CN111325123A - Face angle representation and annotation method, device and computer storage medium - Google Patents

Abstract

A method, device, and computer storage medium for representing and labeling a face angle, comprising: determining the representation data q of the face angle; wherein, the representation data q is a quaternion, q=e ₀ +e ₁ i+e ₂ j+e ₃ k, i ² =j ² =k ² =ijk=-1, e ₀ is called the real part of q, e ₁ i+e ₂ j+e ₃ k is called the imaginary part of q; 3D visualization. The solution in this application solves the problems in the prior art, such as unknown rotation order, unknown rotation of coordinate axes, and universal lock caused by Euler angles representing the angle of the human face, and can provide more accurate facial angle labeling and improve facial The recognition accuracy of the angle recognition network.

Description

Facial angle representation and labeling method, device and computer storage medium

technical field

The present application relates to face recognition technology, and in particular, to a method, device, computer storage medium, and electronic device for representing and labeling a face angle.

Background technique

Face angle recognition is mainly used in scenes such as quality filtering and selection before face recognition, and driver status monitoring. In traditional machine learning methods, the three-dimensional geometric relationship between the key points of the face and the standard key points is usually used to iteratively solve the face angle through the solvePNP function in OpenCV. After the rise of deep learning, more and more deep convolutional networks are used to directly find the face angle.

In academia and industry, face angle representation follows the representation of object rotation, generally represented by three Euler angles: yaw angle (yaw), pitch angle (pitch) and roll angle (roll). As shown in Figure 1, the yaw angle means shaking the head left and right, the pitch angle means nodding up and down, and the roll angle means tilting the head.

Generally, training a deep convolutional network requires a large amount of labeled data. For example, the amount of ImageNet data exceeds 14 million. Therefore, training a face angle network also requires manual labeling of a large amount of face angle data.

Using Euler angles to represent face angles has three disadvantages:

1. The rotation axis and rotation order need to be specified, so there are 12 different rotation orders. When three Euler angles are given, the rotation order must be given at the same time, but the rotation order is easily ignored. The order of face angle is generally Caldan angle, Roll-Pitch-Yaw.

2. It is necessary to agree whether the coordinate axis rotates with the face. Generally, when the coordinate axis rotates with the face, it is called its own coordinate. When the coordinate axis is fixed, it is called the space coordinate. If only three Euler angles and rotation order are given, without specifying whether the coordinate axis is rotated, the resulting angle representation is ambiguous.

3. There is a universal lock phenomenon in Euler angles. Among the three Euler angles, if the second Euler angle rotates so that the first rotation axis coincides with the third rotation axis, the degree of freedom of rotation changes from three to two. Taking Cardan angle as an example, if Pitch is ±π/2, then Roll and Yaw are combined into one degree of freedom. Therefore, using Euler angles to represent the face angle will lead to ambiguous representation due to the universal lock.

Problems existing in the prior art:

The use of Euler angles to represent the face angle is prone to ambiguity, which leads to low recognition accuracy of the face angle.

SUMMARY OF THE INVENTION

Embodiments of the present application provide a method, apparatus, computer storage medium, and electronic device for representing and labeling a face angle, so as to alleviate the above-mentioned technical problems to a certain extent.

According to a first aspect of the embodiments of the present application, a method for representing a face angle is provided, comprising the following steps:

Determine the representation data q of the face angle; wherein, the representation data q is a quaternion, q=e ₀ +e ₁ i+e ₂ j+e ₃ k, i ² =j ² =k ² =ijk=−1 , e ₀ is called the real part of q, e ₁ i+e ₂ j+e ₃ k is called the imaginary part of q;

The representation data q is visualized in three dimensions.

According to a second aspect of the embodiments of the present application, a device for representing a face angle is provided, including:

A data determination module, used for determining the representation data q of the face angle; wherein, the representation data q is a quaternion, q=e ₀ +e ₁ i+e ₂ j+e ₃ k, i ² =j ² =k ² =ijk=-1, e ₀ is called the real part of q, e ₁ i+e ₂ j+e ₃ k is called the imaginary part of q;

The representation module is used for three-dimensional visualization of the representation data q.

According to a third aspect of the embodiments of the present application, a method for labeling a face angle is provided, including:

Determine the facial image to be annotated;

According to the initial face angle representation data q ₀ visualized in advance and the three-dimensional coordinates of the face image to be marked, the representation data q ₁ is obtained;

According to the representation data q ₁ and the representation data q ₀ , obtain the face angle representation data q _g of the facial image to be marked;

Label the face angle of the face image according to the face angle representation data q _g ;

Wherein, the representation data q ₀ , q ₁ , and q _g are all quaternions, including real and imaginary parts.

According to a fourth aspect of the embodiments of the present application, a device for labeling facial angles is provided, including:

an image determination module, used to determine the facial image to be marked;

a first processing module, configured to obtain the representation data q ₁ according to the pre-visualized initial face angle representation data q ₀ and the three-dimensional coordinates of the facial image to be marked;

a second processing module, configured to obtain, according to the representation data q ₁ and the representation data q ₀ , the face angle representation data q _g of the facial image to be marked;

a labeling module, configured to label the facial angle of the facial image according to the facial angle representation data _qg ;

Wherein, the representation data q ₀ , q ₁ , and q _g are all quaternions, including real and imaginary parts.

According to a fifth aspect of the embodiments of the present application, a computer storage medium is provided on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above method are implemented.

According to a sixth aspect of the embodiments of the present application, an electronic device is provided, including a memory and one or more processors, where the memory is used to store one or more programs; the one or more programs are When executed by the one or more processors, the methods described above are implemented.

Using the method, device, computer storage medium, and electronic device for representing and labeling the face angle provided in the embodiments of the present application, and using the quaternion to represent the face angle, the rotation caused by the Euler angle representing the face angle in the prior art is solved. Problems such as unclear order, unknown rotation of the coordinate axis, and universal lock can provide more accurate facial angle annotations and improve the recognition accuracy of the facial angle recognition network.

Description of drawings

The drawings described herein are used to provide further understanding of the present application and constitute a part of the present application. The schematic embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation of the present application. In the attached image:

1 shows a schematic diagram of a method for representing a face angle in the prior art;

FIG. 2 shows a schematic flowchart of the implementation of the method for representing a face angle in Embodiment 1 of the present application;

FIG. 3 shows a schematic flowchart of the implementation of the facial angle labeling method in Embodiment 2 of the present application;

FIG. 4 shows a schematic structural diagram of a device for representing a face angle in Embodiment 3 of the present application;

FIG. 5 shows a schematic structural diagram of a device for labeling a face angle in Embodiment 4 of the present application;

FIG. 6 shows a schematic structural diagram of an electronic device in Embodiment 7 of the present application;

FIG. 7 shows a schematic structural diagram of an electronic device in Embodiment 8 of the present application;

FIG. 8 shows a schematic diagram of an annotation tool interface in Embodiment 9 of the present application;

FIG. 9 shows a schematic diagram of the labeling process in Embodiment 9 of the present application.

Detailed ways

In order to make the technical solutions and advantages of the embodiments of the present application more clear, the exemplary embodiments of the present application will be described in further detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, and Not all embodiments are exhaustive. It should be noted that the embodiments in the present application and the features of the embodiments may be combined with each other in the case of no conflict.

Example 1

FIG. 2 shows a schematic flowchart of the implementation of the method for representing a face angle in Embodiment 1 of the present application.

As shown in the figure, the facial angle representation method includes:

Step 201: Determine the representation data q of the face angle; wherein, the representation data q is a quaternion, q=e ₀ +e ₁ i+e ₂ j+e ₃ k, i ² =j ² =k ² =ijk =-1, e ₀ is called the real part of q, e ₁ i+e ₂ j+e ₃ k is called the imaginary part of q;

Step 202: Perform three-dimensional visualization of the representation data q.

The facial angle representation method provided in the embodiment of the present application is adopted, and the quaternion is used to represent the facial angle, which solves the problems in the prior art, which are caused by the Euler angle representing the face angle. It can provide more accurate facial angle annotation and improve the recognition accuracy of the facial angle recognition network.

In one embodiment, the three-dimensional visualization of the representation data q is: generating a three-dimensional coordinate system according to the representation data q, and the X, Y, and Z axes of the three-dimensional coordinate system are respectively e according to the representation data q ₀ , e ₁ , e ₂ , e ₃ computed vectors.

In one embodiment, the relationship between the representation data q and the three-dimensional coordinate system is:

For any point in the three-dimensional space, the new representation data is obtained after rotating according to the representation data q:

为旋转前三维坐标系下点r＝(r₁，r₂，r₃)的四元数表示；q^*＝e₀-e₁i-e₂j-e₃k；所述新的表示数据对应的三维坐标系下的点为r＇＝(r′₁，r′₂，r′₃)。in,

is the quaternion representation of the point r=(r ₁ , r ₂ , r ₃ ) in the three-dimensional coordinate system before rotation; q ^* = e ₀ -e ₁ ie ₂ je ₃ k; the three-dimensional coordinates corresponding to the new representation data The point under the system is r'=(r' ₁ , r' ₂ , r' ₃ ).

In an embodiment, the determining of the representation data q of the face angle includes: obtaining through a pre-established face angle model or pre-labeled face angle data, or obtaining through a preset.

Embodiment 2

FIG. 3 shows a schematic flowchart of the implementation of the facial angle labeling method in Embodiment 2 of the present application.

As shown in the figure, the facial angle labeling method includes:

Step 301, determine the facial image to be marked;

Step 302, obtaining the representation data q ₁ according to the preliminary three-dimensional visualization of the initial face angle representation data q ₀ and the three-dimensional coordinate system of the facial image to be marked;

Step 303, according to the representation data q ₁ and the representation data q ₀ , obtain the face angle representation data q _g of the facial image to be marked;

Step 304, label the facial angle of the facial image according to the facial angle representation data q _g ;

Wherein, the representation data q ₀ , q ₁ , and q _g are all quaternions, including real and imaginary parts.

Using the face angle labeling method provided in the embodiment of the present application, and using quaternion to represent the face angle, it solves the problems in the prior art due to the Euler angle representing the face angle. It can provide more accurate facial angle annotation and improve the recognition accuracy of the facial angle recognition network.

In one embodiment, the three-dimensional coordinate system of the facial image to be marked can be obtained by preset, for example: according to the direction of the line connecting the two eyes in the facial image, the direction from the midpoint of the two eyes to the chin, and the direction the face is facing The X, Y, and Z axes of the three-dimensional coordinate system of the facial image to be marked are respectively determined.

In one embodiment, the initial face angle unambiguous representation q ₀ is obtained from a pre-established face angle model or pre-labeled face angle data, or the initial face angle unambiguous representation q ₀ is set to q ₀ =1+0i+0j+0k.

In an embodiment, the representation data q ₁ is obtained according to the pre-visualized initial face angle representation data q ₀ and the three-dimensional coordinate system of the face image to be marked, including:

The three-dimensional coordinate system corresponding to the pre-visualized initial face angle representation data q ₀ is rotated so that the three-dimensional coordinate system coincides with the three-dimensional coordinate system of the facial image to be marked, and the representation data q ₁ at this time is obtained.

In an implementation manner, the rotation and even the coincidence of the two three-dimensional coordinate systems can be realized by using the prior art, which will not be described in detail in this application.

In one embodiment, the rotated representation data of each point is:

为旋转前三维坐标系下点r＝(r₁，r₂，r₃)的四元数表示；所述待标注的面部图像的三维坐标系下的点为r＇＝(r′₁，r′₂，r′₃)。in,

is the quaternion representation of the point r=(r ₁ , r ₂ , r ₃ ) under the three-dimensional coordinate system before rotation; the point under the three-dimensional coordinate system of the facial image to be marked is r′=(r′ ₁ , r ' ₂ , r' ₃ ).

In one embodiment, the facial angle of the facial image to be labeled represents data q _g =q ₁ q ₀ .

Embodiment 3

Based on the same inventive concept, an embodiment of the present application provides a device for representing a face angle. The principle of the device for solving the technical problem is similar to that of a method for representing a face angle, and repeated details are not repeated here.

FIG. 4 shows a schematic structural diagram of a device for representing a face angle in Embodiment 3 of the present application.

As shown in the figure, the facial angle representation device includes:

The data determination module 401 is used to determine the representation data q of the face angle; wherein, the representation data q is a quaternion, q=e ₀ +e ₁ i+e ₂ j+e ₃ k, i ² =j ² = k ² =ijk=-1, e ₀ is called the real part of q, e ₁ i+e ₂ j+e ₃ k is called the imaginary part of q;

The representation module 402 is used for three-dimensional visualization of the representation data q.

The facial angle representation device provided in the embodiment of the present application is adopted, and the quaternion is used to represent the facial angle, which solves the problems in the prior art, which are caused by the Euler angle representing the face angle, the unknown rotation order, the unknown rotation of the coordinate axis, and the universal orientation. It can provide more accurate facial angle annotation and improve the recognition accuracy of the facial angle recognition network.

In one embodiment, the representation module is specifically used for any point in the three-dimensional space, after rotating according to the representation data q, the new representation data is calculated according to the following formula:

为旋转前三维坐标系下点r＝(r₁，r₂，r₃)的四元数表示；q^*＝e₀-e₁i-e₂j-e₃k；所述新的表示数据对应的三维坐标系下的点为r＇＝(r′₁，r′₂，r′₃)。in,

is the quaternion representation of the point r=(r ₁ , r ₂ , r ₃ ) in the three-dimensional coordinate system before rotation; q ^* = e ₀ -e ₁ ie ₂ je ₃ k; the three-dimensional coordinates corresponding to the new representation data The point under the system is r'=(r' ₁ , r' ₂ , r' ₃ ).

In an embodiment, the data determination module is specifically configured to obtain the representation data q of the face angle through a pre-established face angle model or pre-labeled face angle data, or obtain the representation data q of the face angle by preset .

Embodiment 4

Based on the same inventive concept, an embodiment of the present application provides a device for labeling a face angle. The principle of the device for solving the technical problem is similar to a method for labeling a face angle, and the repeated details will not be repeated.

FIG. 5 shows a schematic structural diagram of a device for labeling a face angle in Embodiment 4 of the present application.

As shown in the figure, the facial angle labeling device includes:

An image determination module 501, configured to determine the facial image to be marked;

The first processing module 502 is configured to obtain the representation data q ₁ according to the pre-visualized initial face angle representation data q ₀ and the three-dimensional coordinate system of the facial image to be marked;

The second processing module 503 is configured to obtain, according to the representation data q ₁ and the representation data q ₀ , the facial angle representation data q _g of the facial image to be marked;

A labeling module 504, configured to label the facial angle of the facial image according to the facial angle representation data q _g ;

Wherein, the representation data q ₀ , q ₁ , and q _g are all quaternions, including real and imaginary parts.

The facial angle labeling device provided in the embodiment of the present application is adopted, and the quaternion is used to represent the facial angle, which solves the problems caused by the unknown rotation order, unknown rotation of the coordinate axis and universal orientation caused by the Euler angle representing the face angle in the prior art. It can provide more accurate facial angle annotation and improve the recognition accuracy of the facial angle recognition network.

In one embodiment, the initial face angle unambiguous representation q ₀ is obtained from a pre-established face angle model or pre-labeled face angle data, or the initial face angle unambiguous representation q ₀ is set to q ₀ =1+0i+0j+0k.

In one embodiment, the first processing module is specifically configured to rotate the three-dimensional coordinate system corresponding to the pre-visualized initial facial angle representation data q ₀ , so that the three-dimensional coordinate system coincides with the three-dimensional coordinate system of the facial image to be labeled , and obtain the representation data q ₁ at this time.

In one embodiment, the rotated representation data of each point is:

为旋转前三维坐标系下点r＝(r₁，r₂，r₃)的四元数表示；所述待标注的面部图像的三维坐标系下的点为r＇＝(r′₁，r′₂，r′₃)。in,

is the quaternion representation of the point r=(r ₁ , r ₂ , r ₃ ) under the three-dimensional coordinate system before rotation; the point under the three-dimensional coordinate system of the facial image to be marked is r′=(r′ ₁ , r ' ₂ , r' ₃ ).

In one embodiment, the facial angle of the facial image to be labeled represents data q _g =q ₁ q ₀ .

Embodiment 5

Based on the same inventive concept, an embodiment of the present application further provides a computer storage medium, which will be described below.

The computer storage medium stores a computer program thereon, and when the computer program is executed by the processor, implements the steps of the method described in the first embodiment.

Using the computer storage medium provided in the embodiment of the present application, the quaternion is used to represent the face angle, which solves the problems in the prior art, which are caused by the Euler angle representing the face angle, such as the unknown rotation order, the unknown rotation of the coordinate axis, and the universal lock. It can provide more accurate facial angle annotation and improve the recognition accuracy of the facial angle recognition network.

Embodiment 6

Based on the same inventive concept, an embodiment of the present application further provides a computer storage medium, which will be described below.

The computer storage medium has a computer program stored thereon, and when the computer program is executed by the processor, the steps of the method described in Embodiment 2 are implemented.

Using the computer storage medium provided in the embodiment of the present application, the quaternion is used to represent the face angle, which solves the problems in the prior art, which are caused by the Euler angle representing the face angle, such as the unknown rotation order, the unknown rotation of the coordinate axis, and the universal lock. It can provide more accurate facial angle annotation and improve the recognition accuracy of the facial angle recognition network.

Embodiment 7

Based on the same inventive concept, an embodiment of the present application further provides an electronic device, which will be described below.

FIG. 6 shows a schematic structural diagram of an electronic device in Embodiment 7 of the present application.

As shown, the electronic device includes a memory 601 and one or more processors 602, the memory is used to store one or more programs; the one or more programs are stored by the one or more processors When executed, the method described in the first embodiment is implemented.

Using the electronic device provided in the embodiment of the present application, the quaternion is used to represent the face angle, which solves the problems in the prior art, such as the unknown rotation order, the unknown rotation of the coordinate axis and the universal lock caused by the Euler angle representing the face angle. It can provide more accurate facial angle annotation and improve the recognition accuracy of the facial angle recognition network.

Embodiment 8

Based on the same inventive concept, an embodiment of the present application further provides an electronic device, which will be described below.

FIG. 7 shows a schematic structural diagram of an electronic device in Embodiment 8 of the present application.

As shown, the electronic device includes a memory 701 and one or more processors 702, the memory is used to store one or more programs; the one or more programs are stored by the one or more processors When executed, the method described in Embodiment 2 is implemented.

Using the electronic device provided in the embodiment of the present application, the quaternion is used to represent the face angle, which solves the problems in the prior art, such as the unknown rotation order, the unknown rotation of the coordinate axis and the universal lock caused by the Euler angle representing the face angle. It can provide more accurate facial angle annotation and improve the recognition accuracy of the facial angle recognition network.

Embodiment 9

In order to facilitate the implementation of the present application, the embodiment of the present application is described with a specific example.

In the embodiment of the present application, an unambiguous quaternion is used to represent the rotation of the face. A quaternion is a four-dimensional vector that can be represented as q=e ₀ +e ₁ i+e ₂ j+e ₃ k, where i ² =j ² =k ² =ijk=−1. The first item e ₀ in the formula is called the real part of q, and e ₁ i+e ₂ j+e ₃ k is called the imaginary part of q.

The addition of quaternions can be defined as:

p+tq=(d ₀ +te ₀ )+(d ₁ +te ₁ )i+(d ₂ +te ₂ )j+(d ₃ +te ₃ )k

The multiplication of quaternions is:

pq=(d ₀ e ₀ -d ₁ e ₁ -d ₂ e ₂ -d ₃ e ₃ )+(d ₁ e ₀ +d ₀ e ₁ +d ₂ e ₃ -d ₃ e ₂ )i

+(d ₂ e ₀ +d ₀ e ₂ +d ₃ e ₁ -d ₁ e ₃ )j+(d ₃ e ₀ +d ₀ e ₃ +d ₁ e ₂ -d ₂ e ₁ )k

The conjugation of a quaternion is:

q ^* = e ₀ -e ₁ ie ₂ je ₃ k

When |q|=1, define q as a unit quaternion.

那么经过旋转q可以得到新的四元数为：For any rotation in the three-dimensional Euclidean space, it can be represented by a unit quaternion, set as q. Assuming a point r=(r ₁ , r ₂ , r ₃ ) in the three-dimensional Euclidean space, its corresponding quaternion is expressed as

Then after rotating q, the new quaternion can be obtained as:

The rotated point r'=(r' ₁ , r' ₂ , r' ₃ ) corresponding to the quaternion.

FIG. 8 shows a schematic diagram of an interface of an annotation tool in Embodiment 9 of the present application.

As shown in the figure, in the annotation tool:

1 is the directory where the picture to be marked is located, which can include the initialized face angle representation;

2 is an unambiguous representation of the current face angle, taking quaternion as an example;

3 is the reset button, and the marked value returns to the non-rotation state;

4 is the current face image to be labeled;

5 In order to save the unambiguous representation of the current face angle, import the next face image to be labeled;

6 is the visualization of the three-dimensional coordinate axis of the face angle under the current labeling situation.

FIG. 9 shows a schematic diagram of the labeling process in Embodiment 9 of the present application.

As shown in the figure, the labeling process can be as follows:

1. Obtain the unambiguous representation of the face angle through the existing face angle model or the labeled data q ₀ =e ₀ +e ₁ i+e ₂ j+e ₃ k, if there is no initialized representation, set q ₀ to 1+ 0i+0j+0k;

Specifically, a face angle model can be established in advance, and an unambiguous representation q ₀ of the face angle can be directly output through the face angle model; if the face image has been marked, the face image can be read from the disk. An unambiguous representation of the face angle q ₀ .

2. 3D visualization of q ₀ in the annotation tool;

表示，y轴可以用向量

表示，z轴可以用向量

表示，将这三个向量绘制到界面上便可可视化q₀。Specifically, the rotation q ₀ can be represented by a three-dimensional coordinate system, wherein the x-axis can be represented by a vector

Represented, the y-axis can be represented by a vector

Represented, the z-axis can be represented by a vector

represents that q ₀ can be visualized by plotting these three vectors on the interface.

3. The labeler rotates the 3D coordinate system in the labeling tool so that the coordinate system coincides with the assumed 3D coordinates of the face in the image to be labelled, and obtains the rotation representation at this time q ₁ =d ₀ +d ₁ i+d ₂ j+ _d3k ;

During specific implementation, the three-dimensional coordinate system can be obtained by visualization in the above step 2. The assumed three-dimensional coordinate system of the face in the image to be marked can be the horizontal direction of the two eyes in the face image to represent the x-axis direction, and the direction from the midpoint of the two eyes to the chin represents the y-axis. Direction, the direction the face is facing represents the z-axis direction, and the rotation representation can be obtained through Unity3D during rotation, and the rotation representation can be q ₁ =d ₀ +d ₁ i+d ₂ j+d ₃ k.

4. Calculate the unambiguous expression q _g =q ₁ q ₀ of the face angle of the face in the image to be labeled.

Calculated q _g =(d ₀ e ₀ -d ₁ e ₁ -d ₂ e ₂ -d ₃ e ₃ )+(d ₁ e ₀ +d ₀ e ₁ +d ₂ e ₃ -d ₃ e ₂ )i+( d ₂ e ₀ +d ₀ e ₂ +d ₃ e ₁ -d ₁ e ₃ )j+(d ₃ e ₀ +d ₀ e ₃ +d ₁ e ₂ -d ₂ e ₁ )k.

The embodiment of the present application proposes an unambiguous representation of a face angle, which solves the problems of unclear rotation order, unknown rotation of the coordinate axis, and gimbal lock caused by the Euler angle representing the face angle. Based on this face angle representation, more accurate face angle annotation can be provided, and the recognition accuracy of the face angle recognition network can be improved.

As will be appreciated by those skilled in the art, the embodiments of the present application may be provided as a method, a system, or a computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein. The solutions in the embodiments of the present application may be implemented in various computer languages, for example, the object-oriented programming language Java and the literal translation scripting language JavaScript, and the like.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present application. It will be understood that each flow and/or block in the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to the processor of a general purpose computer, special purpose computer, embedded processor or other programmable data processing device to produce a machine such that the instructions executed by the processor of the computer or other programmable data processing device produce Means for implementing the functions specified in a flow or flow of a flowchart and/or a block or blocks of a block diagram.

These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory result in an article of manufacture comprising instruction means, the instructions The apparatus implements the functions specified in the flow or flow of the flowcharts and/or the block or blocks of the block diagrams.

These computer program instructions can also be loaded on a computer or other programmable data processing device to cause a series of operational steps to be performed on the computer or other programmable device to produce a computer-implemented process such that The instructions provide steps for implementing the functions specified in the flow or blocks of the flowcharts and/or the block or blocks of the block diagrams.

While the preferred embodiments of the present application have been described, additional changes and modifications to these embodiments may occur to those skilled in the art once the basic inventive concepts are known. Therefore, the appended claims are intended to be construed to include the preferred embodiment and all changes and modifications that fall within the scope of this application.

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

Claims (12)

1. a face angle representation method, is characterized in that, comprising: Determine the representation data q of the face angle; wherein, the representation data q is a quaternion, q=e ₀ +e ₁ i+e ₂ j+e ₃ k, i ² =j ² =k ² =ijk=−1 , e ₀ is called the real part of q, e ₁ i+e ₂ j+e ₃ k is called the imaginary part of q; The representation data q is visualized in three dimensions. 2. The method according to claim 1, wherein the relationship between the representation data q and the three-dimensional coordinate system is: For any point in the three-dimensional space, the new representation data is obtained after rotating according to the representation data q:

in,

is the quaternion representation of the point r=(r ₁ , r ₂ , r ₃ ) in the three-dimensional coordinate system before rotation; q ^* = e ₀ -e ₁ ie ₂ je ₃ k; the three-dimensional coordinates corresponding to the new representation data The point under the system is r'=(r' ₁ , r' ₂ , r' ₃ ). 3. The method according to claim 1, wherein said determining the representation data q of the face angle comprises: obtaining through a pre-established face angle model or pre-marked face angle data, or obtaining by preset . 4. a face angle labeling method, is characterized in that, comprises: Determine the facial image to be annotated; According to the initial face angle representation data q ₀ visualized in advance and the three-dimensional coordinate system of the face image to be marked, the representation data q ₁ is obtained; According to the representation data q ₁ and the representation data q ₀ , obtain the face angle representation data q _g of the facial image to be marked; Label the face angle of the face image according to the face angle representation data q _g ; Wherein, the representation data q ₀ , q ₁ , and q _g are all quaternions, including real and imaginary parts. 5. The method according to claim 4, wherein the initial facial angle unambiguous representation q ₀ is obtained by a pre-established facial angle model or pre-labeled facial angle data, or the initial facial angle is unambiguous. Ambiguity means that q ₀ is set to q ₀ =1+0i+0j+0k. 6. The method according to claim 4, characterized in that, according to the pre-visualized initial face angle representation data q ₀ and the three-dimensional coordinate system of the facial image to be marked, the representation data q ₁ is obtained, comprising: The three-dimensional coordinate system corresponding to the pre-visualized initial face angle representation data q ₀ is rotated so that the three-dimensional coordinate system coincides with the three-dimensional coordinate system of the facial image to be marked, and the representation data q ₁ at this time is obtained. 7. The method according to claim 6, wherein the representation data after each point is rotated is:

in,

is the quaternion representation of the point r=(r ₁ , r ₂ , r ₃ ) under the three-dimensional coordinate system before rotation; the point under the three-dimensional coordinate system of the facial image to be marked is r′=(r′ ₁ , r ' ₂ , r' ₃ ). 8 . The method according to claim 4 , wherein the face angle representation data of the face image to be marked is q _g =q ₁ q ₀ . 9. A face angle representation device, characterized in that, comprising: A data determination module, used for determining the representation data q of the face angle; wherein, the representation data q is a quaternion, q=e ₀ +e ₁ i+e ₂ j+e ₃ k, i ² =j ² =k ² =ijk=-1, e ₀ is called the real part of q, e ₁ i+e ₂ j+e ₃ k is called the imaginary part of q; The representation module is used for three-dimensional visualization of the representation data q. 10. A device for labeling facial angles, comprising: an image determination module, used to determine the facial image to be marked; a first processing module, configured to obtain the representation data q ₁ according to the pre-visualized initial face angle representation data q ₀ and the three-dimensional coordinate system of the facial image to be marked; a second processing module, configured to obtain, according to the representation data q ₁ and the representation data q ₀ , the face angle representation data q _g of the facial image to be marked; a labeling module, configured to label the facial angle of the facial image according to the facial angle representation data _qg ; Wherein, the representation data q ₀ , q ₁ , and q _g are all quaternions, including real and imaginary parts. 11. A computer storage medium, characterized in that, a computer program is stored thereon, and when the computer program is executed by a processor, the method according to any one of claims 1 to 3 or any one of claims 4 to 8 is implemented. steps of the method described. 12. An electronic device, characterized by comprising a memory and one or more processors, wherein the memory is used to store one or more programs; the one or more programs are controlled by the one or more processors When executed, the method according to any one of claims 1 to 3 or the method according to any one of claims 4 to 8 is implemented.

Images