A multi-biometric fusion identification system and device

Abstract

The present invention provides a multi-biometric feature fusion recognition system and device, belonging to the field of intelligent recognition technology, including: a matrix construction module, used to extract features of each type of biological information according to biological verification points, and construct the user's initial feature matrix; a matching analysis module, used to extract a number of elements from each row vector in the initial feature matrix and perform normalization processing to construct a normalization function; an instruction issuing module, used to issue an unlocking instruction to the vehicle when the normalization function matches the set function required for vehicle unlocking; otherwise, the column information of each column vector in the initial feature matrix is ​​matched with the pre-stored information to obtain the matching coefficient, and the biological image is processed to obtain the unlocking reliability; a reliability determination module, used to pre-process all matching coefficients and unlocking reliability to obtain comprehensive reliability. Effectively improve the recognition reliability and the safety of vehicle door lock control, and reduce the risk of vehicles being used indiscriminately.

Description

Technical Field

[0001] The present invention relates to the field of intelligent identification technology, and in particular to a multi-biometric fusion identification system and device. Background Art

[0002] The vehicle smart key is a common keyless entry system, also known as a smart key system. It is mainly composed of a control system consisting of three receivers, including a transmitter, a remote central lock control module, and a driving authorization system control module, and related wiring harnesses, which greatly facilitates the intelligent control of vehicle door locks.

[0003] However, the vehicle smart key also has a fatal loophole, that is, the user needs to carry it with him to open the vehicle door lock, and when unlocking the vehicle, the vehicle recognizes the key but not the person. No matter who holds the smart key, it can be sensed and read successfully, which will undoubtedly increase the risk of the vehicle being misused.

[0004] Therefore, the present invention proposes a multi-biometric fusion recognition system and device. Summary of the invention

[0005] The present invention provides a multi-biometric fusion recognition system and device, which are used to construct a matrix based on multi-biometric features to provide a basis for identifying users, and subsequently normalize the matrix or pre-process the matching coefficient and unlocking reliability to effectively improve the recognition reliability and the security of vehicle door lock control, thereby reducing the risk of improper use of the vehicle.

[0006] The present invention provides a multi-biometric fusion recognition system, comprising:

[0007] A matrix construction module, used to obtain multiple biological information of a user and extract features for each biological information according to biological verification points to construct an initial feature matrix of the user;

[0008] A matching analysis module, used for extracting a number of elements from each row vector in the initial feature matrix and performing normalization processing to construct a normalization function;

[0009] The command issuing module is used to issue an unlocking command to the vehicle when the normalized function matches the set function required for unlocking the vehicle;

[0010] Otherwise, the column information of each column vector in the initial feature matrix is ​​matched and analyzed with the pre-stored information to obtain the matching coefficient, and at the same time, the biological image of each biological information is processed to obtain the unlocking reliability of the corresponding image;

[0011] A reliability determination module is used to pre-process all matching coefficients and unlocking reliability to obtain a comprehensive reliability;

[0012] If the comprehensive reliability is greater than the preset reliability, an unlocking command is issued to the vehicle;

[0013] Otherwise, the user is prompted to re-identify.

[0014] Preferably, the matrix building module includes:

[0015] A point acquisition unit, used to acquire biological verification points of each biological information from a type-organism comparison table according to the biological species, wherein the number of biological verification points of each biological information is the same;

[0016] A calibration information unit, used to calibrate the biological verification point and the biological image corresponding to the biological information at the same position, and extract calibration information of each calibration point in each biological image;

[0017] A feature extraction unit, used to extract information features of each calibration information to obtain a single row vector corresponding to the biological information;

[0018] The matrix acquisition unit is used to sort and place all single row vectors in sequence to obtain an initial feature matrix.

[0019] Preferably, the matching analysis module includes:

[0020] A set construction unit is used to obtain all unlocking records on the vehicle, and extract historical unlocking records according to corresponding biometric verification points to obtain a record set J based on each biometric verification point = { ,i=1,2,3,...,n1}, where, represents the i-th unlocking record information under the corresponding biometric verification point; n1 represents the number of unlocking records for the biometric verification point;

[0021] a saturation calculation unit, for determining the information saturation of each biological verification point based on the record set;

[0022] ;

[0023] Among them, Xb represents the information saturation of the corresponding biological verification point; Represents the similarity function between the standard biometric information Yk corresponding to the biometric verification point and the i-th unlocking record information; Indicates all The average value of the similarity function that satisfies the normal distribution probability density of 0.8; Indicates all The variance of the similarity function that does not satisfy the normal distribution probability density of 0.8; ln is the sign of the logarithmic function; Nc represents the number of successful verifications of the corresponding biometric verification point in the unlocking process; Ns represents the number of failed verifications of the corresponding biometric verification point in the starting process, and Ns+Nc=n1;

[0024] A quantity calculation unit, used to determine a first quantity N1 of biological verification points whose information saturation is greater than a preset saturation under each biological species, and calculate an extraction quantity Tn under the corresponding biological species;

[0025] ;

[0026] Among them, A0 represents the correction result based on N1; Indicates the rounding up symbol; Indicates the rounding down symbol; N2 indicates the number of biological verification points under the corresponding biological species; exp indicates the exponential function symbol; Yb indicates the corresponding preset saturation;

[0027] The element determination unit is used to sort all information saturations under the corresponding biological species, extract the first Tn biological verification points from the size sorting results, and then obtain the elements corresponding to the extracted first Tn biological verification points from the corresponding row vector of the initial feature matrix.

[0028] Preferably, the matching analysis module further includes:

[0029] A coefficient determination unit, used for normalizing the information features of the screening elements under the same row vector to obtain a normalization coefficient of the corresponding row vector;

[0030] A function building unit, used for building a normalization function based on a normalization coefficient of each biological species;

[0031] Preferably, the instruction issuing module includes:

[0032] A factor determination unit, used to determine a matching loss factor Fc of each column vector;

[0033] ;

[0034] in, A similarity function representing the information xj of the j-th element in the corresponding column vector and the standard information Yj that is consistent with the j-th element in the pre-stored information; represents the cumulative sum function of the similarity functions except the j-th element in the corresponding column vector; max represents the maximum value symbol; represents the flexibility factor, and Represents all the corresponding column vectors The variance of

[0035] A coefficient determination unit, configured to calculate a corresponding matching coefficient Pc based on the matching loss factor;

[0036] ;

[0037] in, Represents the trace factor of the similarity function, and the value range of the trace factor is (0, ), where N2 represents the number of column vectors and is consistent with the number of biological verification points involved in each row in the initial feature matrix.

[0038] Preferably, the instruction issuing module further includes:

[0039] An array construction unit, used for traversing the biological images of each biological type, locking the noise points in the corresponding biological images, and deriving them laterally along the noise points to construct a noise array;

[0040] A single elimination unit is used to determine whether the noise array has periodicity. If so, a matching noise elimination method is obtained from an elimination database according to the noise characteristics of a random period to perform a single noise processing on the corresponding biological image to obtain a current image;

[0041] Otherwise, according to the interference pattern constructed by the noise points, the noise type is matched from the pattern database, and the noise elimination method based on each noise type is matched from the noise database, and the corresponding biological image is subjected to multiple noise processing to obtain the current image;

[0042] The reliability determination unit is used to sequentially obtain the clarity of each pixel in the current image to obtain the unlocking reliability.

[0043] Preferably, the reliability determination module includes:

[0044] The statistical unit is used to count the number of successful verifications of each biological verification point involved in the same biological species during the unlocking process, and obtain the success rate of the same biological species. , where sunNc is the total number of successful verifications of all biological verification points under the same biological species;

[0045] An optimization setting unit, used to set an optimization factor Yy for each biological species according to the success ratio and unlocking reliability under the same row vector;

[0046] ;

[0047] Wherein, Kd represents the unlocking reliability of the corresponding biological species determined based on the biological image; is the reliability threshold;

[0048] A coefficient optimization unit is used to optimize the matching coefficient corresponding to each column vector based on the optimization factors under all biological species to obtain the optimization coefficient Yc;

[0049] ;

[0050] in, Represents the weights of all elements in the corresponding column vector The optimization factor of the biological species to which the largest weight element belongs;

[0051] The average processing unit is used to calculate the average value of all optimization coefficients to obtain the comprehensive reliability.

[0052] Preferably, the reliability determination unit includes:

[0053] A clarity calibration subunit, used to calibrate the clarity level of each pixel in the current image;

[0054] The ratio determination subunit is used to determine the number of pixels that meet the reliable grade standard in the grade calibration result, and obtain the ratio to the total number of pixels corresponding to the current image;

[0055] The table matching subunit is used to obtain the unlocking reliability based on the ratio according to the ratio-reliability comparison table.

[0056] The present invention provides a multi-biometric feature fusion recognition device, which is used to apply any of the multi-biometric feature fusion recognition systems.

[0057] Compared with the prior art, the present invention has the following beneficial effects:

[0058] Building a matrix based on multiple biometric features provides a basis for identifying users, and subsequently normalizing the matrix or preprocessing the matching coefficient and unlocking reliability can effectively improve the recognition reliability and the security of vehicle door lock control, thereby reducing the risk of improper use of the vehicle.

[0059] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0060] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0062] Figure 1 It is a structural diagram of a multi-biometric fusion recognition system in an embodiment of the present invention. DETAILED DESCRIPTION

[0063] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0064] The present invention provides a multi-biometric fusion recognition system. Figure 1 As shown, including:

[0065] A matrix construction module, used to obtain multiple biological information of a user and extract features for each biological information according to biological verification points to construct an initial feature matrix of the user;

[0066] A matching analysis module, used for extracting a number of elements from each row vector in the initial feature matrix and performing normalization processing to construct a normalization function;

[0067] The command issuing module is used to issue an unlocking command to the vehicle when the normalized function matches the set function required for unlocking the vehicle;

[0068] Otherwise, the column information of each column vector in the initial feature matrix is ​​matched with the pre-stored information to obtain the matching coefficient, and at the same time, the biological image of each biological information is processed to obtain the unlocking reliability of the corresponding image;

[0069] A reliability determination module is used to pre-process all matching coefficients and unlocking reliability to obtain a comprehensive reliability;

[0070] If the comprehensive reliability is greater than the preset reliability, an unlocking command is issued to the vehicle;

[0071] Otherwise, the user is prompted to re-identify.

[0072] In this embodiment, multi-biometric information generally refers to fingerprint information, iris information, and face information.

[0073] In this embodiment, the biometric verification point refers to a specific location point based on fingerprint, iris and face, which is predetermined.

[0074] In this embodiment, the initial feature matrix = , where each row vector corresponds to a biological type, and the biological types are: fingerprint type, iris type and face type.

[0075] In this embodiment, each information feature involved in each row vector can be regarded as an element.

[0076] In this embodiment, the normalization process refers to normalizing a number of elements in the same row vector, mainly to determine whether the information of the element is consistent with the standard in the subsequent judgment.

[0077] In this embodiment, the normalization function = ,in, Respectively represent the normalization coefficients based on the corresponding fingerprint, iris, and face.

[0078] In this embodiment, the setting function S0 of the vehicle unlocking requirement is as follows:

[0079] ;

[0080] When the corresponding When both are within the S0 setting range, it is determined to be a match.

[0081] In this embodiment, the unlocking instruction refers to sending a door lock opening instruction to the vehicle to control the door lock to open.

[0082] In this embodiment, each column in the initial feature matrix contains feature information of one biometric verification point each on the fingerprint, iris, and face. The pre-stored information is information pre-recorded for the fingerprint, face, and iris of the user who needs to control the vehicle as a basis for comparison, and the pre-stored information generally includes information in all directions. For example, the face is collected from all angles of the facial contour during the recording process.

[0083] In this embodiment, the matching coefficient is mainly used to determine the matching status between the column information and the pre-stored information, and the value range is generally (0, 1).

[0084] In this embodiment, image processing refers to determining the clarity of the image to obtain the unlocking reliability, thereby optimizing the matching coefficient. In this process, the unlocking reliability is determined to avoid errors in obtaining the matching coefficient due to noise points in the information. That is, the unlocking reliability is used to effectively improve the accuracy of the unlocking judgment process and the reliability of recognition.

[0085] In this embodiment, the preset reliability is pre-set, and the value is generally 0.7.

[0086] In this embodiment, the premise for identifying the user is that the user triggers the door lock sensing device installed on the vehicle door handle. Based on this, the user's relevant information is collected based on the face acquisition module, iris acquisition module and fingerprint acquisition module installed in advance on the vehicle to obtain multi-biometric information.

[0087] The beneficial effects of the above technical solution are: constructing a matrix based on multiple biometric features provides a basis for identifying users, and subsequently normalizing the matrix or preprocessing the matching coefficient and unlocking reliability, effectively improving the recognition reliability and the security of vehicle door lock control, and reducing the risk of improper use of the vehicle.

[0088] The present invention provides a multi-biometric fusion recognition system, wherein the matrix construction module comprises:

[0089] A point acquisition unit, used to acquire biological verification points of each biological information from a type-organism comparison table according to the biological species, wherein the number of biological verification points of each biological information is the same;

[0090] A calibration information unit, used to calibrate the biological verification point and the biological image corresponding to the biological information at the same position, and extract calibration information of each calibration point in each biological image;

[0091] A feature extraction unit, used to extract information features of each calibration information to obtain a single row vector corresponding to the biological information;

[0092] The matrix acquisition unit is used to sort and place all single row vectors in sequence to obtain an initial feature matrix.

[0093] In this embodiment, the type-biometric comparison table includes different biological types and the biological verification points that match them, and the biological information is generally reflected in the form of an image. For example, the biological verification points for the fingerprint are respectively at the center point of the thumb, and points in eight positions based on the center point: up, down, left, right, upper left, upper right, lower left, and lower right.

[0094] In this embodiment, the control calibration of the same position point is to lock the points in eight directions, namely, the center position point, up, down, left, right, upper left, upper right, lower left, and lower right, on the biological information.

[0095] In this embodiment, the calibration information is the biological information of the position (calibration point).

[0096] In this embodiment, the information features are directly extracted based on the feature extraction model, and the feature extraction model belongs to the existing means.

[0097] In this embodiment, a single row vector = {information features of the calibration points involved in the corresponding biological information}.

[0098] In this embodiment, sequentially arranging and placing means randomly arranging and placing the single row vectors of the fingerprint, face, and iris in order from top to bottom.

[0099] In this embodiment, the number of biological verification points under different biological species is consistent in order to ensure the uniformity of matrix construction.

[0100] The beneficial effect of the above technical solution is: by obtaining the biological verification points from the comparison table, it is convenient to perform comparison calibration of the same position point to construct a single row vector, and then obtain the initial feature matrix, which provides a basis for judging whether to issue instructions later.

[0101] The present invention provides a multi-biometric fusion recognition system, wherein the matching analysis module comprises:

[0102] A set construction unit is used to obtain all unlocking records on the vehicle, and extract historical unlocking records according to corresponding biometric verification points to obtain a record set J based on each biometric verification point = { ,i=1,2,3,...,n1}, where, represents the i-th unlocking record information under the corresponding biometric verification point; n1 represents the number of unlocking records for the biometric verification point;

[0103] a saturation calculation unit, for determining the information saturation of each biological verification point based on the record set;

[0104] ;

[0105] Among them, Xb represents the information saturation of the corresponding biological verification point; Represents the similarity function between the standard biometric information Yk corresponding to the biometric verification point and the i-th unlocking record information; Indicates all The average value of the similarity function that satisfies the normal distribution probability density of 0.8; Indicates all The variance of the similarity function that does not satisfy the normal distribution probability density of 0.8; ln is the sign of the logarithmic function; Nc represents the number of successful verifications of the corresponding biometric verification point in the unlocking process; Ns represents the number of failed verifications of the corresponding biometric verification point in the starting process, and Ns+Nc=n1;

[0106] A quantity calculation unit, used to determine a first quantity N1 of biological verification points whose information saturation is greater than a preset saturation under each biological species, and calculate an extraction quantity Tn under the corresponding biological species;

[0107] ;

[0108] Among them, A0 represents the correction result based on N1; Indicates the rounding up symbol; Indicates the rounding down symbol; N2 indicates the number of biological verification points under the corresponding biological species; exp indicates the exponential function symbol; Yb indicates the corresponding preset saturation;

[0109] The element determination unit is used to sort all information saturations under the corresponding biological species, extract the first Tn biological verification points from the size sorting results, and then obtain the elements corresponding to the extracted first Tn biological verification points from the corresponding row vector of the initial feature matrix.

[0110] In this embodiment, the preset saturation value is generally 0.7.

[0111] In this embodiment, the size sorting of information saturation is obtained based on historical records. For example, for a fingerprint, there are: verification point 01, verification point 02, verification point 03 and verification point 04, and the corresponding information saturations are: 0.9, 0.1, 0.8, 0.9 respectively. At this time, the size sorting result is: verification point 01, verification point 04, verification point 03, verification point 02. At this time, the value of Tn is 3. At this time, the biometric information for verification point 01, verification point 04, and verification point 03 is extracted from the initial feature matrix.

[0112] In this embodiment, the unlocking record refers to the record of the vehicle being unlocked using a combination of face, iris, and fingerprint, and the number of records is greater than 1.

[0113] In this embodiment, because the same vehicle may be used by different users who have recorded their faces, irises, and fingerprints, each unlocking record will contain the face, iris, and fingerprint information, and then the record set of each biometric verification point can be obtained by extraction, and the unlocking record information refers to the biometric information of the corresponding biometric verification point.

[0114] In this embodiment, information saturation is mainly used to determine whether the verification point is a high-probability verification point, so as to provide a basis for subsequent element screening, thereby ensuring that the number of verification points in the actual matching process is small and the information corresponding to the number of verification points is reliable, reducing the computational complexity of the matching analysis and improving the efficiency of obtaining results.

[0115] In this embodiment, the verification points for any user for different biological species are the same.

[0116] The beneficial effect of the above technical solution is: by analyzing the historical unlocking records to calculate the information saturation based on each biological verification point, and then by discussing the points, the extraction quantity under each biological species is effectively obtained, which can effectively reduce the process of comparing and analyzing each element one by one, not only improving the unlocking safety but also improving the unlocking speed.

[0117] The present invention provides a multi-biometric fusion recognition system, wherein the matching analysis module further comprises:

[0118] A coefficient determination unit, used for normalizing the information features of the screening elements under the same row vector to obtain a normalization coefficient of the corresponding row vector;

[0119] The function building unit is used to build a normalization function based on the normalization coefficient of each biological species.

[0120] In this embodiment, the screened information features are input into the feature analysis model to obtain the corresponding normalization coefficient, wherein the normalization coefficient is obtained by relying on a model based on fingerprints, irises, and faces recorded in advance, and then training the model based on samples of different information features and normalization coefficients matching multiple features. The value range of the normalization coefficient is 0 to 1, and each row vector corresponds to a normalization coefficient, mainly to determine whether the coefficient is within the specified standard range.

[0121] The beneficial effect of the above technical solution is: by normalizing the features, the normalization coefficients of the row vectors are obtained, thereby facilitating the construction of a normalization function, thereby facilitating subsequent matching with requirements.

[0122] The present invention provides a multi-biometric fusion recognition system, wherein the instruction issuing module comprises:

[0123] A factor determination unit, used to determine a matching loss factor Fc of each column vector;

[0124] ;

[0125] in, A similarity function representing the information xj of the j-th element in the corresponding column vector and the standard information Yj that is consistent with the j-th element in the pre-stored information; represents the cumulative sum function of the similarity functions except the j-th element in the corresponding column vector; max represents the maximum value symbol; represents the flexibility factor, and Represents all the corresponding column vectors The variance of

[0126] A coefficient determination unit, configured to calculate a corresponding matching coefficient Pc based on the matching loss factor;

[0127] ;

[0128] in, Represents the trace factor of the similarity function, and the value range of the trace factor is (0, ), where N2 represents the number of column vectors and is consistent with the number of biological verification points involved in each row in the initial feature matrix.

[0129] The beneficial effect of the above technical solution is: performing similarity analysis between actual information and standard information to determine the loss factor, and then subsequently calculating the matching coefficient to ensure the accuracy of the calculation and provide a reliable basis for unlocking.

[0130] The present invention provides a multi-biometric fusion recognition system, wherein the instruction issuing module further includes:

[0131] An array construction unit, used for traversing the biological images of each biological type, locking the noise points in the corresponding biological images, and deriving them laterally along the noise points to construct a noise array;

[0132] A single elimination unit is used to determine whether the noise array has periodicity. If so, a matching noise elimination method is obtained from an elimination database according to the noise characteristics of a random period to perform a single noise processing on the corresponding biological image to obtain a current image;

[0133] Otherwise, according to the interference pattern constructed by the noise points, the noise type is matched from the pattern database, and the noise elimination method based on each noise type is matched from the noise database, and the corresponding biological image is subjected to multiple noise processing to obtain the current image;

[0134] The reliability determination unit is used to sequentially obtain the clarity of each pixel in the current image to obtain the unlocking reliability.

[0135] In this embodiment, after the noise points are determined, all the noise points are exported horizontally based on the vertical coordinates, that is, all the noise points on the straight line y=a2 are obtained. Because the straight line is a horizontal line, it is horizontal, and all the noise points on the horizontal line are exported, and then the average noise value of all the noise points on the horizontal line is determined.

[0136] In this embodiment, the noise array = {the average value of each of the sequentially adjacent lateral derivation results}.

[0137] In this embodiment, for example, the noise array is: u1, u2, u3, u1, u2, u3, u1, u2, u3. At this time, u1, u2, u3 are regarded as a period, that is, there is periodicity.

[0138] In this embodiment, the elimination database includes different noise features and noise elimination methods matching the features, and the noise features are obtained from a combination-feature database including a combination based on average values ​​and features matching the combination.

[0139] In this embodiment, single noise processing refers to performing noise elimination on an image once using the elimination method.

[0140] In this embodiment, the interference pattern is obtained around the edge contour of the noise point.

[0141] In this embodiment, the pattern database includes different interference patterns and noise types matching the patterns, and the noise types may be Gaussian noise, white noise, and the like.

[0142] In this embodiment, the noise database contains noise elimination methods matching different noise types. If there are several noise types, the image needs to be eliminated several times.

[0143] In this embodiment, the unlocking reliability ranges from 0 to 1.

[0144] The beneficial effect of the above technical solution is: by laterally deriving the noise points in the image to determine whether there is a noise period, and then screening a reasonable number of elimination methods to achieve reasonable processing of the image, and then improving the optimization rationality of the matching coefficient through clarity analysis.

[0145] The present invention provides a multi-biometric fusion recognition system, wherein the reliability determination module comprises:

[0146] The statistical unit is used to count the number of successful verifications of each biological verification point involved in the same biological species during the unlocking process, and obtain the success rate of the same biological species. , where sunNc is the total number of successful verifications of all biological verification points under the same biological species;

[0147] An optimization setting unit, used to set an optimization factor Yy for each biological species according to the success ratio and unlocking reliability under the same row vector;

[0148] ;

[0149] Wherein, Kd represents the unlocking reliability of the corresponding biological species determined based on the biological image; is the reliability threshold;

[0150] A coefficient optimization unit is used to optimize the matching coefficient corresponding to each column vector based on the optimization factors under all biological species to obtain the optimization coefficient Yc;

[0151] ;

[0152] in, Represents the weights of all elements in the corresponding column vector The optimization factor of the biological species to which the largest weight element belongs;

[0153] The average processing unit is used to calculate the average value of all optimization coefficients to obtain the comprehensive reliability.

[0154] The beneficial effect of the above technical solution is: considering the success rate and unlocking reliability, the optimization factor is set to ensure the reliability of the matching coefficient, providing a reliable basis for whether the subsequent instructions can be issued, thereby ensuring the safety of unlocking.

[0155] The present invention provides a multi-biometric fusion recognition system, wherein the reliability determination unit comprises:

[0156] A clarity calibration subunit, used to calibrate the clarity level of each pixel in the current image;

[0157] The ratio determination subunit is used to determine the number of pixels that meet the reliable grade standard in the grade calibration result, and obtain the ratio to the total number of pixels corresponding to the current image;

[0158] The table matching subunit is used to obtain the unlocking reliability based on the ratio according to the ratio-reliability comparison table.

[0159] In this embodiment, the clarity level is preset, and the clarity a1-a2 corresponds to the clarity level 1, the clarity a2-a3 corresponds to the clarity level 2, and so on.

[0160] In this embodiment, meeting the reliability level standard means that the level needs to be level one.

[0161] In this embodiment, the ratio=the number of pixels at the first clarity level / the total number of pixels in the current image.

[0162] In this embodiment, the ratio-reliability comparison table includes different ratios and reliabilities matching the ratios. For example, the ratio is 100 / 103, and the corresponding reliability is 0.9.

[0163] The beneficial effect of the above technical solution is that: by calibrating the clarity level, the number of pixels that meet the standard can be directly determined, and then the ratio can be determined to obtain the unlocking reliability.

[0164] The present invention provides a multi-biometric feature fusion recognition device, which is used to apply any of the multi-biometric feature fusion recognition systems.

[0165] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A multi-biometric fusion recognition system, characterized in that: include: A matrix construction module, used to obtain multiple biological information of a user and extract features for each biological information according to biological verification points to construct an initial feature matrix of the user; A matching analysis module, used for extracting a number of elements from each row vector in the initial feature matrix and performing normalization processing to construct a normalization function; The command issuing module is used to issue an unlocking command to the vehicle when the normalized function matches the set function required for unlocking the vehicle; Otherwise, the column information of each column vector in the initial feature matrix is ​​matched and analyzed with the pre-stored information to obtain the matching coefficient, and at the same time, the biological image of each biological information is processed to obtain the unlocking reliability of the corresponding image; A reliability determination module is used to pre-process all matching coefficients and unlocking reliability to obtain a comprehensive reliability; If the comprehensive reliability is greater than the preset reliability, an unlocking command is issued to the vehicle; Otherwise, remind the user to re-identify; Wherein, the matching analysis module includes: A set construction unit is used to obtain all unlocking records on the vehicle, and extract the historical unlocking records according to the corresponding biometric verification points to obtain a record set based on each biometric verification point. ,in, represents the i-th unlocking record information under the corresponding biometric verification point; n1 represents the number of unlocking records for the biometric verification point; a saturation calculation unit, for determining the information saturation of each biological verification point based on the record set; ; Among them, Xb represents the information saturation of the corresponding biological verification point; Represents the standard biometric information Yk corresponding to the biometric verification point and the i-th unlocking record information Similarity function of ; Indicates all The average value of the similarity function that satisfies the normal distribution probability density of 0.8; Indicates all The variance of the similarity function that does not satisfy the normal distribution probability density of 0.8; ln is the sign of the logarithmic function; Nc represents the number of successful verifications of the corresponding biometric verification point in the unlocking process; Ns represents the number of failed verifications of the corresponding biometric verification point in the starting process, and Ns+Nc=n1; A quantity calculation unit, used to determine a first quantity N1 of biological verification points whose information saturation is greater than a preset saturation under each biological species, and calculate an extraction quantity Tn under the corresponding biological species; ; Among them, A0 represents the correction result based on N1; Indicates the rounding up symbol; Indicates the rounding down symbol; N2 indicates the number of biological verification points under the corresponding biological species; exp indicates the exponential function symbol; Yb indicates the corresponding preset saturation; The element determination unit is used to sort all information saturations under the corresponding biological species, extract the first Tn biological verification points from the size sorting results, and then obtain the elements corresponding to the extracted first Tn biological verification points from the corresponding row vector of the initial feature matrix.

2. The multi-biometric fusion recognition system according to claim 1, characterized in that: The matrix building module comprises: A point acquisition unit, used to acquire biological verification points of each biological information from a type-organism comparison table according to the biological species, wherein the number of biological verification points of each biological information is the same; A calibration information unit, used to calibrate the biological verification point and the biological image corresponding to the biological information at the same position, and extract calibration information of each calibration point in each biological image; A feature extraction unit, used to extract information features of each calibration information to obtain a single row vector corresponding to the biological information; The matrix acquisition unit is used to sort and place all single row vectors in sequence to obtain an initial feature matrix.

3. The multi-biometric fusion recognition system according to claim 1, characterized in that: The matching analysis module further includes: A coefficient determination unit, used for normalizing the information features of the screening elements under the same row vector to obtain a normalization coefficient of the corresponding row vector; The function building unit is used to build a normalization function based on the normalization coefficient of each biological species.

4. The multi-biometric fusion recognition system according to claim 1, characterized in that: The instruction issuing module includes: A factor determination unit, used to determine a matching loss factor Fc of each column vector; ; in, A similarity function representing the information xj of the j-th element in the corresponding column vector and the standard information Yj that is consistent with the j-th element in the pre-stored information; represents the cumulative sum function of the similarity functions except the j-th element in the corresponding column vector; max represents the maximum value symbol; represents the flexibility factor, and Represents all the corresponding column vectors The variance of A coefficient determination unit, configured to calculate a corresponding matching coefficient Pc based on the matching loss factor; ; in, Represents the trace factor of the similarity function, and the value range of the trace factor is (0, ).

5. The multi-biometric fusion recognition system according to claim 4, characterized in that: The instruction issuing module further includes: An array construction unit, used for traversing the biological images of each biological type, locking the noise points in the corresponding biological images, and deriving them laterally along the noise points to construct a noise array; A single elimination unit is used to determine whether the noise array has periodicity. If so, a matching noise elimination method is obtained from an elimination database according to the noise characteristics of a random period to perform a single noise processing on the corresponding biological image to obtain a current image; Otherwise, according to the interference pattern constructed by the noise points, the noise type is matched from the pattern database, and the noise elimination method based on each noise type is matched from the noise database, and the corresponding biological image is subjected to multiple noise processing to obtain the current image; The reliability determination unit is used to sequentially obtain the clarity of each pixel in the current image to obtain the unlocking reliability.

6. The multi-biometric fusion recognition system according to claim 1, characterized in that: The reliability determination module comprises: The statistical unit is used to count the number of successful verifications of each biological verification point involved in the same biological species during the unlocking process, and obtain the success rate of the same biological species. , where sunNc is the total number of successful verifications of all biological verification points under the same biological species; An optimization setting unit, used to set an optimization factor Yy for each biological species according to the success ratio and unlocking reliability under the same row vector; ; Wherein, Kd represents the unlocking reliability of the corresponding biological species determined based on the biological image; is the reliability threshold; A coefficient optimization unit is used to optimize the matching coefficient corresponding to each column vector based on the optimization factors under all biological species to obtain the optimization coefficient Yc; ; in, Represents the weights of all elements in the corresponding column vector The optimization factor of the biological species to which the largest weight element belongs; Pc represents the corresponding matching coefficient; The average processing unit is used to calculate the average value of all optimization coefficients to obtain the comprehensive reliability.

7. The multi-biometric fusion recognition system according to claim 5, characterized in that: The reliability determination unit comprises: A clarity calibration subunit, used to calibrate the clarity level of each pixel in the current image; The ratio determination subunit is used to determine the number of pixels that meet the reliable grade standard in the grade calibration result, and obtain the ratio to the total number of pixels corresponding to the current image; The table matching subunit is used to obtain the unlocking reliability based on the ratio according to the ratio-reliability comparison table.

Images