Biological feature recognition decision fusion method and device, storage medium and equipment

Abstract

The invention discloses a biological feature recognition decision fusion method and device, a storage medium and equipment, and belongs to the field of biological feature recognition. In the biological feature recognition fusion decision, the accuracy and effectiveness of decision fusion are realized through strategy adjustment according to the basic logic of the hierarchical system and the existing problems. Compared with the prior art, all biological characteristic samples are compared, whether recognition is passed or not can be clearly judged through an exclusion method according to the basic logic of a hierarchical system, and the problem that the recognition result is uncertain due to the comparison sequence is solved. And secondly, the effect of each biological characteristic sample is exerted in a mode of reducing a matching threshold value, so that comparison of each biological characteristic sample contributes to biological characteristic recognition of this time instead of being determined only by the last biological characteristic sample, and the overall passing probability is increased.

Description

Technical Field

[0001] This application relates to the field of biometrics, and in particular to a biometrics recognition decision fusion method, apparatus, storage medium and device. Background Technology

[0002] Biometric technology closely integrates computers with high-tech methods such as optics, acoustics, biosensors, and biostatistics to identify individuals using inherent biological or behavioral characteristics of the human body, such as face, iris, voiceprint, palm print, and palm vein.

[0003] With increasing demands for social security and the accuracy and reliability of identity verification, the limitations of single biometric identification in terms of accuracy and reliability are becoming increasingly prominent, failing to meet the needs of product and technology development. As biometric identification technology continues to mature, the research and application of biometric fusion technology will compensate for the limitations and deficiencies of single biometric identification technologies, further reducing the false recognition rate of biometric identification systems and improving identification accuracy.

[0004] Biometric fusion is the process of using multiple biometric types, sensors, samples, instances, and / or algorithms in a certain combination to obtain a specific biometric recognition result. Biometric fusion has four levels: sample fusion, feature fusion, score fusion, and decision fusion. Decision-level fusion, also known as comparison result fusion, is the final fusion level and is a more practical fusion level. It is performed after feature comparison is completed, during the decision-making process. How to perform decision-level fusion or comparison result fusion remains a topic worthy of further research.

[0005] Hierarchical system decision fusion is a form of decision fusion. Taking two biometric samples P1 and P2 as an example, the two biometric samples are compared sequentially. The matching threshold of subsequent biometric samples is adjusted based on the comparison result of the previous biometric sample, thus completing the decision fusion for all biometric sample comparisons. For example, if the comparison of P1 fails, the matching threshold of P2 is increased (matching difficulty increases); if the comparison of P1 succeeds, the matching threshold of P2 is decreased (matching becomes easier).

[0006] The final matching result of hierarchical system decision fusion is related to the comparison order of biometric samples or the judgment order of the comparison results. Uncertainty in the relevant order may lead to uncertainty in the recognition result. Taking the two biometric samples P1 and P2 mentioned above as examples, the final matching result of comparing P1 first and comparing P2 first is likely to be different, leading to uncertainty in the result. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this application provides a biometric recognition decision fusion method, apparatus, storage medium, and device, which solves the problem of uncertain recognition results caused by comparison order and achieves accuracy in decision fusion.

[0008] The technical solution provided in this application is as follows:

[0009] In a first aspect, this application provides a biometric recognition decision fusion method, the method comprising:

[0010] Multiple biometric samples are compared separately to obtain multiple comparison scores;

[0011] Each of the alignment scores is compared with a benchmark threshold in its corresponding pre-set threshold sequence;

[0012] The threshold sequence includes multiple levels of matching thresholds arranged from low to high according to the difficulty of comparison, and the benchmark threshold is one of the multiple levels of matching thresholds.

[0013] When all comparison scores pass the comparison with their corresponding baseline thresholds, the biometric identification is considered successful.

[0014] When all comparison scores fail to meet their corresponding baseline thresholds, the biometric identification is deemed unsuccessful.

[0015] When some comparison scores pass the comparison with their corresponding baseline thresholds, and some comparison scores fail the comparison with their corresponding baseline thresholds, the matching thresholds corresponding to the comparison scores that pass the comparison are increased by at least one level from the baseline thresholds in the direction of increasing difficulty. When at least one comparison score passes the comparison with the matching thresholds after the increase in difficulty, the biometric identification is deemed successful.

[0016] When some comparison scores pass the comparison with their corresponding baseline thresholds, and some comparison scores fail the comparison with their corresponding baseline thresholds, the matching thresholds corresponding to the comparison scores that fail the comparison are lowered by at least one level from the baseline thresholds in the direction of reducing difficulty. When all comparison scores that fail the comparison pass the comparison with the matching thresholds after the level reduction, the biometric recognition is judged to be successful; otherwise, the biometric recognition is judged to be unsuccessful.

[0017] Furthermore, when some comparison scores pass the comparison with their corresponding benchmark thresholds, and some comparison scores fail, the matching thresholds corresponding to the failed comparison scores are lowered by at least one level from the benchmark thresholds in a less difficult direction. When all failed comparison scores pass the comparison with their lowered matching thresholds, the biometric identification is deemed successful; otherwise, the biometric identification is deemed unsuccessful. This includes:

[0018] When some alignment scores pass the comparison with their corresponding benchmark thresholds, and some alignment scores fail the comparison with their corresponding benchmark thresholds, the number of times the comparisons pass is recorded by variable N.

[0019] The matching threshold corresponding to the comparison score that fails the comparison is reduced by M levels from the benchmark threshold in the direction of reducing difficulty. When all the comparison scores that fail the comparison pass the comparison with the matching threshold after the comparison is reduced by M levels, the biometric recognition is judged to be successful; otherwise, the biometric recognition is judged to be unsuccessful.

[0020] Where M = min(N, k), and k is the total number of matching thresholds that are lower than the baseline threshold.

[0021] Furthermore, the step of reducing the matching threshold corresponding to the failed comparison scores from the baseline threshold by M levels in a direction of decreasing difficulty, and determining that biometric recognition is successful when all failed comparison scores pass the comparison with the matching threshold after the M-level reduction, otherwise determining that biometric recognition fails, includes:

[0022] For any comparison score that fails, the corresponding matching threshold is reduced by M levels from the baseline threshold in the direction of reducing difficulty. The comparison score that fails is then compared with the matching threshold after the comparison is reduced by M levels. If the comparison passes, the value of variable N is incremented by 1; otherwise, the biometric identification is deemed to have failed.

[0023] Return to the step of comparing any failed comparison score, until all failed comparison scores have been compared with the matching threshold after being reduced by M levels. When all failed comparison scores have passed the comparison with the matching threshold after being reduced by M levels, the biometric identification is considered successful.

[0024] Furthermore, the threshold sequence includes at least three levels of matching thresholds arranged from low to high according to the difficulty of comparison, wherein the benchmark threshold is one between the highest and lowest level of matching thresholds.

[0025] Furthermore, each biometric sample comes from a different biometric modality; or all biometric samples come from the same biometric modality; or some biometric samples come from different biometric modalities and some biometric samples come from the same biometric modality.

[0026] Furthermore, the biometric modalities include one or more of fingerprints, finger veins, palm prints, palm veins, faces, irises, voiceprints, handwriting, and gait.

[0027] Secondly, this application provides a biometric recognition decision fusion device, the device comprising:

[0028] The comparison score acquisition module is used to perform biometric comparison on multiple biometric samples and obtain multiple comparison scores.

[0029] The comparison module is used to compare each of the alignment scores with a benchmark threshold in its corresponding pre-set threshold sequence;

[0030] The threshold sequence includes multiple levels of matching thresholds arranged from low to high according to the difficulty of comparison, and the benchmark threshold is one of the multiple levels of matching thresholds.

[0031] The first judgment module is used to determine that biometric identification is successful when all comparison scores pass the comparison with their corresponding benchmark thresholds.

[0032] The second judgment module is used to determine that biometric recognition fails when all comparison scores fail to pass the comparison with their corresponding baseline thresholds.

[0033] The third judgment module is used to increase the matching threshold corresponding to the comparison scores that passed the comparison with the corresponding benchmark threshold by at least one level from the benchmark threshold in the direction of increasing difficulty when some comparison scores passed the comparison with the corresponding benchmark threshold and some comparison scores failed the comparison with the benchmark threshold. When at least one comparison score that passed the comparison with the matching threshold after the level was increased passed the comparison, the biometric recognition is judged to be successful.

[0034] The fourth judgment module is used to reduce the matching threshold corresponding to the unsuccessful comparison scores from the benchmark threshold by at least one level in the direction of reducing difficulty when some comparison scores pass the comparison with their corresponding benchmark thresholds and some comparison scores fail the comparison with their corresponding benchmark thresholds. When all comparison scores that fail the comparison pass the comparison with the matching threshold after the reduction, the biometric recognition is judged to be successful; otherwise, the biometric recognition is judged to be unsuccessful.

[0035] Furthermore, the fourth judgment module includes:

[0036] The cumulative unit is used to record the number of successful comparisons by variable N when some comparison scores pass the comparison with their corresponding benchmark thresholds, while others fail.

[0037] The downgrading unit is used to reduce the matching threshold corresponding to the comparison score that fails the comparison by M levels from the benchmark threshold in the direction of reducing difficulty. When all the comparison scores that fail the comparison pass the comparison with the matching threshold after being reduced by M levels, the biometric recognition is judged to be successful; otherwise, the biometric recognition is judged to be unsuccessful.

[0038] Where M = min(N, k), and k is the total number of matching thresholds that are lower than the baseline threshold.

[0039] Furthermore, the downshifting unit includes:

[0040] The downgrade comparison subunit is used to reduce the matching threshold of any comparison score that fails the comparison by M levels from the benchmark threshold in the direction of reducing difficulty, and compare the comparison score that fails the comparison with the matching threshold after reducing the difficulty by M levels. If the comparison passes, the value of variable N is incremented by 1; otherwise, the biometric recognition is judged to have failed.

[0041] The loop subunit is used to return to the downgrade comparison subunit until all the comparison scores that failed have been compared with the matching threshold after being downgraded by M levels. When all the comparison scores that failed have been compared with the matching threshold after being downgraded by M levels, the biometric recognition is considered successful.

[0042] Furthermore, the threshold sequence includes at least three levels of matching thresholds arranged from low to high according to the difficulty of comparison, wherein the benchmark threshold is one between the highest and lowest level of matching thresholds.

[0043] Furthermore, each biometric sample comes from a different biometric modality; or all biometric samples come from the same biometric modality; or some biometric samples come from different biometric modalities and some biometric samples come from the same biometric modality.

[0044] Furthermore, the biometric modalities include one or more of fingerprints, finger veins, palm prints, palm veins, faces, irises, voiceprints, handwriting, and gait.

[0045] Thirdly, this application provides a computer-readable storage medium for biometric identification decision fusion, including a memory for storing processor-executable instructions, which, when executed by the processor, implement the steps of the biometric identification decision fusion method described in the first aspect.

[0046] Fourthly, this application provides an apparatus for biometric recognition decision fusion, characterized in that it includes at least one processor and a memory storing computer-executable instructions, wherein the processor executes the instructions to implement the steps of the biometric recognition decision fusion method described in the first aspect.

[0047] This application has the following beneficial effects:

[0048] In biometric recognition fusion decision-making, this application, based on the fundamental logic and existing problems of hierarchical systems, achieves accuracy and effectiveness in decision fusion through strategy adjustments. This application compares all biometric samples. According to the basic logic of the hierarchical system, it first uses an elimination method to clearly determine whether the recognition passes or fails, solving the problem of uncertainty in recognition results caused by the comparison order. Secondly, this application lowers the matching threshold to leverage the role of each biometric sample, ensuring that the comparison of each biometric sample contributes to the biometric recognition process, rather than being determined solely by the last biometric sample, thus increasing the overall success rate. Attached Figure Description

[0049] Figure 1 A schematic diagram illustrating a specific example of a hierarchical system decision fusion method in the prior art;

[0050] Figure 2 This is a flowchart illustrating the biometric recognition decision fusion method of this application;

[0051] Figure 3 This is a schematic diagram of the biometric recognition decision fusion device of this application. Detailed Implementation

[0052] To make the technical problems, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with the accompanying drawings and specific embodiments. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0053] Taking two biometric samples P1 and P2 as an example, a specific example of an existing hierarchical system decision fusion method is as follows: Figure 1 As shown.

[0054] from Figure 1 It can be seen that the hierarchical system decision fusion method adjusts the matching threshold of subsequent biometric samples P2 based on the comparison result of the previous biometric sample P1. If the P1 comparison fails, the matching threshold of P2 is increased (matching difficulty increases); if the P1 comparison succeeds, the matching threshold of P2 is decreased (matching becomes easier). This method leads to the following problems:

[0055] 1. The matching result is related to the comparison order of P1 and P2. For example, if the first comparison of P1 fails, the matching threshold of P2 will be increased, making it more difficult for P2 to match. Ultimately, the comparison of P2 will also not exceed the threshold, so the final matching result will be unsuccessful.

[0056] 2. The final sample comparison determines whether the matching process passes. Therefore, it is essential to ensure the final comparison passes in order to better utilize the hierarchical system; otherwise, all efforts will be in vain.

[0057] The aforementioned problem arises from the comparison order of samples P1 and P2, or in other words, from the judgment order of the comparison results. Uncertainty in this order can lead to uncertainty in the identification results. Therefore, the solution in this application is to transform uncertainty into certainty, following the basic logic of a hierarchical system.

[0058] To address the aforementioned problems, embodiments of this application provide a biometric recognition decision fusion method, such as... Figure 2 As shown, the method includes:

[0059] S100: Perform biometric comparison on multiple biometric samples to obtain multiple comparison scores.

[0060] Each biometric sample in the plurality of biometric samples described in this application may originate from a different biometric modality, i.e., multimodal biometric recognition. Alternatively, all biometric samples may originate from the same biometric modality, i.e., multi-instance biometric recognition of the same modality. Or, some biometric samples may originate from different biometric modalities, while others may originate from the same biometric modality, i.e., a combination of multimodal biometric recognition and multi-instance biometric recognition.

[0061] For example, multiple biometric samples are represented by P1, P2, P3... P1, P2, P3... are compared with their respective feature templates, and their respective comparison scores are output for subsequent decision-making.

[0062] This application compares multiple biometric samples at the beginning because each sample in the hierarchical system needs to be compared. Therefore, unlike traditional hierarchical systems, the comparisons are not performed sequentially, but are completed all at the beginning.

[0063] S200: Compare each alignment score with the baseline threshold in its corresponding pre-set threshold sequence.

[0064] The threshold sequence includes multiple levels of matching thresholds arranged from low to high according to the difficulty of comparison, and the benchmark threshold is one of the multiple levels of matching thresholds.

[0065] The matching thresholds of the multiple levels arranged from low to high mentioned in this application do not refer to a simple numerical progression from small to large, but rather to a progression from low to high difficulty in passing the comparison.

[0066] In some examples, the higher the comparison score, the higher the similarity between the two biometric samples. In this case, the higher the matching threshold, the more difficult it is to pass the comparison. The multiple matching thresholds in the threshold sequence are arranged from low to high in ascending order of value.

[0067] In other examples, the higher the comparison score, the lower the similarity between the two biometric samples. In this case, the lower the matching threshold, the more difficult it is to pass the comparison. The multiple matching thresholds in the threshold sequence are arranged from low to high in descending order of value.

[0068] The baseline threshold, also known as the standard threshold, is a commonly used threshold in biometric matching, and it is generally located in the middle of the threshold sequence. In one example, the threshold sequence includes at least three matching thresholds arranged from low to high difficulty of comparison, with the baseline threshold being one of the highest and lowest matching thresholds, i.e., a middle one.

[0069] S300: When all comparison scores pass the comparison with their corresponding baseline thresholds, the biometric identification is deemed successful.

[0070] In this application, for cases where a higher alignment score indicates a higher similarity between two biometric samples, the comparison passes if the alignment score is greater than a baseline threshold; otherwise, the comparison fails. Conversely, for cases where a lower alignment score indicates a higher similarity between two biometric samples, the comparison passes if the alignment score is less than a baseline threshold; otherwise, the comparison fails.

[0071] S400: When all comparison scores fail to pass the comparison with their corresponding baseline thresholds, the biometric identification is deemed unsuccessful.

[0072] Based on the basic logic of the hierarchical system, this application first uses the elimination method to clearly determine whether the identification passes (i.e., when all comparisons pass) or fails (i.e., when all comparisons fail), thus solving the problem of uncertain identification results caused by the comparison order.

[0073] S500: When some comparison scores pass the comparison with their corresponding baseline threshold (the biometric samples that pass the comparison are denoted as Px), and some comparison scores fail the comparison with their corresponding baseline threshold (the biometric samples that pass the comparison are denoted as Py), the matching threshold corresponding to the comparison scores that pass the comparison is increased by at least one level from the baseline threshold in the direction of increasing difficulty. When at least one comparison score that passes the comparison passes the comparison with the matching threshold after the level increase, the biometric recognition is judged to be successful.

[0074] In this application, for any Px that passes the comparison, if even one of its biometric samples can still pass the comparison after increasing the matching threshold, then the biometric identification is successful. That is, according to the logical rules of the hierarchical system, the comparison result of Px is placed last for judgment, and the final result will definitely be successful.

[0075] S600: When some comparison scores pass the comparison with their corresponding baseline thresholds, and some comparison scores fail the comparison with their corresponding baseline thresholds, the matching thresholds corresponding to the comparison scores that fail the comparison are lowered by at least one level from the baseline thresholds in the direction of reducing difficulty. When all comparison scores that fail the comparison pass the comparison with the matching thresholds after the reduction, the biometric recognition is judged to be successful; otherwise, the biometric recognition is judged to be unsuccessful.

[0076] In this application, for all biometric samples Py that fail the comparison, if the matching threshold is lowered, the comparison can be passed, and the biometric identification is successful. That is, the biometric samples that pass the comparison are processed first. Since the previous comparisons passed, the matching threshold for subsequent biometric samples can be lowered. In this way, the comparisons of subsequent biometric samples can all pass, and the final result is also successful.

[0077] Finally, for all biometric samples that fail the comparison, if any one still fails the comparison even after lowering the matching threshold, then according to basic logic, the biometric identification is unsuccessful.

[0078] In biometric recognition fusion decision-making, this application, based on the fundamental logic and existing problems of hierarchical systems, achieves accuracy and effectiveness in decision fusion through strategy adjustments. This application compares all biometric samples. According to the basic logic of the hierarchical system, it first uses an elimination method to clearly determine whether the recognition passes or fails, solving the problem of uncertainty in recognition results caused by the comparison order. Secondly, this application lowers the matching threshold to leverage the role of each biometric sample, ensuring that the comparison of each biometric sample contributes to the biometric recognition process, rather than being determined solely by the last biometric sample, thus increasing the overall success rate.

[0079] As an improvement to the embodiments of this application, the aforementioned S600 includes:

[0080] S610: When some alignment scores pass the comparison with their corresponding baseline thresholds, and some alignment scores fail the comparison with their corresponding baseline thresholds, the number of times the comparisons pass is recorded by variable N.

[0081] S620: Reduce the matching threshold corresponding to the comparison score that fails the comparison by M levels from the baseline threshold in the direction of reducing difficulty. When all the comparison scores that fail the comparison pass the comparison with the matching threshold after the comparison is reduced by M levels, the biometric recognition is judged to be successful; otherwise, the biometric recognition is judged to be unsuccessful.

[0082] This application uses a variable to record the number of successful comparisons. Each successful comparison increments the variable by 1. Each 1 point lowers the matching threshold by one level, and 2 points can lower the matching threshold by two levels consecutively. However, the matching threshold must be kept within the minimum safe threshold range.

[0083] For example, if both P1 and P2 pass the comparison, the variable is 2, which allows for continuously lowering the matching threshold of two levels, P3.

[0084] In some examples, the total number of matching thresholds (referred to as low thresholds) that are lower than the baseline threshold may be less than N. For example, when N=2, it is necessary to lower the matching thresholds by two levels, but there may only be one low threshold. In this case, it is difficult to lower the thresholds by two levels. Therefore, M=min(N,k) levels are lowered, where k is the total number of low thresholds.

[0085] Specifically, the S620 includes:

[0086] S621: For any comparison score that fails, reduce its corresponding matching threshold by M levels from the baseline threshold in the direction of reducing difficulty, and compare the comparison score that fails with the matching threshold after reducing the difficulty by M levels. If the comparison passes, the value of variable N is incremented by 1; otherwise, the biometric recognition is judged to have failed.

[0087] S622: Return to step S621 until all failed comparison scores have been compared with the matching threshold after being reduced by M levels. When all failed comparison scores have passed the comparison with the matching threshold after being reduced by M levels, the biometric identification is considered successful.

[0088] For example: if P1 passes the comparison and P2 passes the comparison after the matching threshold is lowered, then the variable is 2, which means that the matching threshold of P3 can be lowered continuously for two levels.

[0089] After multiple adjustments to the matching threshold, the probability of passing the comparison of relevant biometric samples increased significantly. This demonstrates that the more biometric samples involved in the comparison, the greater the overall chance of passing the matching, thus overcoming the problem that the success or failure was determined by the comparison of the last biometric sample.

[0090] The aforementioned biometric modalities apply to all biometric modalities, such as fingerprints (i.e., fingerprint recognition, others are similar), finger veins, palm prints, palm veins, faces, irises, voiceprints, handwriting, gait, hand shape contours, DNA, iris prints, periorbital prints, and lip prints, among one or more of these.

[0091] The following two specific experimental examples illustrate this application in detail.

[0092] Example 1: Fingerprint and finger vein decision fusion

[0093] Preparation process:

[0094] The fingerprint template and finger vein template have been registered.

[0095] The fingerprint matching threshold settings are: 30, 35, 40, 45, and 50. Among them, 40 is the baseline threshold. The larger the matching threshold, the stricter the matching requirements.

[0096] The matching threshold settings for finger veins are: 62, 65, 70, 75, and 80. Among them, 70 is the baseline threshold. The larger the baseline threshold, the stricter the matching requirements.

[0097] Biometric recognition decision fusion process:

[0098] 1) Obtain fingerprint and finger vein samples, which can be obtained simultaneously (e.g., through a fingerprint and finger vein scanner) or sequentially.

[0099] 2) After sample processing, fingerprint features and finger vein features are extracted. The fingerprint features are compared with the fingerprint template, and the comparison score is S1. The finger vein features are compared with the finger vein template, and the comparison score is S2.

[0100] 3) Based on the judgment approach of this application, the judgment can be made according to the following situations:

[0101] Compare the size of S1 with its baseline threshold (40), and compare the size of S2 with its baseline threshold (70).

[0102] a) If S1 < 40 and S2 < 70, that is, S1 and S2 do not exceed their baseline thresholds, then the biometric identification will fail.

[0103] b) If S1≥40 and S2≥70, that is, both S1 and S2 reach their baseline thresholds, then the biometric identification is successful.

[0104] c) If S1≥45 or S2≥75, the biometric identification is successful. That is, as long as one of them reaches the higher matching threshold, the biometric identification is considered successful.

[0105] d) If S1<35 or S2<65, the biometric identification fails. That is, as long as one of the comparison scores does not reach the matching threshold of the next lower level, the biometric identification will be judged as failing.

[0106] e) If S1≥40 and 70>S2≥65 or S2≥70 and 40>S1≥35, then the biometric identification is successful. That is, if one threshold reaches the baseline threshold and the other does not reach the baseline threshold but reaches the matching threshold of the next lower level, the biometric identification is considered successful.

[0107] In this example, based on the above judgment approach, the elimination method can be used to clearly determine whether the biometric recognition passes or fails. Then, if a certain biometric sample passes the comparison, the matching is downgraded by lowering the threshold.

[0108] Example 2: Face, iris, and voiceprint decision fusion

[0109] Preparation process:

[0110] Face templates, iris templates, and voiceprint templates have been registered.

[0111] The face matching threshold settings are: 70, 75, 80, 85, and 95. Among them, 80 is the baseline threshold. The higher the matching threshold, the stricter the matching requirements.

[0112] The matching threshold settings for iris scans are: 60, 70, 75, 85, and 90. Among them, 75 is the baseline threshold. The higher the matching threshold, the stricter the matching requirements.

[0113] The voiceprint matching threshold settings are: 60, 65, 70, 75, and 80. Among them, 70 is the baseline threshold. The larger the matching threshold, the stricter the matching requirements.

[0114] Biometric recognition decision fusion process:

[0115] 1) Acquire face, iris, and voiceprint samples. These can be acquired simultaneously (e.g., through a face and iris recognition all-in-one machine with a microphone) or sequentially.

[0116] 2) After sample processing and feature extraction, the facial features are compared with the facial template, and the comparison score is S1. The iris features are compared with the iris template, and the comparison score is S2. The voiceprint features are compared with the voiceprint template, and the comparison score is S3.

[0117] 3) Based on the judgment logic of this invention, the judgment can be made according to the following situations:

[0118] Compare the size of S1 with its baseline threshold (80), compare the size of S2 with its baseline threshold (75), and compare the size of S3 with its baseline threshold (70).

[0119] a) If S1 < 80, S2 < 75, and S3 < 70, then the biometric identification will fail.

[0120] b) If S1≥80 and S2≥75 and S3≥70, then the biometric identification is successful.

[0121] c) If S1≥85 or S2≥85 or S3≥75, then the biometric identification is successful.

[0122] d) Of S1, S2, and S3, two reach their baseline threshold, while the other does not. Thus, variable N = 2, and the matching threshold for the biometric sample that does not reach its baseline threshold can be lowered by two levels.

[0123] For example: if S1≥80 and S2≥75, and 70>S3≥60, then S3 can be considered to have passed the comparison, and the biometric identification is successful.

[0124] e) If one of S1, S2, and S3 reaches its baseline threshold, variable N = 1, while the other two do not reach their baseline thresholds.

[0125] For example, if S1 ≥ 80 and 70 ≤ S2 < 75, by using variable N = 1, the matching threshold of S2 can be lowered by one level, and the S2 comparison can be determined to be successful. At this time, by using variable N = 2, the matching threshold of S3 can be lowered by two levels. As long as 60 ≤ S3 < 70, the S3 comparison can be determined to be successful, and the biometric identification will be successful.

[0126] Based on the above judgment approach, the elimination method can be used to clearly determine whether biometric identification passes or fails. Then, if a certain biometric sample passes the comparison, the matching is downgraded based on the accumulated variable N and the method of lowering the threshold.

[0127] This application also provides a biometric recognition decision fusion device, such as... Figure 3 As shown, the device includes:

[0128] The comparison score acquisition module 100 is used to perform biometric comparison on multiple biometric samples to obtain multiple comparison scores.

[0129] The comparison module 200 is used to compare each comparison score with a benchmark threshold in its corresponding pre-set threshold sequence.

[0130] The threshold sequence includes multiple levels of matching thresholds arranged from low to high according to the difficulty of comparison, and the benchmark threshold is one of the multiple levels of matching thresholds.

[0131] The first judgment module 300 is used to determine that biometric identification is successful when all comparison scores pass the comparison with their corresponding benchmark thresholds.

[0132] The second judgment module 400 is used to determine that biometric recognition fails when all comparison scores fail to pass the comparison with their corresponding baseline thresholds.

[0133] The third judgment module 500 is used to increase the matching threshold corresponding to the comparison scores that passed the comparison with the corresponding benchmark threshold by at least one level from the benchmark threshold to the direction of increasing difficulty when some comparison scores passed the comparison with the corresponding benchmark threshold, and when at least one comparison score that passed the comparison with the matching threshold after the increase in difficulty passed the comparison with the benchmark threshold, the biometric recognition is judged to be successful.

[0134] The fourth judgment module 600 is used to reduce the matching threshold corresponding to the comparison score that failed to pass the comparison with the corresponding benchmark threshold by at least one level from the benchmark threshold in the direction of reducing difficulty when some comparison scores pass the comparison with the benchmark threshold. When all comparison scores that failed to pass the comparison with the matching threshold after the reduction are passed, the biometric recognition is judged to be successful; otherwise, the biometric recognition is judged to be unsuccessful.

[0135] In biometric recognition fusion decision-making, this application, based on the fundamental logic and existing problems of hierarchical systems, achieves accuracy and effectiveness in decision fusion through strategy adjustments. This application compares all biometric samples. According to the basic logic of the hierarchical system, it first uses an elimination method to clearly determine whether the recognition passes or fails, solving the problem of uncertainty in recognition results caused by the comparison order. Secondly, this application lowers the matching threshold to leverage the role of each biometric sample, ensuring that the comparison of each biometric sample contributes to the biometric recognition process, rather than being determined solely by the last biometric sample, thus increasing the overall success rate.

[0136] As an improvement to the embodiments of this application, the aforementioned fourth determination module includes:

[0137] The cumulative unit is used to record the number of successful comparisons by variable N when some comparison scores pass the comparison with their corresponding benchmark thresholds, while others fail.

[0138] The downgrading unit is used to reduce the matching threshold corresponding to the comparison score that fails the comparison by M levels from the baseline threshold in order to reduce the difficulty. When all the comparison scores that fail the comparison pass the comparison with the matching threshold after being reduced by M levels, the biometric recognition is judged to be successful; otherwise, the biometric recognition is judged to be unsuccessful.

[0139] Specifically, the downshifting unit includes:

[0140] The downgrading comparison subunit is used to reduce the matching threshold of any failed comparison score by M levels from the baseline threshold, and then compare the failed comparison score with the reduced matching threshold. If the comparison passes, the value of variable N is incremented by 1; otherwise, the biometric recognition is considered failed.

[0141] The loop subunit is used to return to the downgrade comparison subunit until all the comparison scores that failed have been compared with the matching threshold after being downgraded by M levels. When all the comparison scores that failed have passed the comparison with the matching threshold after being downgraded by M levels, the biometric recognition is considered successful.

[0142] As an example, the threshold sequence includes at least three levels of matching thresholds arranged from low to high difficulty of comparison, with the baseline threshold being one between the highest and lowest level of matching thresholds.

[0143] In this application, each biometric sample comes from a different biometric modality; or all biometric samples come from the same biometric modality; or some biometric samples come from different biometric modalities and some biometric samples come from the same biometric modality.

[0144] The aforementioned biometric modalities include one or more of fingerprints, finger veins, palm prints, palm veins, faces, irises, voiceprints, handwriting, and gait.

[0145] The apparatus provided in the above embodiments corresponds one-to-one with the embodiments of the aforementioned methods in terms of its implementation principle and the resulting technical effects. For the sake of brevity, any parts of the apparatus not mentioned in the embodiments can be referred to the corresponding content in the embodiments of the aforementioned methods. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the modules and units described in this apparatus can all be referred to the corresponding processes in the embodiments of the aforementioned methods, and will not be repeated here.

[0146] The biometric recognition decision fusion method described in the above embodiments of this application can implement business logic through a computer program and record it on a storage medium. This storage medium can be read and executed by a computer, achieving the effects of the scheme described in the method embodiments of this specification. Therefore, embodiments of this application also provide a computer-readable storage medium for biometric recognition decision fusion, including a memory for storing processor-executable instructions. When these instructions are executed by a processor, they implement the steps of biometric recognition decision fusion as described in the foregoing embodiments.

[0147] The storage medium may include a physical device for storing information, typically digitizing the information and then storing it using electrical, magnetic, or optical methods. The storage medium may include: devices that store information using electrical energy, such as various types of memory, like RAM and ROM; devices that store information using magnetic energy, such as hard disks, floppy disks, magnetic tapes, magnetic core memory, bubble memory, and USB flash drives; and devices that store information using optical methods, such as CDs or DVDs. Of course, there are other readable storage media, such as quantum memories and graphene memories.

[0148] The storage medium described above may also include other implementation methods according to the description of the method embodiments. The implementation principle and technical effects of this embodiment are the same as those of the foregoing method embodiments. For details, please refer to the description of the relevant method embodiments, which will not be repeated here.

[0149] This application also provides an apparatus for biometric recognition decision fusion. The apparatus may be a standalone computer, or it may include an actual operating device that uses one or more of the methods or embodiments described in this specification. The biometric recognition decision fusion apparatus may include at least one processor and a memory storing computer-executable instructions. When the processor executes the instructions, it implements the steps of any one or more of the biometric recognition decision fusion methods described above.

[0150] The device described above may also include other implementation methods according to the method embodiments. The implementation principle and technical effects of this embodiment are the same as those of the foregoing method embodiments. For details, please refer to the description of the relevant method embodiments, which will not be repeated here.

[0151] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The protection scope of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application. All should be covered within the protection scope of this application. Therefore, the protection scope of this application should be determined by the protection scope of the claims.

Claims

1. A biometric recognition and decision fusion method, characterized in that, The method includes: Multiple biometric samples are compared separately to obtain multiple comparison scores; Each of the alignment scores is compared with a benchmark threshold in its corresponding pre-set threshold sequence; The threshold sequence includes multiple levels of matching thresholds arranged from low to high according to the difficulty of comparison, and the benchmark threshold is one of the multiple levels of matching thresholds. When all comparison scores pass the comparison with their corresponding baseline thresholds, the biometric identification is considered successful. When all comparison scores fail to meet their corresponding baseline thresholds, the biometric identification is deemed unsuccessful. When some comparison scores pass the comparison with their corresponding baseline thresholds, and some comparison scores fail the comparison with their corresponding baseline thresholds, the matching thresholds corresponding to the comparison scores that pass the comparison are increased by at least one level from the baseline thresholds in the direction of increasing difficulty. When at least one comparison score passes the comparison with the matching thresholds after the increase in difficulty, the biometric identification is deemed successful. When some comparison scores pass the comparison with their corresponding baseline thresholds, and some comparison scores fail the comparison with their corresponding baseline thresholds, the matching thresholds corresponding to the comparison scores that fail the comparison are lowered by at least one level from the baseline thresholds in the direction of reducing difficulty. When all comparison scores that fail the comparison pass the comparison with the matching thresholds after the level reduction, the biometric recognition is judged to be successful; otherwise, the biometric recognition is judged to be unsuccessful.

2. The biometric recognition decision fusion method according to claim 1, characterized in that, When some comparison scores pass the comparison with their corresponding benchmark thresholds, while others fail, the matching thresholds corresponding to the failing comparison scores are lowered by at least one level from the benchmark thresholds in a less difficult direction. If all failing comparison scores pass the comparison with their lowered matching thresholds, the biometric identification is considered successful; otherwise, the biometric identification is considered unsuccessful. This includes: When some alignment scores pass the comparison with their corresponding benchmark thresholds, and some alignment scores fail the comparison with their corresponding benchmark thresholds, the number of times the comparisons pass is recorded by variable N. The matching threshold corresponding to the comparison score that fails the comparison is reduced by M levels from the benchmark threshold in the direction of reducing difficulty. When all the comparison scores that fail the comparison pass the comparison with the matching threshold after the comparison is reduced by M levels, the biometric recognition is judged to be successful; otherwise, the biometric recognition is judged to be unsuccessful. Where M = min(N, k), and k is the total number of matching thresholds that are lower than the baseline threshold.

3. The biometric recognition decision fusion method according to claim 2, characterized in that, The step of lowering the matching threshold corresponding to the failed comparison scores from the baseline threshold by M levels in a direction of reducing difficulty, and determining that biometric recognition is successful when all failed comparison scores pass the comparison with the matching threshold after the M-level reduction, otherwise determining that biometric recognition fails, includes: For any comparison score that fails, the corresponding matching threshold is reduced by M levels from the baseline threshold in the direction of reducing difficulty. The comparison score that fails is then compared with the matching threshold after the comparison is reduced by M levels. If the comparison passes, the value of variable N is incremented by 1; otherwise, the biometric identification is deemed to have failed. Return to the step of comparing any failed comparison score, until all failed comparison scores have been compared with the matching threshold after being reduced by M levels. When all failed comparison scores have passed the comparison with the matching threshold after being reduced by M levels, the biometric identification is considered successful.

4. The biometric recognition decision fusion method according to claim 1, characterized in that, The threshold sequence includes at least three levels of matching thresholds arranged from low to high difficulty of comparison, with the baseline threshold being one between the highest and lowest level of matching thresholds.

5. The biometric recognition decision fusion method according to any one of claims 1-4, characterized in that, Each biometric sample comes from a different biometric modality; or all biometric samples come from the same biometric modality; or some biometric samples come from different biometric modalities and some biometric samples come from the same biometric modality.

6. A biometric recognition and decision fusion device, characterized in that, The device includes: The comparison score acquisition module is used to perform biometric comparison on multiple biometric samples and obtain multiple comparison scores. The comparison module is used to compare each of the alignment scores with a benchmark threshold in its corresponding pre-set threshold sequence; The threshold sequence includes multiple levels of matching thresholds arranged from low to high according to the difficulty of comparison, and the benchmark threshold is one of the multiple levels of matching thresholds. The first judgment module is used to determine that biometric identification is successful when all comparison scores pass the comparison with their corresponding benchmark thresholds. The second judgment module is used to determine that biometric recognition fails when all comparison scores fail to pass the comparison with their corresponding baseline thresholds. The third judgment module is used to increase the matching threshold corresponding to the comparison scores that passed the comparison with the corresponding benchmark threshold by at least one level from the benchmark threshold in the direction of increasing difficulty when some comparison scores passed the comparison with the corresponding benchmark threshold and some comparison scores failed the comparison with the benchmark threshold. When at least one comparison score that passed the comparison with the matching threshold after the level was increased passed the comparison, the biometric recognition is judged to be successful. The fourth judgment module is used to reduce the matching threshold corresponding to the unsuccessful comparison scores from the benchmark threshold by at least one level in the direction of reducing difficulty when some comparison scores pass the comparison with their corresponding benchmark thresholds and some comparison scores fail the comparison with their corresponding benchmark thresholds. When all comparison scores that fail the comparison pass the comparison with the matching threshold after the reduction, the biometric recognition is judged to be successful; otherwise, the biometric recognition is judged to be unsuccessful.

7. A computer-readable storage medium for biometric recognition decision fusion, characterized in that, It includes a memory for storing processor-executable instructions, which, when executed by the processor, implement the steps of the biometric recognition decision fusion method according to any one of claims 1-5.

8. A device for biometric recognition decision fusion, characterized in that, It includes at least one processor and a memory storing computer-executable instructions, wherein the processor executes the instructions to implement the steps of the biometric recognition decision fusion method according to any one of claims 1-5.

Images