identification model to obtain the calibration features of the waveform magnetic signal. The identification features are used to identify the magnetic security wire, and the calibration features are used to calibrate the identification features. The identification features and calibration features are fused, and the identification result corresponding to the magnetic security wire to be identified is obtained based on the fused features. This application acquires the path features of the waveform magnetic signal corresponding to the magnetic security wire to be identified, as well as the magnetic signal change rate and magnetic signal direction of the aforementioned waveform magnetic signal. Then, using a pre-trained magnetic security wire identification model, it obtains the identification features for identifying the magnetic security wire based on the path features, and obtains the calibration features for calibrating the identification features based on the magnetic signal change rate and magnetic signal direction. Finally, based on the aforementioned identification features and calibration features, the magnetic security wire can be accurately identified.

[0079] In one embodiment, such as Figure 3 As shown, the path features are input into the first branch of the pre-trained magnetic security line recognition model. The recognition features of the waveform magnetic signal are obtained through the first branch, including the following steps:

[0080] Step S301: Input the path features into the first branch and extract the magnetic signal attributes to obtain the extracted path features corresponding to the path features.

[0081] Among them, the extracted path features are the path features after feature extraction.

[0082] Specifically, such as Figure 4 As shown, the path features are input into the first branch. First, the input layer performs the first feature extraction, then the one-dimensional convolutional layer performs the second feature extraction, and finally the fully connected layer performs the third feature extraction at a higher level, resulting in the extracted path features corresponding to the path features.

[0083] Step S302: Obtain the context information of the extracted path features, and based on the context information, obtain the identification features of the waveform magnetic signal.

[0084] Among them, the context information is the feature sequence that is logically more reasonable and closely related to the extracted path features in terms of time sequence.

[0085] Specifically, the extracted path features are input into a bidirectional long short-term memory (BiLSTM) layer to obtain the context information corresponding to the extracted path features. Then, through a self-attention layer, the recognition features of the waveform magnetic signal of the first branch are obtained.

[0086] In this embodiment, by extracting features from the path features multiple times and obtaining the context information corresponding to the path features, the identification features of the path features can be accurately extracted.

[0087] In one embodiment, such as Figure 5 As shown, obtaining the contextual information of the extracted path features includes the following steps:

[0088] Step S501: Obtain the sub-features corresponding to multiple time points of the extracted path features, and obtain the current sub-feature corresponding to the current time point.

[0089] Among them, multiple moments are waveform magnetic signals, that is, multiple time points within the duration of the path feature, while sub-features are the path feature segments corresponding to each moment.

[0090] Specifically, obtain the current sub-feature corresponding to any current time among multiple time segments.

[0091] Step S502: Obtain the weights of each of the remaining sub-features except the current sub-feature, and based on the remaining sub-features and their respective weights, obtain the first branch state at the current time.

[0092] Among them, the remaining sub-features are the sub-features other than the current sub-feature among the above multiple sub-features, and the weight is the contribution weight of the remaining sub-features relative to the current time. As for the first branch state, it is the cell state, temporary cell state and hidden state at the current time in the bidirectional long short-term memory layer (BiLSTM) of the first branch. The first branch state is used to determine whether to memorize the current sub-feature corresponding to the current time.

[0093] Specifically, the contribution values ​​corresponding to each of the remaining sub-features are obtained, and based on the remaining sub-features and their corresponding co-contribution values, the current cell state, temporary cell state, and hidden state in the Bidirectional Long Short-Term Memory (BiLSTM) layer are calculated.

[0094] Step S503: Based on the first branch state corresponding to each time step, obtain the context information of the extracted path features.

[0095] Specifically, the contextual information of the extracted path features is obtained through the current cell state, temporary cell state, and hidden state, as well as the aforementioned sub-features.

[0096] In this embodiment, by utilizing the sub-features of previous and subsequent time points to predict the cell state, temporary cell state, and hidden state at the current time point, the contextual information of the extracted path feature time series can be captured, thereby accurately obtaining the contextual information of the extracted path features.

[0097] In one embodiment, such as Figure 6 As shown, the path characteristics of the waveform magnetic signal are obtained based on the waveform magnetic signal, including the following steps:

[0098] Step S601: Using a pre-obtained preset time window, the waveform magnetic signal is divided into multiple signal segments with the same time.

[0099] The preset time window is the preset time period, and the signal segment is the segment of the waveform magnetic signal corresponding to each time period.

[0100] Specifically, a sliding time window is used to slide across the waveform magnetic signal W, thereby extracting the signature features at each position within the window. Assuming the sliding window size is h = 2Δt + 1, the signal segment at position n within the window can be represented as:

[0101]

[0102] Step S602: Obtain the sub-path features corresponding to each signal segment and the order of each signal segment.

[0103] Among them, the sub-path features are the path feature segments corresponding to the above time periods in the path features, and the order of each signal segment is the time arrangement order of each signal segment.

[0104] Specifically, by using the waveform magnetic signal at each moment and the corresponding magnetic signal value at each moment, the sub-path characteristics of each signal segment and the order of each signal segment are calculated.

[0105] Step S603: Based on the characteristics of each sub-path and the order of each signal segment, the path characteristics of the waveform magnetic signal are obtained.

[0106] Specifically, the path features of each sub-path are arranged in the order of each signal segment to obtain the path features of the waveform magnetic signal.

[0107] In this embodiment, the waveform magnetic signal is divided into multiple signal segments with the same time by using a preset time window, and then the sub-path features corresponding to each signal segment are obtained, so as to accurately obtain the path features of the waveform magnetic signal.

[0108] In one embodiment, obtaining the sub-path features corresponding to each signal segment includes the following steps:

[0109] Obtain multiple time points corresponding to the current signal segment, and multiple magnetic signal values ​​corresponding to each time point; based on the multiple time points and multiple magnetic signal values, obtain multiple iterative integration elements corresponding to the current signal segment; based on the multiple iterative integration elements, obtain the sub-path features corresponding to the current signal segment.

[0110] Among them, multiple time points are multiple moments in the time segment in which the current signal segment lasts, and multiple magnetic signal values ​​are the values ​​of the magnetic signals corresponding to multiple moments. As for the iterative integration elements, they are the results obtained after multiple integrations of the current signal segment.

[0111] Specifically, based on multiple time points and multiple magnetic signal values, multiple coordinates of the current signal segment are first obtained, which can be represented as: The superscript of its k-th order iterative integral can be written as an index sequence i1, i2, ... i k If ,∈{1,...,d}, then the k-th order iterative integral element of the current signal segment path in d dimensions can be represented as:

[0112]

[0113] Then, the number of elements is d. k If there are 1, then the signature of the current signal segment path P in the time interval [0, T] is the set of all iterative integrals of P, expressed as:

[0114]

[0115] Finally, truncate the path. In the m-th layer dimension, the value of m is usually set between [2,5], which can obtain the sub-path features corresponding to the current signal segment.

[0116] In this embodiment, by obtaining multiple iterative integration elements corresponding to the current signal segment, and then based on the multiple iterative integration elements, the sub-path features corresponding to the current signal segment can be accurately obtained.

[0117] In one embodiment, acquiring the magnetic signal change rate and magnetic signal direction corresponding to multiple preset time points of the waveform magnetic signal includes the following steps:

[0118] Obtain the current magnetic signal value of the current waveform magnetic signal at the current preset time point, and obtain the next magnetic signal value of the current waveform magnetic signal at the next time point after the current preset time point; use the difference between the next time point and the current time point, and the difference between the next magnetic signal value and the current magnetic signal value, to obtain the current magnetic signal change rate of the current waveform magnetic signal at the current time point; based on the current magnetic signal change rate, obtain the current magnetic signal direction of the current waveform magnetic signal at the current time point.

[0119] Here, the next time point is the moment after the current time point, and the next magnetic signal value is the magnetic signal value corresponding to the next moment.

[0120] Specifically, the calculation of the current magnetic signal change rate and the current magnetic signal direction is as follows:

[0121]

[0122] Where Δt represents the difference between the next time point and the current time point, v t；Δt This indicates the rate of change of the current magnetic signal, in milliseconds (ms). t+Δt -ms t This represents the difference between the next magnetic signal value and the current magnetic signal value; d t；Δt Indicates the current direction of the magnetic signal.

[0123] In this embodiment, by utilizing the difference between the next time point and the current time point, and the difference between the next magnetic signal value and the current magnetic signal value, the current magnetic signal change rate and the current magnetic signal direction at the current time point of the current waveform magnetic signal can be accurately obtained.

[0124] In one embodiment, obtaining the identification result corresponding to the magnetic security line to be identified based on the fusion features includes the following steps:

[0125] Based on the fusion features, the output result corresponding to the magnetic security line is obtained; the output result is compared with the preset benchmark result to obtain the recognition result corresponding to the magnetic security line to be identified.

[0126] The output results are the fused feature labels corresponding to the magnetic security lines, and the benchmark results are the preset benchmark labels.

[0127] Specifically, based on the fusion feature input, a fully connected layer containing multiple recognition units is used in the magnetic security line recognition model. Each recognition unit contains a reference label to obtain the output result corresponding to the magnetic security line. The output result is compared with the reference label in each recognition unit to obtain the recognition result corresponding to the magnetic security line to be recognized.

[0128] In this embodiment, by comparing the output result with the preset benchmark result, the identification result corresponding to the magnetic security line to be identified can be accurately obtained.

[0129] In one embodiment, such as Figure 7 As shown, a paper resource identification method is provided. This embodiment illustrates the method applied to a terminal. It is understood that this method can also be applied to a server, and further to a system including both a terminal and a server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0130] Step S701: Obtain the recognition result corresponding to the magnetic security thread carried on the paper resource to be identified; the recognition result is obtained according to the magnetic security thread recognition method of the above embodiments;

[0131] Among them, the paper resources to be identified refer to paper resources that can be used for financial transactions.

[0132] Specifically, the identification results corresponding to the magnetic security lines obtained in the above embodiments are obtained.

[0133] Step S702: Use the recognition results to identify the paper resource to be identified.

[0134] Based on the identification results corresponding to the magnetic security thread, the identification results of the paper resources corresponding to the magnetic security thread are obtained.

[0135] In the aforementioned paper resource identification method, the identification result corresponding to the magnetic security thread carried by the paper resource to be identified is obtained; the identification result is obtained according to the magnetic security thread identification method of the above embodiments; based on the identification result corresponding to the magnetic security thread, the identification result of the paper resource corresponding to the magnetic security thread is obtained. This application obtains the path characteristics of the waveform magnetic signal corresponding to the magnetic security thread to be identified, as well as the magnetic signal change rate and magnetic signal direction of the waveform magnetic signal. Then, through a pre-trained magnetic security thread identification model, identification features for identifying the magnetic security thread are obtained based on the path characteristics, and calibration features for calibrating the identification features are obtained based on the magnetic signal change rate and magnetic signal direction. Finally, based on the identification features and calibration features, the identification result corresponding to the magnetic security thread can be accurately obtained, thereby accurately identifying the aforementioned paper resource.

[0136] In one application embodiment, a method for identifying magnetic security threads is provided, comprising the following steps:

[0137] 1. First, a magnetic sensor is used to collect the magnetic signal of the paper resource, and the path signature of the magnetic security thread is calculated. The horizontal axis represents time, and the vertical axis represents the magnetic signal value. The entire magnetic waveform signal of the magnetic security thread is defined as W:

[0138]

[0139] 2. Using the path signature method, the waveform of the magnetic security line is highly generalized, and the geometric features of the waveform are characterized by path signature features. Based on the definition of path signature, for a d-dimensional path P in the time interval [0, T], the coordinates of the path can be represented as follows: The superscript of its k-th order iterative integral can be written as an index sequence i1, i2, ... i k ,∈{1,...,d}. Then the k-th order iterative integral element of the d-dimensional path can be represented as:

[0140]

[0141] The number of elements is d k If there are 1000 unique integrals, then the signature of path P in the time interval [0, T] is the set of all iterative integrals of P, expressed as:

[0142]

[0143] The value of the 0th term is usually set to 1. Since the signature is defined over all possible indices of finite length, the number of elements in the signature is infinite. In practical applications, truncation is performed at level m of S(P) to ensure that the path characteristics are within a reasonable range; such a signature is called a truncated signature. The value of m is usually set between [2, 5] because the amount of information contained in higher-order signatures decays by the factorial of the order, and the computational cost is too high. Truncated Path The dimension of the m-th layer is:

[0144]

[0145] A sliding window is used to slide across the magnetic security line waveform W, and the signature feature at each position in the window is extracted. Assuming the sliding window size is h = 2Δt + 1, the magnetic signal sequence at position n in the window is represented as:

[0146]

[0147] From the magnetic signal sequence w n The path signature features of the magnetic security line waveform can be calculated using a feature vector S(w) n Let m represent this. Therefore, converting the magnetic security wire waveform into a path signature sequence can be represented as:

[0148] P m =(S(w1)|m,S(w2)m,...,S(w N )|m);

[0149] After the above steps, the global geometric features, i.e. path features, of the magnetic security wire waveform signal are extracted using the path signature method.

[0150] 3. Next, calculate the rate of change of the magnetic signal waveform of the magnetic safety wire, the rate of change of the magnetic signal at time t, and its direction:

[0151]

[0152] Where Δt>0 represents the time span. The next step is to obtain the magnetic signal change velocity sequence and direction sequence, and to construct the magnetic signal change velocity sequence and direction sequence from five different time scales:

[0153] V Δt =(v 1；Δt ,v 2；Δt ,...,v N；Δt ) T V = (V4, V8, V) 16 V 32 V 64 );

[0154] D Δt =(d 1；Δt ,d 2；Δt ,...,d N；Δt ) T D = (D4, D8, D) 16 D 32 D 64 );

[0155] 4. Next, we construct a composite neural network (1DCNN+BiLSTM) with two branches. The structure of each branch is the same, only the parameters are different. The branches include an input layer, a one-dimensional convolutional layer, a fully connected layer, a BiLSTM layer, a self-attention layer, and a connection layer. The path features of the magnetic security wire waveform signal are input into the first branch, and the magnetic signal change speed and direction of the magnetic security wire waveform signal are input into the second branch.

[0156] Input Layers: There are two input layers. Each input layer takes a feature sequence or a combination of feature sequences as input, and outputs a label sequence. Convolutional Layers (One-Dimensional Convolutions): Located after the output layers, each convolutional layer contains three one-dimensional convolutions. Batch Normalization (BN) and non-linear activation operations are used in each convolutional layer. The ReLU function is used as the activation function in the convolutional layers. Fully Connected Layers: Used to extract higher-level features, which are then fed into the BiLSTM layer. This layer uses Dropout and the ReLU activation function. BiLSTM Layers: Utilize information from previous and subsequent time steps to predict the current state, thereby capturing the contextual information of the time series. This layer uses the Tanh function as the activation function; the Self-attention layer further assigns contribution values ​​to the information retained in the BiLSTM layer, strengthening the learning of the main features in the retained information, thereby improving the model performance; the Concatenate Layer uses a multi-input neural network, with two branches simultaneously processed by convolutional layers, fully connected layers, and BiLSTM layers. After processing and summarizing in the Concatenate Layer, the features are input into a fully connected layer (Time-distributedDense) containing 5 units. Each of the 5 units in the fully connected layer has five baseline labels. By comparing them with the summarized features, the system can effectively classify and identify genuine banknotes of denominations of 100, 50, 20, and 10 yuan, as well as counterfeit banknotes that cannot be identified.

[0157] In this embodiment, the waveform features of the magnetic security thread are extracted using a path signature algorithm and then input into a neural network for training to achieve the purpose of recognizing the face value of paper resources and authenticating counterfeits. The path signature method effectively extracts the global features of the security thread waveform, improving the accuracy and security of paper resource authentication.

[0158] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but may be executed at different times. The execution order of these steps or stages is not necessarily sequential, but may be performed alternately or in turn with other steps or at least a portion of the steps or stages of other steps.

[0159] Based on the same inventive concept, this application also provides a magnetic security thread identification device for implementing the magnetic security thread identification method described above. The solution provided by this device is similar to the implementation described in the above method; therefore, the specific limitations in one or more embodiments of the magnetic security thread identification device provided below can be found in the limitations of the magnetic security thread identification method described above, and will not be repeated here.

[0160] In one embodiment, such as Figure 8 As shown, a magnetic security thread identification device is provided, comprising: a signal attribute acquisition module 801, a path feature acquisition module 802, an identification feature acquisition module 803, and an identification result acquisition module 804, wherein:

[0161] The signal attribute acquisition module 801 is used to acquire the waveform magnetic signal corresponding to the magnetic security line to be identified, and to acquire the magnetic signal change rate and magnetic signal direction corresponding to multiple preset time points of the waveform magnetic signal.

[0162] The path feature acquisition module 802 is used to obtain the path features of the waveform magnetic signal based on the waveform magnetic signal.

[0163] The feature acquisition module 803 is used to input path features into the first branch of the pre-trained magnetic safety line recognition model, and to acquire the recognition features of the waveform magnetic signal through the first branch. It also inputs the magnetic signal change speed and magnetic signal direction into the second branch of the magnetic safety line recognition model, and to acquire the calibration features of the waveform magnetic signal through the second branch. The recognition features are used to identify the magnetic safety line, and the calibration features are used to calibrate the recognition features.

[0164] The recognition result acquisition module 804 is used to fuse the recognition features with the calibration features and obtain the recognition result corresponding to the magnetic security line to be identified based on the fused features.

[0165] In one embodiment, the identification feature acquisition module 803 is further configured to input the path features into the first branch, extract features from the magnetic signal attributes to obtain the extracted path features corresponding to the path features; acquire the context information of the extracted path features, and obtain the identification features of the waveform magnetic signal based on the context information.

[0166] In one embodiment, the feature acquisition module 803 is further configured to acquire sub-features corresponding to multiple times of the extracted path features, and acquire the current sub-feature corresponding to the current time; acquire the weights corresponding to each of the remaining sub-features other than the current sub-feature, and obtain the first branch state corresponding to the current time based on the remaining sub-features and the weights corresponding to the remaining sub-features; and obtain the context information of the extracted path features based on the first branch states corresponding to each time.

[0167] In one embodiment, the path feature acquisition module 802 is further configured to divide the waveform magnetic signal into multiple signal segments with the same time using a pre-obtained preset time window; acquire the sub-path features corresponding to each signal segment and the order of each signal segment; and obtain the path features of the waveform magnetic signal based on each sub-path feature and the order of each signal segment.

[0168] In one embodiment, the path feature acquisition module 802 is further configured to acquire multiple time points corresponding to the current signal segment, and multiple magnetic signal values ​​corresponding to the multiple time points respectively; based on the multiple time points and the multiple magnetic signal values, obtain multiple iterative integration elements corresponding to the current signal segment; and obtain the sub-path features corresponding to the current signal segment according to the multiple iterative integration elements.

[0169] In one embodiment, the signal attribute acquisition module 801 is further configured to acquire the current magnetic signal value of the current waveform magnetic signal at the current preset time point, and acquire the next magnetic signal value of the current waveform magnetic signal at the next time point of the current preset time point; using the difference between the next time point and the current time point, and the difference between the next magnetic signal value and the current magnetic signal value, the current magnetic signal change rate of the current waveform magnetic signal at the current time point is obtained; based on the current magnetic signal change rate, the current magnetic signal direction of the current waveform magnetic signal at the current time point is obtained.

[0170] In one embodiment, the identification result acquisition module 804 is further used to obtain the output result corresponding to the magnetic security line based on the fusion features; and compare the output result with the preset benchmark result to obtain the identification result corresponding to the magnetic security line to be identified.

[0171] Each module in the aforementioned magnetic security thread identification device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of the paper resource identification device in hardware form or independent of it, or stored in the memory of the paper resource identification device in software form, so that the processor can call and execute the corresponding operations of each module.

[0172] In one embodiment, such as Figure 9 As shown, a paper resource identification device is provided, including: an identification result acquisition module 901 and a paper resource identification module 902, wherein:

[0173] The security thread identification result acquisition module 901 is used to acquire the identification result corresponding to the magnetic security thread carried on the paper resource to be identified; the identification result is obtained according to the magnetic security thread identification method in the above embodiments.

[0174] The paper resource identification module 902 uses the identification results to identify the paper resources to be identified.

[0175] Each module in the aforementioned magnetic security thread identification device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of the paper resource identification device in hardware form or independent of it, or stored in the memory of the paper resource identification device in software form, so that the processor can call and execute the corresponding operations of each module.

[0176] In one embodiment, a paper resource identification device is provided. This device can be a terminal, and its internal structure diagram can be as follows: Figure 10 As shown, the paper resource identification device includes a processor, memory, communication interface, display screen, and input device connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a magnetic security line identification method or a paper resource identification method. The display screen can be an LCD screen or an e-ink screen. The input device can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the device's casing, or an external keyboard, touchpad, or mouse.

[0177] Those skilled in the art will understand that Figure 10 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the paper resource identification device to which the present application is applied. A specific paper resource identification device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0178] In one embodiment, a paper resource identification device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0179] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0180] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0181] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0182] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0183] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0184] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for identifying magnetic security threads, characterized in that, The method includes: Acquire the waveform magnetic signal corresponding to the magnetic security line to be identified, and acquire the magnetic signal change rate and magnetic signal direction corresponding to multiple preset time points of the waveform magnetic signal; Based on the waveform magnetic signal, the path characteristics of the waveform magnetic signal are obtained. The path characteristics are the path integral characteristics of the waveform magnetic signal, and the path characteristics are used to characterize the geometric characteristics of the waveform magnetic signal. The path features are input into the first branch of the pre-trained magnetic security line recognition model to obtain the recognition features of the waveform magnetic signal. The magnetic signal change rate and magnetic signal direction are input into the second branch of the magnetic security line recognition model to obtain the calibration features of the waveform magnetic signal. The recognition features are used to identify the magnetic security line, and the calibration features are used to calibrate the recognition features. The identification features are fused with the calibration features, and the identification result corresponding to the magnetic security line to be identified is obtained based on the fused features; The process of obtaining the path characteristics of the waveform magnetic signal based on the waveform magnetic signal includes: Using a pre-defined time window, the waveform magnetic signal is divided into multiple signal segments with the same time. Obtain the sub-path features corresponding to each of the signal segments, as well as the order of the signal segments; Based on the characteristics of each sub-path and the order of each signal segment, the path characteristics of the waveform magnetic signal are obtained.

2. The method according to claim 1, characterized in that, The step of inputting the path features into the first branch of the pre-trained magnetic security line recognition model, and obtaining the recognition features of the waveform magnetic signal through the first branch, includes: The path features are input into the first branch, and the magnetic signal properties are extracted to obtain the extracted path features corresponding to the path features. Obtain the context information of the extracted path features, and based on the context information, obtain the identification features of the waveform magnetic signal.

3. The method according to claim 2, characterized in that, The step of obtaining the context information of the extracted path features includes: Obtain the sub-features corresponding to multiple time points of the extracted path features, and obtain the current sub-feature corresponding to the current time point; Obtain the weights corresponding to each of the remaining sub-features other than the current sub-feature, and based on the remaining sub-features and their corresponding weights, obtain the first branch state corresponding to the current time. Based on the first branch state corresponding to each of the aforementioned times, the context information of the extracted path features is obtained.

4. The method according to claim 1, characterized in that, The step of obtaining the sub-path features corresponding to each of the signal segments includes: Obtain multiple time points corresponding to the current signal segment, and multiple magnetic signal values ​​corresponding to each of the multiple time points; Based on the multiple time points and the multiple magnetic signal values, multiple iterative integration elements corresponding to the current signal segment are obtained; Based on the multiple iterative integral elements, the sub-path features corresponding to the current signal segment are obtained.

5. The method according to claim 1, characterized in that, The step of acquiring the magnetic signal change rate and magnetic signal direction corresponding to multiple preset time points of the waveform magnetic signal includes: Obtain the current magnetic signal value of the current waveform magnetic signal at the current preset time point, and obtain the next magnetic signal value of the current waveform magnetic signal at the next time point after the current preset time point; By using the difference between the next time point and the current preset time point, and the difference between the next magnetic signal value and the current magnetic signal value, the rate of change of the current waveform magnetic signal at the current preset time point can be obtained; Based on the current rate of change of the magnetic signal, the current magnetic signal direction of the current waveform magnetic signal at the current preset time point is obtained.

6. The method according to claim 1, characterized in that, The step of obtaining the identification result corresponding to the magnetic security line to be identified based on the fusion features includes: Based on the fusion features, the output result corresponding to the magnetic security line is obtained; The output result is compared with the preset benchmark result to obtain the identification result corresponding to the magnetic security line to be identified.

7. A method for identifying paper resources, characterized in that, The method includes: Obtain the identification result corresponding to the magnetic security thread carried on the paper resource to be identified; the identification result is obtained by the magnetic security thread identification method according to any one of claims 1 to 6; The identification results are used to identify the paper resources to be identified.

8. A magnetic security thread identification device, characterized in that, The device includes: The signal attribute acquisition module is used to acquire the waveform magnetic signal corresponding to the magnetic security line to be identified, and to acquire the magnetic signal change rate and magnetic signal direction corresponding to multiple preset time points of the waveform magnetic signal. The path feature acquisition module is used to obtain the path features of the waveform magnetic signal based on the waveform magnetic signal. The path features are the path integral features of the waveform magnetic signal and are used to characterize the geometric features of the waveform magnetic signal. The identification feature acquisition module is used to input the path features into a first branch of a pre-trained magnetic security line identification model, and to acquire the identification features of the waveform magnetic signal through the first branch; and to input the magnetic signal change rate and the magnetic signal direction into a second branch of the magnetic security line identification model, and to acquire the calibration features of the waveform magnetic signal through the second branch; the identification features are used to identify the magnetic security line, and the calibration features are used to calibrate the identification features; The identification result acquisition module is used to fuse the identification feature with the calibration feature, and obtain the identification result corresponding to the magnetic security line to be identified based on the fused feature; The process of obtaining the path characteristics of the waveform magnetic signal based on the waveform magnetic signal includes: Using a pre-defined time window, the waveform magnetic signal is divided into multiple signal segments with the same time. Obtain the sub-path features corresponding to each of the signal segments, as well as the order of the signal segments; Based on the characteristics of each sub-path and the order of each signal segment, the path characteristics of the waveform magnetic signal are obtained.

9. A paper resource identification device, characterized in that, The device includes: The security thread identification result acquisition module is used to acquire the identification result corresponding to the magnetic security thread carried on the paper resource to be identified; the identification result is obtained by the magnetic security thread identification method according to any one of claims 1 to 6. The paper resource identification module uses the identification results to identify the paper resource to be identified.

10. A paper resource identification device, comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

12. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images