CN102087714B - Image recognition login system and method thereof - Google Patents

Abstract

The present invention discloses an image recognition login system and method, which is suitable for verifying the login procedure of a user. A dynamic image acquisition module is used to pre-establish a plurality of first dynamic images in a storage module. When the user wants to perform a login action, the dynamic image acquisition module acquires a second dynamic image of the user and transmits the second dynamic image to a processing module. A comparison module of the processing module compares the first dynamic image with the continuous static images of the second dynamic image to obtain a comparison result, and the processing module performs the next step according to the comparison result. This image recognition login method is safe and can avoid login with forged images.

Description

Image recognition login system and method thereof

technical field

The invention relates to an image recognition login system and its method, in particular to a system with dynamic image recognition function and its method.

Background technique

The operating system of general electronic devices (such as: computers or mobile phones), such as: Microsoft Windows operating system, Linux, UNIX for computers, or Microsoft WindowsMobile, Android for mobile phones, etc., or application software for electronic devices, such as: MSN, When using database software, digital monitoring software, video game software, etc., the login identity and password must be verified by the authentication program before the operating system or application software can be used.

For the Microsoft window operating system, if it is necessary to input authentication data after startup, the operating system loads an authentication program, provides an authentication window screen, and displays an identity field and a password field on the window screen. The user operates an input unit, such as a keyboard, a virtual keyboard, or a touch screen, to input at least one ASCII character, such as A, B, C, . . . , 1, 2, 3 . . . Go to the identity field and the password field to form the authentication data of identity and password for authentication. After confirming that the data is correct, an internal program or other interface is further executed, and then the operating system and its application software can be used.

And currently the most commonly used security management method for online transactions is to use passwords. All transactions on ATM cards and the Internet use passwords for identity identification. However, the mechanism of using passwords has inherent loopholes. Passwords are easy to be peeped and stolen, and data is more likely to be copied by criminals. It is not easy to trace after the event.

Therefore, in recent years, many studies have been devoted to developing new identification methods to improve the security of network transactions. At present, the development of biometric systems is quite mature, and biometric mechanisms including fingerprints, voice, and even iris recognition have been developed. However, the equipment required for these biometric systems is expensive and difficult to operate. Among many biometric identification systems, facial recognition is a more feasible method. The user only needs a camera to cooperate with the system software to operate. In addition to the fact that the cost of the equipment is much lower than fingerprint, voice, and iris recognition methods, etc., It is also much simpler in operation.

However, the aforementioned face recognition is only used for image comparison on a single still face frame. People who are interested can often use pictures or photos to fool the face recognition system, which is a big challenge for the face recognition system. Therefore, it becomes a very important problem to develop a facial image recognition system that can overcome the above-mentioned shortcomings.

Contents of the invention

In view of the above-mentioned problems in the prior art, the object of the present invention is to provide an image recognition system and its method, so as to solve the problem that a single screen login is easily forged.

In order to achieve the above object of the present invention, the present invention provides an image recognition system, the image recognition system includes a storage module, a dynamic image acquisition module and a processing module, the storage module is used to store several first dynamic images in advance, The dynamic image acquisition module is used to acquire a second dynamic image when the user wants to log in. The processing module is connected to the storage module and the dynamic image acquisition module. The processing module is used to receive the second dynamic image acquired by the dynamic image acquisition module and read The first dynamic image of the storage module is compared with the second dynamic image and the first dynamic image to obtain a comparison result, and the processing module provides the next action according to the comparison result.

Wherein, the first dynamic image is composed of several continuous still pictures, which can be pre-acquired and stored in the storage module through the dynamic acquisition module, and the second dynamic picture is composed of several continuous still pictures.

Wherein, the first dynamic image is one of facial movements, head movements and body movements, and the second dynamic image is correspondingly one of face, head and body movements.

Wherein, the processing module includes a comparison module, which compares the continuous still pictures in the first dynamic image and the second dynamic picture, extracts the still pictures according to the time section, and then makes a feature comparison of the still pictures , the feature comparison method can be one of other image comparison methods such as a gray scale comparison method, a feature value comparison method, and a region reduction comparison method (Region Of Interest, ROI), or any two or two more than one.

Wherein, the storage module has a relational database for storing the first dynamic image.

Wherein, the dynamic image acquisition module is a camera, provided to acquire the second dynamic image.

Wherein, the processing module further has a display module, and the display module outputs a command message, which requires the user to perform actions required for login, such as specifying head movements or facial movements, etc., and the display module can further display the first Dynamic image output for user reference.

In order to achieve the above object of the present invention, the present invention also provides an image recognition login method, which is suitable for verifying the user's login program. After the dynamic image acquisition module is used to pre-establish several first dynamic images in the storage module, when the user When it is desired to perform a login action, the dynamic image acquisition module acquires the second dynamic image of the user, and transmits the second dynamic image to the processing module, and the comparison module of the processing module uses the continuous stillness of the first dynamic image and the second dynamic image The features of the images are compared to obtain a comparison result, and the processing module performs the next step according to the comparison result.

Among them, when the comparison result is consistent, the next action is to execute the login action; otherwise, when the comparison result is inconsistent, the next action is an optional operation behavior, such as providing another login method, etc., or once again Execute a verification procedure, or perform a shutdown action.

As mentioned above, according to the image recognition login system and method thereof of the present invention, it may have one or more of the following advantages:

(1) It can solve the situation that when only a single face recognition screen is used, the intruder can log in only with the photo of the original user.

(2) Using various angles of the face or head to achieve features that are difficult to counterfeit, greatly enhancing the security when logging in.

Description of drawings

Fig. 1 is the structural diagram of image recognition login system of the present invention;

Fig. 2 is the flowchart of the image recognition login method of the present invention;

Fig. 3 is a schematic diagram of operating the image recognition login system according to the first embodiment of the present invention;

Fig. 4 is a schematic diagram of operating image recognition and comparison in the first embodiment of the present invention;

5 is a schematic diagram of the second embodiment of the present invention to operate the image recognition login system;

Fig. 6 is a schematic diagram of operating image recognition and comparison in the second embodiment of the present invention;

7 is a schematic diagram of the third embodiment of the present invention to operate the image recognition login system;

FIG. 8 is a schematic diagram of operating image recognition and comparison in the third embodiment of the present invention.

Explanation of main reference signs

10: dynamic image acquisition module;

11: processing module;

110: comparison module;

12: storage module;

13: display module;

130: instruction information;

14: planar photo;

40: the 4th second picture of the first dynamic image;

400: the feature value of the 4th second frame of the first dynamic image;

41: the 8th second picture of the first dynamic image;

410: the feature value of the 8th second frame of the first dynamic image;

42: the 12th second picture of the first dynamic image;

420: Feature value of the 12th second frame of the first dynamic image

43: the 4th second picture of the second dynamic image;

430: the feature value of the 4th second frame of the second dynamic image;

44: the 8th second picture of the second dynamic image;

440: the feature value of the 8th second frame of the second dynamic image;

45: the 12th second picture of the second dynamic image;

450: the 12th second frame feature value of the second dynamic image;

46: the first eigenvalue set;

47: the second eigenvalue set;

60: the 5th second picture of the first dynamic image;

600: the feature value of the 5th second frame of the first dynamic image;

61: the 10th second picture of the first dynamic image;

610: the feature value of the 10th second frame of the first dynamic image;

62: the 15th second picture of the first dynamic image;

620: the feature value of the 15th second frame of the first dynamic image

63: the 5th second picture of the second dynamic image;

630: the feature value of the 5th second frame of the second dynamic image;

64: the 10th second picture of the second dynamic image;

640: the feature value of the 10th second frame of the second dynamic image;

65: the 15th second picture of the second dynamic image;

650: the feature value of the 15th second frame of the second dynamic image;

66: the first eigenvalue set;

67: the second eigenvalue set;

80: the 6th second picture of the first dynamic image;

800: the feature value of the 6th second frame of the first dynamic image;

81: the 12th second picture of the first dynamic image;

810: the feature value of the 12th second frame of the first dynamic image;

82: the 18th second picture of the first dynamic image;

820: the feature value of the 18th second frame of the first dynamic image

83: the 24th second picture of the first dynamic image;

830: the feature value of the 24th second frame of the first dynamic image;

84: the 30th second picture of the first dynamic image;

840: the feature value of the 30th second frame of the first dynamic image;

85: the 6th second picture of the second dynamic image;

850: the feature value of the 6th second frame of the second dynamic image;

86: the 12th second picture of the second dynamic image;

860: the 12th second frame feature value of the second dynamic image;

87: the 18th second picture of the second dynamic image;

870: the feature value of the 18th second frame of the second dynamic image;

88: the 24th second picture of the second dynamic image;

880: the 24th second frame feature value of the second dynamic image;

89: the 30th second picture of the second dynamic image;

890: the feature value of the 30th second frame of the second dynamic image;

90: The first set of eigenvalues

91: a second set of eigenvalues; and

S10～S60: step flow.

Detailed ways

Embodiments of the image recognition login system and method thereof according to the present invention will be described below with reference to related drawings. For ease of understanding, the same components in the following embodiments are described with the same symbols.

Please refer to FIG. 1 , which is a structural diagram of the image recognition login system of the present invention. In the figure, its image recognition login system includes a dynamic image acquisition module 10, a processing module 11 and a storage module 12, the dynamic image acquisition module 10 is connected to the processing module 11, the processing module 11 includes a comparison module 110, and the storage module 12 It is connected to the processing module 11 , and the processing module 11 can further be connected to a display module 13 .

Please refer to FIG. 2 , which is a flow chart of the image recognition login method of the present invention. This method, applicable to the login procedure for authenticating users, includes the following steps:

S10: Create several first dynamic images in advance by the dynamic image acquisition module 10 or other sources, and the first dynamic images are one of facial movements, head movements and body movements, and store these first dynamic images in storage In module 12; in this step, other sources may be a remote image acquisition device or a remote storage device. More specifically, the remote image acquisition device or remote storage device is a memory device connected to the processing module 11 through a wide area network (WAN) or a local area network (LAN), such as: memory (Memory), hard disk (HardDisk) or server ( Server), etc., are not limited to this.

S20: The dynamic image acquisition module 10 acquires a second dynamic image, the second dynamic image is one of facial movements, head movements and body movements, and transmits the second dynamic image to the processing module 11;

S30: The comparison module 110 is included in the processing module 11, and the comparison module 110 extracts the data according to the time zone from the first dynamic image stored in the storage module 12 and the second dynamic image from the dynamic image acquisition module 10 in advance. continuous still pictures, and perform feature comparison based on these continuous still pictures. This feature comparison is one of image comparison methods such as gray scale comparison method, feature value comparison method and region reduction comparison method (ROI). Any two or more, after this feature comparison, the comparison result is obtained;

S40: The processing module 11 performs step S50 according to the comparison result after feature comparison, and when the comparison result is consistent, proceeds to step S50, and when not consistent, proceeds to step S60;

S50: When the processing module 11 confirms that the comparison result is consistent, proceed to the electronic system login procedure or the next step of functional operation; and

S60: When the processing module 11 confirms that the comparison result does not match, it can go to step 20, or end the login procedure, or perform a shutdown, or selectively provide other authentication and login functions.

Please refer to FIG. 3 , which is a schematic diagram of operating the image recognition login system according to the first embodiment of the present invention. In the figure, the dynamic image acquisition module 10 is a video camera arranged on the top of the notebook computer, the processing module 11 is a central processing unit (CPU) and other auxiliary operation components arranged in the notebook computer, and the storage module 12 is arranged on the notebook computer. The hard disk inside is stored in several first dynamic images in advance. The processing module 11 includes a comparison module 110. This comparison module is a pre-written image comparison program, which is loaded into the processing module when the comparison is performed. 11, the processing module is connected to the display module 13 in this embodiment, and the display module 13 is a liquid crystal display on a notebook computer.

When the user performs the login procedure, the display module 13 displays instruction information 130, which informs the user that this is to use the movement of turning the head as identification, and informs the direction of turning the head and the time of turning the head. In this embodiment, turning the head The direction is relative to the user's right side, and this action must be completed within 12 seconds.

Please refer to FIG. 4, which is a schematic diagram of image recognition and comparison in the first embodiment of the present invention, and please refer to FIG. 3 at the same time. 10 to obtain the second dynamic image, the dynamic image acquisition module 10 promptly acquires the second dynamic image of the user and transmits the second dynamic image to the comparison module 110 of the processing module 11 for comparison, and the comparison module 110 will Read several first dynamic images pre-stored in the storage module 12, and perform feature comparison with the second dynamic module acquired by the dynamic image acquisition module 10 and sent to the processing module 11. In this embodiment, the feature comparison Utilize 4 seconds as the time interval to obtain the 4th second frame 40 of the first dynamic image, the 8th second frame 41 of the first dynamic image and the 12th second frame 42 of the first dynamic image, and the 4th second frame 42 of the second dynamic image. The second frame 43, the 8th second frame 44 of the second dynamic image and the 12th second frame 45 of the second dynamic image each carry out feature comparison. In this embodiment, this feature comparison is to utilize the image feature value comparison method, This image feature value comparison method uses a specific algorithm to calculate one or several feature values, and makes a corresponding comparison.

Referring to Fig. 3 and Fig. 4, the eigenvalue 400 of the 4th second frame of the first dynamic image, the eigenvalue 410 of the 8th second frame of the first dynamic image and the eigenvalue 420 of the 12th second frame of the first dynamic image, in In the figure, the first feature value set 46 is formed, and the feature value 430 of the 4th second frame of the second dynamic image, the feature value 440 of the 8th second frame of the second dynamic image and the feature of the 12th second frame of the second dynamic image The value 450 forms the second eigenvalue set 47 in the figure. In this embodiment, the comparison result of the first eigenvalue set 46 and the second eigenvalue set 47 is consistent, and is sent to the processing module 11, then the processing module 11 Perform a login action.

Please refer to FIG. 5 , which is a schematic diagram of the operation of the image recognition login system according to the second embodiment of the present invention. In the figure, the parts that are the same as those in the previous figures will not be repeated, and only the different features will be described. When the user performs the login procedure, the display module 13 will display the instruction information 130, and the instruction information 130 will inform the user that this is a transfer using The head movement is used as identification, and the direction and time of turning the head are notified. In this embodiment, the direction of turning the head is relative to the left direction of the user, and this action must be completed within 15 seconds, but the user uses the planar photo 14 to Fake photography of the user's front face.

Please refer to FIG. 6, which is a schematic diagram of the image recognition comparison of the second embodiment of the present invention, and please refer to FIG. 5 at the same time. In the figure, the same parts as the previous figures will not be repeated, and only the different features will be described. , in this embodiment, this feature comparison uses 5 seconds as the time interval to obtain the 5th second frame 60 of the first dynamic image, the 10th second frame 61 of the first dynamic image, and the 15th second of the first dynamic image The picture 62 is compared with the 5th second picture 63 of the second dynamic image, the 10th second picture 64 of the second dynamic image and the 15th second picture 65 of the second dynamic image. In this embodiment, this feature The comparison is to use the image feature value comparison method. The image feature value comparison method uses a specific algorithm to calculate one or a plurality of feature values, and perform a corresponding comparison.

Referring to Fig. 5 and Fig. 6, the characteristic value 600 of the 5th second frame of the first dynamic image, the characteristic value 610 of the 10th second frame of the first dynamic image and the characteristic value 620 of the 15th second frame of the first dynamic image, in In the figure, the first feature value set 66 is formed, and the feature value 630 of the 5th second frame of the second dynamic image, the feature value 640 of the 10th second frame of the second dynamic image and the feature of the 15th second frame of the second dynamic image The value 650 constitutes the second feature value set 67 in the figure. In this embodiment, the second feature value set 67 is generated by using a fixed photo, so the feature values of each picture in the second dynamic picture are the same, so the second feature value set 67 is the same. The comparison result of the first feature value set 66 and the second feature value set 67 is obviously inconsistent, and is sent to the processing module 11, and the processing module 11 judges that the login fails, and then performs selective verification or repeats the verification procedure.

Please refer to FIG. 7 , which is a schematic diagram of the operation image recognition login system according to the third embodiment of the present invention. In the figure, the same parts as the previous figures will not be repeated, and only the different features will be described. The dynamic image acquisition module 10 is an external video camera connected to the desktop computer, and the processing module 11 is arranged on the desktop computer. The central processing unit (CPU) and other auxiliary operation components in the storage module 12 are hard disks arranged in the desktop computer, and the processing module is connected to the display module 13 in this embodiment, and the display module 13 is a desktop computer on the external LCD monitor.

When the user performs the login procedure, the display module 13 displays instruction information 130, which informs the user that this is to use the movement of turning the head as identification, and informs the direction of turning the head and the time of turning the head. In this embodiment, turning the head The direction is relative to the user's right side, and this action must be completed within 30 seconds.

Please refer to FIG. 8, which is a schematic diagram of image recognition comparison of the third embodiment of the present invention, and please refer to FIG. 7 at the same time. In the figure, the same parts as the previous figures will not be repeated, and only the different features will be described. Note that in this embodiment, in order to enhance the recognition success rate and security, five still images are used for identification. This feature comparison uses 6 seconds as the time interval to obtain the 6th second frame 80 of the first dynamic image, The 12th second frame 81 of the first dynamic image, the 18th second frame 82 of the first dynamic image, the 24th second frame 83 of the first dynamic image and the 30th second frame 84 of the first dynamic image, and the second dynamic image The 6th second frame 85, the 12th second frame 86 of the second moving image, the 18th second frame 87 of the second moving image, the 24th second frame 88 of the second moving image, and the 30th second frame 89 of the second moving image Perform feature comparison. In this embodiment, the feature comparison uses image feature value comparison method. The image feature value comparison method uses a specific algorithm to calculate one or several feature values, and performs corresponding comparisons respectively.

7 and 8, the feature value 800 of the 6th second frame of the first dynamic image, the feature value 810 of the 12th second frame of the first dynamic image, the feature value 820 of the 18th second frame of the first dynamic image, the feature value 820 of the 18th second frame of the first dynamic image, The feature value 830 of the 24th second frame of a dynamic image and the feature value 840 of the 30th second frame of the first dynamic image form the first feature value set 90 in the figure, and the feature value of the 6th second frame of the second dynamic image Value 850, feature value 860 of the 12th second frame of the second dynamic image, feature value 870 of the 18th second frame of the second dynamic image, feature value 880 of the 24th second frame 88 of the second dynamic image, and the second dynamic image The eigenvalues 890 of the 30th second frame form the second eigenvalue set 91 in the figure. In this embodiment, the comparison result of the first eigenvalue set 90 and the second eigenvalue set 91 is consistent and sent to the processing module 11, then the processing module 11 performs a login action.

The above description is for illustration only, not limitation. Any equivalent modification or change without departing from the spirit and scope of the present invention shall be included in the protection scope of the present invention.

Claims (12)

1. An image recognition login system, comprising: A dynamic image acquisition module, pre-establishing a first dynamic image, and acquiring a second dynamic image, the first dynamic image and the second dynamic image are respectively composed of several continuous still pictures, the continuous still pictures of the first dynamic image The picture is a continuous picture of various angles of the face, head or body movements; A storage module, used to store the first dynamic image; and a processing module, connected to the storage module and the dynamic image acquisition module and including a comparison module, the processing module is used to receive the second dynamic image, and Read the first dynamic image in the storage module, and the comparison module performs a feature comparison on the first dynamic image and each of the continuous still pictures in the second dynamic image according to the time segment to obtain a comparison For the result, the processing module provides a next action according to the comparison result. 2. The image recognition login system according to claim 1, wherein when the continuous still images of the second dynamic image correspond to the first dynamic image, it is a continuous frame of various angles of the face, head or body movements, and the When the processing module confirms that the comparison results of the first dynamic image and the second dynamic image match, the processing module performs a login procedure. 3. The image recognition login system according to claim 1, wherein the feature comparison is one, any two or more of gray scale comparison method, feature value comparison method and area reduction comparison method. 4. The image recognition login system as claimed in claim 1, wherein the dynamic image acquisition module is a camera for acquiring the first dynamic image. 5. The image recognition login system according to claim 1, wherein the processing module is connected to a display module, and the display module outputs an instruction message, and the instruction information requires the user to perform a required action of login. 6. The image recognition login system as claimed in claim 5, wherein the display module is capable of outputting the first dynamic image for user's reference. 7. A method for image recognition login, comprising the following steps: A dynamic image acquisition module is used to pre-establish several first dynamic images, and then the first dynamic images are stored in a storage module, the first dynamic images are composed of several continuous still pictures, and the first Continuous still pictures of dynamic images are continuous pictures of various angles of face, head or body movements; The dynamic image acquisition module acquires a second dynamic image and sends it to a processing module; Through a comparison module of the processing module, a feature comparison is performed on the plurality of continuous still pictures in the first dynamic image and the second dynamic image according to the time segment, and a comparison result is obtained; and by The processing module performs a next step according to the comparison result. 8. The image recognition login method according to claim 7, wherein the first dynamic image is obtained by a remote image acquisition device or a remote storage device. 9. The image recognition login method according to claim 7, wherein the processing module is connected to a display module, and the second dynamic image is captured by the dynamic image when the user accepts an instruction message displayed by the display module and generates an action The module is acquired. 10. The image recognition login method according to claim 9, wherein the instruction information is to instruct the user to perform a directional movement, and the directional movement is a head movement, a facial movement or other body movements. 11. The image recognition and login method as claimed in claim 7, wherein the feature comparison is one or both of the image comparison methods of a grayscale comparison method, a feature value comparison method and an area reduction comparison method. or two or more. 12. The image recognition login method according to claim 7, wherein when the comparison result is consistent, the next step is to execute the login action, and when the comparison result is inconsistent, the next step is an optional operation behavior , or execute the verification program again, or execute the shutdown action.