Method for face authentication of a watch wearer

Abstract

The invention relates to a method for facial authentication of a wearer of a watch (1), comprising the following steps:-initiating (10) an authentication process comprising at least one detection sub-step (11) of a triggering motion / gesture performed by the wearer; capturing (13) at least one sequence of images relating to the face of the wearer turning from one direction to another in front of the optical sensor (3); -acquiring (17) surface geometry data of the face associated with each image of said at least one sequence; -generating (18) a three-dimensional model of the face of the wearer from said at least one sequence of images captured and from said geometric data acquired; -determining (19) a recognition index generated on the basis of recognition data relating to a plurality of characteristic characteristics of the face of the wearer of the watch (1) detected on the basis of the three-dimensional model; and identifying (21) the wearer if the identification index is greater than the reference identification index.

Description

[0001] This application is a divisional application of a Chinese patent application with the application number 201911091928.4. The filing date of the original application is November 8, 2019, the priority date is November 9, 2018, and the invention title is "Method for Facial Authentication of a Watch Wearer". Technical Field

[0002] The present invention relates to a method for facial authentication of a watch wearer.

[0003] The present invention also relates to a watch and a computer program for implementing the method. Background Art

[0004] A watch includes a set of functions that can be used by a wearer. In addition to time adjustment or date adjustment, such functions can also involve the use of confidential or private data specific to the wearer and for accessing personalized services. This data is, for example, keys, certificates, authentication codes, passwords, and personal codes that allow for a secure connection of the watch to a company's private network, authentication with a secure server such as a bank server, or a secure messaging service for sending and receiving signed and / or encrypted emails. Therefore, it should be understood that it is important to be able to ensure access to such functions of the watch.

[0005] To this end, methods for authentication are known in the prior art, which are designed to ensure access to the use of these functions of the watch by entering a PIN-type password.

[0006] However, since a third party can relatively easily observe the watch wearer entering the password, this authentication method is not entirely satisfactory. In addition, this mechanism requires the wearer to remember the password, which can be objectionable.

[0007] It can be understood that, therefore, there is a need to find an alternative solution, especially one that does not have the disadvantages of the prior art. Summary of the Invention

[0008] Therefore, an object of the present invention is to propose a method for facial authentication of a watch wearer that provides equally easy-to-implement and robust authentication of the wearer.

[0009] To this end, the present invention relates to a method for facial recognition of a watch wearer, the method comprising the following steps:

[0010] - Initiating an authentication process, the authentication process including a sub-step of detecting at least one triggering movement / gesture performed by the wearer;

[0011] - Capture at least one image sequence related to the face of a wearer who turns from one direction to another in front of an optical sensor;

[0012] - Obtain surface geometry data of the face associated with each image of the at least one sequence;

[0013] - Generate a three-dimensional model of the wearer's face from the captured at least one image sequence and from the obtained geometry data;

[0014] - Determine an identification index generated based on identification data related to a plurality of characteristic features of the face of the wearer of the watch detected based on the three-dimensional model; and

[0015] - Identify the wearer if the identification index is greater than a reference identification index.

[0016] In other embodiments:

[0017] - The initiation step includes a sub-step of determining that the face of the wearer placed opposite the optical sensor belongs to a living being;

[0018] - The capture step includes a sub-step of obtaining a plurality of images forming all parts of the face when the face of the wearer turns from one direction to another in front of the optical sensor.

[0019] - The obtaining sub-step includes a capture phase of at least two images with two different exposures respectively for each identical part of the face: one exposure for visible light and one exposure for infrared light.

[0020] - The obtaining sub-step includes a visual and / or sound guidance phase for the wearer of the watch, aiming to maintain:

[0021] · The watch is in a stable or substantially stable position;

[0022] · The face of the wearer is in an optimal position relative to the optical sensor so that it is included in each image of the at least one image sequence;

[0023] - The obtaining step is performed simultaneously or substantially simultaneously with the capture step;

[0024] - The determination step includes a processing sub-step during which the identification data from the three-dimensional model is compared with pre-recorded identification data relative to a reference three-dimensional model in order to determine the correspondence between the identification data;

[0025] - The at least one triggering movement / gesture is performed by a part of the wearer's body including the watch during the positioning of the face of the wearer opposite the optical sensor of the watch.

[0026] The invention also relates to a particularly intelligent watch for implementing this method, the watch comprising a processing unit, an optical sensor, at least one light source, and a behavior and / or physiological sensor, the processing unit being connected to the optical sensor, the at least one light source, and the behavior and / or physiological sensor.

[0027] The invention also relates to a computer program comprising program code instructions for performing the steps of the method when the computer program is executed by the processing unit of the watch. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Other details and advantages will become apparent from the following description given by way of information and not limitation with reference to the accompanying drawings, in which:

[0029] - Figure 1 shows a flowchart relating to a method for face authentication of a watch wearer according to an embodiment of the invention, and

[0030] - Figure 2 shows a watch implementing the method according to an embodiment of the invention. DETAILED DESCRIPTION

[0031] Figure 1 shows a method for face authentication of the wearer of the watch 1. The aim of this method is to perform a reliable authentication of the wearer of the watch 1 based on a three-dimensional modeling / representation of the wearer's face particularly using the optical sensor 3.

[0032] This method is implemented by the watch 1, in particular a smartwatch. Figure 2 The watch 1 shown in includes, in a non-exhaustive and non-limiting manner:

[0033] - a processing unit 2;

[0034] - an optical sensor 3;

[0035] - a hybrid display dial 4 provided with a first analog display part and a second digital and / or alphanumeric display part;

[0036] - a first light source 5 of the light-emitting diode or laser diode type;

[0037] - a second light source 6 of the infrared type;

[0038] - a third light source 7 designed to be used as a photographic flash;

[0039] - a sound interface, such as a speaker;

[0040] - a communication module;

[0041] - a mechanical or electronic movement, and

[0042] - Behavior and / or physiological sensor 8.

[0043] The processing unit 2 of the watch 1 includes physical and software resources, in particular at least one processor cooperating with memory elements. These memory elements include a reference three-dimensional model of the wearer's face, information data regarding the characteristic features of the wearer's face included in the reference three-dimensional model, and a reference identification index. This processing unit 2 is in particular linked / connected to the optical sensor 3, the display dial 4, the first, second, and third light sources 5, 6, 7, the sound interface, the communication module, the movement, and the environmental, behavior, and / or physiological sensor.

[0044] In this watch 1, the optical sensor 3, which is called a matrix image sensor, can include, for example, a sensor using CCD technology (charge transfer technology) or a sensor using what is called CMOS technology. Such a sensor includes a matrix of photosensitive image points (or pixels). Here, this point matrix is also called a "pixel matrix" and includes N×M unit pixels having excellent sensitivity to light. In an alternative, such a sensor 3 can include a light-sensitive organic film instead of traditional photosites. This optical sensor 3 also includes an analog-to-digital converter. Note that this optical sensor 3 is capable of implementing an autofocus function, as well as an automatic "zoom" function and a function of capturing infrared images. In addition, this optical sensor 3 can operate based on the time-of-flight (TOF) principle in order to measure in real time a three-dimensional scene such as the wearer's face. In this case, in order to eliminate excess ambient light, the optical sensor 3 can include an infrared bandpass filter.

[0045] In addition, the behavior sensor 8 is capable of measuring all types of behavioral characteristics of the wearer of the watch 1, such as, for example, the movements or gestures of the wearer of the watch 1. To this end, these behavior sensors 8 can include one or more inertial sensors of the accelerometer, gyroscope, or micro multi-axis gyroscope type (such as a multi-axis sensor manufactured using MEMS technology), which are capable of detecting angular velocity and linear acceleration according to multiple axes associated with the accelerometer and / or gyroscope. Regarding the physiological sensors 8, they are capable of measuring parameters related to the functions of the wearer's organism, such as, for example, pulse, blood oxygen saturation, skin impedance, blood pressure, respiratory rate, respiratory arrhythmia, skin temperature, sweating frequency rate, infrared radiation of the human body, blood oxygen, or blood flow saturation. Note that the watch 1 can also include environmental sensors specifically adapted to measure environmental parameters (such as, for example, temperature, atmospheric pressure, etc.).

[0046] This method includes an initiation step 10 of an authentication process, which includes at least one detection sub-step 11 of a triggering movement / gesture performed by the wearer. The at least one triggering movement / gesture is performed by a part of the body of the wearer including the watch 1 during the positioning of the wearer's face opposite the optical sensor 3 of the watch 1. The purpose of this step 10 is to initiate the authentication process under certain conditions, which are met by performing the detection sub-step 11 and the determination sub-step 12 of this step 10 described below.

[0047] The detection sub-step 11 allows the pre-identification of the wearer of the watch 1 by ensuring that the wearer is indeed responsible for the authentication process to be initiated. To this end, when the wearer wishes to be authenticated, the wearer then performs at least one predetermined movement / gesture of, for example, the arm and / or the wrist of the wearer configured with the watch 1. Then, after various operations of the processing performed by the processing unit 2, the processing unit 2 identifies the at least one movement / gesture from at least one measurement data of the behavior sensor 7 that has captured the at least one movement / gesture performed by the wearer. Note that these operations particularly involve comparing the at least one measurement data with one or more reference thresholds.

[0048] As described above, this step 10 also includes a determination sub-step 12 that determines that the face of the wearer placed opposite the optical sensor 3 belongs to a living being. This sub-step 12 aims to verify that the face placed opposite the sensor is actually the face of the wearer of the watch who wishes to be authenticated, rather than a replica of their face that can be included, for example, in a photo, video, or sculpture. During this sub-step 12, the processing unit 2 performs various operations of processing on at least one physiological measurement data performed on the face positioned opposite the optical sensor 3, aiming to compare the at least one measurement data with one or more reference thresholds. The measurement can be established based on the infrared radiation emitted from the face, the change in the pupil size of at least one eye of the face after exposure to a third light source 6 used as a photographic flash, the detection of eyelid blinking, etc.

[0049] Note that this determination sub-step 12 can be implemented simultaneously or substantially simultaneously with the detection sub-step 11.

[0050] Once the initiation step 10 has been executed, the method then involves performing a capture step 13 of at least one image sequence of the wearer's face that rotates from one direction to another in front of the optical sensor 3. More precisely, this step 13 includes an acquisition sub-step 14 of obtaining a plurality of images of all parts of the face while the wearer's face rotates from one direction to another in front of the optical sensor 3. It should be understood here that the rotation of the wearer's face in front of the optical sensor 3 is carried out in such a way that all parts of the face are captured. This sub-step 14 includes a capture phase 15 of at least two images with two different exposures for each identical part of the face: one exposure for visible light and one exposure for infrared light.

[0051] Under these conditions, this sub-step 14 allows obtaining at least two image sequences of the wearer's face, a first image sequence of the face whose surface is illuminated by visible light, and a second image sequence of the face whose surface is illuminated by infrared light.

[0052] Note that using images of the face exposed to visible light allows facilitating the detection of facial feature characteristics with respect to the color or texture of the facial skin in these images. Using images of the face exposed to infrared light is advantageous because it is less dependent on lighting conditions compared to using images of the face captured with visible light.

[0053] This sub-step 14 also includes a visual and / or sound guidance phase 16 for the wearer of the watch 1, aimed at maintaining:

[0054] - the watch 1 in a stable or substantially stable position, and / or

[0055] - the wearer's face in an optimal position relative to the optical sensor 3 so that it is included in each image of the at least one image sequence.

[0056] Note that during this sub-step 14, the processing unit 2 can control the focusing and "zoom" functions of the optical sensor 3 to help keep the wearer's face in each image of each captured image sequence.

[0057] Then, the method includes an acquisition step 17 of surface geometric data of the face associated with each image of the at least one sequence during the implementation of the step of capturing at least one image sequence relative to the wearer's face. During this step 17, when the wearer's face rotates from one direction to another in front of the optical sensor 3, this geometric data is generated for each acquired image included in the first sequence and the second sequence. More precisely, during the rotation of the wearer's face, a beam of structured light of the laser type is thus projected onto each captured part of the face present in each image of the first sequence and the second sequence in such a way as to scan the three-dimensional shape of these parts of the face.

[0058] This acquisition step 17 is performed simultaneously or substantially simultaneously with the capture step 13. Then, the images of the first sequence and the second sequence are archived in the memory element of the processing unit 2 together with the corresponding geometric data associated with each image.

[0059] Then, the method includes a generation step 18 of a three-dimensional model of the wearer's face from the at least one captured image sequence and from the acquired geometric data. During this step 18, the processing unit 2 implements an algorithm for constructing a three-dimensional model of the wearer's face based on the at least one image sequence and the geometric data. Such an algorithm may include a phase of preprocessing the images of the first sequence and the second sequence in order to correct the defects they may contain, such as the presence of peaks, valleys or unwanted parts (such as hair or clothing), or the presence of accessories on the face. Thus, the face can be corrected by eliminating peaks, filling valleys and segmenting the useful parts of the face. To this end, a two-dimensional median filter can be applied to each image at the input in order to eliminate noise. This noise can be in the form of pixels that are incoherent with their neighbors. Then, this phase may involve applying two-dimensional interpolation in order to fill the valleys created by the absorption of the laser or structured light by dark areas of the face (such as the iris, eyebrows, or very simply when the mouth is open). A three-dimensional model is generated based on the images obtained for each of these sequences and the corresponding geometric data, in particular by implementing a process of three-dimensional meshing.

[0060] The method then includes a determination step 19 of an identification index based on identification data associated with a plurality of characteristic features of the face of the wearer of the watch 1 detected based on the generated three-dimensional model. These characteristic features of the face (also called point features of the face) allow the structure of the face to be analyzed in three dimensions. In other words, these characteristic features allow the face to be interpreted as a set of characteristic features distributed in height, width and depth. With respect to the following aspects, such characteristic features may be, in a non-limiting and non-exhaustive manner, with regard to:

[0061] - the geometry of the face;

[0062] - The geometry of the facial surface portion;

[0063] - The relative position of the facial portion;

[0064] - The texture of the skin of the face or the texture of that skin area;

[0065] - The hue / color of the skin of the face or the hue / color of that skin area.

[0066] For example, these characteristic features may be wrinkles, chin, freckles, scars, lips, corners of the mouth, nose, tip of the nose, ears, tips of the ears, distance between the eyes, distance between the nasal ridges, distance of the chin, etc.

[0067] Preferably, such characteristic features of the face considered for determining the recognition index are included in the regions / zones of the wearer's face that remain stable during facial expressions.

[0068] During this step 19, a processing sub-step 20 is performed, during which the recognition data from the three-dimensional model is compared with the recognition data pre-recorded in the memory element of unit 2 in order to determine the correspondence between this recognition data. Based on the result of this comparison, a recognition index is thus generated by processing unit 2. The processing sub-step 20 may involve using a limited configurable number of recognition data for its implementation. This number can be parameterized based on the robustness of the method desired with respect to the rapidity of performing such an authentication method.

[0069] Then, if the recognition index is greater than the reference recognition index, the method includes a recognition step 21 of the wearer. The reference index allows verification of the correspondence between the recognition data of the three-dimensional model of the wearer's face constructed during the generation step 18 and the recognition data pre-recorded in the memory element of processing unit 2 and relative to the reference three-dimensional model of the wearer's face. In other words, when the determined recognition index is greater than the reference recognition index, the generated three-dimensional model has a high degree of similarity with the reference three-dimensional model of the wearer's face. On the contrary, when the determined recognition index is less than or equal to the reference recognition index, the number of similarities between the generated three-dimensional model and the reference three-dimensional model is insufficient, and thus the wearer cannot be recognized. In this case, a visual and / or sound message can be broadcast to the wearer to inform the wearer that the wearer has not been recognized.

[0070] The invention also relates to a computer program comprising program code instructions for performing steps 10 to 21 of the method when said computer program is executed by the processing unit 2 of the watch 1.

Claims

1. A method for facial authentication of a wearer of a watch (1), comprising the following steps: - Initiating (10) an authentication process, the authentication process including a detection sub-step (11) of at least one triggering movement / gesture performed by the wearer; - Capturing (13) at least one image sequence related to the face of the wearer that rotates from one direction to another in front of an optical sensor (3); - Scanning the face with a structured light beam of the laser type to obtain (17) surface geometry data of the face associated with each image of the at least one sequence; - Generating (18) a three-dimensional model of the face of the wearer from the captured at least one image sequence and from the obtained surface geometry data, wherein defects included in the images of the at least one image sequence are corrected and / or valleys generated due to absorption of the structured light by dark regions of the face are filled; - Determining (19) an identification index generated based on identification data related to a plurality of characteristic features of the face of the wearer of the watch (1) detected based on the three-dimensional model; and - Identifying (21) the wearer if the identification index is greater than a reference identification index, wherein the initiating step (10) includes a determination sub-step (12) of determining that the face of the wearer placed opposite to the optical sensor (3) is biological, the determination sub-step (12) including comparing at least one measurement data with a reference threshold, the measurement data including infrared radiation emitted by the face, change in the pupil size of the eyes after exposure to a light source, or detection of blinking of the eyelids, and performing different processing operations on at least one physiological measurement data obtained from the face placed opposite to the optical sensor (3) during this sub-step (12), aiming to compare the at least one measurement data with one or more reference thresholds.

2. The method according to claim 1, wherein The capturing step (13) includes an acquisition sub-step (14) of acquiring a plurality of images of all parts forming the face when the face of the wearer rotates from one direction to another in front of the optical sensor (3).

3. The method according to claim 2, wherein, The acquisition sub-step (14) includes a capturing stage (15) of at least two images with two different exposures respectively for each same part of the face: one exposure for visible light and one exposure for infrared light.

4. The method according to any one of claims 2 and 3, wherein, The acquisition sub-step (14) includes a stage of visual and / or voice guidance (16) of the wearer of the watch (1), aiming to maintain: - The watch (1) in a stable or substantially stable position, and / or - The face of the wearer in an optimal position relative to the optical sensor (3) so that it is included in each image of the at least one image sequence.

5. The method according to claim 1, wherein The obtaining step (17) is performed simultaneously or substantially simultaneously with the capturing step (13).

6. The method according to claim 5, wherein The determining step (19) includes a processing sub-step (20) during which the recognition data from the three-dimensional model is compared with pre-recorded recognition data relative to a reference three-dimensional model in order to determine the correspondence between this recognition data.

7. The method according to claim 6, wherein, The at least one triggering movement / gesture is performed by a part of the body of the wearer including the watch (1) during the positioning of the face of the wearer opposite the optical sensor (3) of the watch (1).

8. A watch (1), in particular a smart watch, which implements the method according to any one of claims 1 to 7, the watch (1) comprising a processing unit (2), an optical sensor (3), at least one light source (5, 6, 7) and a behavior and / or physiological sensor (8), the processing unit (2) being connected to the optical sensor (3), the at least one light source (5, 6, 7) and the behavior and / or physiological sensor (8).

9. A computer-readable storage medium having stored thereon a computer program including program code instructions for performing the steps of the method according to any one of claims 1 to 7 when the computer program is executed by a processing unit (2) of a watch (1).

Images