WO2019071663A1

WO2019071663A1 - Electronic apparatus, virtual sample generation method and storage medium

Info

Publication number: WO2019071663A1
Application number: PCT/CN2017/108775
Authority: WO
Inventors: 戴磊
Original assignee: 平安科技（深圳）有限公司
Priority date: 2017-10-09
Filing date: 2017-10-31
Publication date: 2019-04-18
Also published as: CN107784625A; CN107784625B

Abstract

Disclosed are an electronic apparatus, a virtual sample generation method and a storage medium. The method comprises: carrying out wavelet transformation processing on a first human face image, and acquiring a first wavelet coefficient of a low and medium frequency part in a grayscale value corresponding to the first human face image and a second wavelet coefficient of a medium and high frequency part in the grayscale value corresponding to the first human face image; carrying out wavelet transformation processing on an average human face image under a first illumination condition, and acquiring a third wavelet coefficient of a low and medium frequency part in a grayscale value corresponding to the average human face image and a fourth wavelet coefficient of a medium and high frequency part in the grayscale value corresponding to the average human face image; and replacing the first wavelet coefficient with the third wavelet coefficient, and carrying out wavelet inverse transformation processing on the third wavelet coefficient and the second wavelet coefficient so as to obtain a sample of the first human face image under the first illumination condition. By means of the method, the present invention solves the problem that the training sample collection process is tedious, saves a lot of manpower, and improves the efficiency of sample generation.

Description

Electronic device, virtual sample generation method and storage medium

The present application claims priority to Chinese Patent Application No. 201710929640.4, entitled "Electronic Device, Virtual Sample Generation Method, and Storage Medium" on October 9, 2017, the entire contents of which are incorporated by reference. In this application.

Technical field

The invention relates to the field of face recognition, and in particular to an electronic device, a virtual sample generating method and a storage medium.

Background technique

With the advancement of technology and the improvement of social recognition, the application of face recognition technology is more and more extensive. For example, the face recognition access control attendance system is applied in enterprise management, and the face recognition security door application is used in residential security management. Face recognition systems and networks are used by public security, judicial or criminal investigation departments to search for fugitives nationwide.

At present, in the enterprise or residential security management, when face recognition technology is applied for face recognition, the accuracy of face recognition is related to the illumination intensity of the position of the identified person. Therefore, in order to improve the accuracy of face recognition, Images taken at different light intensities need to be collected as a training sample set. Usually, it is difficult to collect pictures of different people under different illumination conditions, and in the case where the number of training samples required is large, for example, the number of samples usually required is tens of thousands or hundreds of thousands, and the collection training is performed. The process of the sample is cumbersome, which not only leads to a waste of manpower, but also has low collection efficiency.

Summary of the invention

In view of this, the present invention provides an electronic device, a virtual sample generating method, and a storage medium, which can generate virtual samples corresponding to illumination conditions according to an average face image under predetermined illumination conditions, thereby solving the cumbersome process of collecting training samples. Problems, a lot of manpower savings, and the efficiency of generating samples.

First, in order to achieve the above object, the present invention provides an electronic device including a memory and a processor connected to the memory, the processor for executing a virtual sample generation program stored on the memory, The following steps are implemented when the virtual sample generation program is executed by the processor:

A. If it is required to obtain a sample of the first face image under the predetermined first illumination condition, the first face image is subjected to wavelet transform processing by using a predetermined wavelet transform rule to obtain the first person. a first wavelet coefficient of a low frequency portion of the gray value corresponding to the face image, and a second wavelet coefficient of the high frequency portion of the gray value corresponding to the first face image, wherein the first wavelet coefficient corresponds to a second illumination condition reflected in the first face image, the second wavelet coefficient corresponding to the first facial contour detail reflected in the first facial image;

B. Using the wavelet transform rule, performing a wavelet transform process on the predetermined average face image in the first illumination condition to obtain a third wavelet in the low frequency part of the gray value corresponding to the average face image a coefficient, and a fourth wavelet coefficient of the high frequency portion of the gray value corresponding to the average face image, wherein the third wavelet coefficient corresponds to the first illumination condition reflected in the average face image, The fourth wavelet coefficient corresponds to the second human face contour detail reflected in the average face image;

C. acquiring the third wavelet coefficient, and replacing the first wavelet coefficient with the third wavelet coefficient;

D. The third wavelet coefficient and the second wavelet coefficient are merged by using an inverse transform rule of the wavelet transform rule to obtain a virtual face of the first face image under the first illumination condition. Image sample.

In addition, to achieve the above object, the present invention further provides a virtual sample generating method, the method comprising the following steps:

B. Using the wavelet transform rule, performing a wavelet transform process on the predetermined average face image in the first illumination condition to obtain a third wavelet coefficient of the low frequency part in the gray value corresponding to the average face image And the average face image corresponds to a fourth wavelet coefficient of the high frequency portion of the gray value, wherein the third wavelet coefficient corresponds to a first illumination condition reflected in the average face image, the fourth wavelet The coefficient corresponds to the second face contour detail reflected in the average face image;

Further, in order to achieve the above object, the present invention also provides a computer readable storage medium storing a virtual sample generation program, the virtual sample generation program being executable by at least one processor such that The at least one processor performs the following steps:

A. If it is required to obtain a sample of the first face image under the predetermined first illumination condition, the first face image is subjected to wavelet transform processing by using a predetermined wavelet transform rule to obtain the first person. a first wavelet coefficient of a low frequency portion of the gray value corresponding to the face image, and a second wavelet coefficient of the high frequency portion of the gray value corresponding to the first face image, wherein the first wavelet coefficient Corresponding to the second illumination condition reflected in the first facial image, the second wavelet coefficient corresponds to the first facial contour detail reflected in the first facial image;

Compared with the prior art, the electronic device, the virtual sample generating method and the storage medium proposed by the present invention firstly perform wavelet transform processing on a first face image by using a predetermined wavelet transform rule to obtain a first face image. Corresponding to the first wavelet coefficient of the low frequency part of the gray value and the second wavelet coefficient of the high frequency part of the corresponding gray value of the first face image; and then performing wavelet transform on the average face image under the predetermined first illumination condition Processing to obtain a third wavelet coefficient of the low frequency portion of the gray value corresponding to the average face image, and a fourth wavelet coefficient of the high frequency portion of the gray value corresponding to the average face image; and finally replacing the first wavelet coefficient with the third wavelet Coefficient, keeping the second wavelet coefficient unchanged, and using the inverse transform of the wavelet transform rule to fuse the third wavelet coefficient with the second wavelet coefficient to obtain a virtual face of the first face image under the predetermined first illumination condition Image sample. The average face image under the first illumination condition and the first face image under the second illumination condition can be processed according to a predetermined wavelet transformation rule to obtain a virtual sample of the first face image under the first illumination condition. It solves the cumbersome problem of collecting training samples, saves a lot of manpower, and has high efficiency in generating samples.

DRAWINGS

1 is a schematic diagram of an optional hardware architecture of an electronic device according to the present invention;

2 is a schematic diagram of virtual functional modules of a computer program of a virtual sample generation system in an electronic device of the present invention;

3 is a schematic flow chart of an implementation of a preferred embodiment of a virtual sample generation method according to the present invention.

The implementation, functional features, and advantages of the present invention will be further described in conjunction with the embodiments.

Detailed ways

The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts are It belongs to the scope of protection of the present invention.

It should be noted that the descriptions of "first", "second" and the like in the present invention are for the purpose of description only, and are not to be construed as indicating or implying their relative importance or implicitly indicating the number of indicated technical features. . Thus, features defining "first" and "second" may include at least one of the features, either explicitly or implicitly. In addition, the technical solutions between the various embodiments may be combined with each other, but must be based on the realization of those skilled in the art, and when the combination of the technical solutions is contradictory or impossible to implement, it should be considered that the combination of the technical solutions does not exist. It is also within the scope of protection required by the present invention.

Referring to FIG. 1 , it is an optional hardware architecture diagram of the electronic device proposed by the present invention. In this embodiment, the electronic device 10 may include, but is not limited to, a memory 101, a processor 102, a network interface 103, and a communication bus 104 that are communicably connected to each other through a system bus. It should be noted that FIG. 1 only shows the electronic device 10 having the components 101-104, but it should be understood that not all illustrated components may be implemented, and more or fewer components may be implemented instead.

The memory 101 includes at least one type of computer readable storage medium including a flash memory, a hard disk, a multimedia card, a card type memory (eg, SD or DX memory, etc.), a random access memory (RAM), and a static Random access memory (SRAM), read only memory (ROM), electrically erasable programmable read only memory (EEPROM), programmable read only memory (PROM), magnetic memory, magnetic disk, optical disk, and the like. In some embodiments, the memory 101 can be an internal storage unit of the electronic device 10, such as a hard disk or memory of the electronic device 10. In other embodiments, the memory 101 may also be an external storage device of the electronic device 10, such as a plug-in hard disk equipped on the electronic device 10, a smart memory card (SMC), and a secure digital (Secure Digital, SD). ) cards, flash cards, etc. Of course, the memory 101 can also include both an internal storage unit of the electronic device 10 and an external storage device thereof. In this embodiment, the memory 101 is generally used to store an operating system installed on the electronic device 10 and various types of application software, such as a virtual sample generation program. Further, the memory 101 can also be used to temporarily store various types of data that have been output or are to be output.

Processor 102 may be a Central Processing Unit (CPU), controller, microcontroller, microprocessor, or other data processing chip in some embodiments. The processor 102 is typically used to control the overall operation of the electronic device 10. In this embodiment, the processor 102 is configured to run program code or processing data stored in the memory 101, such as a running virtual sample generation program or the like.

The network interface 103 can include a wireless network interface or a wired network interface, which is typically used to establish a communication connection between the electronic device 10 and other electronic devices.

Communication bus 104 is used to implement connection communication between components 101-103.

The computer program of the virtual sample generation system 200 stored in the memory 101 is executed by the processor 102 to implement the steps of the virtual sample generation method of various embodiments of the present application.

In the embodiment of the electronic device shown in FIG. 1, the virtual sample generation program stored in the memory 101 is executed by the processor 102 to implement the following steps:

A. If it is required to obtain a sample of the first face image under a predetermined first illumination condition, the first face image is subjected to wavelet transform processing by using a predetermined wavelet transform rule to obtain the first a first wavelet coefficient of a low frequency portion of the gray value corresponding to the face image, and a second wavelet coefficient of the high frequency portion of the gray value corresponding to the first face image, wherein the first wavelet coefficient corresponds to the first face image a second illumination condition reflected in the second wavelet coefficient corresponding to the first facial contour detail reflected in the first facial image;

B. Using wavelet transform rules, the average face image under the predetermined first illumination condition is subjected to wavelet transform processing to obtain the third wavelet coefficient of the low frequency part of the gray value corresponding to the average face image, and the average face The image corresponds to a fourth wavelet coefficient of the high frequency portion of the gray value, wherein the third wavelet coefficient corresponds to the first illumination condition reflected in the average face image, and the fourth wavelet coefficient corresponds to the second person reflected in the average face image Face contour detail;

C. acquiring a third wavelet coefficient, and replacing the first wavelet coefficient with the third wavelet coefficient;

D. Using the inverse transform rule of the wavelet transform rule, the third wavelet coefficient and the second wavelet coefficient are merged to obtain a virtual face image sample of the first face image under the first illumination condition.

In this embodiment, the solution of the present invention is explained by taking a sample under the first illumination condition as an example. When it is required to obtain a sample of the first face image under the first illumination condition, the first face image is subjected to wavelet transform processing using a wavelet transform rule. It can be understood that in the face recognition technology, the face image as the training sample needs to include the contour of the face, so in order to ensure the use value of the obtained sample, the first face image is wavelet transformed by using the wavelet transform rule. Preferably, before processing, it is preferred to detect a facial contour included in the first facial image, and determine that the first facial image includes all facial feature points of the facial contour.

Each face image is composed of pixels of certain data, for example, 128×68 pixels, and the pixel points of each pixel coordinate correspond to one gray value, and a face can be used by using wavelet transform rules. The gray value of the image is separated from the high frequency portion according to the image frequency distribution, and the wavelet coefficient corresponding to the low frequency portion of the gray value and the wavelet coefficient corresponding to the high frequency portion are obtained. In this embodiment, the first face image is subjected to wavelet transform processing by using a 3-layer 2-dimensional wavelet transform rule to obtain a first wavelet coefficient of the low-frequency portion of the gray value corresponding to the first face image, and the first person The second wavelet coefficient of the high frequency portion of the gray value corresponding to the face image.

After wavelet transform processing is performed on the first face image by using the wavelet transform rule, the wavelet transform rule is used to perform wavelet transform processing on the average face image under the first illumination condition to obtain the gray scale corresponding to the average face image. The third wavelet coefficient of the low frequency portion of the value, and the fourth wavelet coefficient of the high frequency portion of the gray value corresponding to the average face image.

The process of generating an average face image under the first lighting condition includes the following steps:

E. respectively acquiring a second face image taken by a preset number of people (for example, 2000 or 500 people) under the first lighting condition, and detecting a facial contour included in each second face image respectively;

F. cutting each second face image by using a predetermined cutting rule, so that each of the cropped second face images includes the detected face contour, and the size of each cropped second face image is made. The specifications are the same;

G. respectively calculating an average value of the gray values of the pixels of the same pixel coordinates of all the cropped second face images, and performing Gaussian filtering on the calculated average values of the gray values of the same pixel coordinates to obtain the Average face image.

Specifically, the steps of respectively calculating the average value of the gray values of each of the same pixel points of each second face image include:

Scanning all the pixels in each second face image separately;

Obtaining gray values of each of the same pixel points in each of the scanned second face images, and obtaining an average value of the acquired pixel gray values.

It should be noted that the predetermined cutting rule is: identifying all face feature points of the second face image to be cropped; determining a preset shape including all the recognized face feature points, for example, a rectangle, the minimum The bounding box, for example, the minimum bounding box refers to the bounding box of the preset shape including all the recognized face feature points, and the smallest bounding box. Next, the geometric center point of the minimum bounding box is determined, and the determined geometric center point is used as a cropping frame of a preset size specification. For example, in this embodiment, the cropping frame of the preset size specification may be x pixels*y pixels. The geometric center point of the cropping frame of the size determines the size and the position of the cropping frame for cropping the second human face image; finally, the second human face image is cropped by using a cropping frame that determines the size and the position of the region.

It can be understood that, in order to further ensure the accuracy of the recognition, before the first face image is subjected to wavelet transform processing by using a predetermined wavelet transform rule, it is required to detect the facial contour included in the first face image; using predetermined cutting. a rule, and the cutting rule for cropping each second face image, cropping the first face image, so that the cropped first face image includes the detected face contour, and the cropped first face image is made The size specifications are the same as the size specifications of the average face image.

The electronic device according to the above embodiment performs wavelet transform processing on the first face image, and acquires a first wavelet coefficient of the low frequency portion of the gray value corresponding to the first face image and a gray value corresponding to the first face image. The second wavelet coefficient of the middle and high frequency portions is subjected to wavelet transform processing on the average face image under the first illumination condition, and the third wavelet coefficient of the low frequency portion corresponding to the gray value corresponding to the average face image is obtained, and the average face image corresponding to the average face image is obtained. The fourth wavelet coefficient of the high frequency part of the gray value is replaced by the third wavelet coefficient, and then the third wavelet coefficient and the second wavelet coefficient are inversely transformed by wavelet, and the first face image is obtained. A sample under light conditions. The invention solves the cumbersome problem of collecting training samples in the above manner, saves a lot of manpower, and has high efficiency in generating samples.

It should be further noted that, as in the embodiment shown in FIG. 2 below, the computer program of the virtual sample generation system may be divided into different logic parts according to different functions implemented by the various parts thereof, and different logic parts may be used. A virtual functional module with different functions is described.

For example, please refer to FIG. 2, which is a schematic diagram of a virtual function module of a computer program of a virtual sample generation system in an electronic device of the present invention.

In this embodiment, each virtual function module of the computer program of the virtual sample generation system is named according to functions implemented by various parts of the computer program of the virtual sample generation system. For example, in FIG. 2, the computer program of the virtual sample generation system can be segmented into a first processing module 201, a second processing module 202, a replacement module 203, and a virtual sample generation module 204. The functions or operation steps implemented by the modules 201-204 are similar to the above, and are not described in detail herein, by way of example, For example, where:

The first processing module 201 is configured to perform wavelet transform processing on the first face image by using a predetermined wavelet transform rule to obtain a sample of the first face image under a predetermined first illumination condition. a first wavelet coefficient of a low frequency portion of the gray value corresponding to the first face image, and a second wavelet coefficient of the high frequency portion of the gray value corresponding to the first face image, wherein the first wavelet The coefficient corresponds to a second illumination condition reflected in the first facial image, and the second wavelet coefficient corresponds to a first facial contour detail reflected in the first facial image;

The second processing module 202 is configured to perform wavelet transform processing on the average face image in the first illumination condition by using the wavelet transform rule to obtain a low frequency in the gray value corresponding to the average face image. a portion of the third wavelet coefficient and a fourth wavelet coefficient of the high frequency portion of the gray value corresponding to the average face image, wherein the third wavelet coefficient corresponds to the first reflected in the average face image a lighting condition, the fourth wavelet coefficient corresponding to the second human face contour reflected in the average face image;

The replacing module 203 is configured to acquire the third wavelet coefficient, and replace the first wavelet coefficient with the third wavelet coefficient;

The virtual sample generation module 204 is configured to fuse the third wavelet coefficient and the second wavelet coefficient by using an inverse transform rule of the wavelet transform rule to obtain the first face image in the first illumination condition. A sample of the virtual face image below.

In addition, the present invention also provides a virtual sample generation method. Please refer to FIG. 3, which is a flowchart of an implementation of a preferred embodiment of the virtual sample generation method of the present invention. The method can be performed by a device that can be implemented by software and/or hardware.

In this embodiment, the virtual sample generation method includes:

Step S301, if it is required to obtain a sample of the first face image under the predetermined first illumination condition, performing the wavelet transform process on the first face image by using a predetermined wavelet transform rule to obtain the first a first wavelet coefficient of a low frequency portion of the gray value corresponding to the face image, and a second wavelet coefficient of the high frequency portion of the gray value corresponding to the first face image, wherein the first wavelet coefficient corresponds to a second illumination condition reflected in the first human face image, the second wavelet coefficient corresponding to the first facial contour detail reflected in the first facial image;

Step S302, using the wavelet transform rule, performing a wavelet transform process on the predetermined average face image under the first illumination condition to obtain a third of the low frequency part of the gray value corresponding to the average face image. a wavelet coefficient, and a fourth wavelet coefficient of the high frequency portion of the gray value corresponding to the average face image, wherein the third wavelet coefficient corresponds to the first illumination condition reflected in the average face image, The fourth wavelet coefficient corresponds to the second human face contour detail reflected in the average face image;

Step S303, acquiring the third wavelet coefficient, and replacing the first wavelet coefficient with the third wavelet coefficient;

Step S304, performing fusion processing on the third wavelet coefficient and the second wavelet coefficient by using an inverse transform rule of the wavelet transform rule to obtain a virtual person of the first face image under the first illumination condition. Face image sample.

In this embodiment, the solution of the present invention is explained by taking a sample under the first illumination condition as an example. When it is required to obtain a sample of the first face image under the first illumination condition, the first face image is subjected to wavelet transform processing using a wavelet transform rule. It can be understood that in the face recognition technology, the face image as the training sample needs to include the contour of the face, so in order to ensure the use value of the obtained sample, the first face image is wavelet transformed by using the wavelet transform rule. Before processing, it is preferable to detect a facial contour included in the first facial image, and after detecting all the facial feature points in the facial contour included in the first facial image, perform the first facial image according to a preset cutting rule. Crop to get the first face image containing all face feature points.

Each face image is composed of pixels of certain data, for example, 128×68 pixels, each pixel point corresponds to a gray value, and the gray value of a face image can be obtained by using wavelet transform rules. The low frequency portion and the high frequency portion are separated according to the image frequency distribution, and the wavelet coefficients corresponding to the low frequency portion of the gray value and the wavelet coefficients corresponding to the high frequency portion are obtained. In this embodiment, the first face image is subjected to wavelet transform processing by using a 3-layer 2-dimensional wavelet transform rule to obtain a first wavelet coefficient of the low-frequency portion of the gray value corresponding to the first face image, and the first person The second wavelet coefficient of the high frequency portion of the gray value corresponding to the face image.

Scanning all the pixels in each second face image separately;

It should be noted that the predetermined cutting rule is: identifying all face feature points of the second face image to be cropped; determining a preset shape including all the recognized face feature points, for example, a rectangle, the minimum The bounding box, for example, the minimum bounding box refers to the bounding box of the preset shape including all the recognized face feature points, and the smallest bounding box. Next, the geometric center point of the minimum bounding box is determined, and the determined geometric center point is used as a cropping frame of a preset size specification. For example, in this embodiment, the cropping frame of the preset size specification may be x pixels*y pixels. Size cropping The geometric center point of the frame determines the size and the position of the crop frame for cropping the second face image; finally, the second face image is cropped by using a crop frame that determines the size and the position of the region.

The virtual sample generating method according to the above embodiment performs wavelet transform processing on the first face image, and acquires a first wavelet coefficient of the low frequency part corresponding to the first human face image and a gray corresponding to the first face image. The second wavelet coefficient of the middle and high frequency portions of the degree value is subjected to wavelet transform processing on the average face image under the first illumination condition, and the third wavelet coefficient and the average face of the low frequency portion corresponding to the gray value corresponding to the average face image are obtained. The fourth wavelet coefficient of the high frequency part corresponding to the gray value of the image, the first wavelet coefficient is replaced by the third wavelet coefficient, and the third wavelet coefficient and the second wavelet coefficient are inversely transformed by the wavelet to obtain the first face image. A sample under the first lighting condition. The invention solves the cumbersome problem of collecting training samples in the above manner, saves a lot of manpower, and has high efficiency in generating samples.

In addition, an embodiment of the present invention further provides a computer readable storage medium, where the virtual sample generation program is stored on the computer readable storage medium, and the virtual sample generation program can be executed by at least one processor to implement the following steps:

A. If it is required to obtain a sample of the first face image under the predetermined first illumination condition, the first face image is subjected to wavelet transform processing by using a predetermined wavelet transform rule to obtain a corresponding image of the first face image. a first wavelet coefficient of the low frequency portion of the gray value, and a second wavelet coefficient of the high frequency portion of the gray value corresponding to the first face image, wherein the first wavelet coefficient corresponds to the second reflected in the first face image a light condition, the second wavelet coefficient corresponding to the first face contour detail reflected in the first face image;

B. Using wavelet transform rules, the average face image under the predetermined first illumination condition is subjected to wavelet transform processing to obtain the third wavelet coefficient and the average face image of the low frequency part of the corresponding gray value of the average face image. Corresponding to a fourth wavelet coefficient of the high frequency portion of the gray value, wherein the third wavelet coefficient corresponds to a first illumination condition reflected in the average face image, and the fourth wavelet coefficient corresponds to the second face reflected in the average face image Contour detail

Further, when the virtual sample generation program is executable by at least one processor, the following steps are further implemented:

E. respectively acquiring a second facial image captured by the preset number of persons under the first lighting condition, and respectively detecting a facial contour included in each of the second facial images;

F. cutting each of the second facial images by using a predetermined cutting rule, so that each of the cropped second facial images includes the detected facial contour, and the cropped second facial images are obtained. Dimensions are the same;

Detecting a facial contour included in the first facial image;

The first face image is cropped by using a predetermined cropping rule, so that the cropped first face image includes the detected face contour, and the size and size of the cropped first face image are The average face image has the same size specifications.

The specific embodiment of the computer readable storage medium of the present invention is substantially the same as the embodiments of the virtual sample generating method described above, and will not be described herein.

It can be understood that the predetermined first lighting conditions, the preset number, the preset size specifications, and the like, which are involved in the foregoing embodiments, need to be preset, and the user can set according to actual conditions.

The serial numbers of the embodiments of the present invention are merely for the description, and do not represent the advantages and disadvantages of the embodiments.

Through the description of the above embodiments, those skilled in the art can clearly understand that the foregoing embodiment method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be through hardware, but in many cases, the former is better. Implementation. Based on such understanding, the technical solution of the present invention, which is essential or contributes to the prior art, may be embodied in the form of a software product stored in a storage medium (such as ROM/RAM, disk, The optical disc includes a number of instructions for causing a terminal device (which may be a cell phone, a computer, a server, an air conditioner, or a network device, etc.) to perform the methods described in various embodiments of the present invention.

The above are only the preferred embodiments of the present invention, and are not intended to limit the scope of the invention, and the equivalent structure or equivalent process transformations made by the description of the present invention and the drawings are directly or indirectly applied to other related technical fields. The same is included in the scope of patent protection of the present invention.

Claims

An electronic device, comprising: a memory, and a processor coupled to the memory, the processor configured to execute a virtual sample generation program stored on the memory, the virtual sample generation program being The processor implements the following steps when executed:

A. If it is required to obtain a sample of the first face image under the predetermined first illumination condition, the first face image is subjected to wavelet transform processing by using a predetermined wavelet transform rule to obtain the first person. a first wavelet coefficient of a low frequency portion of the gray value corresponding to the face image, and a second wavelet coefficient of the high frequency portion of the gray value corresponding to the first face image, wherein the first wavelet coefficient corresponds to a second illumination condition reflected in the first face image, the second wavelet coefficient corresponding to the first facial contour detail reflected in the first facial image;

B. Using the wavelet transform rule, performing a wavelet transform process on the predetermined average face image in the first illumination condition to obtain a third wavelet in the low frequency part of the gray value corresponding to the average face image a coefficient, and a fourth wavelet coefficient of the high frequency portion of the gray value corresponding to the average face image, wherein the third wavelet coefficient corresponds to the first illumination condition reflected in the average face image, The fourth wavelet coefficient corresponds to the second human face contour detail reflected in the average face image;

C. acquiring the third wavelet coefficient, and replacing the first wavelet coefficient with the third wavelet coefficient;

D. The third wavelet coefficient and the second wavelet coefficient are merged by using an inverse transform rule of the wavelet transform rule to obtain a virtual face of the first face image under the first illumination condition. Image sample.
The electronic device according to claim 1, wherein in the step B, the generating process of the average face image comprises the following steps:

E. respectively acquiring a second facial image captured by the preset number of persons under the first lighting condition, and respectively detecting a facial contour included in each of the second facial images;

F. cutting each of the second facial images by using a predetermined cutting rule, so that each of the cropped second facial images includes the detected facial contour, and the cropped second facial images are obtained. Dimensions are the same;

G. respectively calculating an average value of the gray values of the pixels of the same pixel coordinates of all the cropped second face images, and performing Gaussian filtering on the calculated average values of the gray values of the same pixel coordinates to obtain the Average face image.
The electronic device of claim 2, wherein the predetermined cropping rule is:

Identifying all facial feature points of the second face image to be cropped;

Determining a minimum bounding box containing a preset shape of all recognized face feature points;

Determining a geometric center point of the minimum bounding box, and determining the geometric center point as a geometric center point of the trimming frame of the preset size specification, and determining a size and a region position of the cropping frame for cropping the second face image;

The second face image is cropped using a crop frame that determines the size and location of the region.
The electronic device according to claim 3, further comprising the following steps before said step A:

Detecting a facial contour included in the first facial image;

The first face image is cropped by using a predetermined cropping rule, so that the cropped first face image includes the detected face contour, and the size and size of the cropped first face image are The average face image has the same size specifications.
The electronic device according to claim 1, wherein said wavelet transform rule comprises a 3-layer 2-dimensional wavelet transform, and an inverse transform rule of said wavelet transform rule is an inverse transform of said 3-layer 2-dimensional wavelet transform.
The electronic device according to claim 2, wherein said wavelet transform rule comprises a 3-layer 2-dimensional wavelet transform, and an inverse transform rule of said wavelet transform rule is an inverse transform of said 3-layer 2-dimensional wavelet transform.
The electronic device according to claim 3, wherein said wavelet transform rule comprises a 3-layer 2-dimensional wavelet transform, and an inverse transform rule of said wavelet transform rule is an inverse transform of said 3-layer 2-dimensional wavelet transform.
The electronic device according to claim 4, wherein said wavelet transform rule comprises a 3-layer 2-dimensional wavelet transform, and an inverse transform rule of said wavelet transform rule is an inverse transform of said 3-layer 2-dimensional wavelet transform.
A virtual sample generation method, characterized in that the method comprises the following steps:

A. If it is required to obtain a sample of the first face image under the predetermined first illumination condition, the first face image is subjected to wavelet transform processing by using a predetermined wavelet transform rule to obtain the first person. a first wavelet coefficient of a low frequency portion of the gray value corresponding to the face image, and a second wavelet coefficient of the high frequency portion of the gray value corresponding to the first face image, wherein the first wavelet coefficient corresponds to a second illumination condition reflected in the first face image, the second wavelet coefficient corresponding to the first facial contour detail reflected in the first facial image;

B. Using the wavelet transform rule, performing a wavelet transform process on the predetermined average face image in the first illumination condition to obtain a third wavelet coefficient of the low frequency part in the gray value corresponding to the average face image And the average face image corresponds to a fourth wavelet coefficient of the high frequency portion of the gray value, wherein the third wavelet coefficient corresponds to a first illumination condition reflected in the average face image, the fourth wavelet The coefficient corresponds to the second face contour detail reflected in the average face image;

C. acquiring the third wavelet coefficient, and replacing the first wavelet coefficient with the third wavelet coefficient;

D. The third wavelet coefficient and the second wavelet coefficient are merged by using an inverse transform rule of the wavelet transform rule to obtain a virtual face of the first face image under the first illumination condition. Image sample.
The virtual sample generating method according to claim 9, wherein in the step B, the generating process of the average face image comprises the following steps:

E. respectively obtaining a second face map taken by the preset number of people under the first lighting condition And detecting, respectively, a facial contour included in each of the second facial images;

F. cutting each of the second facial images by using a predetermined cutting rule, so that each of the cropped second facial images includes the detected facial contour, and the cropped second facial images are obtained. Dimensions are the same;

G. respectively calculating an average value of the gray values of the pixels of the same pixel coordinates of all the cropped second face images, and performing Gaussian filtering on the calculated average values of the gray values of the same pixel coordinates to obtain the Average face image.
The virtual sample generating method according to claim 10, wherein the predetermined cutting rule is:

Identifying all facial feature points of the second face image to be cropped;

Determining a minimum bounding box containing a preset shape of all recognized face feature points;

Determining a geometric center point of the minimum bounding box, and determining the geometric center point as a geometric center point of the trimming frame of the preset size specification, and determining a size and a region position of the cropping frame for cropping the second face image;

The second face image is cropped using a crop frame that determines the size and location of the region.
The virtual sample generating method according to claim 11, further comprising the following steps before the step A, comprising:

Detecting a facial contour included in the first facial image;

The first face image is cropped by using a predetermined cropping rule, so that the cropped first face image includes the detected face contour, and the size and size of the cropped first face image are The average face image has the same size specifications.
The virtual sample generating method according to claim 9, wherein the wavelet transform rule comprises a 3-layer 2-dimensional wavelet transform, and an inverse transform rule of the wavelet transform rule is an inverse transform of the 3-layer 2-dimensional wavelet transform .
The virtual sample generating method according to claim 10, wherein the wavelet transform rule comprises a 3-layer 2-dimensional wavelet transform, and an inverse transform rule of the wavelet transform rule is an inverse transform of the 3-layer 2-dimensional wavelet transform .
The virtual sample generating method according to claim 11, wherein the wavelet transform rule comprises a 3-layer 2-dimensional wavelet transform, and an inverse transform rule of the wavelet transform rule is an inverse transform of the 3-layer 2-dimensional wavelet transform .
The virtual sample generating method according to claim 12, wherein the wavelet transform rule comprises a 3-layer 2-dimensional wavelet transform, and an inverse transform rule of the wavelet transform rule is an inverse transform of the 3-layer 2-dimensional wavelet transform .
A computer readable storage medium storing a virtual sample generation program, the virtual sample generation program being executable by at least one processor to cause the at least one processor to perform the following steps:

A. If it is required to obtain a sample of the first face image under the predetermined first illumination condition, the first face image is subjected to wavelet transform processing by using a predetermined wavelet transform rule to obtain the first person. a first wavelet coefficient of a low frequency portion of the gray value corresponding to the face image, and the first person a second wavelet coefficient of the high frequency portion of the gray value corresponding to the face image, wherein the first wavelet coefficient corresponds to a second illumination condition reflected in the first face image, and the second wavelet coefficient corresponds to Describe the first face contour detail reflected in the first face image;

B. Using the wavelet transform rule, performing a wavelet transform process on the predetermined average face image in the first illumination condition to obtain a third wavelet coefficient of the low frequency part in the gray value corresponding to the average face image And the average face image corresponds to a fourth wavelet coefficient of the high frequency portion of the gray value, wherein the third wavelet coefficient corresponds to a first illumination condition reflected in the average face image, the fourth wavelet The coefficient corresponds to the second face contour detail reflected in the average face image;

C. acquiring the third wavelet coefficient, and replacing the first wavelet coefficient with the third wavelet coefficient;

D. The third wavelet coefficient and the second wavelet coefficient are merged by using an inverse transform rule of the wavelet transform rule to obtain a virtual face of the first face image under the first illumination condition. Image sample.
The computer readable storage medium according to claim 17, wherein in the step B, the generating process of the average face image comprises the following steps:

E. respectively acquiring a second facial image captured by the preset number of persons under the first lighting condition, and respectively detecting a facial contour included in each of the second facial images;

F. cutting each of the second facial images by using a predetermined cutting rule, so that each of the cropped second facial images includes the detected facial contour, and the cropped second facial images are obtained. Dimensions are the same;

G. respectively calculating an average value of the gray values of the pixels of the same pixel coordinates of all the cropped second face images, and performing Gaussian filtering on the calculated average values of the gray values of the same pixel coordinates to obtain the Average face image.
The computer readable storage medium of claim 18, wherein the predetermined cropping rule is:

Identifying all facial feature points of the second face image to be cropped;

Determining a minimum bounding box containing a preset shape of all recognized face feature points;

Determining a geometric center point of the minimum bounding box, and determining the geometric center point as a geometric center point of the trimming frame of the preset size specification, and determining a size and a region position of the cropping frame for cropping the second face image;

The second face image is cropped using a crop frame that determines the size and location of the region.
The computer readable storage medium according to claim 19, further comprising the following steps, including:

Detecting a facial contour included in the first facial image;

The first face image is cropped by using a predetermined cropping rule, so that the cropped first face image includes the detected face contour, and the size and size of the cropped first face image are The average face image has the same size specifications.