Biometric multimodal fusion recognition method, device, storage medium and equipment

Abstract

The present invention discloses a method, device, storage medium and equipment for multimodal fusion recognition of biometric features. In multimodal recognition, the comparison threshold of the second modality is adjusted according to the quality score of the image of the second modality; when the comparison score of the second modality is not less than the comparison threshold of the adjusted second modality, and the quality score of the biometric image of the second modality is not less than the decision threshold of the second modality, the recognition is judged to be passed; otherwise, the comparison scores of the two modalities are fused to obtain a fusion score, and the recognition is passed when the fusion score is greater than the fusion score threshold. When the score is fused, the comparison scores of the two modalities are normalized to form a score pair, and the score pair is subjected to polynomial kernel mapping to obtain high-dimensional data, and the high-dimensional data is input into the trained logistic regression model to obtain the fused comparison score. The present invention can improve the security and reliability of biometric identity recognition.

Description

Technical Field

[0001] The present invention relates to the field of biometrics, and in particular to a method, device, storage medium and equipment for multi-modal fusion recognition of biometric features. Background Art

[0002] Biometric technology closely combines computers with high-tech means such as optics, acoustics, biosensors and biostatistics principles, and uses the inherent physiological characteristics of the human body such as fingerprints, faces, irises, finger veins, etc. to identify personal identity.

[0003] With the increasing requirements for accuracy and reliability of social security and identity authentication, the limitations of single biometric recognition in terms of accuracy and reliability have become increasingly prominent and can no longer meet the needs of product and technology development. Currently, multimodal biometric recognition is considered to be a research direction with the most potential and advantages. Multimodal biometric recognition can effectively utilize the independence between different modalities to further improve the recognition rate, so that the accuracy of the fused result can be higher than that of any single modality.

[0004] One method of the biometric multimodal fusion method in the prior art is as follows: wherein P1, P2, and P3 are several biometric images, R1, R2, and R3 are biometric templates corresponding to P1, P2, and P3 respectively. During the recognition process, P1 and R1 are compared to obtain a comparison score S1. If the comparison score S1 is greater than a threshold value T1, the comparison threshold corresponding to P2 is adjusted to obtain an adjusted comparison threshold value T2, for example, the score of the original comparison threshold is lowered. Conversely, if the comparison score S1 is less than the threshold value T1, the score of the original comparison threshold value T2 is increased.

[0005] After adjusting the comparison threshold of P2 according to the comparison score of P1 to obtain the comparison threshold T2, P2 is compared with R2 to obtain the comparison score S2. The original comparison threshold T3 of P3 is adjusted according to the comparison score S2 to obtain the comparison threshold T3 corresponding to P3 after adjustment. For example, if the comparison score S2 is greater than the comparison threshold T2, the original comparison threshold T3 corresponding to P3 is adjusted, for example, the score of the original comparison threshold T3 is lowered. Conversely, if the comparison score S2 is less than the threshold T2, the score of the original comparison threshold T3 is increased, and then P3 is compared with R3 to obtain the comparison score S3. If S3 is greater than the comparison threshold T3 obtained after adjustment, it is considered that the comparison passes and the identity recognition is successful. Otherwise, it is considered that the comparison fails and the identity recognition fails.

[0006] That is, the existing technology adjusts the comparison threshold of the subsequent biometric comparison according to the comparison result of the previous biometric. Although this method makes the comparison of the subsequent biometric easier and can improve the pass rate of the comparison, the method of determining the comparison threshold of the subsequent biometric by the comparison result of the previous biometric is uncertain, and the amplitude of the increase or decrease of the comparison threshold is difficult to grasp, resulting in poor security and reliability of biometric identity recognition. At the same time, the method for fusing the comparison score in the existing biometric multimodal fusion recognition method is usually relatively simple, and generally adopts a weighted method, which cannot give full play to the advantages of different modalities and achieve good recognition accuracy. Summary of the invention

[0007] In order to solve the technical problems in the prior art of multimodal biometric recognition that the adjustment of the comparison threshold is uncertain and the comparison score fusion cannot give full play to the advantages of different modes, the present invention provides a biometric multimodal fusion recognition method, device, storage medium and equipment, which dynamically adjust the comparison threshold according to the quality score of the biometric image, and use the quality score of the biometric image to assist in decision-making on the recognition result, and at the same time combine the comparison score fusion to gradually increase the probability of identity authentication, thereby improving the security and reliability of biometric identity recognition.

[0008] The present invention provides the following technical solutions:

[0009] In a first aspect, the present invention provides a biometric multimodal fusion recognition method, the method comprising:

[0010] sequentially identifying the biometric images of the first modality and the second modality;

[0011] When the first modality recognition passes, adjusting the comparison threshold of the second modality according to the quality score of the biometric image of the second modality;

[0012] When the comparison score of the second modality is not less than the minimum threshold of the second modality and the adjusted comparison threshold of the second modality, and the quality score of the biometric image of the second modality is not less than the decision threshold of the second modality, the recognition is judged to be passed;

[0013] When the comparison score of the second modality is not less than the minimum threshold of the second modality, and the comparison score of the second modality is less than the comparison threshold adjusted by the second modality or the quality score of the biometric image of the second modality is less than the decision threshold of the second modality, the comparison score of the first modality and the comparison score of the second modality are fused to obtain a fusion score. If the fusion score is not less than the fusion score threshold, it is judged that the recognition is passed; wherein,

[0014] The step of fusing the comparison score of the first modality and the comparison score of the second modality to obtain a fusion score includes:

[0015] Normalizing the comparison score of the first modality and the comparison score of the second modality;

[0016] The normalized comparison score of the first mode and the comparison score of the second mode are combined into a score pair, and polynomial kernel mapping is performed on the score pair to obtain high-dimensional data;

[0017] The high-dimensional data is input into the trained logistic regression model to obtain a fused alignment score.

[0018] In a second aspect, the present invention provides a biometric multimodal fusion recognition device, the device comprising:

[0019] A recognition module, used to recognize the biometric image of the first modality and the second modality in sequence;

[0020] A comparison threshold adjustment module, used to adjust the comparison threshold of the second modality according to the quality score of the biometric image of the second modality when the first modality recognition passes;

[0021] A first judgment module, configured to judge that the recognition is passed when the comparison score of the second modality is not less than the minimum threshold of the second modality and the adjusted comparison threshold of the second modality, and the quality score of the biometric image of the second modality is not less than the decision threshold of the second modality;

[0022] The second judgment module is used to fuse the comparison score of the first modality and the comparison score of the second modality to obtain a fusion score when the comparison score of the second modality is not less than the minimum threshold of the second modality, and the comparison score of the second modality is less than the comparison threshold adjusted by the second modality or the quality score of the biometric image of the second modality is less than the decision threshold of the second modality, and if the fusion score is not less than the fusion score threshold, it is judged that the recognition is passed; wherein,

[0023] The second judgment module includes:

[0024] A normalization unit, configured to normalize the comparison score of the first modality and the comparison score of the second modality;

[0025] A mapping unit, used for forming a score pair from the normalized comparison score of the first mode and the comparison score of the second mode, and performing polynomial kernel mapping on the score pair to obtain high-dimensional data;

[0026] The fusion unit is used to input the high-dimensional data into the trained logistic regression model to obtain a fused comparison score.

[0027] In a third aspect, the present invention provides a computer-readable storage medium for biometric multimodal fusion recognition, comprising a memory for storing processor-executable instructions, which, when executed by the processor, implements the steps of the biometric multimodal fusion recognition method described in the first aspect.

[0028] In a fourth aspect, the present invention provides a device for biometric multimodal fusion recognition, comprising at least one processor and a memory storing computer executable instructions, wherein when the processor executes the instructions, the steps of the biometric multimodal fusion recognition method described in the first aspect are implemented.

[0029] The present invention has the following beneficial effects:

[0030] The present invention firstly dynamically adjusts the comparison threshold according to the quality score of the biometric image, and also uses the quality score of the biometric image to assist in decision-making on the recognition result. At the same time, combined with the comparison score fusion, the probability of identity authentication is gradually increased, thereby improving the security and reliability of biometric identity recognition.

[0031] When performing multimodal recognition, the obtained multimodal comparison scores are subjected to the same normalization and high-dimensional mapping processing and then input into the logistic regression model with the optimal parameters to obtain the fused comparison scores. The present invention can give full play to the advantages of different modes and achieve good recognition accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a flow chart of the biometric multimodal fusion recognition method of the present invention;

[0033] Figure 2 for Figure 1 A flow chart of the logistic regression model training method in the biometric multimodal fusion recognition method;

[0034] Figure 3 for Figure 1 A flowchart of another specific example of the biometric multimodal fusion recognition method;

[0035] Figure 4 It is a schematic diagram of the biometric multimodal fusion recognition device of the present invention. DETAILED DESCRIPTION

[0036] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings and specific embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present invention.

[0037] Embodiment 1:

[0038] The embodiment of the present invention provides a biometric multi-modal fusion recognition method, such as Figure 1 As shown, the method includes:

[0039] S100: recognizing the biometric feature images of the first modality and the second modality in sequence.

[0040] In this step, a biometric image of a first modality and a biometric image of a second modality of the user to be identified are obtained, and the biometric image of the first modality and the biometric image of the second modality are identified respectively in chronological order, wherein the biometric image of the first modality may be a face image, and the biometric image of the second modality may be an iris image.

[0041] S110: When the first modality recognition passes, adjust the comparison threshold of the second modality according to the quality score of the biometric image of the second modality.

[0042] In this step, the biometric features of the biometric image of the first modality can be extracted, and the extracted biometric features of the first modality are compared with the biometric template of the first modality stored in the database to obtain the comparison score of the first modality, and whether the first modality recognition is passed is determined according to the comparison score of the first modality. In the prior art, if the comparison score of the first modality is generally greater than the threshold score set during normal recognition, it is considered that the comparison is passed. However, it is well known to those skilled in the art that the biometric features are easily affected by various factors such as lighting and user shaking during the recognition process, which can easily lead to the real user failing the recognition due to the influence of the surrounding environment factors. Therefore, the present invention determines whether the recognition is passed by judging whether the comparison score of the first modality is greater than the minimum threshold of the first modality. When the comparison score of the first modality is greater than or equal to the minimum threshold of the first modality, it is determined that the first modality recognition is passed. Among them, the minimum threshold of the first modality is lower than the threshold score set when the modality is normally recognized. For example, under normal circumstances, if it is greater than or equal to 70 points, it is considered that the recognition is passed. In this embodiment, the minimum threshold is set to be less than 70 points. For example, the minimum threshold is set to 60 points. If it is lower than the minimum threshold, it is directly judged as recognition failure. This can prevent real users from misjudgment due to factors such as environment, image effect, shaking, etc.

[0043] When the first modality recognition passes, the quality score of the second modality biometric image is calculated, and the comparison threshold of the second modality biometric recognition is adjusted according to the quality score of the second modality biometric image. For example, the higher the quality score of the second modality biometric image, the higher the comparison threshold of the second modality biometric image is adjusted when performing biometric recognition.

[0044] S120: When the comparison score of the second modality is not less than the minimum threshold of the second modality and the adjusted comparison threshold of the second modality, and the quality score of the biometric image of the second modality is not less than the decision threshold of the second modality, the recognition is judged to be passed.

[0045] In this step, after adjusting the comparison threshold of the second modality according to the quality score of the biometric features of the second modality, the biometric feature image of the second modality of the user to be identified is identified. The biometric features of the biometric feature image of the second modality of the user to be identified are extracted, and the biometric features of the second modality are compared with the biometric feature template of the second modality in the database, and whether the identification is passed is determined according to the comparison score. When the comparison score of the second modality is greater than or equal to the minimum threshold of the second modality and the adjusted comparison threshold of the second modality, and the quality score of the biometric feature image of the second modality is greater than or equal to the decision threshold of the second modality, it is determined that the identification is passed; when the comparison score of the second modality is greater than or equal to the minimum threshold of the second modality, and the comparison score of the second modality is less than the adjusted comparison threshold or the quality score of the biometric feature image of the second modality is less than the decision threshold of the second modality, the comparison score of the first modality and the comparison score of the second modality are fused to obtain a fusion score, and whether the identification is passed is determined according to the fusion score.

[0046] In this step, the comparison score of the second modality is lower than the threshold score set when the second modality passes normal recognition. This can be understood by referring to the minimum threshold of the first modality, and will not be repeated here.

[0047] When the comparison score of the second modality is less than the minimum threshold of the second modality, the recognition is judged to have failed; when the comparison score of the second modality is greater than or equal to the minimum threshold of the second modality, it is judged whether the comparison score of the second modality is greater than or equal to the adjusted comparison threshold of the second modality. If it is greater than the adjusted comparison threshold of the second modality, it is judged whether the quality score of the biometric image of the second modality is greater than the decision threshold of the second modality. If it is greater, it is judged that the biometric recognition has passed.

[0048] The present invention adjusts the comparison threshold according to the quality score of the biometric feature image of the second modality, and adjusts the amplitude of the comparison threshold accordingly according to the quality score, thereby gradually increasing the probability of identity authentication in the process of multimodal biometric identification and improving the security and reliability of biometric identification.

[0049] S130: When the comparison score of the second modality is not less than the minimum threshold of the second modality, and the comparison score of the second modality is less than the adjusted comparison threshold of the second modality or the quality score of the biometric image of the second modality is less than the decision threshold of the second modality, the comparison score of the first modality and the comparison score of the second modality are fused to obtain a fusion score. If the fusion score is not less than the fusion score threshold, the recognition is judged to be passed.

[0050] In this step, when the comparison score of the second modality is greater than or equal to the minimum threshold of the second modality, and the comparison score of the second modality is lower than the adjusted comparison threshold of the second modality, or the quality score of the biometric image of the second modality is less than the decision threshold of the second modality, in order to avoid misidentification, the present invention performs score fusion on the comparison score of the first modality and the comparison score of the second modality, and further determines whether the recognition is passed according to the fusion score, compares the fusion score with the fusion score threshold, and if the fusion score is greater than or equal to the fusion score threshold, then the recognition is judged to be passed, and if the fusion score is less than the fusion score threshold, then the recognition is judged to be failed.

[0051] The step of fusing the comparison score of the first modality and the comparison score of the second modality to obtain a fusion score includes the following steps:

[0052] S131: Normalize the comparison score of the first modality and the comparison score of the second modality.

[0053] S132: The normalized comparison score of the first mode and the comparison score of the second mode are combined into a score pair, and polynomial kernel mapping is performed on the score pair to obtain high-dimensional data.

[0054] S133: Inputting the high-dimensional data into the trained logistic regression model to obtain a fused comparison score.

[0055] The present invention firstly dynamically adjusts the comparison threshold according to the quality score of the biometric image, and also uses the quality score of the biometric image to assist in decision-making on the recognition result, and combines the comparison score fusion to gradually increase the probability of identity authentication. At the same time, when the comparison score of the second modality is lower than the adjusted comparison threshold, or the quality score of the biometric image of the second modality is lower than the decision threshold, the obtained multimodal comparison score is subjected to the same normalization and high-dimensional mapping processing and then input into the logistic regression model with the optimal parameters, so as to obtain the fused comparison score, thereby improving the security and reliability of biometric identity recognition.

[0056] In the present invention, when the first modality comparison passes, the comparison threshold of the second modality is adjusted in the following manner:

[0057] The quality score of the biometric image of the second modality is divided into three intervals [0, Qmin), [Qmin, Qref), [Qref, Qmax]; wherein Qmin, Qref and Qmax are respectively the minimum quality score, reference quality score and maximum quality score of the biometric image of the modality.

[0058] The minimum quality score Qmin corresponds to the lowest threshold value Tmin, and the maximum quality score Qmax corresponds to the highest threshold value Tmax.

[0059] If the quality score of the biometric image of the second modality is within [0, Qmin), it is considered that the quality of the biometric image of the second modality is not up to standard, and the biometric image of the second modality can be discarded.

[0060] If the quality score of the biometric image of the second modality is within [Qmin, Qref), it is considered that the quality of the second biometric image is in an uncertain state, and the comparison threshold of the second modality remains unchanged, that is, the comparison threshold of the second modality is not adjusted.

[0061] If the quality score Q of the biometric image of the second modality is within [Qref, Qmax], the comparison threshold T of the second modality increases with the increase of the quality score Q of the biometric image of the second modality.

[0062] When the quality score is between [Qref, Qmax], it is considered that the quality of the biometric image of the second modality meets the requirements. As the quality score of the biometric image of the second modality increases, the comparison threshold of the biometric image of the second modality during identification also increases.

[0063] The calculation formula is:

[0064] T=Tmin+(Tmax-Tmin)*f((Q-Qref) / (QMax-Qref))

[0065] Where Q∈[Qmin,Qmax]

[0066]

[0067] In the present invention, when fusing the comparison score of the first modality and the comparison score of the second modality, the comparison score of the first modality and the comparison score of the second modality are first normalized; the normalized comparison score of the first modality and the comparison score of the second modality are composed into a score pair, and the score pair is subjected to polynomial kernel mapping to obtain high-dimensional data; the high-dimensional data is input into the trained logistic regression model to obtain the fused comparison score.

[0068] Among them, the logistic regression model needs to be trained in advance, such as Figure 2 As shown, Figure 2 for Figure 1 The flowchart of the logistic regression model training method in the biometric multimodal fusion recognition method is as follows:

[0069] S131': Obtain a training set including a plurality of first modality comparison score samples and second modality comparison score samples, and normalize the first modality comparison score samples and the second modality comparison score samples.

[0070] When training the logistic regression model, the samples containing the first modality comparison scores and the second modality comparison scores are first normalized.

[0071] S132': construct a sample pair consisting of a normalized first modality comparison score sample and a second modality comparison score sample, and perform polynomial kernel mapping on the sample pair to obtain a high-dimensional data sample.

[0072] In this step, a polynomial kernel method can map low-dimensional data to a high-dimensional space.

[0073] S133': Use the logistic regression algorithm to model the high-dimensional data samples, and use the loss function training to obtain the optimal parameters of the logistic regression model to complete the training of the logistic regression model.

[0074] This step is used to establish a logistic regression model, and train the logistic regression model through the training set to obtain the optimal parameters and complete the training. The first modality comparison score and the second modality comparison score of the user to be identified are input, and the fused comparison score can be obtained according to the logistic regression model with the optimal parameters obtained through training.

[0075] The present invention first normalizes and aligns the multimodal alignment scores, then uses polynomial kernel mapping to perform high-dimensional mapping on the multimodal alignment scores, and finally uses a logistic regression algorithm to model the mapped scores, and calculates the optimal parameters of the logistic regression model through a loss function. When performing multimodal recognition, the obtained multimodal alignment scores are subjected to the same normalization and high-dimensional mapping processing and then input into the logistic regression model with the optimal parameters to obtain the fused alignment scores.

[0076] The fused comparison score represents the probability of the same person. Exemplarily, the range may be [0, 1]. The closer it is to 1, the greater the probability of the same person. Conversely, the smaller the probability of the same person.

[0077] As an improvement of the embodiment of the present invention, the aforementioned step S131′ normalizes the comparison score of the first modality and the comparison score of the second modality, which may include the following steps:

[0078] S1311': Obtain the first modality comparison score samples F of N people i Compare the score samples G with the second modality i , and are divided into pairs of samples of the same person And different samples

[0079] S1312′: Calculate the mean Mean_F and standard deviation Std_F of the first modality comparison score samples in different person sets and the mean Mean_G and standard deviation Std_G of the second modality comparison score samples in different person sets.

[0080] S1313': Compare the score samples F of the first modality using the following formula i Compare the score samples G with the second modality i Normalize:

[0081] f i =(F i -Mean_F) / std_F

[0082] g i =(G i -Mean_G) / std_G.

[0083] Among them, F i and G i are the first modality comparison score sample and the second modality comparison score sample of the i-th person, respectively, i and g i are the normalized first modality comparison score samples and second modality comparison score samples of the i-th person, i∈[1,N]; and are the first modality comparison score samples and the second modality comparison score samples of the i-th pair of the same person in the training set, respectively. and are the first modality comparison score samples and second modality comparison score samples of the i-th pair of different people in the training set, respectively.

[0084]

[0085]

[0086]

[0087]

[0088] Correspondingly, in step S131, the comparison score of the first modality and the comparison score of the second modality are normalized to:

[0089] Obtain the first modal comparison score F and the second modal comparison score G, and normalize the first modal comparison score F and the second modal comparison score G using the following formula:

[0090] f=(F-Mean_F) / std_F

[0091] g=(G-Mean_G) / std_G

[0092] Among them, f and g are the normalized first modality comparison scores and second modality comparison scores, respectively.

[0093] Furthermore, step S132' constructs a sample pair consisting of the normalized first modality comparison score sample and the second modality comparison score sample, and performs polynomial kernel mapping on the sample pair to obtain a high-dimensional data sample including:

[0094] S1321': Construct the normalized first modality comparison score sample f i Compare the score samples g with the second modality i The sample pairs (f i ,g i ).

[0095] S1322': Perform polynomial kernel mapping on the sample pair by the following formula, map the two-dimensional face iris score to multiple dimensions, and obtain a high-dimensional data sample. In this embodiment, the polynomial kernel is 2 as an example for description:

[0096] (1,f i , g i , f i 2 , f i g i , g i 2 )=poly_kernel((f i ,g i ))

[0097] Among them, poly_kernel() is a polynomial kernel mapping, (1,f i , g i , f i 2 , f i g i , g i 2 ) is a high-dimensional data sample. When the polynomial kernel is selected as 2, the two-dimensional multimodal alignment score is transformed into six-dimensional data through the polynomial kernel method.

[0098] Correspondingly, step S132 forms a score pair with the normalized comparison score of the first mode and the comparison score of the second mode, and performs polynomial kernel mapping on the score pair to obtain high-dimensional data including:

[0099] S1321: The normalized first modal comparison score f and the second modal comparison score g are combined into a score pair (f, g).

[0100] S1322: Perform polynomial kernel mapping on the score pairs using the following formula to obtain high-dimensional data:

[0101] (1,f,g,f 2 ,fg,g 2 ) = poly_kernel((f,g))

[0102] Among them, (1,f,g,f 2 ,fg,g 2 ) is high-dimensional data.

[0103] The aforementioned S133' uses a logistic regression algorithm to model the high-dimensional data sample, and uses a loss function to train to obtain the optimal parameters of the logistic regression model. The training of the logistic regression model includes:

[0104] S1331': Use the logistic regression algorithm to model the high-dimensional data samples as follows:

[0105] X i =θ1+θ2f i +θ3g i +θ4f i 2 +θ5f i g i +θ6g i 2

[0106] Among them, if the sample pair is the same person, then its label y (i) =1, if the sample pair is from different people, then its label y (i) =0.

[0107] S1332': Use the loss function to train the SGD algorithm to obtain the optimal parameter θ of the logistic regression model.

[0108] The logistic regression model is as follows:

[0109]

[0110] in, θ T =[θ1,θ2,θ3,θ4,θ5,θ6], h θ (x i ) is the probability of predicting the same person.

[0111] After the training is completed, the fused comparison score is obtained through the logistic regression model in S133. The trained logistic regression model is:

[0112]

[0113] in, h θ(x) represents the probability of the same person.

[0114] The present invention does not limit the form of the loss function. Exemplarily, the loss function is a mean square error function or a cross entropy function. The expressions of the mean square error function and the cross entropy function are:

[0115]

[0116]

[0117] Among them, J(θ0,θ1) is the mean square error function, J(θ) represents the cross entropy function, and y (i) Indicates whether the sample pair is the same person's label. If the sample pair is the same person, then y (i) =1, if the sample pairs are different people, then y (i) =0, m represents the number of all samples in the training set, that is, the mean square error function and the cross entropy function are used to train the optimal model parameters θ1, θ2, θ3, θ4, θ5, θ6 through the SGD algorithm (stochastic gradient descent algorithm).

[0118] Therefore, when the comparison score of the second modality is less than the adjusted comparison threshold, or the quality score of the second modality biometric image is lower than the decision threshold, the first modality comparison score and the second modality comparison score are normalized to form a score pair, and high-dimensional data is obtained after polynomial kernel mapping. The high-dimensional data is input into the trained optimal parameter logistic regression model to obtain the fused comparison score, and whether the recognition is passed is determined based on the fused comparison score. If the fused comparison score is greater than the preset threshold, the recognition is considered to be passed, and if the fused comparison score is less than the preset threshold, the recognition is considered to be failed.

[0119] The present invention firstly dynamically adjusts the comparison threshold according to the quality score of the biometric image, and also uses the quality score of the biometric image to assist in decision-making on the recognition result. At the same time, combined with the comparison score fusion, the probability of identity authentication is gradually increased, thereby improving the security and reliability of biometric identity recognition.

[0120] When performing multimodal recognition, the obtained multimodal comparison scores are subjected to the same normalization and high-dimensional mapping processing and then input into the logistic regression model with the optimal parameters to obtain the fused comparison scores. The present invention can give full play to the advantages of different modes and achieve good recognition accuracy.

[0121] Based on a specific example, Figure 3 As shown, Figure 3 for Figure 1 A flowchart of another specific example of the biometric multimodal fusion recognition method, wherein the method comprises:

[0122] S10: Identify the biometric image P1 of the first modality with the biometric template R1 of the first modality to obtain a comparison score S1 of the first modality, and determine whether the comparison score S1 of the first modality is not less than the minimum threshold T1min of the first modality. If so, execute S20, otherwise determine that the recognition fails.

[0123] P1 is a biometric image representing a human individual identity, such as a face or iris, which can be acquired by a dedicated image acquisition device such as a high-definition camera, a finger vein acquisition head, etc. R1 is a registration template of the user's biometric pre-stored in the template library.

[0124] In this step, the acquired biometric image P1 of the first modality is compared for similarity with the biometric registration template R1 of the user pre-stored in the template library to obtain a comparison score S1 of the first modality.

[0125] The minimum threshold value T1min of the first mode is lower than the threshold score set when P1 is normally recognized. For example, under normal circumstances, if S1 is greater than or equal to 70 points, it is considered that the recognition is successful. In this embodiment, T1min is less than 70 points. For example, T1min is set to 60 points. This can prevent misjudgment of real users due to factors such as environment, image effect, and shaking.

[0126] When the comparison score S1 of the user's first modality is less than 60 points, it is determined that the biometric image matching of the current user fails and the recognition fails, otherwise, step S20 is continued.

[0127] S20: Calculate the quality score Q2 of the biometric image P2 of the second modality, and adjust the comparison threshold T2 of the second modality according to the quality score Q2 of the biometric image of the second modality, wherein the higher the quality score Q2 of the biometric image of the second modality, the higher the comparison threshold T2 of the second modality, and conversely, if Q2 is lower, T2 is lower.

[0128] In this step, when S1 is greater than or equal to T1min, it is necessary to further analyze the quality score Q2 of P2 through the image quality algorithm, and adjust the comparison threshold T2 accordingly according to the quality score Q2.

[0129] S30: Identify the biometric image P2 of the second modality with the biometric template R2 of the second modality to obtain a comparison score S2 of the second modality, and determine whether the comparison score S2 of the second modality is not less than the minimum threshold T2min of the second modality. If so, execute S40, otherwise determine that the recognition fails.

[0130] In this step, after the comparison of the biometric image P1 of the first modality is completed and the comparison threshold T2 of the biometric is adjusted according to the quality score Q2 of the biometric image P2 of the second modality, the biometric image P2 of the second modality is compared with the biometric registration template R2 of the user pre-stored in the template library for similarity to obtain a comparison score S2.

[0131] The minimum threshold value T2min of the second mode is lower than the threshold score set when P2 is normally recognized. For example, under normal circumstances, if S2 is greater than or equal to 80 points, it is considered that the recognition is passed. In this embodiment, T2min is less than 80 points, for example, T2min is 70 points. This setting can prevent real users from being misjudged due to factors such as environment, image effect, and shaking.

[0132] When the comparison score S2 of the second modality of the user is less than 70 points, it is determined that the biometric image matching of the current user fails, otherwise, step S40 is continued.

[0133] It should be noted that T1min and T2min can be the same or different. For example, when P1 is a face image and P2 is an iris image, T1min can be set to 60 points. Since the iris comparison accuracy is higher, T2min can be greater than T1min, for example, T2min is set to 70 points.

[0134] S40: Determine whether the comparison score S2 of the second modality is not less than the adjusted comparison threshold T2 of the second modality and the quality score Q2 of the biometric image of the second modality is not less than the decision threshold Q of the second modality. If both conditions are met, determine that the recognition is passed, otherwise execute S50.

[0135] In this step, when the comparison score S2 of the second modality is greater than or equal to the minimum threshold T2min of the second modality, it is necessary to further determine whether the comparison score S2 of the second modality is greater than or equal to the adjusted second comparison threshold T2 to prevent misidentification. If it is greater than or equal to, it is further determined whether the quality score Q2 of P2 is greater than or equal to the decision threshold of the second modality, which is the set quality score threshold. If it is greater than or equal to, the recognition is determined to be passed.

[0136] The purpose of judging whether the quality score Q2 of P2 is greater than or equal to Q is to judge the reliability of the second comparison threshold T2 adjusted according to Q2. If it is greater than or equal to Q, the adjusted T2 is reliable and the recognition is considered to be successful.

[0137] S50: Fusing the comparison score S1 of the first modality and the comparison score S2 of the second modality to obtain a first fusion score S12, and determining whether the first fusion score is less than a first fusion score threshold T12. If so, determining that the recognition fails.

[0138] In this step, the comparison score S1 corresponding to P1 and the comparison score S2 corresponding to P2 are fused to obtain a first fusion score threshold T12.

[0139] The method of fusing the comparison score S1 and the comparison score S2 can be understood with reference to the above embodiment, and will not be described in detail here.

[0140] S60: If the first fusion score S12 is not less than the first fusion score threshold T12, it is determined that the recognition is passed.

[0141] Embodiment 2:

[0142] The embodiment of the present invention provides a biometric multi-modal fusion recognition device, the device comprising:

[0143] The recognition module 100 is used to recognize the biometric image of each modality in turn.

[0144] The comparison threshold adjustment module 110 is used to adjust the comparison threshold of the second modality according to the quality score of the biometric image of the second modality when the first modality recognition passes.

[0145] The first judgment module 120 is used to judge that the recognition is passed when the comparison score of the second modality is not less than the minimum threshold of the second modality and the adjusted comparison threshold of the second modality, and the quality score of the biometric image of the second modality is not less than the decision threshold of the second modality.

[0146] The second judgment module 130 is used to fuse the comparison score of the first modality and the comparison score of the second modality to obtain a fusion score when the comparison score of the second modality is not less than the minimum threshold of the second modality, and the comparison score of the second modality is less than the comparison threshold adjusted by the second modality or the quality score of the biometric image of the second modality is less than the decision threshold of the second modality, and if the fusion score is not less than the fusion score threshold, it is judged that the recognition is passed, wherein the second judgment module includes:

[0147] A normalization unit 131, configured to normalize the comparison score of the first modality and the comparison score of the second modality;

[0148] A mapping unit 132, configured to form a score pair from the normalized comparison score of the first mode and the comparison score of the second mode, and perform polynomial kernel mapping on the score pair to obtain high-dimensional data;

[0149] The fusion unit 133 is used to input the high-dimensional data into the trained logistic regression model to obtain a fused comparison score.

[0150] The present invention firstly dynamically adjusts the comparison threshold according to the quality score of the biometric image, and also uses the quality score of the biometric image to assist in decision-making on the recognition result. At the same time, combined with the comparison score fusion, the probability of identity authentication is gradually increased, thereby improving the security and reliability of biometric identity recognition.

[0151] When performing multimodal recognition, the obtained multimodal comparison scores are subjected to the same normalization and high-dimensional mapping processing and then input into the logistic regression model with the optimal parameters to obtain the fused comparison scores. The present invention can give full play to the advantages of different modalities and achieve good recognition accuracy.

[0152] For the first mode and the second mode, the lowest threshold is lower than the threshold when the corresponding mode passes normal recognition.

[0153] For the second modality, the higher the quality score of the biometric image of the modality is, the higher the comparison threshold of the modality is adjusted.

[0154] The comparison threshold of the second modality is adjusted according to the quality score of the biometric image of the second modality by:

[0155] The quality score of the biometric image of the second modality is divided into three intervals [0, Qmin), [Qmin, Qref), and [Qref, Qmax]; wherein Qmin, Qref, and Qmax are respectively the minimum quality score, reference quality score, and maximum quality score of the biometric image of the second modality.

[0156] If the quality score of the biometric image of the second modality is within [0, Qmin), the biometric image of the second modality is discarded.

[0157] If the quality score of the biometric image of the second modality is within [Qmin, Qref), the comparison threshold of the second modality remains unchanged.

[0158] If the quality score of the biometric image of the second modality is within [Qref, Qmax], the comparison threshold of the second modality increases with the increase of the quality score of the biometric image of the second modality.

[0159] The device provided in the embodiment of the present invention has the same implementation principle and technical effects as those in the aforementioned method embodiment 1. For the sake of brief description, for matters not mentioned in the device embodiment, reference may be made to the corresponding contents in the aforementioned method embodiment 1. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the devices and units described above can all refer to the corresponding processes in the aforementioned method embodiment 1, and will not be repeated here.

[0160] Embodiment 3:

[0161] The method described in the above-mentioned embodiment 1 provided in this specification can realize the business logic through a computer program and record it on a storage medium, and the storage medium can be read and executed by a computer to realize the effect of the solution described in embodiment 1 of this specification. Therefore, the present invention also provides a computer-readable storage medium for biometric multimodal fusion recognition, including a memory for storing processor-executable instructions, and when the instructions are executed by the processor, the steps of the biometric multimodal fusion recognition method including embodiment 1 are realized.

[0162] The present invention firstly dynamically adjusts the comparison threshold according to the quality score of the biometric image, and also uses the quality score of the biometric image to assist in decision-making on the recognition result. At the same time, combined with the comparison score fusion, the probability of identity authentication is gradually increased, thereby improving the security and reliability of biometric identity recognition.

[0163] When performing multimodal recognition, the obtained multimodal comparison scores are subjected to the same normalization and high-dimensional mapping processing and then input into the logistic regression model with the optimal parameters to obtain the fused comparison scores. The present invention can give full play to the advantages of different modes and achieve good recognition accuracy.

[0164] The storage medium may include a physical device for storing information, which is usually a medium that digitizes the information and then stores it in an electrical, magnetic or optical manner. The storage medium may include: a device that stores information in an electrical energy manner, such as various memories, such as RAM, ROM, etc.; a device that stores information in a magnetic energy manner, such as a hard disk, a floppy disk, a magnetic tape, a magnetic core memory, a magnetic bubble memory, a USB flash drive; a device that stores information in an optical manner, such as a CD or a DVD. Of course, there are other readable storage media, such as quantum memory, graphene memory, etc.

[0165] The above-mentioned device may also include other implementations according to the description of method embodiment 1. The specific implementation methods may refer to the description of the relevant method embodiment 1, and will not be described one by one here.

[0166] Embodiment 4:

[0167] The present invention also provides a device for biometric multimodal fusion recognition, which may be a separate computer, or may include an actual operating device that uses one or more of the methods or one or more of the embodiments of this specification. The device for biometric multimodal fusion recognition may include at least one processor and a memory storing computer executable instructions, and when the processor executes the instructions, the steps of the biometric multimodal fusion recognition method described in any one or more of Embodiment 1 are implemented.

[0168] The present invention firstly dynamically adjusts the comparison threshold according to the quality score of the biometric image, and also uses the quality score of the biometric image to assist in decision-making on the recognition result. At the same time, combined with the comparison score fusion, the probability of identity authentication is gradually increased, thereby improving the security and reliability of biometric identity recognition.

[0169] When performing multimodal recognition, the obtained multimodal comparison scores are subjected to the same normalization and high-dimensional mapping processing and then input into the logistic regression model with the optimal parameters to obtain the fused comparison scores. The present invention can give full play to the advantages of different modes and achieve good recognition accuracy.

[0170] The above-mentioned device may also include other implementation modes according to the description of the method or device embodiment. The specific implementation mode may refer to the description of the relevant method embodiment 1, and will not be described one by one here.

[0171] It should be noted that the device or system described above in this specification may also include other implementation methods according to the description of the relevant method embodiments. The specific implementation methods can refer to the description of the method embodiments, which will not be described one by one here. The various embodiments in this specification are described in a progressive manner, and the same and similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the hardware + program type and storage medium + program embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0172] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0173] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a vehicle-mounted human-computer interaction device, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0174] For the convenience of description, the above devices are described in various modules according to their functions. Of course, when implementing one or more of the present specification, the functions of each module can be implemented in the same or more software and / or hardware, or the module implementing the same function can be implemented by a combination of multiple sub-modules or sub-units, etc. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0175] Those skilled in the art also know that, in addition to implementing the controller in a purely computer-readable program code, the controller can be made to implement the same function in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, such a controller can be considered as a hardware component, and the devices for implementing various functions included therein can also be considered as structures within the hardware component. Or even, the devices for implementing various functions can be considered as both software modules for implementing the method and structures within the hardware component.

[0176] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0177] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0178] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0179] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0180] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method or device including the element.

[0181] It should be understood by those skilled in the art that one or more embodiments of the present specification may be provided as a method, system or computer program product. Therefore, one or more embodiments of the present specification may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, one or more embodiments of the present specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0182] One or more embodiments of this specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. One or more embodiments of this specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0183] Each embodiment in this specification is described in a progressive manner, and the same and similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. In the description of this specification, the description of the reference terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of this specification. In this specification, the schematic representation of the above terms does not necessarily target the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples without contradiction.

[0184] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. They should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A biometric multimodal fusion recognition method, characterized in that: The method comprises: sequentially identifying the biometric images of the first modality and the second modality; When the first modality recognition passes, adjusting the comparison threshold of the second modality according to the quality score of the biometric image of the second modality; When the comparison score of the second modality is not less than the minimum threshold of the second modality and the adjusted comparison threshold of the second modality, and the quality score of the biometric image of the second modality is not less than the decision threshold of the second modality, the recognition is judged to be passed; When the comparison score of the second modality is not less than the minimum threshold of the second modality, and the comparison score of the second modality is less than the comparison threshold adjusted by the second modality or the quality score of the biometric image of the second modality is less than the decision threshold of the second modality, the comparison score of the first modality and the comparison score of the second modality are fused to obtain a fusion score. If the fusion score is not less than the fusion score threshold, it is judged that the recognition is passed; wherein, The step of fusing the comparison score of the first modality and the comparison score of the second modality to obtain a fusion score includes: Normalizing the comparison score of the first modality and the comparison score of the second modality; The normalized comparison score of the first mode and the comparison score of the second mode are combined into a score pair, and polynomial kernel mapping is performed on the score pair to obtain high-dimensional data; The high-dimensional data is input into the trained logistic regression model to obtain a fused alignment score.

2. The biometric multimodal fusion recognition method according to claim 1, characterized in that: The logistic regression model is trained by the following method: Constructing a training set including a plurality of first modality comparison score samples and second modality comparison score samples, and normalizing the first modality comparison score samples and the second modality comparison score samples; Constructing a sample pair consisting of a normalized first modal comparison score sample and a second modal comparison score sample, and performing polynomial kernel mapping on the sample pair to obtain a high-dimensional data sample; The high-dimensional data samples are modeled using a logistic regression algorithm, and the optimal parameters of the logistic regression model are obtained using loss function training to complete the training of the logistic regression model.

3. The biometric multimodal fusion recognition method according to claim 2, characterized in that: The construction includes a training set of a plurality of first modality comparison score samples and second modality comparison score samples, and normalizing the first modality comparison score samples and the second modality comparison score samples, including: Get the first modality comparison score sample F of N people i Compare the score samples G with the second modality i , and are divided into pairs of samples of the same person And different samples Calculate the mean Mean_F and standard deviation Std_F of the first modality comparison score samples in different person sets and the mean Mean_G and standard deviation Std_G of the second modality comparison score samples in different person sets; Use the following formula to compare the first modality score sample F i Compare the score samples G with the second modality i Normalize: f i =(F i -Mean_F) / std_F g i =(G i -Mean_G) / std_G Among them, F i and G i are the first modality comparison score sample and the second modality comparison score sample of the i-th person, respectively, i and g i are the normalized first modality comparison score samples and second modality comparison score samples of the i-th person, i∈[1,N]; F i gen and are the first modality comparison score samples and the second modality comparison score samples of the i-th pair of the same person in the training set, respectively. and They are the first modality comparison score samples and the second modality comparison score samples of the i-th pair of different people in the training set respectively; Obtaining a first modality comparison score and a second modality comparison score, and normalizing the first modality comparison score and the second modality comparison score, includes: Obtain the first modal comparison score F and the second modal comparison score G, and normalize the first modal comparison score F and the second modal comparison score G using the following formula: f=(F-Mean_F) / std_F g=(G-Mean_G) / std_G Among them, f and g are the normalized first modality comparison scores and second modality comparison scores, respectively.

4. The biometric multimodal fusion recognition method according to claim 2, characterized in that: The loss function is a mean square error function or a cross entropy function.

5. The biometric multimodal fusion recognition method according to claim 1, characterized in that: The higher the quality score of the biometric image of the second modality is, the higher the comparison threshold of the second modality is adjusted.

6. The biometric multimodal fusion recognition method according to claim 5, characterized in that: The comparison threshold of the second mode is adjusted as follows: The quality score of the biometric image of the second modality is divided into three intervals [0, Qmin), [Qmin, Qref), [Qref, Qmax]; wherein Qmin, Qref and Qmax are the minimum quality score, reference quality score and maximum quality score of the biometric image of the second modality respectively; If the quality score of the biometric image of the second modality is within [0, Qmin), the biometric image of the second modality is discarded; If the quality score of the biometric image of the second modality is within [Qmin, Qref), the comparison threshold of the second modality remains unchanged; If the quality score of the biometric image of the second modality is within [Qref, Qmax], the comparison threshold of the second modality increases with the increase of the quality score of the biometric image of the modality; its calculation formula is: T=Tmin+(Tmax-Tmin)*f((Q-Qref) / (QMax-Qref)) Where Q∈[Qmin,Qmax] 7. A biometric multimodal fusion recognition device, characterized in that: The device comprises: A recognition module, used to recognize the biometric image of the first modality and the second modality in sequence; A comparison threshold adjustment module, used to adjust the comparison threshold of the second modality according to the quality score of the biometric image of the second modality when the first modality recognition passes; A first judgment module, configured to judge that the recognition is passed when the comparison score of the second modality is not less than the minimum threshold of the second modality and the adjusted comparison threshold of the second modality, and the quality score of the biometric image of the second modality is not less than the decision threshold of the second modality; The second judgment module is used to fuse the comparison score of the first modality and the comparison score of the second modality to obtain a fusion score when the comparison score of the second modality is not less than the minimum threshold of the second modality, and the comparison score of the second modality is less than the comparison threshold adjusted by the second modality or the quality score of the biometric image of the second modality is less than the decision threshold of the second modality, and if the fusion score is not less than the fusion score threshold, it is judged that the recognition is passed; wherein, The second judgment module includes: A normalization unit, configured to normalize the comparison score of the first modality and the comparison score of the second modality; A mapping unit, used for forming a score pair from the normalized comparison score of the first mode and the comparison score of the second mode, and performing polynomial kernel mapping on the score pair to obtain high-dimensional data; The fusion unit is used to input the high-dimensional data into the trained logistic regression model to obtain a fused comparison score.

8. The biometric multimodal fusion recognition device according to claim 7, characterized in that: The logistic regression model in the fusion unit is trained by the following method: Constructing a training set including a plurality of first modality comparison score samples and second modality comparison score samples, and normalizing the first modality comparison score samples and the second modality comparison score samples; Constructing a sample pair consisting of a normalized first modal comparison score sample and a second modal comparison score sample, and performing polynomial kernel mapping on the sample pair to obtain a high-dimensional data sample; The high-dimensional data samples are modeled using a logistic regression algorithm, and the optimal parameters of the logistic regression model are obtained using loss function training to complete the training of the logistic regression model.

9. A computer-readable storage medium for biometric multimodal fusion recognition, characterized in that: It comprises a memory for storing processor executable instructions, and when the instructions are executed by the processor, the steps of the biometric multimodal fusion recognition method according to any one of claims 1 to 6 are implemented.

10. A device for multimodal fusion recognition of biometric features, characterized in that: The method comprises at least one processor and a memory storing computer executable instructions, wherein when the processor executes the instructions, the steps of the biometric multimodal fusion recognition method described in any one of claims 1 to 6 are implemented.

Images