A face optimization recognition method and terminal

Abstract

The present invention discloses a face optimization recognition method and terminal, and the method includes: S1, obtaining face data to be recognized, performing similarity comparison between the face data to be recognized and the stored user face data, obtaining a first similarity, if the first similarity exceeds the recognition threshold, the face recognition is successful, otherwise executing step S2; S2, judging whether the first similarity exceeds the approximate threshold, if so, calling the user face data for three-dimensional reconstruction into a user face three-dimensional model; S3, obtaining the first face orientation, first environmental information and first occluding object information corresponding to the face data to be recognized, so as to obtain a simulated current user face three-dimensional model; S4, performing similarity comparison between the face data to be recognized and the current user face three-dimensional model, obtaining a second similarity, if the second similarity exceeds the recognition threshold, the face recognition is successful, otherwise the face recognition fails. The present invention can improve the flexibility and accuracy of face recognition.

Description

Technical Field

[0001] The present invention relates to the technical field of face recognition, and in particular to a face optimization recognition method and a terminal. Background Art

[0002] Face recognition is a biometric technology that identifies people based on their facial features. It uses a camera or camcorder to collect images or video streams containing faces, automatically detects and tracks faces in images, and then performs facial recognition on the detected faces. It is also commonly called portrait recognition or facial recognition.

[0003] Compared with fingerprint recognition and iris recognition, face recognition technology is more natural and safer. The main reason for its safety is that it is more difficult to be deceived by disguise. However, because the difference between faces of different individuals is small, the stability of faces is poor, and face data is relatively complex, and the two important standards of biometric recognition are speed and accuracy, face recognition is also a relatively difficult direction in the field of biometric recognition.

[0004] With the development of face recognition technology, face recognition is now widely used in people's daily life, including unlocking mobile phones, door locks, subway entry and exit, face payment, etc. However, the current face recognition technology still relies on good collection conditions, user cooperation and no changes in the user's face. If the user performs actions that affect facial features such as haircuts, makeup, and wearing masks, the accuracy of face recognition will drop sharply. In severe cases, face recognition may not be completed, resulting in the need to enter passwords or press fingerprints and other biometrics for assistance, thus restricting face recognition technology. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a face optimization recognition method and a terminal to improve the flexibility and accuracy of face recognition.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0007] A face optimization recognition method comprises the steps of:

[0008] S1. Acquire face data to be identified, compare the face data to be identified with the stored user face data to obtain a first similarity. If the first similarity exceeds a recognition threshold, face recognition is successful. Otherwise, execute step S2.

[0009] S2. Determine whether the first similarity exceeds an approximate threshold, and if so, call the user face data to perform three-dimensional reconstruction into a three-dimensional face model of the user;

[0010] S3, obtaining a first face orientation, first environment information, and first obstruction information corresponding to the face data to be recognized, adjusting the user face 3D model according to the first face orientation, simulating the space where the user face 3D model is located according to the first environment information, adding the first obstruction information to the position of the user face 3D model corresponding to the first obstruction information and removing the face data of the obstructed area, to obtain a simulated current user face 3D model;

[0011] S4. Perform a similarity comparison between the face data to be recognized and the three-dimensional face model of the current user to obtain a second similarity. If the second similarity exceeds a recognition threshold, face recognition is successful; otherwise, face recognition fails.

[0012] In order to solve the above technical problems, another technical solution adopted by the present invention is:

[0013] A face optimization recognition terminal includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program:

[0014] S1. Acquire face data to be identified, compare the face data to be identified with the stored user face data to obtain a first similarity. If the first similarity exceeds a recognition threshold, face recognition is successful. Otherwise, execute step S2.

[0015] S2. Determine whether the first similarity exceeds an approximate threshold, and if so, call the user face data to perform three-dimensional reconstruction into a three-dimensional face model of the user;

[0016] S3, obtaining a first face orientation, first environment information, and first obstruction information corresponding to the face data to be recognized, adjusting the user face 3D model according to the first face orientation, simulating the space where the user face 3D model is located according to the first environment information, adding the first obstruction information to the position of the user face 3D model corresponding to the first obstruction information and removing the face data of the obstructed area, to obtain a simulated current user face 3D model;

[0017] S4. Perform a similarity comparison between the face data to be recognized and the three-dimensional face model of the current user to obtain a second similarity. If the second similarity exceeds a recognition threshold, face recognition is successful; otherwise, face recognition fails.

[0018] The beneficial effects of the present invention are: a face optimization recognition method and terminal, when normal face recognition fails, considering that the user may be affected by environmental factors such as light, weather, brightness, or the user may have performed behaviors such as haircut, makeup, wearing a mask, etc. that may obscure facial features, an approximate threshold is used as a basis for judging whether it is a possible user. On the premise that it is possible to be a user, a three-dimensional simulation is performed on the user's current face orientation, environmental information, and obstruction information, and then further face recognition is performed. This allows the user to adapt to different environments and adapt to the user's behavior of obscuring facial features on the basis of ensuring the accuracy of face recognition, so as to improve the flexibility and accuracy of face recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A schematic diagram of a flow chart of a face optimization recognition method according to an embodiment of the present invention;

[0020] Figure 2 The present invention is a schematic diagram of the structure of a face optimization recognition terminal according to an embodiment of the present invention.

[0021] Description of labels:

[0022] 1. A face optimization recognition terminal; 2. A processor; 3. A memory. DETAILED DESCRIPTION

[0023] In order to explain the technical content, achieved objectives and effects of the present invention in detail, the following is an explanation in combination with the implementation modes and the accompanying drawings.

[0024] Please refer to Figure 1 , a face optimization recognition method, comprising the steps of:

[0025] S1. Acquire face data to be identified, compare the face data to be identified with the stored user face data to obtain a first similarity. If the first similarity exceeds a recognition threshold, face recognition is successful. Otherwise, execute step S2.

[0026] S2. Determine whether the first similarity exceeds an approximate threshold, and if so, call the user face data to perform three-dimensional reconstruction into a three-dimensional face model of the user;

[0027] S3, obtaining a first face orientation, first environment information, and first obstruction information corresponding to the face data to be recognized, adjusting the user face 3D model according to the first face orientation, simulating the space where the user face 3D model is located according to the first environment information, adding the first obstruction information to the position of the user face 3D model corresponding to the first obstruction information and removing the face data of the obstructed area, to obtain a simulated current user face 3D model;

[0028] S4. Perform a similarity comparison between the face data to be recognized and the three-dimensional face model of the current user to obtain a second similarity. If the second similarity exceeds a recognition threshold, face recognition is successful; otherwise, face recognition fails.

[0029] From the above description, it can be seen that the beneficial effect of the present invention is that when normal face recognition fails, considering that the user may be affected by environmental factors such as light, weather, brightness, or the user may have performed behaviors such as haircuts, makeup, wearing masks, etc. that will obscure facial features, the approximate threshold is used as the basis for judging whether it is likely to be a user. On the premise that it is likely to be a user, the user's current face orientation, environmental information, and obstruction information are three-dimensionally simulated, and then further face recognition is performed. This allows the system to adapt to different environments and adapt to the user's behavior of obscuring facial features while ensuring the accuracy of face recognition, so as to improve the flexibility and accuracy of face recognition.

[0030] Furthermore, the step S3 further comprises the steps of:

[0031] Acquire in real time auxiliary face data whose face orientation difference with the face data to be identified exceeds an orientation threshold, acquire a second face orientation, second environment information, and second obstruction information corresponding to the auxiliary face data, adjust the user face 3D model according to the second face orientation, simulate the space where the user face 3D model is located according to the second environment information, add the first obstruction information to the user face 3D model at a position corresponding to the second obstruction information and remove the face data of the obstructed area, so as to obtain a simulated auxiliary user face 3D model;

[0032] The step S4 specifically comprises the following steps:

[0033] Comparing the to-be-recognized face data with the current user's face three-dimensional model for similarity to obtain a second similarity;

[0034] Comparing the auxiliary face data with the auxiliary user face three-dimensional model for similarity to obtain a third similarity;

[0035] If both the second similarity and the third similarity exceed the recognition threshold, the face recognition is successful, otherwise the face recognition fails.

[0036] From the above description, it can be seen that the face recognition verification performed after abandoning some features due to obstruction of facial features or distortion of some facial features due to environmental factors reduces the security of face recognition to a certain extent. Therefore, the auxiliary face data corresponding to different face orientations is restricted to obtain more face information for secondary face recognition verification, so as to improve the security performance of face recognition.

[0037] Furthermore, in the step S2, the user face three-dimensional model is three-dimensionally reconstructed and generated when the user face data is input.

[0038] From the above description, it can be seen that three-dimensional reconstruction is performed when the user's face data is entered, ensuring that the three-dimensional face model can be quickly extracted for face recognition when needed, so as to improve the face recognition speed.

[0039] Furthermore, in step S3, if the occlusion ratio of the first occlusion information exceeds the occlusion threshold, it is directly considered that the face recognition has failed.

[0040] From the above description, it can be seen that if the occlusion ratio exceeds the occlusion threshold, for example, more than 50% of the facial information, this phenomenon exceeds the normal operation behavior of the user. Therefore, no subsequent face recognition is performed and it is directly considered as face recognition, thereby preventing criminals from deceiving face recognition through this loophole and improving the security of face recognition.

[0041] Furthermore, if face recognition fails in step S4, adjustment prompt information is output to guide the user to cooperate in achieving face recognition.

[0042] From the above description, it can be seen that the adjustment prompt information is output to guide the user to cooperate in realizing face recognition, thereby improving the success rate of face recognition.

[0043] Please refer to Figure 2 , a face optimization recognition terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program:

[0044] S1. Acquire face data to be identified, compare the face data to be identified with the stored user face data to obtain a first similarity. If the first similarity exceeds a recognition threshold, face recognition is successful. Otherwise, execute step S2.

[0045] S2. Determine whether the first similarity exceeds an approximate threshold, and if so, call the user face data to perform three-dimensional reconstruction into a three-dimensional face model of the user;

[0046] S3, obtaining a first face orientation, first environment information, and first obstruction information corresponding to the face data to be recognized, adjusting the user face 3D model according to the first face orientation, simulating the space where the user face 3D model is located according to the first environment information, adding the first obstruction information to the position of the user face 3D model corresponding to the first obstruction information and removing the face data of the obstructed area, to obtain a simulated current user face 3D model;

[0047] S4. Perform a similarity comparison between the face data to be recognized and the three-dimensional face model of the current user to obtain a second similarity. If the second similarity exceeds a recognition threshold, face recognition is successful; otherwise, face recognition fails.

[0048] From the above description, it can be seen that the beneficial effect of the present invention is that when normal face recognition fails, considering that the user may be affected by environmental factors such as light, weather, brightness, or the user may have performed behaviors such as haircuts, makeup, wearing masks, etc. that will obscure facial features, the approximate threshold is used as the basis for judging whether it is likely to be a user. On the premise that it is likely to be a user, the user's current face orientation, environmental information, and obstruction information are three-dimensionally simulated, and then further face recognition is performed. This allows the system to adapt to different environments and adapt to the user's behavior of obscuring facial features while ensuring the accuracy of face recognition, so as to improve the flexibility and accuracy of face recognition.

[0049] Furthermore, the step S3 further comprises the steps of:

[0050] Acquire in real time auxiliary face data whose face orientation difference with the face data to be identified exceeds an orientation threshold, acquire a second face orientation, second environment information, and second obstruction information corresponding to the auxiliary face data, adjust the user face 3D model according to the second face orientation, simulate the space where the user face 3D model is located according to the second environment information, add the first obstruction information to the user face 3D model at a position corresponding to the second obstruction information and remove the face data of the obstructed area, so as to obtain a simulated auxiliary user face 3D model;

[0051] The step S4 specifically comprises the following steps:

[0052] Comparing the to-be-recognized face data with the current user's face three-dimensional model for similarity to obtain a second similarity;

[0053] Comparing the auxiliary face data with the auxiliary user face three-dimensional model for similarity to obtain a third similarity;

[0054] If both the second similarity and the third similarity exceed the recognition threshold, the face recognition is successful, otherwise the face recognition fails.

[0055] From the above description, it can be seen that the face recognition verification performed after abandoning some features due to obstruction of facial features or distortion of some facial features due to environmental factors reduces the security of face recognition to a certain extent. Therefore, the auxiliary face data corresponding to different face orientations is restricted to obtain more face information for secondary face recognition verification, so as to improve the security performance of face recognition.

[0056] Furthermore, in the step S2, the user face three-dimensional model is three-dimensionally reconstructed and generated when the user face data is input.

[0057] From the above description, it can be seen that three-dimensional reconstruction is performed when the user's face data is entered, ensuring that the three-dimensional face model can be quickly extracted for face recognition when needed, so as to improve the face recognition speed.

[0058] Furthermore, in step S3, if the occlusion ratio of the first occlusion information exceeds the occlusion threshold, it is directly considered that the face recognition has failed.

[0059] From the above description, it can be seen that if the occlusion ratio exceeds the occlusion threshold, for example, more than 50% of the facial information, this phenomenon exceeds the normal operation behavior of the user. Therefore, no subsequent face recognition is performed and it is directly considered as face recognition, thereby preventing criminals from deceiving face recognition through this loophole and improving the security of face recognition.

[0060] Furthermore, if face recognition fails in step S4, adjustment prompt information is output to guide the user to cooperate in achieving face recognition.

[0061] From the above description, it can be seen that the adjustment prompt information is output to guide the user to cooperate in realizing face recognition, thereby improving the success rate of face recognition.

[0062] The face optimization recognition method and terminal of the present invention can be applied to scenarios where face recognition is required, and are described below through specific implementation methods:

[0063] Please refer to Figure 1 , Embodiment 1 of the present invention is:

[0064] A face optimization recognition method comprises the steps of:

[0065] S1. Obtain the face data to be identified, compare the face data to be identified with the stored user face data to obtain a first similarity. If the first similarity exceeds the recognition threshold, the face recognition is successful, otherwise, execute step S2.

[0066] The stored user face data is the standard face data entered by the user after identity verification. In this embodiment, the face data to be identified is compared with the stored user face data for similarity using an existing mature algorithm, that is, the existing face recognition is performed normally, and if the recognition is unsuccessful, the subsequent steps are performed.

[0067] S2, determining whether the first similarity exceeds an approximate threshold, and if so, calling the user's face data to perform three-dimensional reconstruction into a three-dimensional face model of the user;

[0068] For example, if the recognition threshold is 80%, the approximate threshold can be set to 60%, that is, the approximate threshold must be smaller than the recognition threshold.

[0069] Among them, the user's face three-dimensional model is reconstructed and generated when the user's face data is entered, and can be directly called later.

[0070] S3, obtaining the first face orientation, first environment information, and first obstruction information corresponding to the face data to be recognized, adjusting the user face 3D model according to the first face orientation, simulating the space where the user face 3D model is located according to the first environment information, adding the first obstruction information to the position of the user face 3D model corresponding to the first obstruction information and removing the face data of the obstructed area, to obtain the simulated current user face 3D model;

[0071] Acquire in real time auxiliary face data whose face orientation difference with the face data to be identified exceeds an orientation threshold, acquire a second face orientation, second environment information, and second occlusion information corresponding to the auxiliary face data, adjust the user face 3D model according to the second face orientation, simulate the space where the user face 3D model is located according to the second environment information, add the first occlusion information to the user face 3D model at a position corresponding to the second occlusion information and remove the face data of the occluded area, so as to obtain a simulated auxiliary user face 3D model;

[0072] In this embodiment, the face orientation, environmental information and obstruction information can be obtained through the collection camera, where the face orientation is the direction of the user relative to the collection camera, the environmental information includes lighting, weather, brightness, etc. that may affect the expression of facial features, and the obstruction information is pre-trained objects that may affect facial features, such as changes in hairstyle, some obvious cosmetic products, and wearing masks, etc.

[0073] Among them, the face orientation difference between the face data to be identified and the auxiliary face data exceeds the orientation threshold, which can be 30° or other angles. Generally speaking, the initial face orientation is straight, so the front face information is relatively complete. When the face orientation is different from that of the collection camera, the face features located at the edge will be further collected to obtain more facial features to make up for the facial features missing due to the environment or obstructions.

[0074] In this embodiment, if the occlusion ratio of the first occlusion information exceeds the occlusion threshold, it is directly considered that the face recognition fails.

[0075] S4. Perform a similarity comparison between the face data to be recognized and the three-dimensional face model of the current user to obtain a second similarity. If the second similarity exceeds a recognition threshold, the face recognition is successful, otherwise the face recognition fails.

[0076] In this embodiment, step S4 specifically includes the following steps:

[0077] S41, performing a similarity comparison between the face data to be recognized and the three-dimensional face model of the current user to obtain a second similarity;

[0078] S42, performing a similarity comparison between the auxiliary face data and the auxiliary user face three-dimensional model to obtain a third similarity;

[0079] S43: If both the second similarity and the third similarity exceed the recognition threshold, the face recognition is successful; otherwise, the face recognition fails.

[0080] In this embodiment, if face recognition fails, adjustment prompt information is output to guide the user to cooperate in achieving face recognition.

[0081] Please refer to Figure 2 , Embodiment 2 of the present invention is:

[0082] A face optimization recognition terminal 1 includes a memory 3, a processor 2, and a computer program stored in the memory 3 and executable on the processor 2. When the processor 2 executes the computer program, the steps of the above-mentioned embodiment 1 are implemented.

[0083] To sum up, the present invention provides an optimized face recognition method and terminal. When normal face recognition fails, an approximate threshold is used as the basis for judging whether it is likely to be a user. On the premise that it is likely to be a user, a three-dimensional simulation is performed on the user's current face orientation, environmental information, and obstruction information. Then, further secondary face recognition is performed. This allows the method and terminal to adapt to different environments and to adapt to the user's behavior of obscuring facial features while ensuring the accuracy of face recognition, so as to improve the flexibility, security, and accuracy of face recognition.

[0084] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent transformations made using the contents of the present invention's specification and drawings, or directly or indirectly applied in related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A face optimization recognition method, It is characterized in that Includes steps: S1. Acquire face data to be identified, compare the face data to be identified with the stored user face data to obtain a first similarity. If the first similarity exceeds a recognition threshold, face recognition is successful. Otherwise, execute step S2. S2. Determine whether the first similarity exceeds an approximate threshold, and if so, call the user face data to perform three-dimensional reconstruction into a three-dimensional face model of the user; S3, obtaining a first face orientation, first environment information, and first obstruction information corresponding to the face data to be recognized, adjusting the user face 3D model according to the first face orientation, simulating the space where the user face 3D model is located according to the first environment information, adding the first obstruction information to the position of the user face 3D model corresponding to the first obstruction information and removing the face data of the obstructed area, to obtain a simulated current user face 3D model; Acquire in real time auxiliary face data whose face orientation difference with the face data to be identified exceeds an orientation threshold, acquire a second face orientation, second environment information, and second obstruction information corresponding to the auxiliary face data, adjust the user face 3D model according to the second face orientation, simulate the space where the user face 3D model is located according to the second environment information, add the first obstruction information to the user face 3D model at a position corresponding to the second obstruction information and remove the face data of the obstructed area, so as to obtain a simulated auxiliary user face 3D model; S4. Perform a similarity comparison between the face data to be recognized and the three-dimensional face model of the current user to obtain a second similarity, and perform a similarity comparison between the auxiliary face data and the auxiliary user face three-dimensional model to obtain a third similarity. If both the second similarity and the third similarity exceed the recognition threshold, face recognition is successful, otherwise face recognition fails.

2. A face optimization recognition method according to claim 1, It is characterized in that In step S2, the user face three-dimensional model is three-dimensionally reconstructed and generated when the user face data is input.

3. The face optimization recognition method according to claim 1, It is characterized in that In step S3, if the occlusion ratio of the first occlusion information exceeds the occlusion threshold, it is directly considered that the face recognition has failed.

4. The face optimization recognition method according to claim 1, It is characterized in that If the face recognition fails in step S4, adjustment prompt information is output to guide the user to cooperate in realizing the face recognition.

5. A face optimization recognition terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, It is characterized in that When the processor executes the computer program, the following steps are implemented: S1. Acquire face data to be identified, compare the face data to be identified with the stored user face data to obtain a first similarity. If the first similarity exceeds a recognition threshold, face recognition is successful. Otherwise, execute step S2. S2. Determine whether the first similarity exceeds an approximate threshold, and if so, call the user face data to perform three-dimensional reconstruction into a three-dimensional face model of the user; S3, obtaining a first face orientation, first environment information, and first obstruction information corresponding to the face data to be recognized, adjusting the user face 3D model according to the first face orientation, simulating the space where the user face 3D model is located according to the first environment information, adding the first obstruction information to the position of the user face 3D model corresponding to the first obstruction information and removing the face data of the obstructed area, to obtain a simulated current user face 3D model; Acquire in real time auxiliary face data whose face orientation difference with the face data to be identified exceeds an orientation threshold, acquire a second face orientation, second environment information, and second obstruction information corresponding to the auxiliary face data, adjust the user face 3D model according to the second face orientation, simulate the space where the user face 3D model is located according to the second environment information, add the first obstruction information to the user face 3D model at a position corresponding to the second obstruction information and remove the face data of the obstructed area, so as to obtain a simulated auxiliary user face 3D model; S4. Perform a similarity comparison between the face data to be recognized and the three-dimensional face model of the current user to obtain a second similarity, and perform a similarity comparison between the auxiliary face data and the auxiliary user face three-dimensional model to obtain a third similarity. If both the second similarity and the third similarity exceed the recognition threshold, face recognition is successful, otherwise face recognition fails.

6. The face optimization recognition terminal according to claim 5, It is characterized in that In step S2, the user face three-dimensional model is three-dimensionally reconstructed and generated when the user face data is input.

7. The face optimization recognition terminal according to claim 5, It is characterized in that In step S3, if the occlusion ratio of the first occlusion information exceeds the occlusion threshold, it is directly considered that the face recognition has failed.

8. The face optimization recognition terminal according to claim 5, It is characterized in that If the face recognition fails in step S4, adjustment prompt information is output to guide the user to cooperate in realizing the face recognition.

Images