Smart watch payment method based on face recognition

文档序号：1521867 发布日期：2020-02-11 浏览：14次中文

阅读说明：本技术 一种基于人脸识别的智能手表支付方法 (Smart watch payment method based on face recognition ) 是由杨杰于 2019-10-23 设计创作，主要内容包括：本发明涉及一种基于人脸识别的智能手表支付方法,预设参考人脸几何特征和所述参考人脸几何特征的距离；当处理器判断支付程序启动时,启动心率传感器,判断所述心率传感器采集的心率曲线是否与预设参考曲线拟合,若拟合度达到拟合阈值,则：启动摄像头和距离传感器,采集人脸几何特征和距离,当所述采集得到的人脸几何特征与所述参考人脸几何特征基于采集距离和参考人脸几何特征的距离满足几何相似性,则身份确认,完成支付。(The invention relates to an intelligent watch payment method based on face recognition, which is characterized in that the distance between a reference face geometric characteristic and the reference face geometric characteristic is preset; when the processor judges that the payment program is started, the heart rate sensor is started, whether the heart rate curve collected by the heart rate sensor is fitted with a preset reference curve or not is judged, and if the fitting degree reaches a fitting threshold value, the method comprises the following steps: and starting the camera and the distance sensor, acquiring the geometric features and the distance of the face, and when the acquired geometric features of the face and the geometric features of the reference face meet geometric similarity based on the acquisition distance and the geometric features of the reference face, confirming the identity and completing payment.)

1. A smart watch payment method based on face recognition, the method comprising:

presetting the distances between the geometric features of the reference face and the geometric features of the reference face;

when the processor judges that the payment program is started, the heart rate sensor is started, whether the heart rate curve collected by the heart rate sensor is fitted with a preset reference curve or not is judged, and if the fitting degree reaches a fitting threshold value, the method comprises the following steps:

and starting the camera and the distance sensor, acquiring the geometric features and the distance of the face, and when the acquired geometric features of the face and the geometric features of the reference face meet geometric similarity based on the acquisition distance and the geometric features of the reference face, confirming the identity and completing payment.

2. The payment method of claim 1, wherein the facial geometric features comprise facial geometric features.

3. The payment method according to claim 1, wherein the distance of one of the reference face geometric features corresponds to a set of preset reference face geometric features;

the distance of the geometric features of the reference face is smaller than a distance threshold value;

the set of preset reference face geometric features is a set of preset reference face geometric features with the nose tip as the center of a circle, the distance between the reference face geometric features as a radius, a face symmetry plane including the midpoint of two eyes as a symmetry plane, a left-right total angle based on the symmetry plane smaller than an angle threshold and a height between the chin and the nose tip for sampling, and not less than 20 reference face geometric features are set.

4. The payment method according to claim 3, wherein the identity is confirmed when the collected geometric features of the face satisfy geometric similarity with any one of the set of predetermined geometric features of the reference face based on the collection distance and the distance of the geometric features of the reference face.

5. Payment method according to claim 4, wherein the geometric similarity is preferably a geometric coincidence of 90% or more.

6. The payment method as recited in claim 1, wherein the collected heart rate curve is filtered, a second derivative process is performed to obtain a second derivative curve, and the second derivative curve is fitted with the preset reference curve to determine whether the fitting degree reaches a fitting threshold.

7. The payment method according to claim 6, wherein the preset reference curve is subjected to tolerance processing, and the tolerance processing is to expand the numerical value corresponding to each time coordinate in the reference curve according to a proportional threshold value, so as to form a reference curve band.

8. The payment method of claim 7, wherein the predetermined reference curve band is fitted, and if the time ratio contained in the reference curve band exceeds a fitting threshold, the identity is determined to be identifiable.

9. The payment method according to claim 1, wherein the reference curve is a learning curve, and the learning curve obtaining method comprises:

s11, taking a heart rate curve at least having N wave crests and N wave troughs, taking a wave crest curve in a first time length before and after each wave crest is a midpoint, and taking a wave trough curve in a first time length before and after each wave trough is a midpoint;

s12, performing second derivative processing on all the peak curves to obtain a first peak second derivative curve; performing second derivative processing on all the trough curves to obtain a first trough second derivative curve;

s13, performing scatter processing on all first peak second-order derivative curves, and performing standard deviation operation on all scatters with the same abscissa; performing scatter processing on all first trough second-order derivative curves, and performing standard deviation operation on all scatter points with the same abscissa;

s14, performing interpolation processing to obtain a first reference peak second derivative curve and a first reference trough second derivative curve which are formed by standard deviation and interpolation; the first reference crest second derivative curve and the first reference trough second derivative curve form an initial learning curve or a first learning curve.

10. A payment method as claimed in claim 9, characterized in that, when the watch is taken off and worn the ith time, the following method is implemented:

si0, if the identity can be identified, the process is switched to Si 1;

si1, taking a heart rate curve at least having N wave crests and N wave troughs, taking a wave crest curve in a first time length before and after each wave crest is a midpoint, and taking a wave trough curve in a first time length before and after each wave trough is a midpoint;

si2, performing second derivative processing on all the wave crest curves to obtain an ith wave crest second derivative curve; performing second derivative processing on all the trough curves to obtain an ith trough second derivative curve;

si3, performing standard deviation operation on scattered points based on the same time abscissa on the ith peak second order derivative curve and the ith-1 reference peak second order derivative curve in the ith peak second order derivative curve and the ith-1 learning curve; performing standard deviation operation on scattered points based on the same time abscissa on the ith-1 reference trough second derivative curve in the ith trough second derivative curve and the ith-1 initial learning curve;

si4, performing interpolation processing to obtain an ith reference crest second derivative curve and an ith reference trough second derivative curve which are formed by standard deviation and interpolation; and the ith reference crest second derivative curve and the ith reference trough second derivative curve form an ith learning curve.

Technical Field

The invention relates to a payment method, in particular to a payment method of a smart watch based on face recognition.

Background

At present, the mobile phone payment generally adopts the biological characteristics for confirmation besides the password confirmation during the payment, and the confirmation adopts a single confirmation form, such as fingerprints, face appearances, irises and the like. The mobile phone identifies the input biological characteristics based on a powerful processor, thereby completing payment confirmation.

The smart watch is as the biggest wearable equipment of branch, has adopted smart watch to carry out the technique of paying at present, but these techniques all are based on built-in chip generally, like NFC carries out the expense of subway, bus. Because the chip needs to be additionally arranged and the processor of the watch is comparatively good, no perfect payment mode which does not depend on hardware of the watch is available on the market at present.

The applicant's prior application CN2019110031403 proposes a heart rate identification method, which can complete identification with a high success rate only by relying on a heart rate sensor of a smart watch. On the basis, the applicant proposes a utilization of the method, and in some smart watches provided with a camera, multiple verification is adopted during payment, so that the identity is confirmed for payment.

Disclosure of Invention

In view of the above, the present invention provides a face recognition based smart watch payment method, by which a user's identity can be more accurately recognized without adding additional hardware on the basis of a smart watch having a camera, thereby completing payment.

The specific technical scheme of the invention is as follows:

a smart watch payment method based on face recognition, the method comprising:

presetting the distances between the geometric features of the reference face and the geometric features of the reference face;

Further, the geometric features of the human face comprise geometric features of five sense organs.

Further, the distance of one reference face geometric feature corresponds to a group of preset reference face geometric features;

the distance of the geometric features of the reference face is smaller than a distance threshold value;

Further, when the acquired geometric features of the face and any one of the preset geometric features of the preset reference face satisfy geometric similarity based on the acquisition distance and the distance of the geometric features of the reference face, the identity is confirmed.

Further, the geometric similarity is preferably a geometric coincidence degree of 90% or more.

And further, filtering the acquired heart rate curve, performing second derivative processing to obtain a second derivative curve, fitting the second derivative curve with the preset reference curve, and judging whether the fitting degree reaches a fitting threshold value.

Further, carrying out tolerance processing on the preset reference curve, wherein the tolerance processing is to enlarge a numerical value corresponding to each time coordinate in the reference curve according to a proportional threshold value, so as to form a reference curve band.

And further fitting the reference curve band, and if the time proportion contained in the reference curve band exceeds a fitting threshold value, judging that the identity is identifiable.

Further, the preset reference curve is a learning curve, and the learning curve obtaining method includes:

Further, when the watch is taken off and worn for the ith time, the following method is implemented:

si0, if the identity can be identified, the process is switched to Si 1;

Through the technical means, firstly, the second derivative value of the heart rate curve is creatively used as reference, secondly, the heart rate and the face recognition are combined for identity judgment, and the recognition success rate can reach more than 95% on the basis of only depending on the success rate of 80% of the heart rate.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below, and it is obvious that the described embodiments are a part of the embodiments of the present invention, but not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The terminology used in the embodiments of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used in the examples of the present invention and the appended claims, the singular forms "a", "an", and "the" are intended to include the plural forms as well, and "a" and "an" generally include at least two, but do not exclude at least one, unless the context clearly dictates otherwise.

It should be understood that the term "and/or" as used herein is merely one type of association that describes an associated object, meaning that three relationships may exist, e.g., a and/or B may mean: a exists alone, A and B exist simultaneously, and B exists alone. In addition, the character "/" herein generally indicates that the former and latter related objects are in an "or" relationship.

It should be understood that although the terms first, second, third, etc. may be used to describe … … in embodiments of the present invention, these … … should not be limited to these terms. These terms are used only to distinguish … …. For example, the first … … can also be referred to as the second … … and similarly the second … … can also be referred to as the first … … without departing from the scope of embodiments of the present invention.

The words "if", as used herein, may be interpreted as "at … …" or "at … …" or "in response to a determination" or "in response to a detection", depending on the context. Similarly, the phrases "if determined" or "if detected (a stated condition or event)" may be interpreted as "when determined" or "in response to a determination" or "when detected (a stated condition or event)" or "in response to a detection (a stated condition or event)", depending on the context.

It is also noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a good or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such good or system. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a commodity or system that includes the element.

The camera on the smart watch is generally used for video and voice calls, and certainly can also be used for taking pictures and the like, but due to the limitation of the size of the smart watch, the camera equipped on the smart watch cannot reach the same hardware standard as a mobile phone in the current technology. Therefore, the camera on the intelligent watch is not applied to an identity recognition scene at present.

The applicant previously proposed a method of using heart rate for identification, which can be used for unlocking a smart watch. However, to use heart rate for monetary payments, its 80% success rate is clearly insufficient, and developing new algorithms, or providing more advanced heart rate sensors, based on simple heart rate monitoring, would require a lot of creative work. Based on the current situation, the method and the device are creatively provided for organically combining two completely different monitoring modes of the heart rate and the face together so as to achieve high-success-rate identity recognition and pay.

The camera of the intelligent watch monitors geometric features of the face, the distance sensor detects the distance between the camera and the nose tip, the features of the five sense organs in the geometric features are extracted to form geometric curves of the five sense organs, and the formed geometric curves of the five sense organs form geometric features of a reference face.

Since the user cannot be in the same position every time when performing face recognition, or before or after, or left or right, a set of reference face geometric features needs to be set according to a certain trajectory at the same distance. The trajectory is obtained as follows:

and drawing a circle or a spherical surface by taking the nose tip as a circle center and the distance as a radius, taking a face symmetric plane including the midpoint of the two eyes as a symmetric plane, taking a region with an angle smaller than an angle threshold (preferably 20 degrees) and a height between the chin and the nose tip as a sample on the basis of the symmetric plane, and setting at least 20 reference face geometric features.

The distance and all reference face geometric features are recorded.

Meanwhile, a reference identification basis of the heart rate is set.

The user wears a watch with a heart rate sensor, preferably a PPG heart rate identification watch, the measured heart rate curve being a continuous curve, similar to a sine, but largely different from a sine curve, mainly in that between the peaks and troughs of the curve there is an irregular curve of a discontinuity-like nature. Generally, we go through the curve including the peak or the trough in 0.1-0.2 seconds for study.

Every person's crest and trough all are different, have very big relation with user's blood pressure, cardiac function, and some people's crest is more steady, explains that whole qi and blood is not too enough, and some people's crests present sharp-pointed state in the place at summit, and is very powerful, and qi and blood are flourishing. Of course, these are not within the scope of the present application, but these phenomena show that the factors behind the curve of each person's heart rate are different, and accurate identification can be made by these phenomena.

A section of heart rate curve of a user is taken, the heart rate curve at least comprises N wave crests and N wave troughs, and in order to balance accuracy and comprehensiveness of data, the value of N is generally larger than 10. The peaks and troughs of the heart rate curve of each user are also greatly different, and therefore, the peaks and the troughs need to be processed respectively.

For the peak and the trough, the time period of the first duration centered on the peak point and the trough point is taken, generally 0.1 to 0.2 seconds. At least N peak curves and N valley curves are formed.

And performing second derivative processing on all the peak curves and all the valley curves to obtain a first peak second derivative curve and a first valley second derivative curve.

Because each first duration is short in time, but there are infinite points in the continuous curve, we process starting scatter points, and in order to balance accuracy and calculation amount, we generally take the number of scatter points to be 50-100 points. And the number of scattered points and the corresponding time point are the same for the peak or the trough of each first time length. If the time of the first point of each peak curve or each trough curve is set to be 0, the first point is taken at 0.001 second of each peak, the second point is taken at 0.002 second of each peak, and each time point is ensured to have a corresponding scattered point on each peak curve or each trough curve.

And performing standard deviation operation on scattered points with the same time abscissa of all the peak curves, and performing interpolation processing to obtain a reference peak second derivative curve. The same operation is also performed on the trough curve to obtain a reference trough second derivative curve.

And the reference peak second-order derivative curve and the reference trough second-order derivative curve form an initial learning curve.

And on the next day or when the user wears the watch again, identifying the identity of the user based on the initial learning curve, if the fitting degree of the second derivative curve obtained by measurement and calculation and the initial learning curve exceeds a fitting threshold value, generally setting the fitting threshold value to be 95%, considering that the user who wears the watch at present and the user who wears the watch at the last time are the same user, and identifying the identity at the moment.

In fact, in the identification process, the fitting degree can be set in a simpler mode. The method comprises the following steps:

first, the initial learning curve is subjected to a tolerance process, and the tolerance process is an expansion process to obtain a learning curve band. This expansion is based on the measurement error of the user at the time of measurement. For example, the second derivative value corresponding to each time coordinate is expanded within the proportional threshold. The proportional threshold may be set at 1% -5%. Thus, the comparison is convenient. And comparing the measured and calculated second derivative curve with the learning curve band, and if the proportion of the time length of extending the learning curve band to the total time length (the first time length) exceeds a proportion threshold value, judging that the identity is not identifiable, namely the person worn at the time is different from the person worn at the last time.

The data fed back by the second derivative of the heart rate curve may in fact be slowly changing due to factors such as exercise therapy, and therefore, the learning curve is named as an initial learning curve, which is convenient for the same user to continuously perform learning optimization after wearing. The learning mode is as follows:

s14, performing interpolation processing to obtain a first reference peak second derivative curve and a first reference trough second derivative curve which are formed by standard deviation and interpolation; the first reference peak second derivative curve and the first reference trough second derivative curve form an initial learning curve.

When the user wears the clothes, the following operations are carried out:

s20, if the identity is recognizable, the process goes to S21;

s21, taking a heart rate curve at least having N wave crests and N wave troughs, taking a wave crest curve in a first time length before and after each wave crest is a midpoint, and taking a wave trough curve in a first time length before and after each wave trough is a midpoint;

s22, performing second derivative processing on all the peak curves to obtain a second peak second derivative curve; performing second derivative processing on all the trough curves to obtain a second trough second derivative curve;

s23, performing standard deviation operation on the second crest second-order derivative curve and the first reference crest second-order derivative curve in the initial learning curve based on the scattered points of the same time abscissa; performing standard deviation operation on scattered points based on the same time abscissa on the second trough second-order derivative curve and the first reference trough second-order derivative curve in the initial learning curve;

s24, performing interpolation processing to obtain a second reference peak second derivative curve and a second reference trough second derivative curve which are formed by standard deviation and interpolation; and the second reference peak second derivative curve and the second reference trough second derivative curve form a second learning curve.

When the ith watch is worn, the following operations are performed:

si0, if the identity is recognizable, the process goes to S21;

s24, performing interpolation processing to obtain an ith reference crest second derivative curve and an ith reference trough second derivative curve which are formed by standard deviation and interpolation; and the ith reference crest second derivative curve and the ith reference trough second derivative curve form an ith learning curve.

Therefore, the reference curve used as comparison can be continuously learned, and the user's continuously changing self requirements can be better met.

Thus, reference data of the face and reference data of the heart rate are recorded simultaneously. On the basis, the following method is implemented:

when the processor judges that the payment program is started, the heart rate sensor is started, whether a heart rate curve collected by the heart rate sensor is fitted with a preset reference curve or not is judged, if the fitting degree reaches a fitting threshold value, the optimal fitting threshold value is 90% -95%, then:

And when the acquired geometric features of the face and any one of the preset geometric features of the reference face meet geometric similarity based on the acquisition distance and the distance of the geometric features of the reference face, the identity is confirmed. The geometric similarity is preferably a geometric coincidence of 90% or more.

Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.

9页详细技术资料下载

上一篇：一种医用注射器针头装配设备

下一篇：支付方法、支付装置、移动终端及支付终端

Smart watch payment method based on face recognition

相关技术

网友询问留言