CN103106385B - Face image recognition method and device - Google Patents

Abstract

The invention discloses a face image identification device and method. The processor of the face image recognition device calculates a red component statistical information, a green component statistical information and a blue component statistical information for each of a plurality of face images. The processor further processes at least two color component statistics of the red, green and blue component statistics by an independent component analysis to obtain a first component information and a second component information. The processor converts the first component information and the second component information into a frequency domain to obtain a first frequency domain information and a second frequency domain information. The processor calculates an energy value of the first frequency domain information in a preset frequency range, and determines that the face images are captured from a real person or a dummy according to the energy value.

Description

Face image recognition method and device

technical field

The present invention relates to a face image recognition method and device; in detail, it relates to a face image recognition method and device using independent component analysis.

Background technique

In recent years, due to the rapid development of multimedia technology, face recognition technology has been widely used in different application fields, such as computer games, monitoring systems and so on. In face recognition technology, in addition to identifying whether an image contains a human face, it is more important to determine whether the face in the image is taken from a "real person" or a "dummy".

The prior art uses a plurality of face images to judge whether the faces in them are "real people" or "dummy people". Specifically, the conventional technology first judges whether these human face images contain human eyes, and then judges whether there is an action of "blinking" from these human face images. If the result of the judgment is "with blinking action", it means that the face image in the image is a real face image; if the result of the judgment is "no blinking action", it means that the face image in the image is a dummy face images.

Since the conventional technology needs to judge whether the human face image contains human eyes, a high-resolution image is required to obtain a more accurate result. When the resolution is low, the probability of misjudgment will be greatly increased.

In view of this, there is still an urgent need in the art for a technology capable of judging whether a face image is captured from a real person or a dummy without being affected by the image resolution.

Contents of the invention

To solve the aforementioned problems, the present invention provides a face image recognition method and a face image recognition device.

The face image recognition method provided by the present invention is applicable to an electronic device. The electronic device includes a processor and a memory unit. The memory unit stores a plurality of face images. The face image recognition method includes the following steps: (a) making the processor calculate a red component statistical information, a green component statistical information and a blue component statistical information for each of the human face images; (b) making the processor Using an independent component analysis (IndependentComponentAnalysis) to process at least two color component statistical information among the red component statistical information, the green component statistical information, and the blue component statistical information, thereby obtaining a first component information and a second component information; (c) causing the processor to convert the first component information and the second component information into a frequency domain, thereby obtaining a first frequency domain information and a second frequency domain information (d) causing the processor to calculate an energy value of the first frequency domain information in a predetermined frequency range; and (e) causing the processor to compare the energy value with a preset value, thereby determining the The face image is extracted from a real person or a dummy.

The face image recognition device provided by the present invention includes a memory unit and a processor. The memory unit stores a plurality of face images, and the processor is electrically connected to the memory unit. The processor calculates a red component statistical information, a green component statistical information and a blue component statistical information for each of the face images. The processor then uses an independent component analysis method to process at least two color component statistical information among the red component statistical information, the green component statistical information, and the blue component statistical information, thereby obtaining a first Composition information and a second composition information. The processor further transforms the first component information and the second component information into a frequency domain, thereby obtaining a first frequency domain information and a second frequency domain information. The processor further calculates an energy value of the first frequency domain information in a preset frequency range, and compares the energy value with a preset value, thereby determining that the face images are captured from a real person or a person dummy.

The face image recognition method and the face image recognition device provided by the present invention use the independent component analysis method to analyze the statistical information of different color components of a plurality of face images, thereby obtaining the first component information and the second component information . Afterwards, the first component information and the second component information are converted to the frequency domain to analyze the energy value of the two component information in the frequency domain, and then judge whether the face images are taken from real people or dummy according to the energy value. In short, the present invention analyzes the statistical information of the color components in the face image for subsequent processing and judgment. From the perspective of each face image, since each color component comes from the face image rather than a small part of the image, the information covered is more accurate and will not be affected by poor resolution of the image to subsequent processing and judgment results.

In order to make the above-mentioned purpose, technical features and advantages of the present invention more comprehensible, the following is a detailed description of preferred embodiments with accompanying drawings.

Description of drawings

Fig. 1 is a schematic diagram depicting the internal components of the face image recognition device of the first embodiment;

2A and 2B are flow charts depicting the face image recognition method of the third embodiment;

Figure 2C is a detailed flowchart depicting step 21 of the third embodiment; and

FIG. 3 is a flow chart depicting the face image recognition method of the fourth embodiment.

Description of main component symbols:

1 Facial image recognition device

11 image sensor

13 processors

15 memory cells

17 receiving interface

detailed description

The following will illustrate the face image recognition device and the face image recognition method provided by the present invention through the embodiments. However, the embodiments of the present invention are not intended to limit the present invention to be implemented in any environment, application or manner as described in the embodiments. Therefore, the descriptions about the embodiments are only for the purpose of illustrating the present invention, rather than directly limiting the present invention. It should be noted that in the following embodiments and illustrations, elements not directly related to the present invention have been omitted and not shown.

The first embodiment of the present invention is a face image recognition device 1 , the schematic diagram of its internal components is shown in FIG. 1 . The face image recognition device 1 includes an image sensor 11, a processor 13, a memory unit 15 and a receiving interface 17, wherein the processor 13 is electrically connected to the image sensor 11, the memory unit 15 and the receiving interface 17 . The image sensor 11 can be a video camera (Webcam) or other devices capable of image capture that can be easily imagined by those skilled in the art. The processor 13 can be any one of various processors, central processing units, microprocessors or other computing devices known to those skilled in the art to which the present invention pertains. The memory unit 15 can be a memory, a floppy disk, a hard disk, an optical disk, a pen drive, a magnetic tape, a database accessible from the network, or a storage medium that can be easily imagined by those skilled in the art to have the same function.

First, the image sensor 11 captures a plurality of images within a time interval. For each image captured by the image sensor 11 , the processor 13 performs face detection. If the detected result is that the image contains a human face, the processor 13 will regard the image as a human face image and store it in the memory unit 15; if the detected result does not contain a human face in the image, the processor 13 will 13 discards the image. The face detection technology of the image by the processor 13 is common knowledge in the technical field of the present invention, and is not the focus of the present invention, so details thereof will not be repeated. It is assumed that a plurality of face images have been stored in the memory unit 15 after the aforementioned processing is performed by the image sensor 11 and the processor 13 .

It should be noted that other face image recognition devices implemented according to the present invention may not be equipped with the image sensor 11 . In this case, the user can first store a plurality of face images in the memory unit.

Next, the processor 13 calculates a red component statistical information, a green component statistical information and a blue component statistical information for each face image. An example is hereby used to illustrate the calculation methods of the statistical information of the red component, the statistical information of the green component and the statistical information of the blue component. For example, the processor 13 can first separate the red component, green component and blue component of each face image, and after processing all the face images, a plurality of red component images, a plurality of green component images and Multiple blue component images.

The processor 13 can then perform calculations on each of the red component images to obtain a piece of red component information. After processing all the red component images, a plurality of red component information can be obtained. Similarly, the processor 13 may perform calculations on each of the green component images to obtain a piece of green component information, and obtain a plurality of green component information after processing all the green component images. Similarly, the processor 13 may perform calculations on each of the blue component images to obtain a piece of blue component information, and obtain a plurality of blue component information after processing all the blue component images.

For example, each piece of red component information can be the average luminance value, maximum luminance value, minimum luminance value or other statistical values of the corresponding red component image; each piece of green component information can be the average luminance value, Maximum brightness value, minimum brightness value or other statistical values; each blue component information can be the average brightness value, maximum brightness value, minimum brightness value or other statistical values of the corresponding blue component image.

After obtaining the red component information, the green component information and the blue component information, the processor 13 selects two of the color component information for subsequent processing, assuming that the processor 13 selects the red component information and the green component information Ingredient information. Afterwards, the processor 13 processes the red component information and the green component information by independent component analysis, and obtains a first component information and a second component information after processing. It should be noted that the independent component analysis method is an existing algorithm. When there are two sets of input data, the independent component analysis method will generate two sets of output data. This is the common knowledge in the technical field of the present invention. In addition, "first" and "second" in the first component information and the second component information are only used to distinguish the first component information and the second component information as different component information.

The purpose of adopting the independent component analysis method is to separate the human heartbeat signal and other signals in the red component information and the green component information. In other words, one of the first component information and the second component information can be regarded as a human heartbeat signal. The human heartbeat signal has special characteristics in the frequency domain, that is, the signal energy is stronger in a specific frequency range. In this embodiment, the specific frequency range is set as a preset frequency range for subsequent processing. Generally speaking, the frequencies of most human heartbeat signals are relatively low, so the preset frequency range can be set at low frequencies.

Specifically, the processor 13 then converts the first component information and the second component information into a frequency domain, thereby obtaining a first frequency domain information and a second frequency domain information. For example, the processor 13 may adopt Fast Fourier Transform (FFT) to convert the first component information and the second component information into first frequency domain information and second frequency domain information respectively. Other face image recognition devices implemented according to the present invention may also adopt other frequency domain conversion methods.

Afterwards, the processor 13 calculates an energy value of the first frequency domain information in a preset frequency range, and then compares the energy value with a preset value. If the energy value of the first frequency domain information in the predetermined frequency range is greater than the predetermined value, the face image stored in the memory unit 15 is captured from a real person. If the energy value of the first frequency domain information in the preset frequency range is smaller than the preset value, the processor 13 recalculates an energy value of the second frequency domain information in the preset frequency range, and then compares the energy value with the preset value. Compare. If the energy value of the second frequency domain information in the predetermined frequency range is greater than the predetermined value, the face image stored in the memory unit 15 can also be identified as being captured from a real person. If the energy value of the first frequency domain information in the preset frequency range is smaller than the preset value, and the energy value of the second frequency domain information in the preset frequency range is also smaller than the preset value, the face image stored in the memory unit 15 would be assumed to be taken from a dummy. In short, as long as the energy value of one of the first frequency domain information or the second frequency domain information is greater than a preset value, the face image stored in the memory unit 15 will be identified as being captured from a real person.

If after the aforementioned processing, the processor 13 determines that the face images are captured from a real person, the face image recognition device 1 will continue to perform the login procedure. Specifically, the receiving interface 17 receives login information, and the processor 13 further processes the login information. For example, if the login information includes the user's account number and password, the processor 13 will determine whether the account number and password are correct. However, if after the aforementioned processing, the processor 13 determines that the face images are captured from a dummy, the face image recognition device 1 will not perform the login process.

The second embodiment of the present invention is also a face image recognition device 1 , but its detailed operation is slightly different from that of the first embodiment. Only the differences between the second embodiment and the first embodiment will be described below.

In this embodiment, after the processor 13 obtains the red component information, the green component information and the blue component information, the processor 13 selects all the color component information for subsequent processing. Afterwards, the processor 13 processes the red component information, the green component information and the blue component information by independent component analysis, and obtains a first component information, a second component information and a first component information after processing. Three component information. It should be noted that the independent component analysis method is an existing algorithm. When there are three sets of input data, the independent component analysis method will generate three sets of output data. This is a common knowledge person in the technical field of the present invention. In addition, "first", "second" and "third" in the first component information, the second component information and the third component information are only used to distinguish the first component information, the second component information and the third component information as Different ingredient information only.

Since the processor 13 of this embodiment selects three kinds of color component information for subsequent processing, the independent component analysis method will separate the human heartbeat signal, human heartbeat signal, human Moving (or shaking) signals with other signals. In other words, one of the first component information, the second component information and the third component information can be regarded as a human heartbeat signal.

The processor 13 then converts the first component information, the second component information and the third component information into a frequency domain, thereby obtaining a first frequency domain information, a second frequency domain information and a third frequency domain information . Afterwards, the processor 13 calculates an energy value of the first frequency domain information in the preset frequency range, an energy value of the second frequency domain information in the preset frequency range, and an energy value of the third frequency domain information in the preset frequency range value. As long as the energy value of one of the first frequency domain information, the second frequency domain information or the third frequency domain information is greater than a preset value, the face image stored in the memory unit 15 is identified as being captured from a real person.

In addition to the above steps, the second embodiment can also perform all the operations and functions described in the first embodiment, and those skilled in the art can directly understand how the second embodiment performs these operations and functions based on the above-mentioned first embodiment function, so I won’t go into details.

The third embodiment of the present invention is a face image recognition method, the flow chart of which is shown in FIG. 2A and FIG. 2B . The face image recognition method is applicable to an electronic device, such as the face image recognition device 1 of the first embodiment and the second embodiment. The electronic device includes a processor and a memory unit, and the memory unit stores a plurality of face images.

First, the face image recognition method executes step S21, so that the processor calculates a red component statistical information, a green component statistical information and a blue component statistical information for each of the human face images.

For example, step S21 can be accomplished through the detailed flowchart depicted in FIG. 2C . In step S211, the face image recognition method makes the processor separate the red component, green component and blue component of each face image, thereby obtaining a plurality of red component images, a plurality of green component images and a plurality of blue components image. Next, step S212 enables the processor to perform calculations on each of the red component images to obtain the red component statistical information. Step S213 enables the processor to perform calculations on each of the green component images to obtain the green component statistical information. Step S214 enables the processor to perform calculations on each of the blue component images to obtain the blue component statistical information. It should be noted that the present embodiment does not limit the execution sequence of steps S212 , S213 and S214 . Furthermore, each statistical information of the red component can be the average luminance value, maximum luminance value, minimum luminance value or other statistical values of the corresponding red component image, and each of the green component statistical information is the corresponding green component image. The average brightness value, maximum brightness value, minimum brightness value or other statistical values, and each of the blue component statistical information is the average brightness value, maximum brightness value, minimum brightness value or other statistical values of the corresponding blue component image.

After step S21, the face image recognition method executes step S22, so that the processor uses the independent component analysis method to analyze at least two colors in the red component statistical information, the green component statistical information and the blue component statistical information The component statistical information is processed to obtain a first component information and a second component information.

Next, step S23 enables the processor to convert the first component information and the second component information into a frequency domain, thereby obtaining a first frequency domain information and a second frequency domain information. For example, step S23 can enable the processor to adopt the Fourier transform method.

The face image recognition method then executes step S24 to enable the processor to calculate an energy value of the first frequency domain information in a preset frequency range. Then step 25 is executed to enable the processor to determine whether the energy value of the first frequency domain information in the predetermined frequency range is greater than a predetermined value. If the judgment result of step S25 is yes, the face image recognition method executes step S29 to make the processor judge that the face images are captured from real people. If the result of step S25 is negative, step S26 is executed to enable the processor to calculate an energy value of the second frequency domain information in a predetermined frequency range. Then step S27 enables the processor to determine whether the energy value of the second frequency domain information in the preset frequency range is greater than a preset value. If the judgment result of step S27 is yes, the face image recognition method executes step S29 to make the processor judge that the face images are captured from real people. If the judgment result of step S25 is negative, the face image recognition method executes step S28 to make the processor judge that the face images are taken from dummy persons.

It should be noted that steps S24 to S29 can also be implemented by other steps instead. For example, an energy value of the first frequency domain information in a preset frequency range may be calculated first, and an energy value of the second frequency domain information in a preset frequency range may be calculated. Then, the one with the larger energy value is selected for subsequent comparison. If the larger of the two energy values is greater than the preset value, the processor can determine that the face images are taken from real people. If the larger of the two energy values is smaller than the preset value, the processor can determine that the face images are taken from dummy people.

If the determination result of the aforementioned step S27 is yes and the face image recognition method executes step S29 accordingly, the face image recognition method can continue to perform the login procedure. Specifically, the face image recognition method can further execute step S291, so that a receiving interface of the electronic device receives a login information; then, the face image recognition method executes step S293, so that the processor processes the login information. For example, if the login information includes the user's account number and password, step S293 enables the processor to determine whether the account number and password are correct. It should be noted that if the determination result of the aforementioned step S27 is negative and the face image recognition method executes step S28 accordingly, the face image recognition method will not perform the login procedure.

In addition to the above steps, the third embodiment can also perform all the operations and functions described in the first embodiment, and those skilled in the art can directly understand how the third embodiment performs these operations and functions based on the above-mentioned first embodiment function, so I won’t go into details.

The fourth embodiment of the present invention is a face image recognition method, the flowchart of which is shown in FIG. 3 . The face image recognition method is applicable to an electronic device, such as the face image recognition device 1 of the first embodiment and the second embodiment. The electronic device includes a processor and a memory unit, and the memory unit stores a plurality of face images. The detailed flow of the fourth embodiment is slightly different from that of the third embodiment. Only the differences between the fourth embodiment and the third embodiment will be described in detail below.

Same as the third embodiment, the face image recognition method of the fourth embodiment first executes step S21, so that the processor calculates a red component statistical information, a green component statistical information and a blue component statistical information for each of the human face images . Step S21 can also be completed by steps S211 , S212 , S213 and S214 shown in FIG. 2C .

The fourth embodiment then executes step S32, so that the processor uses the independent component analysis method to process all the statistical information of the red components, the statistical information of the green components, and the statistical information of the blue components. This obtains a first component information, a second component information and a third component information. Afterwards, step S33 enables the processor to convert the first component information, the second component information and the third component information into a frequency domain, thereby obtaining a first frequency domain information, a second frequency domain information and a Third frequency domain information.

Next, steps S34 , S35 and S36 are executed to enable the processor to calculate an energy value of each of the first frequency domain information, the second frequency domain information and the third frequency domain information in a preset frequency range. It should be noted that the fourth embodiment does not limit the execution order of steps S34 , S35 and S36 . In step S37, the processor selects the larger of the three energy values in steps S34, S35 and S36 for subsequent processing.

The fourth embodiment then executes step S38 to determine whether the energy value selected in step S37 is greater than a preset value. If the judging result of step S37 is yes, then step S39 is executed to make the processor judge that the face images are captured from real people. If the judgment result of step S37 is negative, the face image recognition method executes step S40 to make the processor judge that the face images are captured from dummy persons.

Similarly, steps S34 to S40 of the fourth embodiment can also be implemented by other steps instead. For example, the fourth embodiment can first calculate

An energy value of one of the first frequency domain information, the second frequency domain information and the third frequency domain information in a preset frequency range, and judging whether the energy value is greater than a preset value. If yes, the face images are taken from real people, and the operation can be stopped; if not, other frequency domain information is selected to calculate energy values and subsequent judgments. Subsequent processing and so on. If the energy values of all the frequency domain information are not greater than the preset value, then the face images are taken from dummy people.

In addition to the above-mentioned steps, the fourth embodiment can also perform all the operations and functions described in the second embodiment, and those skilled in the art can directly understand how the fourth embodiment performs these operations and functions based on the above-mentioned second embodiment function, so I won’t go into details.

In addition, the facial image recognition method described in the third embodiment and the fourth embodiment can be executed by a computer program product, when the electronic device or face recognition device loads the computer program product and executes the computer program product contained in the computer program product After a plurality of instructions, the projection method described in the third implementation and the fourth embodiment can be completed. The aforementioned computer program product may be stored in a computer-readable recording medium, such as read-only memory (ROM), flash memory, floppy disk, hard disk, optical disk, flash drive, magnetic tape, and data accessible through the Internet. library or any other storage medium known to those skilled in the art that can perform the same function.

From the above descriptions of the first to fourth embodiments, it can be seen that the face image recognition method and face image recognition device provided by the present invention use the independent component analysis method to analyze the statistical information of different color components of a plurality of face images , so as to obtain at least the first component information and the second component information. One of the first component information and the second component information can be regarded as a human heartbeat signal. The present invention then converts the first component information and the second component information to the frequency domain to analyze the energy value of the two component information in the frequency domain, and then judges based on the energy value whether the face images are taken from real people or dummy people .

In short, the present invention analyzes the statistical information of the color components in the face image for subsequent processing and judgment. From the perspective of each face image, since each color component comes from the entire image, not a small part of the image, the information covered is more accurate and will not be affected by the poor resolution of the image. to subsequent processing and judgment results.

The above-mentioned embodiments are only used to illustrate the implementation of the present invention and illustrate the technical features of the present invention, and are not intended to limit the scope of protection of the present invention. Any changes or equivalence arrangements that can be easily accomplished by those skilled in the art belong to the scope of the present invention, and the protection scope of the present invention should be based on the scope of the claims of the present invention.

Claims (17)

1. A face image recognition method, suitable for an electronic device, the electronic device includes a processor and a memory unit, the memory unit stores a plurality of face images, it is characterized in that the person The face image recognition method includes the following steps: (a) causing the processor to calculate a red component statistical information, a green component statistical information and a blue component statistical information for each of the described human face images; (b) causing the processor to perform an independent component analysis on at least two color component statistical information among the red component statistical information, the green component statistical information, and the blue component statistical information processing, thereby obtaining a first component information and a second component information; (c) causing the processor to respectively convert the first component information and the second component information into a frequency domain, thereby obtaining a first frequency domain information and a second frequency domain information; (d) causing the processor to calculate an energy value of the first frequency domain information in a predetermined frequency range; and (e) making the processor compare the energy value with a preset value, thereby determining that the face image is captured from a real person or a dummy. 2. the facial image recognition method as claimed in claim 1, is characterized in that, described step (a) comprises the following steps: (a1) causing the processor to separate the red component, green component and blue component of each of the human face images, thereby obtaining a plurality of red component images, a plurality of green component images and a plurality of blue component images; (a2) causing the processor to calculate each of the red component images to obtain the statistical information of the red component; (a3) causing the processor to calculate each of the green component images to obtain the green component statistical information; and (a4) enabling the processor to perform calculations on each of the blue component images to obtain the blue component statistical information. 3. The face image recognition method according to claim 2, wherein each of the statistical information of the red components is an average brightness value of the corresponding red component image, and each of the statistical information of the green components is The information is an average brightness value of the corresponding green component image, and each of the blue component statistical information is an average brightness value of the corresponding blue component image. 4. The face image recognition method as claimed in claim 1, wherein said step (b) is to make said processor utilize said independent component analysis method to analyze said red component statistical information, The green component statistical information and the blue component statistical information are processed to obtain the first component information, the second component information and a third component information, and the step (c) Further enabling the processor to convert the third component information into the frequency domain, thereby obtaining a third frequency domain information. 5. The facial image recognition method as claimed in claim 1, wherein said step (c) is to make said processor respectively analyze said first component information and said second component information performing a fast Fourier transform to obtain the first frequency domain information and the second frequency domain information. 6. The face image recognition method as claimed in claim 1, wherein when said energy value is greater than said preset value, said step (e) determines said face image to be retrieved Taken from said real person. 7. The face image recognition method as claimed in claim 1, wherein when said energy value is less than said preset value, said step (e) determines said face image to be retrieved Taken from the dummies described. 8. The face image recognition method as claimed in claim 1, wherein said step (e) determines that said face image is captured from a real person, and said face image recognition method is further Contains the following steps: making a receiving interface of the electronic device receive a login message; and causing the processor to process the login information. 9. A human face image recognition device, characterized in that the human face image recognition device comprises: A memory unit for storing multiple face images; A processor electrically connected to the memory unit, the processor comprising: The first module is used to calculate a red component statistical information, a green component statistical information and a blue component statistical information for each of the human face images, The second module is used to use an independent component analysis method to process at least two kinds of color component statistical information among the red component statistical information, the green component statistical information and the blue component statistical information, by This obtains a first component information and a second component information, A third module, configured to respectively convert the first component information and the second component information into a frequency domain, thereby obtaining a first frequency domain information and a second frequency domain information, A fourth module, configured to calculate an energy value of the first frequency domain information in a preset frequency range, and The fifth module is used for comparing the energy value with a preset value, so as to determine that the face image is captured from a real person or a dummy. 10. The human face image recognition device according to claim 9, wherein the first module separates the red component, green component and blue component of each of the human face images, thereby obtaining a plurality of red component images, a plurality of green component images and a plurality of blue component images, the first module calculates each of the red component images to obtain the red component statistical information, and the first module calculates each The green component image is calculated to obtain the green component statistical information, and the first module is calculated for each of the blue component images to obtain the blue component statistical information. 11. The face image recognition device according to claim 10, wherein each of the statistical information of the red components is an average brightness value of the corresponding red component image, and each of the statistical information of the green components is The information is an average brightness value of the corresponding green component image, and each of the blue component statistical information is an average brightness value of the corresponding blue component image. 12. The human face image recognition device according to claim 9, wherein said second module utilizes said independent component analysis method to analyze said red component statistical information, said green component statistical information and The blue component statistical information is processed to obtain the first component information, the second component information and a third component information, and the second module further converts the third component information Convert to the frequency domain, thereby obtaining a third frequency domain information. 13. The face image recognition device according to claim 9, wherein said third module performs a fast Fourier transform on said first component information and said second component information respectively, In this way, the first frequency domain information and the second frequency domain information are obtained. 14. The face image recognition device according to claim 9, wherein when the energy value is greater than the preset value, the fifth module determines that the face image is to be captured from the real person described. 15. The face image recognition device according to claim 9, wherein when the energy value is smaller than the preset value, the fifth module determines that the face image is to be captured from the dummies described. 16. The human face image recognition device as claimed in claim 9, wherein the human face image recognition device further comprises: an image sensor for capturing a plurality of images; and The sixth module is used to perform face detection on each of the images to obtain the face images. 17. The human face image recognition device as claimed in claim 9, wherein the human face image recognition device further comprises: a receiving interface, electrically connected to the processor, and receiving a login message after the fifth module determines that the face image is captured from a real person; and The seventh module is used for processing the login information.