Multi-modal fusion method and system of fingerprint identification chip

Abstract

The invention relates to the technical field of fingerprint identification, and discloses a multi-modal fusion method and system for a fingerprint identification chip, and the method comprises the steps: collecting target fingerprint information through a multi-fingerprint identification mode, carrying out the authenticity factor analysis of the collected multi-path information, obtaining authenticity verification features, and carrying out the verification of the authenticity. Security verification of a fingerprint disguise mode is carried out based on the verification factor set, security information is obtained, when the security information meets the standard, feature vectors are extracted and fused, and finally a fingerprint identification result is obtained.According to the method, through multi-modal fusion, the accuracy and security of fingerprint identification are improved, fake fingerprint attacks are effectively prevented, and the security of the fingerprint identification is improved. And the robustness and the anti-counterfeiting capability of the system are enhanced, and the problem that fingerprints are easy to counterfeit in a single identification mode in the prior art is solved.

Description

Technical Field

[0001] This invention relates to the field of fingerprint recognition technology, and in particular to a multimodal fusion method and system for a fingerprint recognition chip. Background Technology

[0002] With the development of information technology, fingerprint recognition, as a commonly used biometric technology, is widely used in fields such as identity verification, information security, financial payment, and smart devices. However, traditional fingerprint recognition systems mainly rely on a single fingerprint feature for identification, which is easily affected by factors such as forged fingerprints, damage, and dirt, thus threatening the accuracy and security of the recognition results. Summary of the Invention

[0003] The purpose of this invention is to provide a multimodal fusion method and system for fingerprint recognition chips, aiming to solve the problem that fingerprints are easily forged using a single recognition method in the prior art.

[0004] The present invention is implemented as follows: In a first aspect, the present invention provides a multimodal fusion method for a fingerprint recognition chip, comprising: Information about the target fingerprint is collected through various fingerprint recognition methods to obtain the recognition information of the target fingerprint for each fingerprint recognition method. Based on the identification information described above, the target fingerprint is analyzed for key features to verify its authenticity, so as to obtain the authenticity verification features in each of the identification information descriptions. Based on the authenticity verification features, the security of the target fingerprint is verified for various fingerprint spoofing methods to obtain the security information of each of the identification information. When the security information meets the preset standard, the effective information of each identification information item is extracted and fused to obtain the fingerprint recognition result. In a second aspect, the present invention provides a multimodal fusion system for a fingerprint recognition chip, used to implement the multimodal fusion method for a fingerprint recognition chip as described in any one of the first aspects, comprising: The information acquisition module is used to acquire information from the target fingerprint through various fingerprint recognition methods, so as to obtain the recognition information of the target fingerprint corresponding to each fingerprint recognition method; The factor analysis module is used to analyze the key features of the fingerprint authenticity of the target fingerprint based on each of the identification information to obtain the authenticity verification features in each of the identification information. The security verification module is used to perform security verification on the target fingerprint for various fingerprint spoofing methods based on the authenticity verification features, so as to obtain the security information of each of the identification information. The information fusion module is used to extract and fuse effective information from each of the identification information items when the security information meets the preset standard, so as to obtain the fingerprint recognition result.

[0005] This invention provides a multimodal fusion method for fingerprint recognition chips, which has the following beneficial effects: This invention collects target fingerprint information through multiple fingerprint recognition methods, performs authenticity factor analysis on the collected multi-path information to obtain authenticity verification features, and performs security verification of fingerprint spoofing methods based on the corroboration factor set to obtain security information. When the security information meets the standard, feature vectors are extracted and fused to obtain the final fingerprint recognition result. This method improves the accuracy and security of fingerprint recognition through multimodal fusion, effectively prevents forged fingerprint attacks, and enhances the robustness and anti-counterfeiting capabilities of the system, solving the problem that single recognition methods in existing technologies are prone to fingerprint forgery. Attached Figure Description

[0006] Figure 1 This is a schematic diagram illustrating the steps of a multimodal fusion method for a fingerprint recognition chip provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a multimodal fusion system for a fingerprint recognition chip provided in an embodiment of the present invention. Detailed Implementation

[0007] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0008] The implementation of the present invention will be described in detail below with reference to specific embodiments.

[0009] Reference Figure 1 , Figure 2 The diagram shows a preferred embodiment of the present invention.

[0010] In a first aspect, the present invention provides a multimodal fusion method for a fingerprint recognition chip, comprising: S1: Collect information from the target fingerprint using various fingerprint recognition methods to obtain the recognition information of the target fingerprint corresponding to each fingerprint recognition method; S2: Based on the identification information described in each item, perform key feature analysis on the target fingerprint to obtain the authenticity verification features in each item of the identification information; S3: Based on the authenticity verification features, perform security verification on the target fingerprint for various fingerprint spoofing methods to obtain security information for each of the identification information; S4: When the security information meets the preset standard, the effective information of each identification information is extracted and fused to obtain the fingerprint recognition result.

[0011] Specifically, in step S1 of the embodiment provided by the present invention, multiple fingerprint recognition technologies are selected and configured, such as optical recognition, capacitive recognition, ultrasonic recognition, thermal recognition, and pressure-sensitive recognition. These methods collect fingerprint information through different types of sensors. Using multiple methods to collect information can significantly improve recognition accuracy and avoid the limitations that may occur with a single method (for example, optical recognition is prone to failure in humid environments). Each recognition method has different advantages: optical methods are suitable for providing clear images, capacitive methods are suitable for providing details, ultrasonic methods can penetrate surfaces to obtain more in-depth information, thermal methods can identify through temperature differences, and pressure-sensitive methods can sense changes in touch and pressure. The coordinated work of multiple methods can overcome the defects of a single sensor, enhance the anti-interference ability in the recognition process, ensure that the collection of fingerprint information is more comprehensive and accurate, and improve the fault tolerance of recognition. Even if one method has an error or is invalid, other methods can still provide valid information.

[0012] More specifically, each fingerprint recognition method collects information from the target fingerprint separately, acquiring specific data for each path. Each recognition method uses different sensors and technologies for data collection, ensuring that the feature information of the target fingerprint is captured from different angles. Through parallel operation and independent collection of data from multiple paths, the error rate can be further reduced, avoiding the bias of a single data source. Each path can work independently, reducing recognition time and providing more comprehensive data support for subsequent feature fusion.

[0013] More specifically, by integrating data collected from various identification methods, multiple identification information of the target fingerprint is obtained. This information will be used for subsequent authenticity verification and feature fusion. The identification information provided by each method is an independent interpretation of the target fingerprint, which can be comprehensively judged from multiple dimensions and features, thereby improving the accuracy of fingerprint recognition. The identification information of each method complements each other, which helps to identify genuine fingerprints or determine whether they are counterfeit. After collecting multi-dimensional and multi-level fingerprint data, we can gain a more comprehensive understanding of the characteristic information of the target fingerprint, providing more reliable data support for subsequent authenticity verification, feature extraction and security analysis, and providing a strong foundation for further verification of the authenticity of the fingerprint and the security of the identification information.

[0014] Understandably, through these steps, the pre-deployed multiple fingerprint recognition methods make the fingerprint recognition process more accurate and secure. The multi-path acquisition method ensures the comprehensiveness of fingerprint information, effectively overcoming the limitations of a single path. Multiple recognition methods can provide fingerprint information at different levels, providing diverse data support for subsequent security analysis and spoofing detection. The collaborative work of multiple paths makes the system perform more stably in different environments, especially in complex or interfered environments, providing more reliable recognition results. In general, this multi-path information acquisition method greatly enhances the comprehensive capabilities of the fingerprint recognition system, making it more efficient and accurate in handling complex and ever-changing situations.

[0015] Specifically, in step S2 of the embodiment provided by the present invention, the fingerprint information collected by each recognition method is analyzed in detail, and key feature factors in the data of each path are extracted, such as ridges, sweat gland positions, and the depth and width of fingerprint ridges. The features obtained by each fingerprint recognition technology (such as optical, capacitive, ultrasonic, etc.) are different in terms of accuracy and acquisition dimensions. By analyzing the feature factors of each path in detail, the uniqueness of fingerprints can be understood more comprehensively. By extracting features from multiple dimensions, more evidence can be provided for subsequent authenticity verification. Detailed feature extraction enhances the depth of fingerprint recognition and helps to detect even subtle changes that are not easily revealed when forging fingerprints. This ensures that the information obtained from different technical paths is reliable and provides a more accurate analytical basis for subsequent authenticity verification.

[0016] More specifically, the extracted key factors are matched against a pre-set database of real and spoofed fingerprint features to identify whether a fingerprint is genuine or forged. Comparing known patterns of real and spoofed fingerprint features helps verify whether the target fingerprint conforms to the standards of a genuine fingerprint. Through this comparison, the authenticity of a fingerprint can be quickly identified, and forgery or other fraudulent activities can be promptly determined, achieving rapid and accurate fingerprint authenticity verification. By comparing with real and spoofed patterns, genuine and fake fingerprints can be effectively distinguished, avoiding misidentification, preventing forged fingerprints from deceiving the system, and ensuring the security and reliability of the fingerprint recognition system.

[0017] More specifically, based on the matching results and comparative analysis, fingerprint authenticity verification features are calculated. These factors are a quantitative representation of fingerprint authenticity and can be used for subsequent security verification. By calculating authenticity verification features, the authenticity of fingerprints can be quantified, facilitating further system analysis and decision-making. This factor calculation provides objective data support, avoiding errors that may arise from relying solely on manual judgment. It transforms fingerprint authenticity from qualitative analysis to quantitative assessment, making the system more accurate and automated. Authenticity verification features provide objective evidence for subsequent verification processes, reducing human error and judgment mistakes.

[0018] More specifically, the authenticity verification features obtained from various methods are integrated into a factor set to represent the authenticity level of the target fingerprint. Integrating the authenticity verification features from various methods helps to build a comprehensive fingerprint authenticity profile. The factor set provides a comprehensive authenticity assessment, and the integrated factor set can provide more authoritative and comprehensive data support for subsequent decision-making. The integrated factor set enhances the comprehensive judgment capability of the fingerprint recognition system and provides a more objective basis for further decision-making. By combining factors from different methods, the limitations of relying on a single method are avoided, and the accuracy and security of the recognition results are improved.

[0019] Understandably, through these steps, fingerprint authenticity analysis and the generation of corroborating factors can greatly improve the security and accuracy of fingerprint recognition. The analysis of information from multiple sources can analyze fingerprint features from different levels and dimensions, ensuring higher accuracy. By comparing with real and fake fingerprint features, the system can effectively identify forged fingerprints and avoid fraudulent behavior. The generation and integration of authenticity verification features make fingerprint recognition results more quantitative, improve the automation and intelligence level of the system, and integrate data from multiple sources to avoid errors from a single source, thereby improving the reliability and stability of fingerprint recognition.

[0020] Specifically, in step S3 of the embodiment provided by the present invention, common fingerprint spoofing methods are identified and classified, such as fake fingerprints, copied fingerprints, 3D printed fingerprints, optical simulation fingerprints, etc. Each spoofing method has its specific spoofing characteristics and manifestations. Fingerprint spoofing methods are ever-changing, and different forgery methods exhibit different characteristics in the biometric identification process. Understanding and identifying these spoofing methods helps to design targeted identification strategies. Analyzing the characteristics of spoofing methods can help the system make judgments from multiple dimensions to ensure that forged fingerprints are identified in a timely manner. Through classification analysis, the system can formulate targeted verification strategies for different spoofing methods, improve the accuracy of spoofing identification, enhance the flexibility of the identification system, and enable it to cope with a variety of complex forgery methods.

[0021] More specifically, the previously generated authenticity verification features are compared with the features of the forged fingerprint. By verifying whether each factor conforms to the feature range of a normal fingerprint, the system determines whether the target fingerprint is forged. By comparing and analyzing the features of the real fingerprint and the forged fingerprint, the authenticity verification features can help accurately determine the security of the fingerprint. If the values ​​in the factor set are significantly different from the features of the forged fingerprint, the system can confirm that the fingerprint is forged. This comparison method can effectively avoid misidentification or missed identification, enhancing the security of fingerprint verification. The authenticity verification features provide a quantitative standard that can effectively identify forged fingerprints. Through precise numerical comparison, the system can quickly and accurately identify forged fingerprints, enhancing the security protection capability of the identification system and preventing the risk of breaking through the fingerprint recognition system through simple forgery methods.

[0022] More specifically, the security information obtained from multiple identification methods is analyzed to differentiate the risk levels of different camouflage methods. This includes analyzing which camouflage methods pose a greater security threat to the system and which are easier for the system to identify. Different camouflage methods have different impacts on the identification system, and some camouflage methods are more difficult to identify than others. Therefore, prioritizing and security assessing these camouflage methods can help the system take appropriate measures. The purpose of analyzing security information is to identify potential security risks and provide a basis for system optimization. Through security analysis of different camouflage methods, the system can effectively identify high-risk forged fingerprints, prevent possible security attacks in advance, and provide a security grading scheme, enabling the system to specifically strengthen its ability to identify high-risk camouflage methods.

[0023] More specifically, by integrating information collected from various sources and corroborating factors verified through security checks, the final security information is generated. This information is used to assess the true security of the target fingerprint. Through multi-path analysis and comprehensive factor verification, a comprehensive assessment of the target fingerprint's security can be achieved. The identification information from each path and the analysis results of each factor further enhance the assessment capability of the target fingerprint. This method helps the system determine whether to accept the fingerprint based on the security information during the recognition process. By integrating security information from multiple sources, the system can make a more accurate security assessment, thereby effectively preventing the system from being deceived by forged fingerprints, strengthening the protection capability of the fingerprint recognition system, and improving the overall accuracy and reliability of recognition.

[0024] More specifically, a verification report is generated based on the final security information to provide feedback to the user, explaining the security level of the target fingerprint, whether it is a forged fingerprint, and its corresponding risk level. By generating the report, the security verification results of the fingerprint can be clearly conveyed to the user, and relevant risk assessments can be provided. This helps to further enhance the credibility of the system. Security feedback can help users understand the risks in a timely manner and take appropriate countermeasures, such as re-collecting the fingerprint or taking additional verification steps. The intuitive and clear security assessment report helps users understand the risks of fingerprint recognition and make corresponding decisions, enhancing users' trust in the system and improving the user experience of the fingerprint recognition system.

[0025] Understandably, through these steps, security verification based on authenticity verification features can significantly improve the security protection capabilities of fingerprint recognition systems. By comparing the characteristics of real fingerprints and spoofed fingerprints, the system can effectively identify various spoofing methods, prevent fraud attacks, and the combination of multiple approaches and factors provides a more accurate security assessment, reducing the possibility of false identification and missed identification. The system can dynamically adjust according to the security level of different spoofing methods, thereby strengthening security protection in a targeted manner, generating detailed security verification reports, helping users understand the security status of fingerprint recognition, and improving decision-making efficiency.

[0026] Specifically, in step S4 of the embodiment provided by the present invention, it is checked whether the security information extracted from the multi-path identification information meets the preset standards. These standards include the accuracy and stability of the fingerprint data, as well as whether it meets the authenticity requirements of biometrics. Only when the security information meets the preset standards can the credibility of the fingerprint information be ensured. If the information does not meet the standards, there is a risk of forgery, so it needs to be verified first. This step can ensure that the system has eliminated unreliable fingerprint data before processing feature extraction and fusion, thereby ensuring the effectiveness of identification, improving the security of the system, avoiding subsequent processing of false or inaccurate fingerprint information, thereby improving the accuracy of fingerprint recognition, ensuring that subsequent processing is carried out on the basis of valid and reliable input data, and avoiding unqualified data from affecting subsequent processing steps.

[0027] More specifically, representative feature vectors are extracted from the multi-path identification information that conforms to the standards. These feature vectors are high-dimensional representations of fingerprint data, typically containing information such as the unique geometric shape, texture features, ridge density, and curvature of the fingerprint. Feature vector extraction is a core step in fingerprint recognition, transforming complex fingerprint information into a numerical representation that a computer can process. Through feature vectors, the system can effectively compare and identify different fingerprints. Identification information from different paths provides features of different dimensions. These feature vectors represent the details of the fingerprint and can reflect the individual differences of the fingerprint. The extracted feature vectors are a condensation and simplification of fingerprint data, containing the key information that best characterizes fingerprint identity, improving the accuracy of the identification process. Feature vectors provide a unified data representation for subsequent fusion and comparison, facilitating efficient feature matching and fusion.

[0028] More specifically, feature vectors extracted from multiple approaches are fused to generate a comprehensive feature vector. Common fusion methods include weighted average, principal component analysis (PCA), and support vector machine (SVM). The identification information from multiple approaches comes from different technologies or sensors, and the features extracted from these approaches have a certain degree of complementarity. By fusing these feature vectors, the advantages of different approaches can be combined to enhance the robustness and accuracy of identification. Feature fusion can integrate the advantages of each approach, avoid the error of a certain path from having a significant impact on the identification result, and thus improve the overall performance of the system. The fused feature vector can more comprehensively represent the information of the target fingerprint, improve the stability and accuracy of fingerprint recognition, and enhance the system's adaptability to different environments and conditions by fusing features from different approaches, reducing the identification error caused by the limitations of a single approach.

[0029] More specifically, the fused feature vector is compared with templates in the fingerprint database. By calculating similarity or distance, the final fingerprint recognition result is obtained, i.e., whether the target fingerprint matches a pre-stored template or is identified as a new identity. The final fingerprint recognition result is obtained by comparing the fused feature vector with existing fingerprint templates. This step is the core objective of fingerprint recognition, which is to determine the matching degree between the input fingerprint and the templates in the database. The comparison result provides authentication information to the user and determines whether access or other subsequent operations are allowed. Through feature vector comparison, the system can quickly and accurately determine whether the target fingerprint is a pre-stored fingerprint, thereby obtaining the identity authentication result. Efficient feature comparison algorithms can reduce computational complexity, improve system response speed, and enhance user experience.

[0030] More specifically, based on the final identification results, feedback information and security reports are generated. Feedback information includes the accuracy of fingerprint matching and the confidence level of the identification results; the security report records any anomalies or security issues during the authentication process. The feedback helps users understand the details of the fingerprint recognition process, especially in cases of recognition failure or risk. The report provides detailed security analysis and helps detect and monitor potential security threats, such as fingerprint forgery attacks, providing a basis for subsequent security optimization. The feedback and security reports increase the transparency of the system, enabling users to clearly understand the authentication process and respond to anomalies. They also provide security assessment reports, enhance the system's protection capabilities, and promptly identify potential security risks.

[0031] Understandably, through these steps, the system can extract representative features from information identified through multiple pathways and fuse them to obtain accurate fingerprint recognition results. The extraction and fusion of feature vectors enable the system to more comprehensively represent fingerprint features and improve recognition accuracy. Through feature fusion from multiple pathways, the system can effectively cope with recognition challenges under various environmental conditions and reduce the impact of errors from a single pathway. The simplified representation after feature fusion can speed up the fingerprint recognition process and improve system response speed. By generating result feedback and security reports, the system can provide detailed analysis of the authentication process and enhance security monitoring capabilities.

[0032] This invention provides a multimodal fusion method for fingerprint recognition chips, which has the following beneficial effects: This invention collects target fingerprint information through multiple fingerprint recognition methods, performs authenticity factor analysis on the collected multi-path information to obtain authenticity verification features, and performs security verification of fingerprint spoofing methods based on the corroboration factor set to obtain security information. When the security information meets the standard, feature vectors are extracted and fused to obtain the final fingerprint recognition result. This method improves the accuracy and security of fingerprint recognition through multimodal fusion, effectively prevents forged fingerprint attacks, and enhances the robustness and anti-counterfeiting capabilities of the system, solving the problem that single recognition methods in existing technologies are prone to fingerprint forgery.

[0033] Preferably, the fingerprint recognition method includes optical recognition, capacitive recognition, ultrasonic recognition, thermal recognition, and pressure-sensitive tactile recognition. The optical recognition method collects information through an optical sensor, the capacitive recognition method collects information through a capacitive sensor array, the ultrasonic recognition method collects information through an ultrasonic sensor, the thermal recognition method collects information through a miniature temperature sensor, and the pressure tactile sensor collects information through a pressure sensor array or a piezoelectric film.

[0034] Specifically, fingerprint information is collected through optical sensors. Optical sensors can capture optical images of the fingerprint surface and obtain the texture features of the fingerprint using reflected or scattered light. Optical sensors can quickly acquire fingerprint image data and are suitable for most standard fingerprint recognition applications. Optical recognition has high resolution and clarity, can accurately capture the detailed features of the fingerprint, and provide high-quality fingerprint images, ensuring image clarity and detail, making subsequent feature extraction and comparison more accurate. In well-lit environments, optical sensors provide stable and reliable recognition results, and the technology is mature and relatively low-cost.

[0035] More specifically, fingerprint information is collected through a capacitive sensor array. The capacitive sensor uses the principle of electric field change to sense the capacitance change at different locations on the fingerprint surface, thereby obtaining information such as the fingerprint's geometric shape and ridges. The capacitive sensor has high sensitivity and can capture minute changes in the fingerprint. Compared with optical sensors, the capacitive sensor is not sensitive to the external environment (such as light) and is suitable for working under different lighting conditions. It provides a fine capacitance image of the fingerprint, which can accurately reflect the fine structure of the fingerprint, thereby improving the accuracy of fingerprint recognition and enhancing the system's adaptability. In particular, it can still ensure stable recognition results in low light or strong light environments.

[0036] More specifically, information is acquired through ultrasonic sensors. These sensors emit high-frequency sound waves and receive the echoes reflected from the fingerprint surface. The three-dimensional structure of the fingerprint is identified based on the time and intensity of the echoes. Ultrasonic recognition obtains the fingerprint surface morphology through sound wave propagation and reflection, thus enabling it to identify fingerprint depth information. It is unaffected by external lighting conditions, making it suitable for use in adverse environments such as wet fingers, dirt, or other situations that might affect optical sensor recognition. By acquiring fingerprint depth and three-dimensional data, it improves fingerprint recognition accuracy. Especially when there are stains or dirt on the fingerprint surface, ultrasonic recognition is highly adaptable, unaffected by lighting conditions, and offers greater environmental adaptability and stability.

[0037] More specifically, information is collected through miniature temperature sensors. Thermistors detect the temperature distribution on the fingerprint surface and identify fingerprints by utilizing the differences in thermal conductivity between different areas of the human body. Since the temperature distribution on the fingerprint surface varies slightly from person to person, the thermistor captures these differences and forms a thermal image. During fingerprint acquisition, the temperature information of the finger effectively reflects the subtle differences in the fingerprint, helping to improve recognition accuracy. Thermistor recognition can provide reliable identification under different environmental conditions, especially in low or high ambient temperatures and when optical sensors fail. When combined with other recognition methods, it adds a unique dimension to the system, further improving recognition accuracy.

[0038] More specifically, information is acquired through pressure sensor arrays or piezoelectric films. Pressure tactile sensors obtain detailed fingerprint features by sensing pressure changes at different parts of the fingerprint surface. The fingerprint surface produces different pressure distributions upon contact, and these pressure distributions have individual differences, providing unique information about the fingerprint. Pressure sensor arrays can accurately measure minute pressure changes in the fingerprint, providing additional feature data for fingerprint recognition and offering another way to perceive fingerprint surface features. This can effectively enhance the accuracy of recognition, especially since the minute pressure differences generated when the finger touches the sensor can complement other methods (such as optical and ultrasonic methods) and improve the overall recognition robustness of the system.

[0039] Understandably, multi-path fingerprint recognition systems can combine the advantages of different sensors, improving the accuracy and robustness of fingerprint recognition through complementary cooperation. Each approach has its unique technical characteristics and adaptability to different environments. They complement each other and can provide stable and accurate recognition results in different scenarios. Different recognition methods can provide stable and reliable recognition under various environmental conditions (such as poor lighting, wet fingerprints, etc.). By fusing the recognition information from different sensors, fingerprint features can be more comprehensively characterized, reducing false recognition and missed recognition. Different recognition methods can complement each other, avoiding the impact of damage to a single approach or changes in the environment, thereby improving the overall robustness of the system. The system can better adapt to the fingerprint collection needs of users in various states (such as wet fingers, dry fingers, etc.), ensuring high-quality recognition results in different usage scenarios.

[0040] Preferably, the step of analyzing the key features for fingerprint authenticity of the target fingerprint based on each of the identification information items to obtain the authenticity verification features in each of the identification information items includes: S21: Obtain the identification method corresponding to each of the identification information items, and retrieve the corresponding real information pattern and various disguised information patterns from the preset database according to the identification method corresponding to each of the identification information items; S22: Based on the real information pattern and various disguised information patterns, perform information representation pattern matching processing on the identification information to obtain the real pattern matching features and disguised pattern matching features of the identification information; S23: Based on the real pattern matching features and the various spoofing pattern matching features, perform authenticity feedback semantic parsing on each of the identification information to generate authenticity verification features for each of the identification information.

[0041] Specifically, by decomposing the identification information from different sensors (such as optical, capacitive, ultrasonic, thermal, and pressure-sensitive sensors), the data from each pathway is extracted independently to obtain their respective identification information. Each identification method provides multi-dimensional information, and the decomposed information helps to analyze the data characteristics of each identification method separately. This decomposition facilitates targeted comparison and analysis in subsequent steps, improves the flexibility of data processing, and allows for independent and detailed analysis between different identification methods. It provides a clear and concise data structure for subsequent matching and analysis, making the authenticity judgment more targeted.

[0042] More specifically, based on the identification method corresponding to each identification information, the system retrieves the real information pattern and various disguised information patterns of that identification method from a preset database. The database stores the real fingerprint information model and the corresponding disguised information model for each identification method. When the system obtains identification information from various methods, it retrieves the relevant real and disguised patterns from the database. The preset real and disguised information patterns provide a standard comparison benchmark for the fingerprint recognition system. By comparing with these patterns, the authenticity and forgery of fingerprint information can be effectively distinguished. Through comparison based on standard patterns, a more accurate judgment of fingerprint authenticity can be provided, improving the accuracy of the system and enabling the identification of various disguise methods (such as fake fingerprints, artificial smearing, etc.), thus enhancing the system's anti-deception capability.

[0043] More specifically, based on real information patterns and various disguised information patterns, the identification information undergoes information representation pattern matching processing. The identification information obtained from each identification method is matched with the corresponding real information patterns and disguised information patterns. By comparing their similarity, error, etc., the status characteristics of information pattern matching are generated. Pattern matching can quantify the authenticity of the identification information. By finding similarity between different patterns, it is possible to detect whether the fingerprint matches the expected real fingerprint or whether it may be disguised. Pattern matching processing can improve the identification accuracy and accurately distinguish between different types of disguised information and real information. By quantifying the matching results, it can provide an objective basis for subsequent authenticity judgment and reduce the interference of subjective judgment.

[0044] More specifically, based on the information pattern matching characteristics of real information patterns and the matching characteristics of various camouflage patterns, the identification information of various identification methods through multiple pathways is subjected to semantic analysis for authenticity feedback. Based on the matching results in step 3, the fingerprint information of each identification method is semantically analyzed. Real patterns with good matching are verified and feedback is provided, while camouflage patterns with poor matching are warned or marked. This process also includes semantic analysis, that is, analyzing the relationship between the matching status and fingerprint recognition features. Through semantic feedback analysis, the authenticity of fingerprint information can be further confirmed. For example, if the matching of a certain pathway is similar to the camouflage information, its authenticity is low; otherwise, it is a real fingerprint. This provides more accurate authenticity feedback, helps the identification system to better judge whether the fingerprint is real or not, improves the intelligence of the system, and avoids incorrect authenticity judgments due to false signals or interference signals from a certain pathway.

[0045] More specifically, authenticity verification features are generated for identification information corresponding to various identification methods. These features are then combined to obtain the authenticity verification features. After performing authenticity feedback semantic parsing, authenticity verification features are generated for each identification method. These factors include the matching degree, error, and feedback information of each approach. Finally, these factors are combined into a set of corroborating factors for comprehensive judgment of fingerprint authenticity. Each identification method's fingerprint information has an independent authenticity judgment standard. Combining their corroborating factors can improve the accuracy of the judgment. Through comprehensive analysis of multiple approaches, the authenticity of fingerprints can be evaluated more comprehensively. Through comprehensive feedback of information from multiple approaches, the accuracy and robustness of the system are enhanced, making the final authenticity judgment more reliable. The system has a strong ability to identify various forgery methods, especially under complex camouflage methods. By combining the set of corroborating factors, the system's sensitivity to camouflage can be improved.

[0046] Understandably, through the above steps, the system can comprehensively analyze fingerprint information under multiple identification methods to generate authenticity verification features. This factor set serves as the core basis for judging the authenticity of fingerprints, which can effectively improve the accuracy of fingerprint recognition, anti-counterfeiting capabilities, and adaptability to different environments. The multi-path and multi-dimensional verification mechanism ensures that the system can accurately combat various spoofing and attack methods, providing reliable fingerprint identity verification.

[0047] Preferably, the step of performing authenticity feedback semantic parsing on the identification information of various identification methods for each type of identification information based on the real pattern matching features and various spoofing pattern matching features to generate authenticity verification features corresponding to the identification information of various identification methods includes: S241: Vectorize the real pattern matching features and spoofing pattern matching features of the identification information to obtain the real feature matrix and each spoofing feature matrix of the identification information; S242: Construct an adjacency matrix for the authenticity feature matrix and each disguised feature matrix of the identification information, and perform vector clustering processing on the authenticity feature matrix and each disguised feature matrix based on the adjacency matrix to extract authenticity feature vector clusters and each disguised feature vector cluster; S243: Calculate cluster similarity and cluster separation based on the real feature vector cluster and each disguised feature vector cluster to generate a probability score for each vector cluster, and perform key semantic transformation on each vector cluster based on the probability score of each vector cluster to generate a feedback semantic factor corresponding to each vector cluster. S244: Analyze the conflict relationships between the feedback semantic factors, configure the credibility weights of the feedback semantic factors, perform threshold analysis on the feedback semantic factors with credibility weights, eliminate feedback semantic factors whose credibility weights do not meet the requirements, and combine the remaining feedback semantic factors to generate authenticity verification features.

[0048] Specifically, the matching results of identification information with real information patterns and various masquerading information patterns are transformed into vector representations. Specifically, by calculating the similarity, error, or matching degree of each match, these results are represented in vector form, forming a real feature matrix and a masquerading feature matrix. Vectorization is an effective means of simplifying and structuring complex data, making subsequent analysis more efficient. Through vectorization, the matching features of fingerprints can be converted into numerical values, facilitating subsequent calculations, clustering, and comparisons. The vectorization process makes the data more standardized, which is beneficial for subsequent algorithm processing, ensuring that the feature information of each identification method can be uniformly evaluated, improving data processing efficiency, and providing a foundation for subsequent clustering and similarity calculations.

[0049] More specifically, an adjacency matrix is ​​constructed for the authentic feature matrix and each camouflage feature matrix of the identification information. Based on the adjacency matrix, vector clustering is performed on the authentic and camouflage feature matrices to extract authentic feature vector clusters and camouflage feature vector clusters. An adjacency matrix is ​​constructed, and related feature vectors are connected by calculating the similarity between vectors. Then, vector clustering is performed based on the adjacency matrix. The goal of clustering is to group authentic features and camouflage features separately to form feature clusters (authentic feature clusters and camouflage feature clusters). The adjacency matrix helps to reveal the correlation between features, while the clustering process can group similar features together, making authentic and camouflage features more clearly distinguishable. This step helps to extract reliable fingerprint features. Vector clustering enables the system to distinguish authentic and camouflage features through group features, improving the system's ability to identify the authenticity of fingerprints, optimizing the feature organization structure, and facilitating subsequent calculation of cluster similarity and cluster separation, thereby improving the accuracy and precision of the system.

[0050] More specifically, based on the genuine feature vector clusters and each spoofed feature vector cluster, cluster similarity and cluster separation are calculated to generate a probability score for each vector cluster. Then, based on the probability scores, key semantic transformations are performed on each vector cluster to generate feedback semantic factors for the corresponding vector clusters. The similarity (representing the consistency of features within a cluster) and separation (representing the difference between different clusters) of each cluster are calculated. Based on these calculation results, a probability score for each cluster is generated. Then, semantic transformation is performed on the clusters to extract key semantic information and generate feedback semantic factors. The calculation of cluster similarity and separation is a key indicator for evaluating the quality of each cluster. Through this calculation, the system can identify the most representative and credible feature clusters. These clusters are then transformed into meaningful semantic information, providing support for subsequent trust assessment and analysis. This provides a quantitative assessment of intra-cluster feature consistency and inter-cluster discriminability, enhancing the fingerprint recognition system's identification capabilities. Semantic transformation converts feature data into easily understandable and analyzable semantic information, helping to better understand the authenticity of fingerprints and thus improving the accuracy of decision-making.

[0051] More specifically, conflict relationship analysis is performed on the feedback semantic factors, credible weights are configured for the feedback semantic factors, and threshold analysis is performed on the feedback semantic factors with credible weights to eliminate unqualified feedback semantic factors. Credible feedback semantic factors are retained and combined to generate authenticity verification features. Conflict relationship analysis is performed on each feedback semantic factor to ensure consistency among different semantic factors. Based on the analysis results, credible weights are configured for each feedback semantic factor. Subsequently, threshold analysis is performed on semantic factors with credible weights to eliminate factors that do not meet the standards. Finally, the retained factors are combined to generate authenticity verification features. Conflict relationship analysis and weight configuration help to handle and eliminate contradictions between different feedback semantic factors, ensuring that the finally selected factors are reliable. Threshold analysis can eliminate invalid factors that do not meet the requirements, further improving the accuracy of the analysis. Conflict relationship analysis and weight configuration can effectively improve the credibility of feedback semantic factors, avoiding interference from erroneous or unstable factors in the system's authenticity judgment. Threshold analysis improves the system's anti-interference ability, making the finally generated authenticity verification features more reliable and accurate.

[0052] Understandably, through these steps, the system can not only accurately identify the authenticity of fingerprints, but also intelligently process multiple information sources. It optimizes the results through a series of steps such as clustering, similarity calculation, semantic transformation, and conflict analysis. This comprehensive and multi-dimensional processing method can effectively improve the accuracy and anti-forgery ability of fingerprint recognition, and enhance the robustness and intelligence of the system.

[0053] Preferably, the step of performing security verification on the target fingerprint using various fingerprint spoofing methods based on the authenticity verification features to obtain security information for each of the identification information includes: S31: Perform information correlation analysis on the various identification methods corresponding to each of the identification information to obtain the information verification relationship between the various identification methods of each of the identification information; S32: Based on the information verification relationship between various identification methods, interactive information verification is performed on each authenticity verification feature in the authenticity verification features to optimize each authenticity verification feature into an interactive verification factor. S33: Adjust the parameters of the pre-constructed digital twin model according to each of the interactive verification factors to obtain a fingerprint digital reproduction model, and perform information matching on the fingerprint digital reproduction model based on various fingerprint spoofing methods to obtain the confidence score of the fingerprint digital reproduction model relative to various fingerprint spoofing methods, which serves as the security information of each of the identification information.

[0054] Specifically, information correlation analysis is performed on the various identification methods corresponding to the multi-path identification information to obtain the information verification relationship between the various identification methods of the multi-path identification information. By analyzing the output of different identification methods, the consistency and correlation between them are evaluated. This step constructs an information verification relationship network by calculating the cross information of each identification method, ensuring that the output of each path can mutually verify the output of other paths. The multi-path identification method verifies the same target through multiple independent paths. Information correlation analysis helps to evaluate the reliability and complementarity of each path. Ensuring good information interaction between paths is the key to verifying the authenticity of fingerprints. This step enhances the accuracy and robustness of the multi-path identification method. Through information verification relationships, potential conflicts and consistency between multiple paths can be identified, further improving the credibility of fingerprint identification.

[0055] More specifically, based on the information verification relationships between various identification methods, interactive information verification is performed on each authenticity verification feature to optimize each authenticity verification feature into interactive verification factors. According to the obtained verification relationships, the authenticity verification features are further interactively verified. This process involves mutual verification between multiple factors to ensure that each factor reflects the target fingerprint consistently. Through interactive information verification, the original authenticity verification features are optimized to generate more accurate interactive verification factors. Interactive verification can further verify the independence and consistency of each factor, remove redundant information and possible errors, and improve the accuracy and reliability of the factors. Interactive verification optimizes the quality of each factor, reduces errors and inaccuracies, and through mutual verification, ensures that each factor is more representative and credible, thereby improving the accuracy of the final security verification result.

[0056] More specifically, based on various corroborating factors, a digital reconstruction model of the target fingerprint is constructed to reflect its actual condition. This model applies the corroborating factors to the target fingerprint's real-world context, and uses mathematical models or algorithms to convert the fingerprint's features into a digital reconstruction model. This model accurately reflects the fingerprint's performance and characteristics in real-world scenarios. The digital reconstruction model provides a quantitative basis for fingerprint verification, facilitating simulation, evaluation, and comparison. It exhibits strong adaptability and scalability, and its construction is fundamental to achieving automated verification and evaluation. The digital model can more accurately simulate fingerprint features, facilitating further verification of various fingerprint camouflage methods.

[0057] More specifically, information matching is performed on fingerprint digital reproduction models based on various fingerprint spoofing methods to obtain confidence scores of the fingerprint digital reproduction models relative to various fingerprint spoofing methods. Information matching is performed on fingerprint digital reproduction models using fingerprint spoofing methods to detect the model's performance in spoofing scenarios, and confidence scores are calculated and generated to represent the model's adaptability and stability under different spoofing methods. Matching fingerprint spoofing methods helps to evaluate the security and anti-spoofing ability of the reproduction model. Through confidence scores, the credibility of fingerprints under spoofing can be quantified, and the accuracy of the model under different conditions can be intuitively reflected. Confidence scores provide a quantitative basis for the security analysis of fingerprint spoofing methods, enhance the system's ability to cope with spoofing technologies, and clearly indicate which spoofing methods have a greater impact on the fingerprint recognition system and which have a smaller impact, thereby strengthening protection in a targeted manner.

[0058] More specifically, the confidence score is used as security information for multi-path identification. Finally, based on the confidence score, the final security information is generated. This security information can be used to evaluate the overall security of the target fingerprint under multiple identification methods, serving as the basis for the system to judge the reliability of the target fingerprint. Using the confidence score as security information can provide a comprehensive security assessment for multi-path identification. Through this comprehensive assessment, the system can more accurately identify spoofed fingerprints and take effective countermeasures. The confidence score ultimately provides a security level assessment system, providing an effective basis for protecting against fingerprint spoofing methods and enhancing the overall security of the system.

[0059] Understandably, through the above steps, the system can accurately identify the security of fingerprint spoofing techniques and generate security assessment information based on multiple identification methods. Each step provides comprehensive security verification for fingerprint recognition through information verification, interaction optimization, digital construction, and confidence scoring. Ultimately, the system can generate security information with high reliability and accuracy, providing strong support for the detection and protection against fingerprint spoofing.

[0060] Preferably, the step of extracting and fusing effective information from each of the identification information items to obtain the fingerprint recognition result includes: S41: Vectorize the identification information corresponding to each identification method to obtain the fingerprint identification feature matrix corresponding to the identification information of each identification method; S42: Calculate the importance of the feature vectors of each fingerprint recognition feature matrix to remove low-contribution feature vectors and compress the dimensions of the retained high-contribution feature vectors to generate principal component representative information for various recognition methods. S43: Based on the principal component representative information of various identification methods, feature fusion is performed to directly fuse the consistent content and differentially fuse the inconsistent content of the principal component representative information of various identification methods to obtain the fingerprint recognition result.

[0061] Specifically, the identification information corresponding to each identification method in the multi-path identification information is vectorized to obtain the fingerprint identification feature matrix corresponding to each identification method. The identification information of each identification method is expressed through vectorization methods, such as principal component analysis (PCA) and convolutional neural networks (CNN), to transform the original fingerprint image, signal, or features into a fixed-length feature vector, forming the fingerprint identification feature matrix for each path. Vectorization transforms different forms of fingerprint information into a numerical format that can be processed uniformly. The feature matrix provides standardized input for subsequent calculations and analysis, facilitating feature fusion and dimensionality reduction. Through vectorization, information from different paths is uniformly represented, facilitating subsequent comparison, analysis, and fusion of features from each path. The vectorized feature matrix provides the foundation for feature selection and dimensionality reduction.

[0062] More specifically, the importance of eigenvectors in each fingerprint recognition feature matrix is ​​calculated to remove low-contribution eigenvectors, and the dimensionality of the retained high-contribution eigenvectors is compressed to generate principal component representative information for various recognition methods. Feature selection algorithms (such as information gain, chi-square test, etc.) are used to calculate the importance of eigenvectors in each fingerprint recognition feature matrix, removing low-contribution eigenvectors. For the retained high-contribution eigenvectors, dimensionality reduction algorithms (such as PCA, t-SNE, Autoencoder, etc.) are used for dimensionality compression to generate representative eigenvectors for each path. The importance calculation of eigenvectors helps to identify features that contribute significantly to the fingerprint recognition results, thereby removing redundant or irrelevant features and improving feature effectiveness. Dimensionality compression reduces data complexity while retaining key features, improving computational efficiency. Removing low-contribution eigenvectors and dimensionality compression effectively reduce computational load and improve model accuracy and training speed. By retaining high-contribution features, the key information of each path can be better reflected, improving the accuracy of the recognition results.

[0063] More specifically, feature fusion is performed based on principal component representative information from various recognition methods. This involves direct fusion of consistent content and differential fusion of inconsistent content from the principal component representative information of various recognition methods to obtain fingerprint recognition results. Feature fusion algorithms (such as weighted averaging, concatenation, and concatenation followed by PCA) are used to fuse the principal component representative information from various recognition methods. For consistent content, direct fusion is performed; for inconsistent content, differential fusion methods (such as weighted fusion based on difference metrics) are used. Feature fusion can integrate effective information from different approaches, improving the overall recognition effect. Direct fusion of consistent content strengthens the common features among different approaches, while differential fusion of inconsistent content helps retain the unique information of each approach, further improving recognition accuracy. Through feature fusion, the useful information of each approach can be retained to the maximum extent, improving the robustness and accuracy of the overall recognition. Direct fusion of consistent content strengthens the common features of different approaches, while differential fusion retains the unique information of each approach, effectively improving the model's discriminative ability.

[0064] More specifically, the final fingerprint recognition result is obtained through the feature fusion process described above. This result is obtained by fusing information from various approaches and optimizing consistency and difference features. The final fingerprint recognition result is generated based on the feature fusion of multiple approaches, which can make full use of the advantages of different recognition methods and improve the overall recognition performance. This step ensures the high accuracy and robustness of the recognition result. By fusing feature information from different approaches, the limitations of a single approach can be overcome and the performance of fingerprint recognition in complex environments can be improved.

[0065] Understandably, through the above steps, the system can effectively extract fingerprint recognition features from various identification methods, perform feature selection and compression, and combine the consistency and difference information of each approach to ultimately obtain a highly efficient and accurate fingerprint recognition result that integrates information from multiple approaches. Each step improves the overall performance and robustness of the fingerprint recognition system by reducing redundancy, improving feature quality, and rationally integrating information. This multi-approach feature fusion method not only improves recognition accuracy but also enhances the system's adaptability to different camouflage methods.

[0066] Preferably, the step of fusing the differences in the inconsistencies of principal component representation information from various identification methods includes: S401: Analyze the deviation relationship of the inconsistent content of the principal component representative information of various identification methods to obtain the deviation relationship of the inconsistent content; S402: Based on the information identification pattern of various identification methods, trace the cause of the deviation relationship to obtain the potential cause of the deviation, and correct the erroneous part of the inconsistent content according to the potential cause of the deviation, so as to achieve the difference fusion of the inconsistent content.

[0067] Specifically, the deviation relationship of the inconsistencies in the principal component representative information of various identification methods is analyzed to obtain the deviation relationship of the inconsistencies. The inconsistencies in the feature vectors extracted by multiple re-identification methods are compared, and a deviation matrix or difference tensor is constructed to quantify the feature differences between the various approaches. Euclidean distance, Manhattan distance, cosine similarity and other calculation methods can be used for comparison to determine whether the deviation is numerical, directional or structural, and to clarify which dimensions or combinations of dimensions the differences are reflected in. This is beneficial for source tracing analysis. Quantifying the deviation relationship is the basis for subsequent deviation cause investigation and error correction, enabling the system to automatically identify and locate the conflict points of different identification methods, improve the system's ability to perceive anomalies, and provide structured deviation information input for dynamic fusion strategies.

[0068] More specifically, based on the information recognition patterns of various recognition methods, the deviation relationships are traced to identify potential causes of deviation. Combining the input sources, acquisition methods, and modeling algorithms of each recognition method (e.g., image path recognition vs. capacitance signal recognition vs. 3D reconstruction), possible sources of discrepancies are analyzed. Knowledge graphs, causal modeling, or source-tracing decision trees are used for deviation tracking, distinguishing between deviation types caused by poor sensor quality, external interference, camouflage interference, or model recognition ambiguity. Not all deviations require correction. It is essential to first understand whether the deviation is a reasonably existing difference or a problem caused by recognition errors to prevent miscorrection. This provides a basis for credibility weighting in the fusion strategy, improving the intelligence level of the recognition system, reducing the risk of misjudgment, and enhancing the system's self-explanatory ability.

[0069] More specifically, the system corrects the erroneous parts of the inconsistent content based on the potential causes of deviation. If the source of deviation is identified as an error or interference (such as noise contamination of a certain path), the feature vector of that path is locally corrected or reweighted. Correction methods include: replacing or deleting outliers (based on confidence intervals or Z-scores), using data from other paths for interpolation completion, and performing weighted attenuation processing on the affected features to retain the effective differences of inconsistency. This helps the system learn in multiple dimensions, accurately corrects the source of errors, optimizes the final recognition result while preserving data to the maximum extent, significantly improves the accuracy of multi-path fusion, and enhances the fault tolerance and stability of the recognition system in complex or harsh environments.

[0070] More specifically, the differential fusion of the non-consistent content is achieved. The corrected feature vector is finally integrated through differential fusion algorithms (such as attention mechanism, weighted average, multimodal residual fusion). Unlike the simple splicing or weighting of consistent features, this stage focuses more on the mining and reconstruction of residual information. Differential fusion does not eliminate differences, but selectively retains and strengthens valuable differential information. The final fused fingerprint features are more comprehensive and robust, and can effectively cope with complex scenarios such as camouflage, blurring, and defects, supporting subsequent classifiers or discrimination models to make more stable fingerprint recognition decisions.

[0071] Preferably, when a target fingerprint is detected, detection commands for facial recognition and iris recognition are triggered simultaneously, and the detection commands are sent to the facial recognition module and iris recognition module electrically connected to the fingerprint recognition chip to perform facial recognition and iris recognition.

[0072] Specifically, when a target fingerprint is detected, a detection command for facial and iris recognition is triggered. The system first identifies the target fingerprint through the fingerprint recognition module. Upon detection of the target fingerprint, simultaneous facial and iris recognition detection commands are triggered. This triggering mechanism can be implemented through the system's internal control unit (such as a microcontroller or embedded processing unit). When the fingerprint recognition module successfully identifies the target fingerprint, it issues a command signal. The multimodal recognition system can simultaneously utilize multiple biometrics (fingerprint, face, and iris) to enhance the accuracy and security of recognition. Simultaneous triggering of facial and iris recognition can improve the system's anti-counterfeiting capabilities. Especially when fingerprint recognition is damaged or forged, facial and iris verification can serve as an effective supplement, improving the system's security and accuracy. It avoids the possibility of misidentification or spoofing attacks that may occur when relying on only a single recognition method, and can quickly switch to other verification methods. Even when fingerprint recognition is interfered with, the entire system can still operate efficiently.

[0073] More specifically, detection commands are sent to the face recognition module and the iris recognition module. These commands are sent via electrical connections, and trigger signals are typically transmitted to these two modules using bus communication (such as I2C, SPI, or UART). Upon receiving the detection commands, these modules begin their respective biometric acquisition and analysis processes. The face recognition module begins acquiring facial images, and the iris recognition module begins capturing and processing iris images. Unified control and multi-module collaboration enable simultaneous multimodal biometric tasks, avoiding waiting for individual modules to complete processing and reducing overall response time. Face and iris recognition, as supplements to fingerprint recognition, can improve the multiple confirmation of identification by running simultaneously, reducing the risks caused by single modalities, quickly initiating multi-channel data acquisition, improving recognition efficiency, effectively shortening the overall recognition time, and ensuring improved recognition accuracy, especially in complex environments where face and iris recognition can further verify the accuracy of fingerprints.

[0074] More specifically, in performing facial recognition and iris recognition, after the fingerprint recognition system triggers the detection command, the facial recognition module and the iris recognition module independently perform the recognition tasks: the facial recognition module uses a camera to capture facial images and extracts and matches facial features, while the iris recognition module uses a dedicated iris scanner to capture iris images and analyzes and matches iris patterns. The recognition results of the face and iris are fed back to the central processing unit through the transmission interface for final judgment and result output. The independent execution of their respective recognition tasks ensures that the recognition accuracy of each modality is not affected by multi-task processing. As redundant verification methods, facial recognition and iris recognition can effectively reduce the false recognition rate of any single mode. Facial and iris recognition complement the limitations of fingerprint recognition, enhance the system's comprehensive recognition capabilities, provide more efficient multi-verification, and effectively prevent identity impersonation and forgery attacks.

[0075] More specifically, after each biometric module completes its work, the system synthesizes the final identification result. Then, based on a pre-defined fusion algorithm (such as weighted average, voting mechanism, or weighted sum), the system comprehensively analyzes each identification result. The synthesized result can be used to determine the final identity authentication result through decision tree or multimodal matching algorithms. Employing multiple biometric features simultaneously greatly improves the accuracy and robustness of identity verification, ensuring the system is not affected by errors in a single feature. By fusing multimodal results through algorithms, a more accurate identity verification mechanism can be formed, providing a more secure and reliable identity authentication system. This avoids vulnerabilities that may arise from relying on a single biometric system and significantly enhances the system's anti-counterfeiting capabilities without affecting user experience.

[0076] Reference Figure 2 As shown, in a second aspect, the present invention provides a multimodal fusion system for a fingerprint recognition chip, used to implement the multimodal fusion method for a fingerprint recognition chip as described in any one of the first aspects, comprising: The information acquisition module is used to acquire information from the target fingerprint through various fingerprint recognition methods, so as to obtain the recognition information of the target fingerprint corresponding to each fingerprint recognition method; The factor analysis module is used to analyze the key features of the fingerprint authenticity of the target fingerprint based on each of the identification information to obtain the authenticity verification features in each of the identification information. The security verification module is used to perform security verification on the target fingerprint for various fingerprint spoofing methods based on the authenticity verification features, so as to obtain the security information of each of the identification information. The information fusion module is used to extract and fuse effective information from each of the identification information items when the security information meets the preset standard, so as to obtain the fingerprint recognition result.

[0077] In this embodiment, the specific implementation of each module in the above system embodiment is described in the above method embodiment, and will not be repeated here.

[0078] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A multimodal fusion method for a fingerprint recognition chip, characterized in that, include: Information about the target fingerprint is collected through various fingerprint recognition methods to obtain the recognition information of the target fingerprint for each fingerprint recognition method. Based on the identification information described above, the target fingerprint is analyzed to determine its authenticity features, thereby obtaining the authenticity verification features in each of the identification information items. Based on the authenticity verification features, the security of the target fingerprint is verified for various fingerprint spoofing methods to obtain the security information of each of the identification information. When the security information meets the preset standard, the effective information of each identification information is extracted and fused to obtain the fingerprint recognition result.

2. The multimodal fusion method for fingerprint recognition chips as described in claim 1, characterized in that, The fingerprint recognition methods include optical recognition, capacitive recognition, ultrasonic recognition, thermal recognition, and pressure-sensitive tactile recognition. The optical recognition method collects information through an optical sensor; the capacitive recognition method collects information through a capacitive sensor array; the ultrasonic recognition method collects information through an ultrasonic sensor; the thermal recognition method collects information through a miniature temperature sensor; and the pressure tactile recognition method collects information through a pressure sensor array or a piezoelectric film.

3. The multimodal fusion method for fingerprint recognition chips as described in claim 1, characterized in that, The steps of analyzing the key features for fingerprint authenticity of the target fingerprint based on the identification information to obtain the authenticity verification features in the identification information include: Obtain the identification method corresponding to each of the identification information items, and retrieve the corresponding real information pattern and various disguised information patterns from the preset database according to the identification method corresponding to each of the identification information items; Based on the real information pattern and the various disguised information patterns, the identification information is subjected to information representation pattern matching processing to obtain the real pattern matching features and disguised pattern matching features of the identification information. Based on the real pattern matching features and the various spoofing pattern matching features, the authenticity feedback semantic parsing is performed on each of the identification information to generate authenticity verification features for each identification information.

4. The multimodal fusion method for fingerprint recognition chips as described in claim 3, characterized in that, The steps of performing authenticity feedback semantic parsing on the identification information of each identification information according to the real pattern matching features and the various spoofing pattern matching features to generate authenticity verification features corresponding to the identification information of each identification method include: The real pattern matching features and spoofing pattern matching features of the identification information are vectorized to obtain the real feature matrix and each spoofing feature matrix of the identification information; An adjacency matrix is ​​constructed for the authenticity feature matrix and each disguised feature matrix of the identification information, and vector clustering is performed on the authenticity feature matrix and each disguised feature matrix based on the adjacency matrix to extract the authenticity feature vector cluster and each disguised feature vector cluster. Based on the real feature vector clusters and each disguised feature vector cluster, the cluster similarity and cluster separation are calculated to generate a probability score for each vector cluster. Based on the probability score of each vector cluster, key semantic transformation is performed on each vector cluster to generate a feedback semantic factor for each vector cluster. The conflict relationship analysis between each of the feedback semantic factors is performed to configure the credibility weight of each of the feedback semantic factors, and the threshold analysis is performed on each of the feedback semantic factors with credibility weight to eliminate feedback semantic factors whose credibility weight does not meet the requirements. The remaining feedback semantic factors are then combined to generate authenticity verification features.

5. The multimodal fusion method for fingerprint recognition chips as described in claim 1, characterized in that, The steps of performing security verification on the target fingerprint based on the authenticity verification features to obtain security information for each of the identification information include: An information correlation analysis is performed on the various identification methods corresponding to each of the aforementioned identification information to obtain the information verification relationship between the various identification methods of each of the aforementioned identification information; Based on the information verification relationship between various identification methods, interactive information verification is performed on each authenticity verification feature in the authenticity verification feature to optimize each authenticity verification feature into an interactive verification factor. The parameters of the pre-constructed digital twin model are adjusted according to the interactive verification factors to obtain a fingerprint digital reproduction model. Information matching is performed on the fingerprint digital reproduction model based on various fingerprint spoofing methods to obtain the confidence score of the fingerprint digital reproduction model relative to various fingerprint spoofing methods, which serves as the security information of each of the identification information.

6. The multimodal fusion method for fingerprint recognition chips as described in claim 1, characterized in that, The steps for extracting and fusing effective information from each of the aforementioned identification information to obtain the fingerprint recognition result include: The identification information corresponding to each identification method is vectorized to obtain the fingerprint identification feature matrix corresponding to the identification information of each identification method. The importance of the feature vectors of each fingerprint recognition feature matrix is ​​calculated to remove low-contribution feature vectors, and the dimensionality of the retained high-contribution feature vectors is compressed to generate principal component representative information for various recognition methods. Feature fusion is performed based on principal component representative information from various identification methods to directly fuse consistent content and differentially fuse inconsistent content of principal component representative information from various identification methods, so as to obtain fingerprint recognition results.

7. The multimodal fusion method for fingerprint recognition chips as described in claim 6, characterized in that, The steps for fusing the inconsistencies in principal component representation information from various identification methods include: The deviation relationship of the inconsistent content of the principal component representative information of various identification methods is analyzed to obtain the deviation relationship of the inconsistent content; Based on information identification patterns using various identification methods, the deviation relationships are traced to identify potential causes of deviation. Then, based on these potential causes, errors in the inconsistent content are corrected to achieve the fusion of the differences in the inconsistent content.

8. The multimodal fusion method for fingerprint recognition chips as described in claim 1, characterized in that, When a target fingerprint is detected, detection commands for facial recognition and iris recognition are triggered simultaneously, and the detection commands are sent to the facial recognition module and iris recognition module that are electrically connected to the fingerprint recognition chip to perform facial recognition and iris recognition.

9. A multimodal fusion system for a fingerprint recognition chip, characterized in that, A multimodal fusion method for implementing a fingerprint recognition chip according to any one of claims 1-8 includes: The information acquisition module is used to acquire information from the target fingerprint through various fingerprint recognition methods, so as to obtain the recognition information of the target fingerprint corresponding to each fingerprint recognition method; The factor analysis module is used to analyze the key features of the fingerprint authenticity of the target fingerprint based on each of the identification information to obtain the authenticity verification features in each of the identification information. The security verification module is used to perform security verification on the target fingerprint for various fingerprint spoofing methods based on the authenticity verification features, so as to obtain the security information of each of the identification information. The information fusion module is used to extract and fuse effective information from each of the identification information items when the security information meets the preset standard, so as to obtain the fingerprint recognition result.

Images