TWI466034B - Methods to Improve Face Recognition - Google Patents

Description

Method for improving face recognition rate

The present invention relates to a method for improving face recognition rate, and more particularly to a method for improving face recognition rate, which is suitable for a digital camera having a storage unit.

In today's daily life, digital products that replace traditional analog technology have become very popular, and digital cameras are a good example. The digital camera uses the light sensor to capture the image and convert it into a digital signal, and then save it in the format of the electronic image file. By adjusting various parameters, the user can shoot the desired digital image as desired. Most digital cameras today also provide users with many convenient functions, such as Auto Focusing, various scene modes, or Facial Recognition. With the features provided by the digital camera, users can more easily take satisfactory photos.

Among them, the technique of face recognition has been quite common in recent years, but there are still many shortcomings. Face recognition refers to the technique of analyzing and comparing the visual characteristics of faces to identify the identity. It is considered to be one of the most difficult research topics in the field of biometrics and even artificial intelligence. In fact, the shape of the face is very unstable, because people can produce a lot of expressions through changes in facial muscles. And at different viewing angles, the look of the face looks very different. In addition, face recognition is also affected by various factors such as lighting conditions (such as day and night, indoor and outdoor, etc.), coverings on people's faces (such as masks, sunglasses, hair, beards, etc.) and age.

At present, face recognition used by digital cameras mostly captures images from the front of a human face, and the accuracy of the same face with different expressions, shapes, lighting conditions, capturing angles or capturing distances is insufficient. Therefore, when a character as a target changes expression, changes shape, moves, or turns a head, the conventional face recognition method cannot be correctly recognized.

In order to solve the above problem of insufficient recognition rate, the present invention provides a method for improving the face recognition rate.

The method for improving the face recognition rate provided by the present invention can improve the face recognition rate without bothering the user.

The method for improving the face recognition rate provided by the present invention is applicable to a digital camera having a storage unit.

The storage unit of the digital camera stores at least one face information, and each face information includes at least one face feature.

The method for improving the face recognition rate comprises: capturing a first face image of a target person; performing a face recognition program according to the storage unit, and determining, by the face recognition program, whether the captured first face image corresponds to a person One of the face information; and when the first face image corresponds to one of the face information, an update procedure is performed on the face information to increase the character of the face information corresponding to the face information stored in the storage unit according to the target person Face features.

The updating program may include: when the first face image corresponds to one of the face information, the at least one second face image is tracked by the face tracking means by the face tracking means; and the second face image is analyzed to obtain at least one a new face feature; and a face feature that stores the new face feature to the storage unit as the face information corresponding to the first face image.

According to an embodiment of the present invention, the method for improving the face recognition rate may further include: when the first face image corresponds to one of the face information, the target object is captured by the storage unit in the corresponding face information. The message, and the image of the subject is displayed in the image corresponding to the subject.

When the first face image does not correspond to any of the face information, an additional program may be performed, and the newly added program includes: obtaining at least one new face feature according to the first face image; and adding a new face The feature is stored to the storage unit as a new face information. Here, the target person can be continuously tracked and the update program is repeatedly executed to obtain more facial features, thereby further improving the face recognition rate.

In addition, the face recognition program described above may include: calculating a face similarity according to the first face image and each face information; and determining whether any face similarity is greater than a first threshold value.

According to an embodiment of the present invention, when capturing the first face image and the second face image, different shooting parameters are used, and the shooting parameters are an exposure value, a focal length, a resolution, or a capture. The condition at which at least one of the angle and the captured distance captures the image. The face feature can be used to identify various features of a face, such as a face contour, a face skin color, or a facial face.

In summary, the method for improving the face recognition rate according to the present invention automatically captures new face features by the face tracking means, and then updates the corresponding face in the storage unit of the digital camera with the obtained face features. Face information is used as a follow-up face recognition program. Because the method of improving the face recognition rate can obtain the face features captured by the target characters in different capturing environments or different capturing angles, the face recognition rate of the target characters can be effectively improved. And the update program can be executed without the user's awareness.

The detailed features and advantages of the present invention are set forth in the Detailed Description of the Detailed Description of the <RTIgt; </ RTI> <RTIgt; </ RTI> </ RTI> </ RTI> <RTIgt; The objects and advantages associated with the present invention can be readily understood by those skilled in the art.

The method for improving the face recognition rate according to an embodiment of the present invention is applicable to a digital camera having a storage unit.

"FIG. 1" is a schematic diagram of the architecture of a digital camera to which the present invention is applied. The digital camera to which the present invention is applied may be, but is not limited to, the architecture shown in "FIG. 1".

Referring to FIG. 1 , the digital camera 100 can include a lens device 102 , a photosensitive element 106 , a sampling circuit 108 , a storage unit 112 , a driving motor 114 , and a processing unit 116 . The light reflected by the scene in front of the lens device 102 of the digital camera 100 enters the photosensitive element 106 via the lens device 102 and an aperture device (not shown), and the photosensitive element 106 converts the incoming light into an image signal and transmits it to the sampling. After the circuit 108, the image can be recorded in the storage unit 112.

At the time of capture, the processing unit 116 actuates the drive motor 114 to move the lens unit 102 to a specified focal length position, and then captures the shutter value and a capture aperture value. The photosensitive element 106 corresponds to the lens device 102 and converts a picture having a front scene into an electrical signal of a digital image. The sampling circuit 108 transmits the image received by the photosensitive element 106 to the storage unit 112 via the driving of the processing unit 116.

Next, a method of improving the face recognition rate according to an embodiment of the present invention will be described by the digital camera 100 constructed as described above.

Please refer to FIG. 2A, which is a schematic flowchart of a method for improving the face recognition rate according to an embodiment of the present invention.

Here, the storage unit 112 of the digital camera 100 includes at least one face information 120, and each face information 120 includes at least one facial feature 122. In more detail, the storage unit 112 can include a feature database as a basis for face recognition. The face information 120 is stored in the feature database of the storage unit 112, and each face information 120 corresponds to a target person.

Each of the subject characters may have its own different face features 122. For example, the facial features 122 can be a facial contour, a facial skin color, or a facial facial feature. The facial features 122 can also be facial features, nose shapes, freckles, or iris colors, etc., but the present invention Not limited to this. For example, the target person corresponding to the face information 120a has the face feature 122 whose content is "face contour: long ellipse" and "face skin color: brown", and the target character corresponding to the face information 120b The face feature 122 having the content "Iris Color: Blue". In this way, the face recognition program of the digital camera 100 can distinguish the face information 122 corresponding to the target person in the image by comparing the face feature 122 in the image and the face feature 122 in the feature database.

More preferably, the face information 120 may further include a subject information (not shown) for describing the corresponding person. The subject matter message can be the name, nickname or group category of the subject person. For example, the content of the subject matter message may be "Wang Xiao Ming", "Xiao Ming", "Classmate" or "Minister". The subject information may be input by the user, or may be automatically filled in by the digital camera 100 according to information such as the face feature 122 or the photographing time, such as "Blond" or "20091010-1".

In this embodiment, the digital camera 100 captures a first face image from at least one target person (step S30). The first face image is an image having a face of a person, and the target person is a person who is photographed by the first face image of the face. In other words, the digital camera 100 captures the scene of the person having the subject in front through the lens device 102 to obtain the first face image corresponding to the subject person.

After capturing the first face image, the digital camera 100 performs a face recognition process on the first face image to determine whether the captured first face image corresponds to the face information in the storage unit 112 through the face recognition program. One of 120 (step S40). Based on the feature database in the storage unit 112, the digital camera 100 determines whether the target person captured in the first face image corresponds to any one of the face information 120 in the feature database. In other words, the face information 120 belonging to the subject in the feature database is found through the face recognition program.

When the first face image corresponds to one of the face information 120, an update procedure is performed on the face information 120 (step S50). Wherein, when the face information 120 belonging to the target person is found in the feature database, the face information 120 may be updated in the face information 120 to add more face features 122 corresponding to the target person.

Furthermore, when the first face image does not correspond to any of the face information 120, the digital camera 100 can execute a new program (step S60), as shown in "FIG. 2B". In other words, when there is no face information 120 belonging to the target person in the feature database, an additional program may be executed to create the face information 120 of the person belonging to the target in the feature database.

Please refer to "FIG. 3". The face recognition program used in step S40 may include the following steps.

First, the digital camera 100 calculates a face similarity based on the first face image and each face information 120 in the face recognition program (step S42). The digital camera 100 extracts the features of the face of the target person from the first face image, and calculates the face similarity according to the feature of the first face image and each face data 120 in the feature database. In more detail, in the face recognition program, the digital camera 100 compares the features of the first face image with the face features 122 in the face data 120, and for example, the reciprocal of the difference obtained by the comparison is similar to the face. degree. In addition, different kinds of facial features 122 may be given different weights and weighted to calculate facial similarity. For example, hair color or iris color is easy to use to identify whether the face is the same, and can give a higher weight value. Therefore, for each face data 120, a face similarity can be calculated.

The digital camera 100 then compares each face similarity with the first threshold value in the face recognition program (step S44) to determine whether any of the face similarities are greater than the first threshold. For example, the features extracted from the first face image are "face contour: long ellipse" and "iris color: black", and the face information 120a includes "face contour: long oval" and "face" Skin color: brown" face feature 122, and face information 120b includes "iris color: blue" face feature 122. When the face similarity corresponding to the face information 120a is calculated to be higher than the first threshold, and the face similarity corresponding to the face information 120b is lower than the first threshold, the digital camera 100 determines the first of the captures. The face image corresponds to the face information 120a.

After the step S40 determines that the character of the first face image corresponds to one of the face information 120, the method of increasing the face recognition rate performs an update procedure in step S50. If it is determined in step S40 that the target person of the first face image does not correspond to any face information 120, then a new program is executed in step S60, or a method of increasing the face recognition rate is ended.

Please refer to "FIG. 4", and the update procedure executed in step S50 may include the following steps.

After the digital camera 100 captures the first face image for the subject person, the target person or digital camera 100 may move. However, when the target person is still in the range that can be captured by the digital camera 100, the digital camera 100 can track at least one second face image (step S52) by tracking the target person by a face tracking method in the update program. Tracking methods can use techniques such as motion detection and motion estimation to lock the position of the person in the picture range. The digital camera 100 captures at least one second face image from the subject person to obtain more information related to the target person.

The digital camera 100 analyzes the acquired second face image in the update program to obtain at least one new face feature (step S54). The new face feature also corresponds to the face information 120 of the corresponding target character in the feature database.

The second face image captured in step S52 may have a different shooting parameter than the first face image, that is, different shooting parameters are used when capturing the first face image and the second face image. The parameter is at least one of an exposure value, a focal length, a resolution, a capture angle, and a capture distance. The second face image can be captured during any period of tracking the target person. Therefore, when the digital camera 100 captures the second face image, it may use a different exposure value, a focal length, or a resolution than the first face image. The digital camera 100 can also capture the second face image from the target person at different angles of capture, and capture the side faces of different angles of the target person. Therefore, the second face image can extract new face features different from those stored in the feature database.

The face of the subject person has a great difference in different expressions, different angles of capture, or different lighting conditions (such as sunny, night, or cloudy), but the digital camera 100 can be updated in the above various situations. The second face image is taken, and the second face image is analyzed to obtain various new face features of the target person to improve the face recognition rate. Then, the new face feature obtained by analyzing the second face image is stored in the storage unit 112 as the face feature 122 of the face information 120 corresponding to the first face image (step S56), that is, the new face feature is newly added. The face information 120 corresponding to the subject is enhanced. By collecting various different facial features 122 of the subject person, a higher face recognition rate can be obtained. In this way, in addition to the face feature 122 of the corresponding face information 120, a face feature 122 as a new face feature can be used to execute the face recognition program.

More preferably, when the new face feature is updated into the face information 120, the environmental conditions (such as the exposure value or white balance) when the second face image is captured may also be recorded. In this way, when face recognition is required, the face feature 122 can be selected according to the current environmental conditions to perform face recognition, which can further improve the face recognition rate. For example, when performing a face recognition program at night, the applicable face feature 122 is a face feature 122 obtained from a second face image captured at night or under low exposure.

Again, the update procedure can be repeated to obtain more facial features 122. Referring to "figure 5", after the execution of step 56 in the update program, it is possible to continue to determine whether or not there is an image of the person who has tracked the target (step S58). When there is a person tracking the target, the digital camera 100 continues to capture the second face image (step S52), and analyzes the second face image (step S54) to add facial features to the corresponding face information 120. 122 (step S56).

As long as the face tracking means of the digital camera 100 also tracks the target person, the method of improving the face recognition rate can automatically track the target person, capture the second face image, and add a new face feature into the corresponding Face information 120. That is to say, in the case that the user does not know, the method of improving the face recognition rate can also obtain the corresponding face feature 122 anytime, anywhere and continuously, and is used to improve the face recognition rate.

In addition, when it is determined that the first facial image corresponds to one of the facial information 120, the method for increasing the facial recognition rate displays the target information in the facial information 120 corresponding to the first facial image. In more detail, the above-described step S30 to step S56 may be performed when the digital camera 100 is still in the state S _0, the message subject matter may be provided to the user in the state S _0.

The S ₀ is a preview mode (Preview) in which the digital camera 100 is turned on. The general two-stage shutter digital camera 100 can be divided into three states (modes) of S ₀ , S ₁ and S ₂ when in use. S ₀ is a preview mode, and is generally used by the digital camera 100 to capture a composition of an image or perform a function of adjusting a shooting parameter. S ₁ is a state in which the shutter button is half pressed (Half Shutter). The digital camera 100 entering this state performs auto-focus (AF) and is ready to enter the S ₂ state at any time. S ₂ is a state in which the shutter button is fully pressed (Full Shutter). When the user fully presses the shutter button, the digital camera 100 performs a final preparation such as fine-tuning the focus and officially shoots the image.

The digital camera 100 can display the corresponding target information of the target person identified in step S40 on the display screen for the user to refer to or confirm the result of the face recognition. And regardless of whether the digital camera 100 is in the state of S ₀ , S ₁ or S ₂ , the method of increasing the face recognition rate can be performed.

Please refer to "Figure 6". The new program can include the following steps. First, at least one new face feature is obtained according to the first face image (step S62), and the newly added face feature obtained according to the first face image is stored in the storage unit 112 as a new face information 120 ( Step S64). That is to say, in the newly added program, the face feature 122 obtained by analyzing the first face image is taken as a new face information new promotion storage unit 112. That is, in the new program, the digital camera 100 adds a face information 120 data to the target character in the feature database. The digital camera 100 can be added to the display screen to confirm the window before the addition, for the user to confirm whether to add the data of the target person. If the user determines that the target person is to be the new face information 120 new promotion feature database, the user may also input the target information of the new face information 120. The subject information of the new face information 120 may also be automatically filled in by the digital camera 100 according to the information such as the newly added face feature or the photographing time, and the automatically filled in subject matter message can be manually changed by the user.

In addition, the new program does not have to be executed when capturing the first face image. If the first face image is stored in the storage unit 112, the user can select whether to update the feature database with the new character when the image is subsequently viewed. If the first face image or the second face image has multiple target characters, the method for improving the face recognition rate may perform face recognition programs, update programs, and new programs on the target characters individually to improve The face recognition rate of these target characters.

In summary, the updating procedure of the method for improving the face recognition rate according to the present invention can automatically capture new facial features by the face tracking means without the user's perception, and then obtain the new facial features. The face feature updates the corresponding face information in the storage unit of the digital camera for use as a subsequent face recognition program. Since the method for improving the face recognition rate can obtain the face features of the target person under various capturing environments anytime, anywhere and continuously, the face recognition rate of the target person can be effectively improved. And the action of capturing the second face image and updating the face information corresponding to the target person can be performed without the user's knowledge.

While the present invention has been described above in terms of the preferred embodiments thereof, it is not intended to limit the invention, and the invention may be modified and modified without departing from the spirit and scope of the invention. The patent protection scope of the invention is subject to the definition of the scope of the patent application attached to the specification.

100. . . Digital camera

102. . . Lens device

106. . . Photosensitive element

108. . . Sample hold circuit

112. . . Storage unit

114. . . Drive motor

116. . . Processing unit

120, 120a, 120b. . . Face information

122. . . Face feature

1 is a schematic diagram of the architecture of a digital camera to which the present invention is applied;

2A is a schematic flow chart of a method for improving a face recognition rate according to an embodiment of the present invention;

2B is a diagram for improving face recognition rate according to another embodiment of the present invention.

2B is a schematic flow chart of a method for improving a face recognition rate according to another embodiment of the present invention;

3 is a schematic flow chart of a face recognition program according to an embodiment of the present invention;

4 is a schematic flow chart of an update procedure according to an embodiment of the present invention;

Figure 5 is a flow chart showing an update procedure according to another embodiment of the present invention;

Figure 6 is a flow chart showing a new procedure according to an embodiment of the present invention.

Claims (8)

A method for improving the face recognition rate is applicable to a digital camera, wherein the digital camera has a storage unit, the storage unit stores at least one face information, and the face information includes at least one face feature, and the face recognition is improved. The method includes: capturing a first face image of a target person; performing a face recognition program on the first face image, and determining, by the face recognition program, whether the captured first face image corresponds to One of the face information in the storage unit; and when the first face image corresponds to one of the face information, an update procedure is performed on the face information to be added to the target person according to the target person Corresponding to the face feature in the face information in the storage unit, wherein the updating process comprises: tracking a target person by using a face tracking means to capture at least one second face image; analyzing the second face image Obtaining at least one new face feature, wherein the new face feature is not present in the storage unit before the execution of the update program ; And wherein the new face to the storage unit is stored as the plurality of facial features of the first facial image corresponding to the face information. The method for improving the face recognition rate as described in claim 1 of the patent application scope includes: When the first facial image corresponds to one of the facial information, the storage unit captures a target information in the corresponding facial information, and displays the target information on the image of the target person. The method for improving face recognition rate according to claim 1, wherein the face recognition program comprises: calculating a face similarity according to the first face image and each of the face information; and determining whether Any one of the face similarities is greater than a first threshold value phase. The method for improving the face recognition rate according to the first aspect of the patent application, further comprising: when the first face image does not correspond to one of the face information, performing a new program, the new program includes : obtaining at least one new facial feature according to the first facial image; and storing the newly added facial feature to the storage unit as a new facial information. The method for improving the face recognition rate as described in claim 1 of the patent application further includes: repeatedly performing the update procedure. The method for improving face recognition rate according to claim 1, wherein the first face image and the second face image system use different shooting parameters. A method for improving a face recognition rate as described in claim 6 of the patent application, wherein The shooting parameters are at least one of an exposure value, a focal length, a resolution, a capture angle, and a capture distance. The method for improving the face recognition rate according to the first aspect of the patent application, wherein the face feature is a face contour, a face color or a face facial position.