CN111709272A - A fingerprint collection method, identity authentication method and electronic device based on small-area fingerprints - Google Patents

Abstract

The invention provides a fingerprint acquisition method, an identity authentication method and an electronic device based on a small-area fingerprint, which take all feature point data in the small-area fingerprint of a user, the distance between the small-area fingerprint and a full fingerprint central point, the number of full fingerprints, the sequence of acquiring the full fingerprints and a user identifier as user fingerprint information and authenticate the identity of the user. The invention starts from the full fingerprint, selects the optimal part of the whole full fingerprint as the authentication proof as much as possible, and has simple processing process and high processing speed.

Description

A fingerprint collection method, identity authentication method and electronic device based on small area fingerprints

technical field

The invention relates to the field of computer terminal security, in particular to a fingerprint collection method, an identity authentication method and an electronic device based on a small-area fingerprint.

Background technique

Fingerprint authentication is an authentication technology with a long history, which can be traced back to the Tang Dynasty in China. Fingerprints, knuckle prints or palm prints were found on many documents, contracts, and wills of the Tang Dynasty discovered after the liberation of China, as an important means of identifying individuals. In the following dynasties, the habit of taking fingerprints and palm molds as a guide has been used in documents. The ancient army of our country had the "Jidou Book", that is, the fingerprints of soldiers were registered for inspection. This shows that the fingerprints can be correctly classified according to the shape and structure at that time, and the classification characteristics and knowledge can be applied to social practice. The records of my country's use of fingerprints to solve crimes can also be traced back to the Qin Dynasty more than 2,000 years ago. Therefore, the feasibility of fingerprints in identity authentication is beyond doubt.

Traditional fingerprint collection methods can be divided into two categories, physical methods and chemical methods. Physical methods use powder, magnetic powder or laser to visualize and record fingerprints, while chemical methods use chemical agents to obtain special images and fluorescence generated by chemical reactions. fingerprint.

With the development of computer technology, the active collection and identification of fingerprints has become more and more convenient. For example, optical identification technology has been widely used, and fingerprint-based identity authentication is also used in access control punching, unlocking, mobile phones and other mobile devices. locks, and even bank authentication during account registration.

As fingerprint recognition devices become smaller and smaller, many mobile devices such as mobile phones have added fingerprint recognition functions, and fingerprint recognition-based identity authentication has become the preferred identity authentication method for many applications. At present, WeChat payment, Alipay payment, etc. have already realized the support for fingerprint authentication. For Android devices, the addition of Trusted Execution Environment (TEE) also provides hardware support for the identification and protection of user fingerprints. However, different fingerprint identification devices have different identification methods, different accuracy of collected fingerprints, and different identification algorithms, resulting in different degrees of loopholes in the existing technology.

Chinese patent application CN2017109201495 discloses a fingerprint image matching method based on selective extension, which uses minutiae and direction field information to match small fingerprint images, but it uses overlapping fingerprint image blocks to form a DC matrix, which increases the difficulty of the algorithm. affects accuracy.

Researchers from New York University used GAN to successfully create a "universal fingerprint", which can achieve a maximum cracking rate of 76.67% for the device. The reason for this universal fingerprint is that the device must have a certain fault tolerance rate. If If the fault tolerance rate is too low, even the slightest deviation of the fingerprints entered by the user will lead to a recognition error, and the fault tolerance rate is too high, which is unrealistic.

Since the user's fingerprint will remain on the surface of some smooth objects, for some scenarios with high security level requirements, the security factor of the fingerprint is not even as high as that of a 4-6-digit pure digital password.

At present, many existing small-area fingerprint technologies lose a lot of information of the whole fingerprint in the processing method, so it is difficult to guarantee the uniqueness of the small-area fingerprint after the processing is completed. Many existing technologies do not consider the distribution characteristics of feature points on small-area fingerprints, so the data collected on some fingerprints may be ineffective and contain less user feature information, resulting in lower recognition accuracy and lower security.

SUMMARY OF THE INVENTION

In view of the above situation and existing problems, the present invention provides a fingerprint collection method, an identity authentication method and an electronic device based on a small-area fingerprint, which are suitable for application scenarios such as single-device authentication with few users, such as mobile phones, personal computers ; Or each user can only verify on a few specific devices, such as scientific research laboratories, etc., which can maximize the use of full fingerprint information and speed up the processing speed of fingerprint verification.

A fingerprint collection method based on small-area fingerprints, the steps of which include:

1) Collect several full fingerprints of the user in turn, obtain all feature point data in the full fingerprint for each full fingerprint, and make an inscribed matrix with the center point of the full fingerprint as the center and containing most of the feature points;

2) Convolving the inscribed matrix, using the corresponding position of each convolution result obtained as each small area fingerprint, and selecting some or all of the small area fingerprints, and calculating the distance between each selected small area fingerprint and the center point of the full fingerprint;

3) Select all feature point data in the small area fingerprint and the distance between the small area fingerprint and the center point of the full fingerprint, the number of full fingerprints, the order of collecting the full fingerprints and the user ID as the user fingerprint information.

Further, the ORB algorithm is used to obtain all feature point data in the full fingerprint.

Further, the feature point data includes the position and type information of the feature point; the position of the feature point is calculated according to a two-dimensional coordinate system centered on the center point of the full fingerprint.

Further, the inscribed matrix is divided into several sub-matrices of the same shape and size, and the inscribed matrix is convolved with the set receptive field and stride.

Further, select some small area fingerprints through the following steps:

1) According to the number of feature points in each small-area fingerprint, sort all small-area fingerprints from more to less;

2) Calculate the average value of the Euclidean distance between each feature point in the small-area fingerprints of the same number of feature points and other features in the small-area fingerprint, and sort the small-area fingerprints with the same number of feature points from large to small;

3) Several small-area fingerprints before sorting are regarded as part of the small-area fingerprints.

Further, the average value of the Euclidean distance between each feature point in the selected small-area fingerprint and the center point of the full fingerprint is calculated, and the distance between the selected small-area fingerprint and the center point of the full fingerprint is obtained.

A small-area fingerprint-based identity authentication method, the steps of which include:

1) Acquire a number of full fingerprints of the user to be authenticated in turn, search for the user ID corresponding to the same number of full fingerprints in the above user fingerprint information according to the number of full fingerprints of the user to be authenticated, collect the sequence of full fingerprints, and select all feature point data in the small area fingerprint And select the distance between the small area fingerprint and the full fingerprint center point;

2) According to the sequence of collecting the full fingerprint, selecting the distance between the small area fingerprint and the center point of the full fingerprint, and obtaining the full fingerprint of the user to be authenticated, match all feature point data in the selected small area fingerprint with the full fingerprint of the user to be authenticated, and based on the matching Successfully select the number of small-area fingerprints, and determine whether the full fingerprint of the user to be verified is successfully matched;

3) If there is a full fingerprint of the user to be authenticated that has been successfully matched, it is determined that the user to be authenticated has passed the identity authentication, and the user logs in according to the user ID.

Further, if the number of successfully selected small-area fingerprints is greater than or equal to the set number, the corresponding full-fingerprint matching is successful; if the number of successfully matched small-area fingerprints is less than the set number, the corresponding full-fingerprint matching fails.

A storage medium in which a computer program is stored, wherein the computer program executes each step of the above method.

An electronic device comprising a memory and a processor having a computer program stored in the memory, the processor being arranged to run the computer program to perform the above method.

Compared with the prior art, the method proposed by the present invention has the following advantages and effects:

The invention starts from the whole fingerprint, selects the best part of the whole fingerprint as the authentication credential as much as possible, the processing process is relatively simple, and the processing speed is relatively fast.

Description of drawings

Figure 1 is an overall flow chart of an identity authentication method based on small area fingerprints.

FIG. 2 is a schematic diagram of a fingerprint collection process.

Figure 3 is a schematic diagram of the convolution process.

FIG. 4 is a schematic diagram of a fingerprint verification process.

Detailed ways

In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further elaborated below through specific embodiments and accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present invention, but not to limit the present invention.

The overall flow of the present invention is shown in FIG. 1 .

The present invention first performs fingerprint collection based on small-area fingerprints, and establishes a small-area fingerprint information database, as shown in FIG. 2 .

When the user enters the fingerprint, first locate the center point of the fingerprint, determine the rectangular boundary of the fingerprint, divide the rectangular area into a 4*4 area, and convolve the rectangle with a 2*2 receptive field and a step size of (1,1). The fingerprint feature points of the user are sampled.

During fingerprint entry, the user will be required to enter multiple fingerprints, and the number of times is set by the user. The fingerprints entered each time can be different or the same. If it is different, the order of entry should also be used during fingerprint verification. The recommended number of entries is 4-6 times, the maximum is 8 times, and 3 different fingerprints are recommended.

Each entered full fingerprint corresponds to 6 small-area fingerprints. The database needs to store the location information of the small-area fingerprint and the small-area fingerprint relative to the center of the fingerprint, the number of full fingerprints, the order in which the full fingerprints are collected, and the user ID.

The steps of fingerprint collection are as follows:

The first step is to collect the full fingerprints of the user. The number and order are determined by the user, and the number is denoted as fn.

The second step is to take out a full fingerprint entered by the user, identify the center point of the fingerprint, use the center point of the fingerprint as a reference point, and use the ORB (Oriented Brief) algorithm to identify all the feature points on the fingerprint. ORB algorithm is a more general image feature recognition algorithm.

In the third step, the feature points in the fingerprint are usually concentrated in an inverted U-shaped area. In this area, the center point identified in the second step is used as the center to make the inscribed rectangle of the fingerprint, so that 90% of the feature points can be included.

The fourth step is to divide the length and width of the rectangle into 4 equal parts, and divide the rectangle according to this standard to obtain 4*4 small rectangular areas of the same size. Using the receptive field of the size of 2*2 rectangular area, convolve it according to the step size of (1, 1), and obtain 9 convolution results. As shown in Figure 3, R0, R1, R2, B0, B1, B2, G0, G1, G2 have a total of 9 convolution results.

The fifth step is to count the distribution characteristics of the feature points in the 9 convolution results. It is preferred to sort by the number of feature points. For the same number of feature points, calculate each feature point in the convolution result and other features in the convolution result. The Euclidean distance average M of points, whichever is larger. Information about the location of each sample relative to the center is recorded. which is:

There are r feature points, {N ₀ , N ₁ , N ₂ ,...N _r-1 }

其中d(x,y)是两点之前的欧氏距离，Nn和Np分别代表两个不同的特征点，n不等于p。依照此指标筛选出特征点较为分散的小面积指纹，可提高系统容错率。选择前6个小面积指纹，将其相对于中心点的位置信息记录。

where d(x,y) is the Euclidean distance before the two points, Nn and Np represent two different feature points respectively, and n is not equal to p. According to this indicator, small-area fingerprints with scattered feature points can be screened out, which can improve the fault tolerance rate of the system. Select the first 6 small-area fingerprints and record their position information relative to the center point.

Repeat steps 2 to 5 until all fingerprints are processed.

The sixth step is to group the small-area fingerprint information, each group is 6: <small-area fingerprint 1 data, location>, <small-area fingerprint data, location>...<small-area fingerprint 6 data, location>, for each The group stores a tag information, which can be set to empty for single-user authentication. Among them, the small area fingerprint data refers to the location and type information of all feature points on the small area fingerprint, which is given by the ORB algorithm, and the location refers to the location information of the entire small area fingerprint relative to the fingerprint center reference point.

The user's multiple pairs of fingerprint data are stored in the database.

After the user's fingerprint data is collected, processed and stored, the user's identity can be verified based on the user's small-area fingerprint data. During fingerprint verification, the verification module will receive up to 8 full fingerprint information input from the user, and then store it according to the database. If the number of user entries does not match the corresponding stored number, or a certain small-area fingerprint group does not match successfully, the verification fails. Validation succeeds if and only if all succeed.

The authentication steps are shown in Figure 4:

The first step is to obtain the user's full fingerprint according to the user's input. The number and entry order are determined by the user's input, and the number is recorded as fn. There are no restrictions here.

The second step is to identify the fingerprint center for the full fingerprint input by each user, and use the ORB algorithm to identify all feature points.

The third step is to traverse the fingerprint data of all users in the database, and take out the pre-stored fingerprint data of the user. If the number of fingerprints fn entered by the user is not equal to the number of fingerprints pre-stored by the current user, the matching is skipped.

The fourth step, if the number is equal, traverse each group of small area fingerprint data of the pre-stored user fingerprint data, start from the first group, try to match, if the match fails, go back to the third step, select the next user to try to match . Since the data and position relative to the center of each small-area fingerprint in each group are recorded at the time of acquisition, there is no need to perform convolutional sampling on the fingerprint to be verified again. When matching, directly according to the position of the pre-stored small-area fingerprint data, the fingerprint to be verified can be sampled according to the center.

The fifth step is to use the Brute-Force Hamming Distance algorithm to calculate the matching score between the six pre-stored small-area fingerprints in each group and the small-area fingerprints sampled from the user input. This algorithm was introduced by Dr. Macleod in "Telecommunications Engineer's Reference Book" in 1993. It is proposed in the book to represent the number of different characters in the corresponding positions of two equal-length strings. In this example, the two strings are the descriptors for a feature point finally given by the ORB algorithm. The threshold is set to 30 (the optimal threshold observed in the experiment). When the score is higher than 30, it is considered that the current small-area fingerprint pair does not match. When there is a group with more than 1 pair of small-area fingerprints that do not match, the user to be verified is considered to be unmatched. The fingerprint does not match the pre-stored user, try the next user. If each group of small-area fingerprints are successfully matched, the authentication process ends and the authentication is successful.

The advantages of using the ORB algorithm combined with the Brute-Force Hamming Distance algorithm are that the feature extraction speed is faster, and the matching results are more reliable than other similar algorithms, taking into account reliability and performance. At the same time, the fault tolerance rate can be achieved by adjusting the threshold. The threshold is the average Hamming Distance between each matching feature, and the ORB algorithm's description of the feature point output is used as the output parameter for calculating the Hamming distance.

In the sixth step, if the fingerprint entered by the user does not correspond to a set of fingerprints pre-stored in the database, the authentication fails. If there is a set of pre-stored fingerprint data in the database that matches the entry, the authentication is successful.

The above embodiments are only used to illustrate the technical solutions of the present invention rather than limit them. Those of ordinary skill in the art can modify or equivalently replace the technical solutions of the present invention without departing from the spirit and scope of the present invention. The scope of protection shall be subject to what is stated in the claims.

Claims (10)

1. A fingerprint collection method based on a small area fingerprint, the steps comprising: 1) Collect several full fingerprints of the user in turn, obtain all feature point data in the full fingerprint for each full fingerprint, and make an inscribed matrix with the center point of the full fingerprint as the center and containing most of the feature points; 2) Convolving the inscribed matrix, using the corresponding position of each convolution result obtained as each small area fingerprint, and selecting some or all of the small area fingerprints, and calculating the distance between each selected small area fingerprint and the center point of the full fingerprint; 3) Select all feature point data in the small area fingerprint and the distance between the small area fingerprint and the center point of the full fingerprint, the number of full fingerprints, the order of collecting the full fingerprints and the user ID as the user fingerprint information. 2 . The method of claim 1 , wherein the ORB algorithm is used to obtain all feature point data in the full fingerprint. 3 . 3 . The method of claim 1 , wherein the feature point data includes position and type information of the feature point; the position of the feature point is calculated according to a two-dimensional coordinate system centered on the center point of the full fingerprint. 4 . 4. The method of claim 1, wherein the inscribed matrix is divided into several sub-matrices of the same shape and size, and the inscribed matrix is convolved with a set receptive field and stride. 5. The method of claim 1, wherein part small area fingerprints are selected by the following steps: 1) According to the number of feature points in each small-area fingerprint, sort all small-area fingerprints from more to less; 2) Calculate the average value of the Euclidean distance between each feature point in the small-area fingerprints of the same number of feature points and other features in the small-area fingerprint, and sort the small-area fingerprints with the same number of feature points from large to small; 3) Several small-area fingerprints before sorting are regarded as part of the small-area fingerprints. 6. The method according to claim 1, wherein the average value of the Euclidean distance between each feature point in the selected small-area fingerprint and the center point of the full fingerprint is calculated to obtain the distance between the selected small-area fingerprint and the center point of the full fingerprint. 7. An identity authentication method based on small-area fingerprints, the steps comprising: 1) Obtain several full fingerprints of the user to be authenticated in turn, search for the user identity corresponding to the same full fingerprint quantity in the user fingerprint information obtained by any method in claims 1-6 according to the obtained full fingerprint quantity of the user to be authenticated, and collect the full fingerprints. Order, select all feature point data in the small area fingerprint and select the distance between the small area fingerprint and the center point of the full fingerprint; 2) According to the sequence of collecting the full fingerprint, selecting the distance between the small area fingerprint and the center point of the full fingerprint, and obtaining the full fingerprint of the user to be authenticated, match all feature point data in the selected small area fingerprint with the full fingerprint of the user to be authenticated, and based on the matching Successfully select the number of small-area fingerprints, and determine whether the full fingerprint of the user to be authenticated is successfully matched; 3) If there is a successful matching full fingerprint of the user to be authenticated, it is determined that the user to be authenticated has passed the identity authentication. 8. method as claimed in claim 7, it is characterized in that, if matching successful selection small area fingerprint quantity is greater than or equal to set quantity, then to be authenticated user full fingerprint matching is successful; If matching successful selection small area fingerprint quantity is less than set quantity. If the number of users to be authenticated fails, the full fingerprint matching of the user to be authenticated fails. 9. A storage medium in which a computer program is stored, wherein the computer program is configured to execute the method of any one of claims 1-8 when run. 10. An electronic device comprising a memory and a processor having a computer program stored in the memory, the processor being arranged to run the computer program to perform the method of any of claims 1-8.

Images