WO2018082011A1

WO2018082011A1 - Living fingerprint recognition method and device

Info

Publication number: WO2018082011A1
Application number: PCT/CN2016/104566
Authority: WO
Inventors: 王信亮; 覃耀辉; 吴东承
Original assignee: 深圳市汇顶科技股份有限公司
Priority date: 2016-11-04
Filing date: 2016-11-04
Publication date: 2018-05-11
Also published as: CN106663203A; CN106663203B

Abstract

A living fingerprint recognition method and device. The living fingerprint recognition method comprises: extracting an image feature of a real-time fingerprint from a collected real-time fingerprint image (S11); comparing the image feature of the real-time fingerprint with stored fingerprint image features of a living body and a non-living body, so as to generate the credibility of the real-time fingerprint (S12); and determining, according to the value of the credibility of the real-time fingerprint, whether the real-time fingerprint is collected from the living body (S13). In the living fingerprint recognition method, by comparing an image feature of a real-time fingerprint with stored fingerprint image features of a living body and a non-living body, the credibility is generated to identify whether the obtained real-time fingerprint comes from the living body or the non-living body, thereby effectively filtering out real-time fingerprint data collected from the non-living body; the method is used in the fingerprint template update and fingerprint validity authentication, thereby improving the security of fingerprint recognition; and the method does not require any additional hardware device, thereby reducing costs.

Description

Living fingerprint identification method and device

Technical field

The embodiments of the present invention relate to the field of fingerprint identification technologies, and in particular, to a living body fingerprint identification method and device.

Background technique

The mobile terminal uses a password, which is safe but inconvenient; the mobile terminal does not use a password, but the security is reduced. However, as the mobile terminal introduces the fingerprint recognition technology, the fingerprint unlocking replaces the password unlocking, sliding unlocking, etc., so that the unlocking no longer needs other operations, and only the finger and the sensor are required to contact, and the mobile terminal is guaranteed to be safe at the same time. Increased convenience of use of mobile terminals. With the application of fingerprint recognition of mobile terminals, corresponding fingerprint recognition technologies are also increasing, such as fingerprint identification payment.

However, since there is often a large amount of personal information in the mobile terminal, the security of the user's property and privacy is involved. After the fingerprint recognition is applied to the mobile terminal, the criminals create a fake fingerprint by stealing the user fingerprint to crack the security system of the user, thereby obtaining the information of the user in the mobile terminal, and increasing the probability that the fingerprint password of the mobile terminal is recognized, and the mobile terminal is Information security poses a greater threat.

Therefore, providing a fingerprint identification technology that can identify and filter fake fingerprint patterns has become a technical problem that needs to be solved in the prior art.

Summary of the invention

The purpose of the embodiments of the present application is to provide a living body fingerprint identification method and device, which can increase the living fingerprint identification step on the existing fingerprint recognition technology, and increase the accuracy of fingerprint recognition of the mobile terminal.

The technical solutions adopted by the embodiments of the present application are as follows:

The embodiment of the present application provides a living fingerprint identification method, which is applied to a mobile terminal, and includes:

Extracting image features of real-time fingerprints from the collected real-time fingerprint images;

Comparing the image features of the real-time fingerprint with the pre-existing living and non-living fingerprint image features to generate a credibility of the real-time fingerprint;

Whether the real-time fingerprint is collected from the living body is determined according to the size of the credibility of the real-time fingerprint.

Optionally, in the embodiment of the present application, the image feature of the real-time fingerprint is compared with the pre-existing living and non-living fingerprint image features, and the credibility of generating the real-time fingerprint includes:

The authenticity of the real-time fingerprint is generated by comparing the image features of the real-time fingerprint with the pre-existing live and non-living fingerprint image features by a classifier.

Optionally, in the embodiment of the present application, the method further includes: displaying an image quality according to the living body and the non-living fingerprint image And grading the pre-existing living and non-living fingerprint images to obtain a level of the living and non-living fingerprint images, and extracting image feature vectors of each of the living and non-living fingerprint images to establish the classifier.

Optionally, in the embodiment of the present application, the classifier includes a plurality of sub-classifiers, and an image quality of each of the live-level and non-living fingerprint images is corresponding to the sub-classifier.

Optionally, in the embodiment of the present application, different image feature vectors of the living body and the non-living fingerprint image respectively have corresponding sub-classifiers, and the sub-classifiers are established according to different image feature vectors.

Optionally, in the embodiment of the present application, the comparing, by the classifier, the image features of the real-time fingerprint and the pre-existing living and non-living fingerprint image features, the credibility of the real-time fingerprint is further included:

And classifying the classifier parameters of each level of the living body and the non-living fingerprint image according to the image feature vector of each level of the living and non-living fingerprint images and the type of the classifier to establish the classifier;

Comparing the image features of the real-time fingerprint with the pre-existing living and non-living fingerprint image features, and generating the credibility of the real-time fingerprint includes:

Determining a level of the real-time fingerprint image according to an image quality of the real-time fingerprint image;

The credibility of the real-time fingerprint is generated according to the classifier parameter of the living and non-living fingerprint images having the same level as the real-time fingerprint image and the feature vector of the real-time fingerprint image.

Optionally, in the embodiment of the present application, the classifier parameters of each level of the living body and the non-living fingerprint image are trained by the support vector machine to establish the classifier.

Optionally, in the embodiment of the present application, the classifier is a hyperplane classifier, and the classifier parameter is a hyperplane classifier parameter.

Optionally, in the embodiment of the present application, determining whether the real-time fingerprint is collected from the living body according to the size of the credibility of the real-time fingerprint includes:

Whether the real-time fingerprint is collected from the living body is determined according to the size of the credibility of the real-time fingerprint and the set threshold.

Optionally, in the embodiment of the present application, after determining, according to the size of the credibility of the real-time fingerprint, whether the real-time fingerprint is collected from a living body, the method further includes:

If the real-time fingerprint is collected from the living body and matched with the preset fingerprint template, the real-time fingerprint is updated to the preset fingerprint template, otherwise it is not updated.

Optionally, in the embodiment of the present application, if the real-time fingerprint is collected from a living body and matched with a preset fingerprint template, the real-time fingerprint is updated to the preset fingerprint template, otherwise the update is not included. :

After determining that the real-time fingerprint is matched with the preset fingerprint template, determining whether the real-time fingerprint is collected from the living body according to the size of the real-time fingerprint and the set threshold, the credibility is greater than the first threshold. And updating the real-time fingerprint to the preset template, otherwise it is not updated.

If the real-time fingerprint matches the preset fingerprint template and is collected from the living body, the real-time fingerprint is determined to be a legal fingerprint, otherwise the real-time fingerprint is determined to be an illegal fingerprint.

Optionally, in the embodiment of the present application, if the real-time fingerprint matches a preset fingerprint template and is collected from a living body, determining that the real-time fingerprint is a legal fingerprint, otherwise determining that the real-time fingerprint is an illegal fingerprint includes :

After determining that the real-time fingerprint is matched with the preset fingerprint template, determining whether the real-time fingerprint is collected from the living body according to the size of the real-time fingerprint and the set threshold, the credibility is greater than the second threshold. And determining that the real-time fingerprint is a legal fingerprint, otherwise determining that the real-time fingerprint is an illegal fingerprint.

Optionally, in the embodiment of the present application, a third threshold that is greater than the second threshold is further set.

If the credibility is greater than the second threshold, less than the third threshold, and the real-time fingerprint matches the preset fingerprint template, determining, according to the security level of the mobile terminal, the real-time fingerprint Whether it is a legal fingerprint.

Optionally, in the embodiment of the present application, determining, according to the security level of the mobile terminal, whether the real-time fingerprint is a legal fingerprint includes:

If the security level of the mobile terminal is high, determining that the real-time fingerprint is an illegal fingerprint, and if the security level of the mobile terminal is low, determining that the real-time fingerprint is a legal fingerprint.

The embodiment of the present application provides a living fingerprint identification device, including:

An extraction module configured to extract an image feature of a real-time fingerprint from the collected real-time fingerprint image;

The comparison module, the image feature set as the real-time fingerprint is compared with the pre-stored live and non-living fingerprint image features, and the credibility of the real-time fingerprint is generated;

The determining module is configured to determine whether the real-time fingerprint is collected from the living body according to the size of the credibility of the real-time fingerprint.

Optionally, in the embodiment of the present application, the method further includes: an update module, configured to: if the real-time fingerprint matches a preset fingerprint template and is collected from a living body, the real-time fingerprint is updated to the fingerprint template, otherwise Update.

Optionally, in the embodiment of the present application, the method further includes: a legality authentication module, configured to: if the real-time fingerprint If the preset fingerprint template is matched and collected from the living body, the real-time fingerprint is determined to be a legal fingerprint, otherwise the real-time fingerprint is determined to be an illegal fingerprint.

The technical solution of the embodiment of the present application has the following advantages:

1) By comparing the image features of the real-time fingerprint with the pre-existing live and non-living fingerprint image features, generating credibility to verify whether the collected real-time fingerprint is from a living body or a non-living body, effectively filtering out the collected non-living body Real-time fingerprint data improves the security of fingerprint recognition.

2) The method and device provided by the present application do not need to add any hardware equipment, which reduces the cost.

DRAWINGS

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief description of the drawings used in the embodiments or the prior art description will be briefly described below. Obviously, the drawings in the following description It is a certain embodiment of the present application, and other drawings can be obtained according to the drawings without any creative work for those skilled in the art.

1 is a schematic flow chart of a living body fingerprint identification method according to an embodiment of the present application;

2 is a schematic flow chart of a specific method of step S12 in FIG. 1;

3 is a schematic structural diagram of a living body fingerprint identification device according to an embodiment of the present application;

4 is a schematic flow chart of a living body fingerprint identification method according to an embodiment of the present application;

FIG. 5 is a schematic structural diagram of a living body fingerprint identification device according to an embodiment of the present application; FIG.

6 is a schematic flowchart of a living body fingerprint identification method according to an embodiment of the present application;

7 is a schematic structural diagram of a living body fingerprint identification device according to an embodiment of the present application;

FIG. 8 is a schematic diagram of a living fingerprint identification application scenario according to an embodiment of the present application.

detailed description

The technical solutions in the embodiments of the present application are clearly and completely described in the following with reference to the accompanying drawings in the embodiments of the present application. It is a part of the embodiments of the present application, and not all of the embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present application without departing from the inventive scope are the scope of the present application.

The technical solutions of the present application are further described below through specific implementation manners.

FIG. 1 is a schematic flowchart of a method for identifying a living body fingerprint according to an embodiment of the present invention. As shown in FIG. 1 , the method is applied to a mobile terminal, such as a smart phone, a tablet computer, a notebook computer, etc., and specifically includes:

S11. Extract image features of real-time fingerprints from the collected real-time fingerprint images.

The collected real-time fingerprint image includes a real-time fingerprint image of the user acquired by the fingerprint sensor set on the mobile terminal when the user uses the fingerprint recognition. The real-time fingerprint image is collected and processed by the processor of the mobile terminal, and the fingerprint image processing may include fingerprint image enhancement processing, denoising processing, binarization processing, refinement processing, curve-wave transformation, etc., through this series of processing methods. The image features of the real-time fingerprint can be obtained, such as the texture segment of the fingerprint, the node position of the fingerprint, and the trend of the fingerprint ridge.

S12: Comparing the image features of the real-time fingerprint with the pre-existing living and non-living fingerprint image features, and generating the credibility of the real-time fingerprint.

Specifically, in this embodiment, the pre-stored non-living fingerprint image may include an offline fingerprint image obtained from the network and having an invisible fingerprint characteristic, and extracting an image feature of the non-living fingerprint image from the offline fingerprint image, the processor Taking a large number of non-living fingerprint image features of the same process as samples, the characteristics of the fake fingerprint of each process can be obtained. The more pre-stored non-living fingerprint images, the larger the sample capacity, and the more non-living fingerprint images are obtained. Accurate, so that the more accurate the comparison result in step S12, the higher the accuracy of the obtained credibility.

In this embodiment, the non-living fingerprints produced by any of the processes are significantly different from the live fingerprints collected online, such as the gray scale changes of the pseudo-fingerprints made by the silica gel are relatively uniform, and the ridge width changes regularly; The fake fingerprints made of wood glue are more scattered and smooth.

For example, about 2000 non-living fingerprint images may be acquired first, in which different manufacturing processes occupy approximately the same proportion in the total sample. According to different manufacturing processes, it classifies, summarizes and acquires corresponding fingerprint image features, and generates a non-living fingerprint image feature database. The non-living fingerprint image feature database may be generated locally by the processor of the mobile terminal, or may be updated by the external processor to update the non-living fingerprint image feature database to the mobile terminal.

Compared with the pre-stored non-living fingerprint, the pre-stored live fingerprint image is a fingerprint image directly collected from the human body, and the image sample of such fingerprint can be obtained through an offline image obtained by the online fingerprint feature; or can be directly used online. The preset fingerprint template of the mobile terminal; or both. Similarly, the larger the capacity of the pre-stored live fingerprint image sample, the richer the feature database of the living fingerprint image generated according to the sample, and the comparison of the image feature of the real-time fingerprint with the pre-existing living and non-living fingerprint image features. Accurate, the accuracy of the generated credibility is higher.

In this embodiment, the image feature of the real-time fingerprint is compared with the pre-existing living and non-living fingerprint image features in step S12, and the credibility of the real-time fingerprint may specifically include:

The image features of the extracted real-time fingerprints are compared with the pre-existing living and non-living fingerprint image features, and the similarity between the real-time fingerprint image features and the living body and non-living fingerprint image features is obtained; according to the similarity and the corresponding fingerprint features The contribution degree is combined to obtain the credibility of the real-time fingerprint to determine whether the real-time fingerprint is collected from the living body or collected from the non-living body.

Specifically, the different fingerprint feature contributions may refer to the influence of different fingerprint features on the judgment result. For example, the similarity of the real-time fingerprint collection from the living body is 80% according to the texture dispersion degree feature, and the contribution degree of the texture dispersion degree feature is obtained. 0.3, according to the variation of ridge width characteristics, the similarity of real-time fingerprint acquisition from the living body is 30%, and the contribution of the ridge width variation feature is 0.7, then the reliability obtained by the combination is 80%*0.3+30. %*0.7=45%.

Alternatively, in this embodiment, the authenticity of the real-time fingerprint may also be generated by comparing the image features of the real-time fingerprint with the pre-existing live and non-living fingerprint image features by the classifier.

The classifier is a function or model. After the function or model is established, the collected real-time fingerprint data can be mapped to the living fingerprint type and the non-living fingerprint category, and the credibility of the real-time fingerprint is output according to the mapping result.

Specifically, the living and non-living fingerprint image features are divided into a basic sample, a training sample, a test sample, a classifier is established according to the basic sample, and a classifier basic function is obtained; and the classifier is trained according to the training sample to determine an accurate parameter of the classifier, The classifier is executed on the test sample, the test result is generated, and the test result is compared with the real result to determine the accuracy. If the accuracy of the classifier can meet the requirements of fingerprint recognition, the determined classifier can be used to classify the collected real-time fingerprints, map them to their corresponding live fingerprint categories, non-living fingerprint categories, and output real-time fingerprints. The credibility to determine whether the real-time fingerprint is collected from the living body or from the non-living body.

S13. Determine, according to the size of the credibility of the real-time fingerprint, whether the real-time fingerprint is collected from a living body.

The credibility represents the probability that the collected real-time fingerprint is a living fingerprint, and whether the real-time fingerprint is collected from the living body is determined according to the size of the credibility of the real-time fingerprint and the set threshold.

The real-time fingerprint credibility is compared with the set threshold. If the real-time fingerprint credibility is greater than the preset threshold, the fingerprint is determined to be collected from the living body, otherwise it is collected from the non-living body.

In this embodiment, the threshold may be preset according to the accuracy of the classifier, or the threshold may be calculated according to the usage habit of the mobile terminal user, or may include a preset threshold and a threshold calculated according to the usage habit of the user.

The user's usage habits include the security level of the mobile terminal, etc. The higher the security level, the larger the threshold determined according to the security level, and the threshold is generated according to the user's usage habits, which is more in line with the user's habits, and improves the mobile terminal's habits. safety.

The threshold may be multiple. Different thresholds divide the credibility into different levels, and then the judgment result is obtained according to the level at which the credibility is located. For example, setting two thresholds divides the credibility into three levels, respectively High, medium, and low, when the level is low, the real-time fingerprint is directly judged from the non-living body. When the level is high, the real-time fingerprint is directly collected from the living body. When the level is medium, the real-time fingerprint can be judged from the living body or the non-living body. According to the user's habit of using the mobile terminal, such as the security level, it is determined whether the real-time fingerprint is collected from the living body. If the security level is high, the real-time fingerprint is collected from the non-living body. If the security level is low, the real-time fingerprint is determined to be collected from the living body.

Specifically, in the embodiment shown in FIG. 1 or any other embodiment, the establishment of the classifier may be implemented by:

According to the image quality of the living body and the non-living fingerprint image, the pre-existing living body and the non-living fingerprint image are graded to obtain the level of the living body and the non-living fingerprint image, and the image feature vector of each level of the living body and the non-living fingerprint image is extracted to establish a classifier. .

The above steps of establishing the classifier may be included in the above step S12, either before step S12, after S11, or before step S11.

Image quality is an evaluation of the visual perception of an image. The quality of the fingerprint image can be determined according to the characteristics of the acquired fingerprint image such as contrast and texture definition. Grading the fingerprint image according to the image quality can make the established classifier more detailed, and the generated credibility accuracy rate is higher.

For example, the fingerprint image can be divided into five levels according to the image quality, and the image quality is excellent, good, standard, poor, and extremely poor.

It is also possible to classify living and non-living fingerprint images according to spatial relationships, as long as the accuracy of credibility can be improved.

Specifically, in this embodiment, the classifier may include a plurality of sub-classifiers, and the image quality of each of the living-level and non-living fingerprint images is corresponding to the sub-classifier. If the image level includes n levels, the number of sub-classifiers is at least n.

According to the image quality of the living body and the non-living fingerprint image, the pre-existing living body and the non-living fingerprint image are classified to obtain the level of the living body and the non-living fingerprint image, and then the image feature is used as a sample to establish a classifier, and the image feature includes the texture segment of the fingerprint. , the node position of the fingerprint, the trend of the fingerprint ridge line, and the like. However, image features are not specific values and cannot be directly used as samplers to build classifiers. Therefore, it is necessary to statistically obtain image feature vectors for image features, and use the image feature vectors as samples to establish a classifier.

Specifically, in this embodiment, the image feature vector includes a vector for characterizing the image feature obtained after the image feature of the fingerprint image is extracted, such as a binary statistical feature vector obtained according to the binary statistical feature, according to the phase. The statistical feature vector obtained by the statistical feature. An image feature vector of each level of the living body and the non-living fingerprint image is extracted to establish a classifier, and specifically includes:

Since different image feature vectors are acquired in different ways and have little relationship with each other, different sub-classifiers with different parameters should be established according to different image feature vectors, and the sub-classifiers are combined to obtain a classifier.

Specifically, in this embodiment, each level includes at least one sub-classifier that is established according to image feature vectors of different fingerprint images. If the type of image feature vectors is m, the sub-category corresponding to each image feature vector The number of devices is at least n, and the total number of sub-classifiers is at most m*n.

2 is a schematic flowchart of a specific method in step S12 of FIG. 1. As shown in FIG. 2, after the fingerprint image is classified by image quality, step S12 is specifically illustrated by using a hyperplane classifier as an example, and specifically includes:

S121. Train the classifier parameters of each level of the living body and the non-living fingerprint image according to the image feature vector of each level of the living body and the non-living fingerprint image and the type of the classifier to establish a classifier.

After the classifier is established according to the image feature vector of each level of the living body and the non-living fingerprint image, the classifier is trained according to the image feature vector of each level of the living body and the non-living fingerprint image and the type of the classifier, and each level is obtained. The classifier parameter Wt of the living and non-living fingerprint images, the higher the level, the larger the classifier parameters, and the greater the credibility obtained.

Similarly, the classifier parameters Wt obtained by the sub-classifiers established according to different image feature vectors v at the t level are also different. A sub-classifier combination will be established to obtain a hyperplane classifier suitable for generating real-time fingerprint image credibility.

S122. Determine a level of the real-time fingerprint image according to the image quality of the real-time fingerprint image.

The criterion for determining the quality level at which the image quality is based on the image quality of the real-time fingerprint is the same as in step S121.

Different levels of classifiers are established according to different image quality, and after determining the image quality of the real-time fingerprint, since the classifier parameters corresponding to different image quality levels are different, the sub-corresponding sub-image is determined according to the level t of the image quality. The classifier can get higher accuracy.

S123. Generate a credibility of the real-time fingerprint according to the classifier parameter of the live and non-living fingerprint images and the feature vector of the real-time fingerprint image having the same level as the real-time fingerprint image.

Specifically, in this embodiment, the classifier parameters of each level of the living body and the non-living fingerprint image are trained by the support vector machine to establish a classifier. After establishing the classifier, according to the real-time fingerprint image feature vector D and the classifier The parameter Wt generates a credibility Yt=∑(Wt*D) of the real-time fingerprint image.

Specifically, in this embodiment, a Support Vector Machine (SVM) supervises a learning model, analyzes data, identifies a model, and uses classification and regression through related algorithms. The parameters of the classifier are obtained by the support vector machine training, which can be better applied to high-dimensional recognition, and is especially suitable for the classifier which is composed of sub-classifiers established by multiple levels and multiple image feature vectors in the present application.

Specifically, in this embodiment, the fingerprint image feature includes a fingerprint texture feature, such as a binary statistical feature of the fingerprint image, a phase statistical feature, and the like.

The classifier may be pre-established, and may be established before the real-time fingerprint is acquired, and the image features of the real-time fingerprint are compared with the pre-existing live and non-living fingerprint image features. This embodiment is not limited herein.

The living body fingerprint identification method provided by the embodiment provides a credibility by comparing the image features of the real-time fingerprint with the pre-existing live and non-living fingerprint image features to verify whether the collected real-time fingerprint is from a living body or a non-living body. The real-time fingerprint data collected from the non-living body is effectively filtered out, and the living fingerprint identification method provided in the embodiment does not need to add any hardware device, thereby reducing the cost.

FIG. 3 is a schematic structural diagram of a living body fingerprint identification device according to an embodiment of the present application, as shown in FIG. 3, including:

The extraction module 301 is configured to extract image features of the real-time fingerprint from the collected real-time fingerprint images.

The extraction module processes the acquired real-time fingerprint image to obtain image features of the real-time fingerprint for comparison.

The comparison module 302 compares the image features set as real-time fingerprints with the pre-stored live and non-living fingerprint image features to generate the credibility of the real-time fingerprint.

The comparison module compares the image features of the real-time fingerprint with the pre-existing live and non-living fingerprint image features, and determines the reliability according to the comparison result.

The determining module 303 is configured to determine whether the real-time fingerprint is collected from the living body according to the size of the credibility of the real-time fingerprint.

The living body fingerprint identification device provided by the embodiment provides a credibility by comparing the image features of the real-time fingerprint with the pre-existing live and non-living fingerprint image features to verify whether the collected real-time fingerprint is from a living body or a non-living body. The real-time fingerprint data collected from the non-living body is effectively filtered out, and the living fingerprint identification device provided in the embodiment does not need to add any hardware device, thereby reducing the cost.

FIG. 4 is a schematic flowchart of a live fingerprint identification method according to an embodiment of the present disclosure. The live fingerprint identification method provided in this embodiment may be used to update a fingerprint template, as shown in FIG. 4, which includes:

S41. Extract image features of real-time fingerprints from the collected real-time fingerprint images.

Specifically, in the embodiment, extracting the image feature specifically includes: first performing wavelet transform on the image to obtain an image of different frequencies; and performing denoising on the image after wavelet transform, and reconstructing the image after de-drying to obtain The image after denoising; subtracting the image after decontamination from the original image to obtain a noise map; calculating the standard deviation of the noise map, obtaining a standard deviation map, and obtaining an image feature vector according to the standard deviation graph.

S42. The image features of the real-time fingerprint are compared with the pre-existing live and non-living fingerprint image features, and the credibility of the real-time fingerprint is generated.

Specifically, in the embodiment, the authenticity is outputted by the classifier comparing the image features of the real-time fingerprint image with the pre-existing live and non-living fingerprint image features.

Credibility is used to measure whether the collected real-time fingerprints are collected from living data. By comparing the image features of real-time fingerprints with the pre-existing credibility of live and non-living fingerprint image features, the real-time fingerprints collected can be visually represented. Close to the living fingerprint or non-living fingerprint, the judgment result is more simple.

S43. Determine, according to the size of the credibility of the real-time fingerprint, whether the real-time fingerprint is collected from a living body.

Specifically, in this embodiment, whether the real-time fingerprint is collected from the living body is determined by the size of the credibility and the threshold of the credibility.

The steps S41, S42, and S43 of the method for identifying the living body fingerprint provided in the embodiment of the present application are the same as the embodiment of the method for identifying the living body fingerprint provided in FIG. 1, and are not described herein.

S44. If the real-time fingerprint matches the preset fingerprint template and is collected from the living body, the real-time fingerprint is updated to the fingerprint template, otherwise it is not updated.

Specifically, in this embodiment, the first real-time fingerprint is collected from the living body, and then the real-time fingerprint is matched with the preset template, and the real-time fingerprint is matched with the preset template, and then the real-time fingerprint is collected from the living body, thereby saving Fingerprint recognition time and reduce the false positive rate of live fingerprint recognition.

Specifically, in this embodiment, the matching between the real-time fingerprint and the preset template includes:

Obtain real-time fingerprint images;

Processing real-time fingerprint images, including image enhancement, denoising, etc.;

Feature extraction is performed on the processed image to obtain the same type of feature value as the preset fingerprint template, and the feature categories mainly include fingerprint nodes and fingerprint distribution rules;

The feature value of the obtained real-time fingerprint is compared with the feature value of the pre-stored fingerprint template, and the similarity between the real-time fingerprint and the pre-stored fingerprint template is obtained, and whether the real-time fingerprint matches the preset template is determined according to the similarity.

Specifically, in the embodiment, the similarity includes a feature point matching number, a histogram distribution similarity, and the like, and a metric parameter that can represent the similarity between the images.

Specifically, in this embodiment, after determining that the real-time fingerprint matches the preset fingerprint template, determining whether the real-time fingerprint is collected from the living body according to the size of the real-time fingerprint and the set threshold value, If the credibility is greater than the first threshold, the real-time fingerprint is updated to the preset template, otherwise it is not updated. The fingerprint template refers to the user fingerprint information input by the user before using the fingerprint recognition of the mobile terminal as a preset fingerprint template, and is used for subsequent fingerprint identification.

Updating the fingerprint template means that after the fingerprint recognition is passed, the collected real-time fingerprint data is merged with the original, and the fingerprint image features that are not present or unclear in the original fingerprint template are updated to the preset fingerprint template to form a new one. The fingerprint template improves the fingerprint data of the user and reduces the rejection rate of subsequent fingerprint recognition.

When the fingerprint template is updated, the real-time fingerprint is directly updated into the fingerprint template after the real-time fingerprint matches the preset fingerprint template. Therefore, if the fingerprint collected from the non-living body is once updated, it will be updated to the fingerprint template; if the fingerprint feature collected from the non-living body is updated to the fingerprint template, it will be easy to collect subsequent fingerprints from the non-living fingerprint. The identification is passed, resulting in a decrease in the security of fingerprint recognition of the mobile terminal, which brings a large privacy hazard to the user.

The living body fingerprint identification method provided by the embodiment determines that the real-time fingerprint matches the preset fingerprint template and the real-time fingerprint is collected from the living body, effectively filtering out the real-time fingerprint data collected from the non-living body, and filtering out the fingerprint collected from the non-living body. After the data is updated to the fingerprint template, the rejection rate is reduced on the basis of ensuring the accuracy of the fingerprint recognition, and the living fingerprint identification method provided in the embodiment does not need to add any hardware equipment, thereby reducing the cost.

FIG. 5 is a schematic structural diagram of a living body fingerprint identification device according to an embodiment of the present disclosure. The living body fingerprint identification device provided in this embodiment may be used to update a fingerprint template, as shown in FIG. 5, which includes:

The extraction module 501 is configured to extract image features of the real-time fingerprint from the collected real-time fingerprint images.

The comparison module 502 compares the image features set as real-time fingerprints with the pre-stored live and non-living fingerprint image features to generate a credibility of the real-time fingerprint.

The determining module 503 is configured to determine whether the real-time fingerprint is collected from the living body according to the size of the credibility of the real-time fingerprint.

The update module 504 is configured to update the real-time fingerprint to the fingerprint template if the real-time fingerprint matches the preset fingerprint template and is collected from the living template, and is not updated.

The living body fingerprint identification device provided by the embodiment determines that the real-time fingerprint matches the preset fingerprint template and the real-time fingerprint is collected from the living body, effectively filtering out the real-time fingerprint data collected from the non-living body, and filtering out the fingerprint collected from the non-living body. After the data is updated to the fingerprint template, the rejection rate is reduced on the basis of the accuracy of the fingerprint recognition, and the living fingerprint identification device provided in the embodiment does not need to add any hardware equipment, thereby reducing the cost.

FIG. 6 is a schematic flowchart of a living body fingerprint identification method according to an embodiment of the present disclosure. The living body fingerprint identification method provided in this embodiment can be used for authenticating a fingerprint, as shown in FIG. 6 , which includes:

S61. Extract image features of real-time fingerprints from the collected real-time fingerprint images.

S62. Compare the image features of the real-time fingerprint with the pre-existing living and non-living fingerprint image features to generate a credibility of the real-time fingerprint.

The image features of the real-time fingerprint image are compared with pre-existing live and non-living fingerprint image features, and credibility is generated.

S63. Determine, according to the magnitude of the credibility of the real-time fingerprint, whether the real-time fingerprint is collected from a living body.

Specifically, in the application embodiment, whether the real-time fingerprint is collected from the living body is determined according to the size of the credibility of the real-time fingerprint and the set threshold.

The steps S61, S62, and S63 of the method for identifying the living body fingerprint provided in the embodiment of the present application are the same as the embodiment of the method for identifying the living body fingerprint provided in FIG. 1, and are not described herein.

S64. If the real-time fingerprint matches the preset fingerprint template and is collected from the living body, the real-time fingerprint is determined to be a legal fingerprint, otherwise the real-time fingerprint is determined to be an illegal fingerprint.

The order in which the real-time fingerprint is matched with the preset template and the order in which the real-time fingerprint is collected from the living body can be adjusted, and no limitation is imposed here.

Specifically, in this embodiment, the real-time fingerprint is first matched with the preset template, and then the real-time fingerprint is collected from the living body to reduce the false positive rate of the living fingerprint identification.

Specifically, in this embodiment, the legality authentication is used to determine whether the fingerprint is correct. If the fingerprint legality authentication is passed, it may be determined that the collected real-time fingerprint is from the user and performs subsequent operations, such as unlocking, paying, etc. If the fingerprint legality authentication fails, it can be determined that the collected real-time fingerprint does not come from the user himself, and the fingerprint is restarted.

Specifically, in this embodiment, after determining that the real-time fingerprint matches the preset fingerprint template, determining whether the real-time fingerprint is collected from the living body according to the size of the real-time fingerprint and the set threshold value, If the credibility is greater than the second threshold, the real-time fingerprint is determined to be a legal fingerprint, otherwise the real-time fingerprint is determined to be an illegal fingerprint.

Alternatively, in this embodiment, the second threshold is greater than the third threshold, and if the reliability is greater than the second threshold, less than the third threshold, and the real time is If the fingerprint matches the preset fingerprint template, it is determined whether the real-time fingerprint is a legal fingerprint according to the security level of the mobile terminal.

Specifically, in this embodiment, the security level of the mobile terminal represents the degree of protection of the mobile terminal. According to the level of security, real-time fingerprint credibility can be set. Different security levels represent different security requirements. Therefore, the corresponding credibility thresholds are also different.

Specifically, in this embodiment, the security level is that the user performs security settings on his mobile terminal according to factors such as his own user habits and the environment in which he or she is located. The level of security level can include multiple to accommodate different situations. According to the credibility, it is determined whether the real-time fingerprint is a legal fingerprint, and the security level of the mobile terminal can be considered.

If the security level is high, the user has high security requirements for the mobile terminal. When verifying the legality of the fingerprint, the passing condition should be increased, and the corresponding second/third threshold is higher. Otherwise, the passing condition is lowered. Its corresponding threshold is lower.

The living body fingerprint identification method provided by the embodiment further determines that the real-time fingerprint matches the preset fingerprint template and the real-time fingerprint is collected from the living body, effectively filtering out the real-time fingerprint data collected from the non-living body, and then authenticating the legality of the fingerprint, further The security of the fingerprint identification of the mobile terminal is improved, and the living fingerprint identification method provided by the embodiment does not need to add any hardware equipment, thereby reducing the cost.

FIG. 7 is a schematic structural diagram of a living body fingerprint identification device according to an embodiment of the present disclosure. The living body fingerprint identification device provided in this embodiment can be used for authenticating a fingerprint, as shown in FIG. 7 , which includes:

The extraction module 701 is configured to extract image features of the real-time fingerprint from the collected real-time fingerprint images.

The comparison module 702 compares the image features set as real-time fingerprints with the pre-stored live and non-living fingerprint image features to generate a credibility of the real-time fingerprint.

The determining module 703 is configured to determine whether the real-time fingerprint is collected from the living body according to the size of the credibility of the real-time fingerprint.

The legality authentication module 704 is configured to determine that the real-time fingerprint is a legal fingerprint if the real-time fingerprint is matched with a preset fingerprint template and is collected from a living body, otherwise the real-time fingerprint is determined to be an illegal fingerprint.

The living body fingerprint identification device provided by the embodiment further determines that the real-time fingerprint matches the preset fingerprint template and the real-time fingerprint is collected from the living body, effectively filtering out the real-time fingerprint data collected from the non-living body, and then authenticating the legality of the fingerprint, further The security of the fingerprint recognition of the mobile terminal is improved, and the living fingerprint identification device provided by the embodiment does not need to add any hardware device, thereby reducing the cost.

FIG. 8 is a schematic diagram of a live fingerprint identification application scenario according to an embodiment of the present application, as shown in FIG. 8 , including:

S801. Collect real-time fingerprint data. The real-time fingerprint data refers to real-time fingerprint data collected by the mobile terminal through the fingerprint sensor when the user performs fingerprint recognition.

The collected fingerprint data is processed by enhancement, denoising, etc., and the pre-processed image is obtained. The pre-processed image is subjected to curvelet transform and other methods to extract image features.

S802. Determine whether the real-time fingerprint is recognized. If it is identified, proceed to step S803, otherwise terminate. It is judged whether the real-time fingerprint is recognized or not, and the similarity is obtained by comparing the collected real-time fingerprint with the preset fingerprint template, and the similarity is obtained according to the similarity degree.

Specifically, in this embodiment, the collected real-time fingerprint and the preset fingerprint template are compared with the fingerprint image feature of the fingerprint template and the image feature of the real-time fingerprint, and the similarity obtained according to the image feature includes: feature point matching The number, histogram distribution similarity, etc., can represent metric parameters of similarity between images. The greater the similarity, the greater the image matching probability.

S803. Extract texture features of real-time fingerprints, and calculate credibility. Credibility refers to the probability of real-time fingerprint collection from the living body. The credibility can be obtained by comparing the image features of the real-time fingerprint with the pre-existing live and non-living fingerprint image features.

Specifically, in the embodiment, the credibility is obtained by comparing the image features of the real-time fingerprint with the texture features of the pre-existing living and non-living fingerprints by the classifier, and the texture feature is an important component of the image feature, including: a fingerprint image. Binary statistical features, phase statistical features, etc. For example, by extracting LBP or LPQ and other related features, the SVM is used for classification training to establish a classification hyperplane. The farther away from the hyperplane, the higher the credibility, and the farther away from the hyperplane, the more credible. The worse the degree.

S804. Determine a credibility level of the real-time fingerprint according to the credibility and the threshold, and set two thresholds to classify the credibility level into three levels: high, medium, and low.

Specifically, in this embodiment, the two thresholds include a second threshold and a third threshold used when authenticating the fingerprint, wherein the third threshold is multiplexed into a first threshold of the fingerprint template update.

Specifically, in this embodiment, the credibility is greater than the third threshold, and the credibility level is high, indicating that the real-time fingerprint must be collected from the living body, and the credibility is greater than the second threshold is less than the third threshold, and the credibility level is medium. , indicating that the real-time fingerprint is collected from the living body or the non-living body. If the reliability is less than the second threshold, the credibility level is low, indicating that the real-time fingerprint must be collected from the non-living body.

If the reliability is high, step S805 is performed. If the reliability is low, step S806 is performed. If the reliability is medium, step S807 is performed.

S805, determining that the fingerprint is a legal fingerprint and the fingerprint template is updated.

When the credibility level is high, it is determined that the real-time fingerprint is collected from the living body, and the fingerprint is a legal fingerprint and the fingerprint template is updated.

S806. The fingerprint is determined to be an illegal fingerprint, and the fingerprint template is not updated.

When the credibility level is low, it is determined that the real-time fingerprint collection is not self-living, and the fingerprint is an illegal fingerprint, and the fingerprint template is not updated.

S807. Determine a security level of the mobile terminal. If the security level is low, step S808 is performed, and if the security level is high, step S809 is performed.

S808: The fingerprint is determined to be a legal fingerprint, but the fingerprint template is not updated.

S809: The fingerprint is determined to be an illegal fingerprint, and the fingerprint template is not updated.

The living fingerprint identification application scenario provided by the embodiment includes the addition of the living fingerprint identification The fingerprint template update and fingerprint legality authentication greatly improve the security of the mobile terminal without increasing any hardware equipment and reduce the cost by increasing the recognition step of the living body.

The device embodiments described above are merely illustrative, wherein the modules described as separate components may or may not be physically separate, and the components displayed as modules may or may not be physical modules, ie may be located A place, or it can be distributed to multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the embodiment. Those of ordinary skill in the art can understand and implement without deliberate labor.

Through the description of the above embodiments, those skilled in the art can clearly understand that the various embodiments can be implemented by means of software plus a necessary general hardware platform, and of course, by hardware. Based on such understanding, the above-described technical solutions may be embodied in the form of software products in essence or in the form of software products, which may be stored in a computer readable storage medium such as ROM/RAM, magnetic Discs, optical discs, etc., include instructions for causing a computer device (which may be a personal computer, server, or network device, etc.) to perform the methods described in various embodiments or portions of the embodiments.

Finally, it should be noted that the above embodiments are only used to explain the technical solutions of the present application, and are not limited thereto; although the present application is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still The technical solutions described in the foregoing embodiments are modified, or the equivalents of the technical features are replaced by the equivalents. The modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

A living fingerprint identification method is applied to a mobile terminal, which is characterized in that it comprises:

Extracting image features of real-time fingerprints from the collected real-time fingerprint images;

Comparing the image features of the real-time fingerprint with the pre-existing living and non-living fingerprint image features to generate a credibility of the real-time fingerprint;

Whether the real-time fingerprint is collected from the living body is determined according to the size of the credibility of the real-time fingerprint.
The method according to claim 1, wherein the comparing the image features of the real-time fingerprint with the pre-existing living and non-living fingerprint image features, and generating the credibility of the real-time fingerprint comprises:

The authenticity of the real-time fingerprint is generated by comparing the image features of the real-time fingerprint with the pre-existing live and non-living fingerprint image features by a classifier.
The method according to claim 2, further comprising: classifying the pre-existing living body and the non-living fingerprint image according to image quality of the living body and the non-living fingerprint image to obtain a level of the living body and the non-living fingerprint image, Image feature vectors for each level of live and non-living fingerprint images are extracted to establish the classifier.
The method according to claim 3, wherein the classifier comprises a plurality of sub-classifiers, and image quality of each of the living-level and non-living fingerprint images is corresponding to the sub-classifier.
The method according to claim 4, wherein different image feature vectors of the living body and the non-living fingerprint image respectively have corresponding sub-classifiers, and the sub-classifiers are established according to different image feature vectors.
The method according to claim 3, wherein the comparing the image features of the real-time fingerprint with the pre-existing live and non-living fingerprint image features by the classifier to generate the credibility of the real-time fingerprint further comprises:

And classifying the classifier parameters of each level of the living body and the non-living fingerprint image according to the image feature vector of each level of the living and non-living fingerprint images and the type of the classifier to establish the classifier;

Comparing the image features of the real-time fingerprint with the pre-existing living and non-living fingerprint image features, and generating the credibility of the real-time fingerprint includes:

Determining a level of the real-time fingerprint image according to an image quality of the real-time fingerprint image;

The credibility of the real-time fingerprint is generated according to the classifier parameter of the living and non-living fingerprint images having the same level as the real-time fingerprint image and the feature vector of the real-time fingerprint image.
The method of claim 6 wherein each level is trained by a support vector machine The classifier parameters of the live and non-living fingerprint images to establish the classifier.
The method of claim 6 wherein said classifier is a hyperplane classifier and said classifier parameter is a hyperplane classifier parameter.
The method according to claim 1, wherein determining whether the real-time fingerprint is collected from the living body according to the size of the credibility of the real-time fingerprint comprises:

Whether the real-time fingerprint is collected from the living body is determined according to the size of the credibility of the real-time fingerprint and the set threshold.
The method according to any one of claims 1 to 9, wherein, after determining whether the real-time fingerprint is collected from a living body according to the magnitude of the credibility of the real-time fingerprint, the method further comprises:

If the real-time fingerprint is collected from the living body and matched with the preset fingerprint template, the real-time fingerprint is updated to the preset fingerprint template, otherwise it is not updated.
The method according to claim 10, wherein if the real-time fingerprint is collected from a living body and matched with a preset fingerprint template, the real-time fingerprint is updated to the preset fingerprint template, otherwise Updates include:

After determining that the real-time fingerprint is matched with the preset fingerprint template, determining whether the real-time fingerprint is collected from the living body according to the size of the real-time fingerprint and the set threshold, the credibility is greater than the first threshold. And updating the real-time fingerprint to the preset template, otherwise it is not updated.
The method according to any one of claims 1 to 9, wherein, after determining whether the real-time fingerprint is collected from a living body according to the magnitude of the credibility of the real-time fingerprint, the method further comprises:

If the real-time fingerprint matches the preset fingerprint template and is collected from the living body, the real-time fingerprint is determined to be a legal fingerprint, otherwise the real-time fingerprint is determined to be an illegal fingerprint.
The method according to claim 12, wherein if the real-time fingerprint matches a preset fingerprint template and is collected from a living body, determining that the real-time fingerprint is a legal fingerprint, otherwise determining that the real-time fingerprint is illegal Fingerprints include:

After determining that the real-time fingerprint is matched with the preset fingerprint template, determining whether the real-time fingerprint is collected from the living body according to the size of the real-time fingerprint and the set threshold, the credibility is greater than the second threshold. And determining that the real-time fingerprint is a legal fingerprint, otherwise determining that the real-time fingerprint is an illegal fingerprint.
The method according to claim 13, wherein a third threshold greater than the second threshold is further set,

If the reliability is greater than the second threshold, less than the third threshold, and the real-time fingerprint matches the preset fingerprint template, determining, according to the security level of the mobile terminal, whether the real-time fingerprint is Legal fingerprint.
The method according to claim 14, wherein the determining, according to the security level of the mobile terminal, whether the real-time fingerprint is a legal fingerprint comprises:

If the security level of the mobile terminal is high, determining that the real-time fingerprint is an illegal fingerprint, and if the security level of the mobile terminal is low, determining that the real-time fingerprint is a legal fingerprint.
A living body fingerprint identification device, comprising:

An extraction module configured to extract an image feature of a real-time fingerprint from the collected real-time fingerprint image;

The comparison module, the image feature set as the real-time fingerprint is compared with the pre-stored live and non-living fingerprint image features, and the credibility of the real-time fingerprint is generated;

The determining module is configured to determine whether the real-time fingerprint is collected from the living body according to the size of the credibility of the real-time fingerprint.
The device according to claim 16, further comprising:

And updating the module, if the real-time fingerprint matches the preset fingerprint template and is collected from the living body, the real-time fingerprint is updated to the fingerprint template, otherwise it is not updated.
The device according to claim 16, further comprising:

The legality authentication module is configured to determine that the real-time fingerprint is a legal fingerprint if the real-time fingerprint is matched with a preset fingerprint template and is collected from a living body, otherwise the real-time fingerprint is determined to be an illegal fingerprint.