Check authentic identification method and system based on multispectral image fusion and OCR

Abstract

The invention relates to a check authentic identification method and system based on multispectral image fusion and OCR, and the method comprises the steps: obtaining equipment information of multispectral bill scanning equipment, and loading corresponding equipment parameters according to the equipment information; the method comprises the following steps: scanning a check to be authenticated through a multispectral bill scanning device to obtain a multispectral image, recording device state data, and preprocessing the multispectral image; inputting the preprocessed multispectral image into a preset image fusion model, outputting a feature enhanced image, and fusing various anti-counterfeiting information into the feature enhanced image; performing multi-dimensional authentic identification feature extraction based on the feature enhanced image, and performing analysis by integrating equipment state data to obtain an authentic identification result; positioning and extracting a ticket number area in the reflection color image to obtain a ticket number image; and identifying the ticket number image by adopting a self-adaptive OCR engine to obtain a ticket number identification result. The multi-feature authentication and recognition process is integrated, and the check authentication accuracy and the ticket number recognition robustness are improved.

Description

Technical Field

[0001] This invention belongs to the field of financial instrument authentication technology, and in particular relates to a check authentication method and system based on multispectral image fusion and OCR. Background Technology

[0002] The widespread use of financial instruments makes their authentication crucial. Currently, instrument authentication typically relies on one or a few features (e.g., infrared signatures, ultraviolet fiber filaments), but high-precision counterfeiting techniques can mimic these features, leading to insufficient reliability in authentication.

[0003] Meanwhile, checks are prone to wear, creases, and contamination during circulation, which makes the key information on them unclear, posing a great challenge to optical character recognition (OCR) and resulting in a high error rate. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a method and system for check authentication based on multispectral image fusion and OCR.

[0005] The technical solution adopted in this invention is: Firstly, a method for checking counterfeit detection based on multispectral image fusion and OCR is provided, including: Obtain the equipment information of the multispectral ticket scanning device, and load the corresponding equipment parameters based on the equipment information; Multispectral images are obtained by scanning the check to be authenticated using a multispectral ticket scanning device. The device status data during the scanning process is recorded, and the multispectral images are preprocessed. The multispectral images include reflective color images, reflective infrared images, reflective ultraviolet images, transmitted infrared images, and transmitted ultraviolet images. The preprocessed multispectral image is input into a preset image fusion model, and the output is a feature-enhanced image, which is fused with multiple anti-counterfeiting information. Multi-dimensional anti-counterfeiting features are extracted based on feature-enhanced images, and the anti-counterfeiting results are obtained by analyzing the integrated device status data. The ticket number region in the reflected color image is located and extracted to obtain the ticket number image; An adaptive OCR engine is used to recognize the ticket number image, and the ticket number recognition result is obtained.

[0006] Furthermore, obtain the equipment information of the multispectral ticket scanning device, and load the corresponding equipment parameters based on the equipment information, including: Obtain device information for the currently connected multispectral ticket scanning device, including device model, firmware version number, CIS type, and supported scanning light source types; Based on the device model in the device information, the matching device parameters are dynamically loaded from the preset configuration file, enabling the multispectral ticket scanning device to perform image scanning according to the device parameters.

[0007] Furthermore, a multispectral image is obtained by scanning the check to be authenticated using a multispectral document scanning device. The device status data during the scanning process is recorded. The multispectral image is then preprocessed, including: The check to be authenticated is scanned using a multispectral ticket scanning device to obtain multiple images with different spectra, which are then used as multispectral images. Record the device status data during the scanning process of the multispectral ticket scanning device to authenticate the check;

[0008] Preprocessing is performed on multispectral images, including image registration, real-time bias correction, normalization, and enhancement.

[0009] Furthermore, multi-dimensional anti-counterfeiting features are extracted based on the enhanced image, and the device status data is analyzed to obtain the anti-counterfeiting results, including: Based on feature-enhanced images, multi-dimensional anti-counterfeiting features are extracted using fiber feature detection technology, microtext detection technology, and ink and printing feature analysis technology, respectively, to obtain fiber features, microtext features, and ink and printing features. The fiber filament characteristics are matched with the standard fiber filament characteristics of a genuine check template to obtain the fiber filament confidence level. The existence and correctness of microtext features are verified to obtain the microtext confidence level. The ink properties of the ink printing characteristics are analyzed to obtain the ink printing confidence level; Based on the confidence levels of fiber filaments, microtext, and ink printing, a weighted fusion with preset weights is performed to obtain the counterfeit detection confidence level. Determine if there is any equipment malfunction based on the equipment status data; If there is no equipment malfunction, the authentication confidence level will be used as the authentication result. If there is a device malfunction, the confidence level is determined based on the malfunction level. The confidence level is then multiplied by the confidence level to obtain the authentication result.

[0010] Furthermore, an adaptive OCR engine is used to recognize the ticket number image, obtaining the ticket number recognition results, including: An adaptive OCR engine was used to recognize the ticket number image, and the initial recognition results and corresponding ticket number confidence scores were obtained. Determine whether the confidence level of the ticket number falls within the high confidence interval, medium confidence interval, or low confidence interval; If it is in the high confidence interval, the initial recognition result will be used as the ticket number recognition result; If the result is within the medium confidence range, the initial identification result is verified or corrected using multispectral assistance to obtain the ticket number identification result. If it is in the low confidence range, then query the device error code; Determine if the problem is an image quality issue based on the device error code; If the problem is with image quality, the hardware retry mechanism will be triggered. If the error is not due to image quality, the initial recognition result is determined to be a recognition failure result, and the recognition failure result is used as the ticket number recognition result.

[0011] Secondly, a check authentication system based on multispectral image fusion and OCR is provided, including: The device management and parameter configuration module is used to obtain the device information of the multispectral ticket scanning device and load the corresponding device parameters based on the device information. The image acquisition and status monitoring module is used to obtain multispectral images by scanning the check to be authenticated using a multispectral ticket scanning device, record the device status data during the scanning process, and preprocess the multispectral images. The multispectral images include reflective color images, reflective infrared images, reflective ultraviolet images, transmitted infrared images, and transmitted ultraviolet images. The image fusion module is used to input the preprocessed multispectral image into a preset image fusion model and output a feature-enhanced image, which fuses multiple anti-counterfeiting information. The anti-counterfeiting module is used to extract multi-dimensional anti-counterfeiting features based on feature-enhanced images, analyze comprehensive device status data, and obtain anti-counterfeiting results. The ticket number image extraction module is used to locate and extract the ticket number region in the reflective color image to obtain the ticket number image; The adaptive OCR recognition module is used to recognize ticket number images using an adaptive OCR engine to obtain ticket number recognition results.

[0012] Furthermore, the device management and parameter configuration modules include: The device management unit is used to obtain device information of the currently connected multispectral ticket scanning device, including device model, firmware version number, CIS type and supported scanning light source type. The parameter configuration unit is used to dynamically load matching device parameters from a preset configuration file based on the device model in the device information, so that the multispectral ticket scanning device can perform image scanning according to the device parameters.

[0013] Furthermore, the image acquisition and status monitoring module includes: The image acquisition unit is used to perform multispectral scanning on the check to be authenticated using a multispectral ticket scanning device to obtain multiple images of different spectra, which are used as multispectral images. The equipment status monitoring unit records the equipment status data during the scanning process of the multispectral ticket scanning equipment for counterfeit detection. The image preprocessing unit is used to perform image registration, real-time bias correction, normalization, and enhancement preprocessing on multispectral images.

[0014] Furthermore, the counterfeit detection module specifically extracts multi-dimensional counterfeit detection features based on feature-enhanced images using fiber filament feature detection technology, microtext detection technology, and ink and printing feature analysis technology, obtaining fiber filament features, microtext features, and ink printing features. It then matches the fiber filament features with the standard fiber filament features of a genuine check template to obtain a fiber filament confidence score. It verifies the existence and correctness of the microtext features to obtain a microtext confidence score. It analyzes the ink characteristics of the ink printing features to obtain an ink printing confidence score. Based on the fiber filament confidence score, microtext confidence score, and ink printing confidence score, it performs a weighted fusion with preset weights to obtain a counterfeit detection confidence score. Finally, it determines whether there is an equipment anomaly based on equipment status data. If no equipment anomaly is found, the counterfeit detection confidence score is used as the counterfeit detection result. If an equipment anomaly is found, a downgrade parameter is obtained based on the anomaly level, and this downgrade parameter is multiplied by the counterfeit detection confidence score to obtain the counterfeit detection result.

[0015] Furthermore, the adaptive OCR recognition module is specifically used to recognize the ticket number image using an adaptive OCR engine, obtaining an initial recognition result and the corresponding ticket number confidence level; determining whether the ticket number confidence level falls within a high confidence range, a medium confidence range, or a low confidence range; if it falls within a high confidence range, the initial recognition result is used as the ticket number recognition result; if it falls within a medium confidence range, the initial recognition result is verified or corrected using multispectral assistance to obtain the ticket number recognition result; if it falls within a low confidence range, the device error code is queried; the device error code is used to determine whether it is an image quality problem; if it is an image quality problem, a hardware retry mechanism is triggered; if it is not an image quality error, the initial recognition result is determined to be a recognition failure result, and the recognition failure result is used as the ticket number recognition result.

[0016] The beneficial effects achieved by this invention are as follows: The system acquires equipment information for a multispectral ticket scanning device and loads corresponding equipment parameters based on this information. It then scans the check to be authenticated using the multispectral ticket scanning device to obtain a multispectral image. The system records equipment status data during the scanning process and preprocesses the multispectral image, which includes reflective color, reflective infrared, reflective ultraviolet, transmitted infrared, and transmitted ultraviolet images. The preprocessed multispectral image is input into a pre-set image fusion model, which outputs a feature-enhanced image. This feature-enhanced image incorporates multiple anti-counterfeiting information. Multi-dimensional anti-counterfeiting features are extracted based on the feature-enhanced image, and the system analyzes the combined equipment status data to obtain the authentication result. The system locates and extracts the ticket number region from the reflective color image to obtain the ticket number image. An adaptive OCR engine is then used to recognize the ticket number image, yielding the ticket number recognition result. This integrated multi-feature authentication and recognition process improves the accuracy of check authentication and the robustness of ticket number recognition. Attached Figure Description

[0017] Figure 1 This is a flowchart of the check authentication method based on multispectral image fusion and OCR of the present invention; Figure 2 This is a flowchart of the authentication process of this invention; Figure 3 This is a flowchart of the adaptive OCR recognition process of the present invention; Figure 4 This is the first structural diagram of the check authentication system based on multispectral image fusion and OCR of the present invention; Figure 5 This is the second structural diagram of the check authentication system based on multispectral image fusion and OCR of the present invention. Detailed Implementation

[0018] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0019] like Figure 1 As shown, this embodiment of the invention provides a check authentication method based on multispectral image fusion and OCR, including: 101. Obtain the equipment information of the multispectral ticket scanning device, and load the corresponding equipment parameters based on the equipment information; In this embodiment, the check authentication method based on multispectral image fusion and OCR is applied to the check authentication system based on multispectral image fusion and OCR. After the system starts, it first obtains the device information of the currently connected multispectral ticket scanning device by calling the driver layer interface (e.g., DPM_SP_GetDevInfo). The device information includes the device model, firmware version number, CIS type (e.g., standard 126mm or cost-reduced 108mm) and supported scanning light source type. Based on the device model in the device information, dynamically load matching device parameters from the preset configuration file; For example, if the device model is identified as a cost-reduced CIS model (DPM910H101), the effective pixel width parameter in subsequent image processing will be automatically adjusted to 864 pixels, and dedicated cropping and black-and-white balance calibration parameters will be invoked to ensure optimal image quality on different hardware platforms.

[0020] 102. Obtain a multispectral image by scanning the check to be authenticated using a multispectral ticket scanning device, record the device status data during the scanning process of the multispectral ticket scanning device, and preprocess the multispectral image. In this embodiment, after the multispectral ticket scanning device is in normal working order, a scanning command is initiated through the driver interface (such as DPM_SP_Scan). The multispectral ticket scanning device sequentially activates reflective RGB, reflective infrared (IR), reflective ultraviolet (UV), transmitted infrared (IRtr), and transmitted ultraviolet (UVtr) light sources to scan the front and back of the check to be authenticated, obtaining multiple images of different spectra, which are used as multispectral images. The multispectral images include reflective color images, reflective infrared images, reflective ultraviolet images, transmitted infrared images, and transmitted ultraviolet images. Record the device status data during the scanning process of the multispectral ticket scanning device for counterfeit detection. The device status data consists of data from multiple types of sensors on the multispectral ticket scanning device. Preprocessing of multispectral images, including image registration, real-time bias correction, normalization, and enhancement; Image registration: Since images of different spectra may have slight positional shifts, a feature point-based registration algorithm is used to align all images to the reflectance color image coordinate system; Real-time correction: Read the status value of the detection sensor returned by the firmware in real time; if the check to be counterfeited is seriously deviated (such as exceeding the 5% tolerance), the driver layer's rejection instruction (DPM_SP_Eject) is immediately called to eject the check to be counterfeited and prompt the user to reinsert it; if the deviation is within the tolerance, the software affine transformation algorithm is used for image correction. Normalization and Enhancement: Grayscale and normalization are performed on each spectral image, and the brightness of the image is adaptively enhanced according to the configuration parameters to improve the stability of subsequent feature extraction.

[0021] 103. Input the preprocessed multispectral image into the preset image fusion model, and output the feature-enhanced image. The feature-enhanced image is fused with multiple anti-counterfeiting information. The purpose of the pre-built image fusion model is to fuse multiple images of the same check with different spectra into a single image. During the fusion process, various anti-counterfeiting information is also concentrated in the image as a feature enhancement image.

[0022] 104. Based on the feature-enhanced image, multi-dimensional anti-counterfeiting features are extracted, and the device status data is analyzed to obtain the anti-counterfeiting results; The specific authentication process is as follows: Figure 2 As shown, the steps are as follows: 201. Based on feature-enhanced images, multi-dimensional anti-counterfeiting features are extracted using fiber feature detection technology, microtext detection technology, and ink and printing feature analysis technology, respectively, to obtain fiber features, microtext features, and ink and printing features. Fiber feature detection technology mainly involves analyzing ultraviolet fluorescence images to extract the location, quantity, size, and color characteristics of colored fibers on a check. Microtext detection technology mainly analyzes high-resolution reflective images to locate and identify microtext features such as "BANK" and "OF" on checks; Ink and printing feature analysis technology mainly utilizes the differences between infrared transmission and reflection images to analyze the ink characteristics (e.g., infrared absorption) of specific areas on a check; and uses ultraviolet images to detect fluorescent reaction areas.

[0023] 202. Match the fiber characteristics with the standard fiber characteristics of the genuine check template to obtain the fiber confidence level; 203. Verify the existence and correctness of the microtext features to obtain the microtext confidence score; 204. The ink properties of the ink printing characteristics are analyzed to obtain the ink printing confidence level; 205. Based on the confidence levels of fiber filaments, microtext, and ink printing, a weighted fusion with preset weights is performed to obtain the counterfeit detection confidence level. Specifically, the confidence level for counterfeit detection The expression is: ; in, 1 represents the confidence level of the fiber filament. For microtext confidence, For ink printing confidence level, , and These are preset weights; 206. Determine if there are any equipment malfunctions based on the equipment status data; Analyze the equipment status data. If abnormal states such as "open cover", "ticket box full" or "motor error" are detected, it is determined that there is an equipment abnormality and proceed to step 208; if there is no equipment abnormality, proceed to step 207. 207. The confidence level of the authentication is used as the authentication result; Confidence level for counterfeit detection It is directly used as the result of authentication; 208. Based on the anomaly level of the equipment anomaly, the confidence level is reduced. The confidence level is then multiplied by the confidence level to obtain the authentication result.

[0024] Different anomaly levels are set for equipment malfunctions. Generally, the greater the impact of equipment malfunction on counterfeit detection, the higher the anomaly level; the smaller the impact of equipment malfunction on counterfeit detection, the lower the anomaly level. The higher the anomaly level, the higher the set signal reduction parameter. The smaller the value, the lower the anomaly level, and the lower the set signal reduction parameter. The larger the value of ; ; Authentication results .

[0025] 105. Locate and extract the ticket number region in the reflected color image to obtain the ticket number image; Locate and crop the serial number area of ​​the check to be authenticated from the reflected color image to obtain the serial number image.

[0026] 106. An adaptive OCR engine is used to recognize the ticket number image to obtain the ticket number recognition result.

[0027] The specific process of adaptive OCR recognition is as follows: Figure 3 As shown, the steps include: 301. An adaptive OCR engine is used to recognize the ticket number image, obtaining the initial recognition result and the corresponding ticket number confidence level. ; 302. Determine whether the confidence level of the ticket number falls within the high confidence interval, medium confidence interval, or low confidence interval. Pre-set high-confidence intervals, medium-confidence intervals, or low-confidence intervals. The high-confidence interval is based on a high-confidence threshold. Division, The specific value can be 0.95; the medium confidence interval is based on... and Division, The value can be 0.7; less than or equal to The interval is considered low confidence. like If so, proceed to step 303; if If so, proceed to step 304; Then proceed to step 305; 303, use the initial recognition result as the ticket number recognition result; 304. After verifying or correcting the initial identification results using multispectral assistance, the ticket number identification result is obtained. Multispectral-assisted discrimination specifically involves extracting features of the same ticket number area under infrared and ultraviolet images. For example, under infrared images, handwritten handwriting may disappear while printed text remains. This characteristic can be used to verify or correct the initial recognition result. After verification or correction, the ticket number recognition result is obtained. 305, query device error code; Retrieve the last device error code returned by the driver layer.

[0028] 306, determine whether the problem is related to image quality based on the device error code; Different device error codes can uniquely point to a specific device problem. For example, device error code "0x7F32" points to "uneven brightness of scanned image", which means there is an image quality problem. If it is an image quality problem, proceed to step 307; if it is not an image quality error, proceed to step 308. 307 triggers the hardware retry mechanism; Hardware retry mechanism: Call the return instruction to eject the check to be authenticated and prompt the user "Image is unclear, please reinsert"; 308. The initial recognition result is determined to be a recognition failure result, and the recognition failure result is used as the ticket number recognition result.

[0029] After outputting the failure results, it is also necessary to log them and prompt for manual intervention.

[0030] The beneficial effects achieved by the embodiments of the present invention are as follows: The process involves acquiring equipment information for a multispectral ticket scanning device and loading corresponding equipment parameters. The device scans the check to be authenticated to obtain a multispectral image, recording its status data during the scanning process. The multispectral image is then preprocessed, including reflected color, reflected infrared, reflected ultraviolet, transmitted infrared, and transmitted ultraviolet images. This preprocessed image is input into a pre-defined image fusion model, which outputs a feature-enhanced image. This feature-enhanced image incorporates various anti-counterfeiting information. Multi-dimensional authentication features are extracted based on the feature-enhanced image, and the authentication result is obtained by analyzing the combined equipment status data. The ticket number region in the reflected color image is located and extracted to obtain the ticket number image. An adaptive OCR engine is used to recognize the ticket number image, yielding the ticket number recognition result. This integrated multi-feature authentication and recognition process improves the accuracy of check authentication and the robustness of ticket number recognition. In the process of multi-feature counterfeit detection, the impact of equipment malfunctions on the detection results is considered by combining equipment status data, which further improves the accuracy of check counterfeit detection. In the adaptive OCR recognition process, it can not only use multispectral images to assist in the discrimination of medium confidence intervals, but also link with the retry mechanism of the device at the underlying level for low confidence intervals. When faced with dirty or wrinkled checks, it can actively trigger a rescan, thereby improving the success rate of check number recognition.

[0031] Based on the check authentication method based on multispectral image fusion and OCR described in the above embodiments, the check authentication system based on multispectral image fusion and OCR will be described below through embodiments.

[0032] like Figure 4 As shown, this embodiment of the invention provides a check authentication system based on multispectral image fusion and OCR, comprising: The device management and parameter configuration module 401 is used to obtain the device information of the multispectral ticket scanning device and load the corresponding device parameters according to the device information. The image acquisition and status monitoring module 402 is used to obtain a multispectral image by scanning the check to be authenticated using a multispectral ticket scanning device, record the device status data during the scanning process of the multispectral ticket scanning device, and preprocess the multispectral image; the multispectral image includes a reflected color image, a reflected infrared image, a reflected ultraviolet image, a transmitted infrared image, and a transmitted ultraviolet image. Image fusion module 403 is used to input the preprocessed multispectral image into a preset image fusion model and output a feature-enhanced image, which fuses multiple anti-counterfeiting information. The anti-counterfeiting module 404 is used to extract multi-dimensional anti-counterfeiting features based on feature-enhanced images, analyze the comprehensive device status data, and obtain the anti-counterfeiting result. The ticket number image extraction module 405 is used to locate and extract the ticket number region in the reflective color image to obtain the ticket number image; The adaptive OCR recognition module 406 is used to recognize the ticket number image using an adaptive OCR engine to obtain the ticket number recognition result.

[0033] Combination Figure 4 The illustrated embodiment represents a preferred implementation of the invention, for example... Figure 5 As shown, the device management and parameter configuration module 401 includes: The device management unit 4011 is used to obtain the device information of the currently connected multispectral ticket scanning device. The device information includes the device model, firmware version number, CIS type and supported scanning light source type. The parameter configuration unit 4012 is used to dynamically load matching device parameters from a preset configuration file according to the device model in the device information, so that the multispectral ticket scanning device can perform image scanning according to the device parameters.

[0034] Combination Figure 4 The illustrated embodiment represents a preferred implementation of the invention, for example... Figure 5 As shown, the image acquisition and status monitoring module 402 includes: The image acquisition unit 4021 is used to perform multispectral scanning on the check to be authenticated using a multispectral ticket scanning device to obtain multiple images of different spectra, which are used as multispectral images. The equipment status monitoring unit 4022 records the equipment status data during the scanning process of the multispectral ticket scanning equipment for counterfeit detection. The image preprocessing unit 4023 is used to perform image registration, real-time bias correction, normalization and enhancement preprocessing on multispectral images.

[0035] Combination Figure 4The illustrated embodiment, a preferred embodiment of the present invention, includes an authentication module 404, specifically configured to extract multi-dimensional authentication features based on a feature-enhanced image using fiber filament feature detection technology, microtext detection technology, and ink and printing feature analysis technology, obtaining fiber filament features, microtext features, and ink printing features; matching the fiber filament features with the standard fiber filament features of a genuine check template to obtain a fiber filament confidence score; verifying the existence and correctness of the microtext features to obtain a microtext confidence score; analyzing the ink characteristics of the ink printing features to obtain an ink printing confidence score; performing a weighted fusion of the fiber filament confidence score, microtext confidence score, and ink printing confidence score with preset weights to obtain an authentication confidence score; determining whether there is an equipment anomaly based on equipment status data; if no equipment anomaly exists, using the authentication confidence score as the authentication result; if an equipment anomaly exists, obtaining a downgrade parameter based on the anomaly level, and multiplying the downgrade parameter by the authentication confidence score to obtain the authentication result.

[0036] Combination Figure 4 The illustrated embodiment, a preferred embodiment of the present invention, includes an adaptive OCR recognition module 406, specifically configured to use an adaptive OCR engine to recognize the ticket number image, obtain an initial recognition result and the corresponding ticket number confidence level; determine whether the ticket number confidence level falls within a high confidence range, a medium confidence range, or a low confidence range; if it falls within a high confidence range, the initial recognition result is used as the ticket number recognition result; if it falls within a medium confidence range, the initial recognition result is verified or corrected using multispectral assistance to obtain the ticket number recognition result; if it falls within a low confidence range, the device error code is queried; the device error code is used to determine whether it is an image quality problem; if it is an image quality problem, a hardware retry mechanism is triggered; if it is not an image quality error, the initial recognition result is determined to be a recognition failure result, and the recognition failure result is used as the ticket number recognition result.

[0037] The beneficial effects achieved by the embodiments of the present invention are as follows: The check authentication system based on multispectral image fusion and OCR integrates multi-feature authentication and recognition processes, improving the accuracy of check authentication and the robustness of check number recognition. In the process of multi-feature counterfeit detection, the impact of equipment malfunctions on the detection results is considered by combining equipment status data, which further improves the accuracy of check counterfeit detection. In the adaptive OCR recognition process, it can not only use multispectral images to assist in the discrimination of medium confidence intervals, but also link with the retry mechanism of the device at the underlying level for low confidence intervals. When faced with dirty or wrinkled checks, it can actively trigger a rescan, thereby improving the success rate of check number recognition.

[0038] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0039] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0040] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0041] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0042] The above are merely embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of the claims of the present invention pending approval.

Claims

1. A check authentication method based on multispectral image fusion and OCR, characterized in that, include: Obtain the device information of the multispectral ticket scanning device, and load the corresponding device parameters based on the device information; The multispectral document scanning device scans the check to be authenticated to obtain a multispectral image, records the device status data during the scanning process, and preprocesses the multispectral image; the multispectral image includes a reflected color image, a reflected infrared image, a reflected ultraviolet image, a transmitted infrared image, and a transmitted ultraviolet image. The preprocessed multispectral image is input into a preset image fusion model, and a feature-enhanced image is output, which fuses multiple anti-counterfeiting information. Based on the enhanced image, multi-dimensional anti-counterfeiting features are extracted, and the device status data is analyzed to obtain the anti-counterfeiting result. The ticket number region in the reflected color image is located and extracted to obtain the ticket number image; An adaptive OCR engine is used to recognize the ticket number image to obtain the ticket number recognition result.

2. The check authentication method based on multispectral image fusion and OCR according to claim 1, characterized in that, The process of acquiring device information for a multispectral ticket scanning device and loading corresponding device parameters based on that device information includes: Obtain the device information of the currently connected multispectral ticket scanning device, including device model, firmware version number, CIS type, and supported scanning light source type; Based on the device model in the device information, matching device parameters are dynamically loaded from a preset configuration file, enabling the multispectral ticket scanning device to perform image scanning according to the device parameters.

3. The check authentication method based on multispectral image fusion and OCR according to claim 1, characterized in that, The process of scanning the check to be authenticated using the multispectral ticket scanning device to obtain a multispectral image, recording the device status data during the scanning process, and preprocessing the multispectral image includes: The multispectral ticket scanning device performs multispectral scanning on the check to be authenticated, obtaining multiple images with different spectra, which are used as multispectral images. Record the device status data during the scanning process of the multispectral ticket scanning device on the check to be authenticated; The multispectral image is preprocessed with image registration, real-time bias correction, normalization, and enhancement.

4. The check authentication method based on multispectral image fusion and OCR according to claim 3, characterized in that, The step of extracting multi-dimensional anti-counterfeiting features based on the enhanced image, and analyzing the device status data to obtain the anti-counterfeiting result includes: Based on the feature-enhanced image, multi-dimensional anti-counterfeiting feature extraction is performed using fiber feature detection technology, microtext detection technology, and ink and printing feature analysis technology to obtain fiber features, microtext features, and ink and printing features. The fiber filament characteristics are matched with the standard fiber filament characteristics of a genuine check template to obtain the fiber filament confidence level. The existence and correctness of the microtext features are verified to obtain the microtext confidence level. The ink properties of the ink printing features are analyzed to obtain the ink printing confidence level; Based on the confidence levels of the fiber filaments, the microtext, and the ink printing, a weighted fusion with preset weights is performed to obtain the counterfeit detection confidence level. Determine whether there is a device malfunction based on the device status data; If there is no equipment malfunction, the aforementioned authentication confidence level will be used as the authentication result. If a device malfunction is detected, a confidence level reduction parameter is obtained based on the malfunction level. The confidence level reduction parameter is then multiplied by the authentication confidence level to obtain the authentication result.

5. The check authentication method based on multispectral image fusion and OCR according to claim 1, characterized in that, The step of using an adaptive OCR engine to recognize the ticket number image and obtaining the ticket number recognition result includes: An adaptive OCR engine is used to recognize the ticket number image to obtain the initial recognition result and the corresponding ticket number confidence level; Determine whether the confidence level of the ticket number falls within the high confidence interval, medium confidence interval, or low confidence interval; If the result is in the high confidence interval, the initial identification result will be used as the ticket number identification result. If the initial identification result is within the medium confidence range, the ticket number identification result is obtained after verifying or correcting the initial identification result using multispectral assistance. If it is in the low confidence range, then query the device error code; Determine whether the problem is an image quality issue based on the device error code. If the problem is with image quality, the hardware retry mechanism will be triggered. If the error is not due to image quality, the initial recognition result is determined to be a recognition failure result, and the recognition failure result is used as the ticket number recognition result.

6. A check authentication system based on multispectral image fusion and OCR, characterized in that, include: The device management and parameter configuration module is used to obtain the device information of the multispectral ticket scanning device and load the corresponding device parameters according to the device information. The image acquisition and status monitoring module is used to obtain a multispectral image by scanning the check to be authenticated using the multispectral ticket scanning device, record the device status data during the scanning process of the multispectral ticket scanning device, and preprocess the multispectral image; the multispectral image includes a reflected color image, a reflected infrared image, a reflected ultraviolet image, a transmitted infrared image, and a transmitted ultraviolet image. The image fusion module is used to input the preprocessed multispectral image into a preset image fusion model and output a feature-enhanced image, wherein the feature-enhanced image fuses multiple anti-counterfeiting information. The anti-counterfeiting module is used to extract multi-dimensional anti-counterfeiting features based on the feature-enhanced image, analyze the device status data, and obtain the anti-counterfeiting result. The ticket number image extraction module is used to locate and extract the ticket number region in the reflected color image to obtain the ticket number image; An adaptive OCR recognition module is used to recognize the ticket number image using an adaptive OCR engine to obtain the ticket number recognition result.

7. The check authentication system based on multispectral image fusion and OCR according to claim 6, characterized in that, The device management and parameter configuration module includes: The device management unit is used to obtain the device information of the currently connected multispectral ticket scanning device, including the device model, firmware version number, CIS type, and supported scanning light source type. The parameter configuration unit is used to dynamically load matching device parameters from a preset configuration file according to the device model in the device information, so that the multispectral ticket scanning device can perform image scanning according to the device parameters.

8. The check authentication system based on multispectral image fusion and OCR according to claim 6, characterized in that, The image acquisition and status monitoring module includes: The image acquisition unit is used to perform multispectral scanning on the check to be authenticated using the multispectral ticket scanning device to obtain multiple images of different spectra, which are used as multispectral images. The equipment status monitoring unit records the equipment status data during the process of the multispectral ticket scanning device scanning the check to be authenticated; The image preprocessing unit is used to perform image registration, real-time bias correction, normalization, and enhancement preprocessing on the multispectral image.

9. The check authentication system based on multispectral image fusion and OCR according to claim 8, characterized in that, The counterfeit detection module is specifically used to extract multi-dimensional counterfeit detection features based on the feature-enhanced image using fiber feature detection technology, microtext detection technology, and ink and printing feature analysis technology, respectively, to obtain fiber features, microtext features, and ink printing features; match the fiber features with the standard fiber features of a genuine check template to obtain fiber confidence; verify the existence and correctness of the microtext features to obtain microtext confidence; and analyze the ink characteristics of the ink printing features to obtain ink printing confidence. Based on the confidence levels of the fiber filaments, the microtext, and the ink printing, a weighted fusion with preset weights is performed to obtain the counterfeit detection confidence level. Determine whether there is a device malfunction based on the device status data; If there is no equipment malfunction, the aforementioned authentication confidence level will be used as the authentication result. If a device malfunction is detected, a confidence level reduction parameter is obtained based on the malfunction level. The confidence level reduction parameter is then multiplied by the authentication confidence level to obtain the authentication result.

10. The check authentication system based on multispectral image fusion and OCR according to claim 6, characterized in that, The adaptive OCR recognition module is used to recognize the ticket number image using an adaptive OCR engine to obtain an initial recognition result and the corresponding ticket number confidence level; determine whether the ticket number confidence level is in a high confidence range, a medium confidence range, or a low confidence range; if it is in a high confidence range, the initial recognition result is used as the ticket number recognition result; if it is in a medium confidence range, the initial recognition result is verified or corrected using multispectral assistance to obtain the ticket number recognition result; if it is in a low confidence range, the device error code is queried; the device error code is used to determine whether it is an image quality problem; if it is an image quality problem, a hardware retry mechanism is triggered; if it is not an image quality error, the initial recognition result is determined to be a recognition failure result, and the recognition failure result is used as the ticket number recognition result.

Images