Double-view-angle face identity recognition and authentication method and system

Abstract

According to the double-view-angle face identity recognition and authentication method and system provided by the invention, the height of the camera is adjusted according to the height of the to-be-passed person, so that the face images of the to-be-passed persons with different heights can be conveniently acquired, the to-be-passed persons with diversified heights can be matched, efficient passing and non-inductive passing are realized, and the user experience is improved. And the passing experience feeling is greatly improved. After the first camera and the second camera which are arranged at an interval collect the same face image, height adjustment is carried out, and unnecessary height adjustment caused by passers-by is avoided to a certain extent. When the images collected by the camera are side face images, two images with most exposed side faces are selected to be spliced into a front face image for comparison, and it is guaranteed that passing personnel who get permission in advance pass smoothly.

Description

Technical Field

[0001] The present invention relates to the technical field of identity recognition, and in particular to a dual-viewing angle face identity recognition and authentication method and system. Background Art

[0002] Identity verification technology based on facial biometrics is widely used in scenarios that require real-name authentication, among which facial access control systems with cameras are widely used in tourist attractions, entry and exit, customs, airports, office parks, etc. System suppliers usually collect users' facial information in advance. When using the access control system, passers-by need to constantly adjust their posture so that the camera can collect facial information. The system compares the pre-collected facial information with the facial information collected on site to complete the identity verification process.

[0003] However, during the passage process, passers-by need to constantly adjust their posture to adapt to the camera. Considering the diversity of heights of passers-by, this process of constantly adjusting their posture to adapt to facial information collection makes the passage experience poor. Summary of the invention

[0004] The present invention provides a dual-viewing angle human face identification and authentication method and system, aiming to match diverse passers-by and improve the passing experience.

[0005] The technical solution of the present invention is: A dual-view face identification and authentication method, characterized by comprising: Providing a first camera and a second camera arranged at intervals; Image acquisition of the person to be passed: After the first camera and the second camera acquire the same face image, the heights of the first camera and the second camera are adjusted according to the height H of the person to be passed, and image acquisition continues; Face matching: The face image captured by at least one of the first camera and the second camera is compared with the pre-stored face image. When the captured face image matches the pre-stored face image, the person waiting to pass is allowed to pass.

[0006] In the dual-view face identification and authentication method of the present invention, the height adjustment of the first camera and the second camera according to the height H of the person to be passed includes: Obtain a distance D from a plane where the person to be passed is located to a line connecting the first camera and the second camera, wherein the plane where the person to be passed is located is parallel to the line connecting the first camera and the second camera; Calculate the height H of the person waiting to pass based on the distance D; The heights of the first camera and the second camera are adjusted according to the height H of the person to be passed.

[0007] In the dual-view face identification and authentication method of the present invention, the step of obtaining the distance D from the plane where the person to be passed is located to the line connecting the first camera and the second camera comprises: Obtain the distance L between the first camera and the second camera; Obtain an angle α1 with the first camera as the endpoint, the line connecting the first camera and the second camera as one of the sides, and the line connecting the first camera and the person to be passed as the other side; Obtain an angle α2 with the second camera as the endpoint, the line connecting the second camera and the first camera as one side, and the line connecting the second camera and the person to be passed as the other side; Obtain an angle α3 with the person waiting to pass as an endpoint, a line connecting the person waiting to pass and the first camera as one side, and a line connecting the person waiting to pass and the second camera as the other side; Distance D=L·sinα1·sinα2 / sinα3.

[0008] In the dual-view face identification and authentication method of the present invention, the angle α1, the angle α2 and the angle α3 are obtained in the following manner: A predetermined deflection angle β of the first camera is provided, wherein the predetermined deflection angle of the first camera takes the first camera as an endpoint, takes the line connecting the first camera and the second camera as one of its sides, and the other side faces the detection area for the persons to pass through, and the predetermined deflection angle β of the second camera is consistent with the predetermined deflection angle β of the first camera, wherein the predetermined deflection angle of the second camera takes the second camera as an endpoint, takes the line connecting the second camera and the first camera as one of its sides, and the other side faces the detection area for the persons to pass through; Obtaining a deviation angle θ1 of the first camera relative to a predetermined deflection angle toward a detection area for a person to pass; Obtaining a deviation angle θ2 of the second camera relative to a predetermined deflection angle toward a detection area for a person to pass; The included angle α1=β+θ1, the included angle α2=β+θ2, and the included angle α3=180°-α1-α2.

[0009] In the dual-view face identification and authentication method of the present invention, the step of calculating the height H of the person to be passed according to the distance D comprises: Providing a height h1 of the first camera or the second camera from the tread surface; The height H of the person waiting to pass is calculated based on the height h1 and the distance D.

[0010] In the dual-view face identification and authentication method of the present invention, the calculation of the height H of the person to be passed according to the height h1 and the distance D comprises: Obtain the angle θ3 between the line connecting the first camera and the person to be passed and the horizontal plane where the first camera is located, with the first camera as the endpoint, and the height H=h1+Dtanθ3; or Obtain the angle θ3 between the line connecting the second camera and the person to be passed and the horizontal plane where the second camera is located, with the second camera as the endpoint, and the height H=h1+Dtanθ3.

[0011] In the dual-viewing face identity recognition and authentication method described in the present invention, the face image used for comparison is a frontal face image.

[0012] In the dual-view face identification and authentication method of the present invention, comparing the face image captured by at least one of the first camera and the second camera with the pre-stored face image includes: When the face image captured by one of the first camera and the second camera is a frontal face image, comparing the frontal face image with the pre-stored face image; When the facial images captured by the first camera and the second camera are both side face images, the side face images captured by the first camera and the second camera are spliced ​​into a front face image, and the spliced ​​front face image is compared with the pre-stored facial image.

[0013] In the dual-view facial identity recognition and authentication method described in the present invention, when the facial images captured by the first camera and the second camera are both side face images, the image in which the side face is exposed the most in the image captured by the first camera and the image in which the side face is exposed the most in the image captured by the second camera are spliced ​​into a frontal face image.

[0014] Another technical solution of the present invention is: A dual-view face identification and authentication system, characterized in that it includes: a gate body, a first adjustment mechanism, a first camera, a second adjustment mechanism, a second camera, a blocking member and a controller, wherein the first camera is arranged on the gate body through the first adjustment mechanism, the first adjustment mechanism is used to adjust the height of the first camera, the second camera is arranged on the gate body through the second adjustment mechanism, the second adjustment mechanism is used to adjust the height of the second camera, the blocking member is movably arranged on the gate body, the blocking member is used to block people waiting to pass, the first camera, the second camera and the blocking member are all connected to the controller, and the controller is used to adjust the height of the first camera and the second camera according to the height H of the person waiting to pass when the first camera and the second camera capture the same face image, and at least compare the face image captured by one of the first camera and the second camera with the pre-stored face image, and after the captured face image matches the pre-stored face image, control the blocking member to move to allow the person waiting to pass.

[0015] In the dual-view facial identification and authentication system described in the present invention, the distance L between the first camera and the second camera ranges from 0.8m to 1.5m, the field of view angles θ of the first camera and the second camera range from 45° to 75°, and the predetermined deflection angle β of the first camera toward the person to pass and the predetermined deflection angle β of the second camera toward the person to pass range from 30° to 50°.

[0016] Compared with the prior art, the present invention has the following beneficial effects: The height adjustment is performed only after the first camera and the second camera have captured the same facial image, which can avoid the height adjustment caused by passers-by to a certain extent and reduce the energy consumption of the equipment. The height adjustment of the first camera and the second camera is performed according to the height H of the person to be passed, so as to facilitate the collection of facial images of people to be passed with different heights. It can match a variety of passers-by and achieve efficient passage. It also solves the problem in the prior art that the person to be passed needs to tiptoe or squat before his facial image can be collected, and realizes senseless passage, which greatly improves the passage experience. After the collected facial image matches the pre-stored facial image, the person to be passed is allowed to pass, which can prevent unauthorized persons from entering and achieve the purpose of security protection. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. The drawings described below are only drawings corresponding to some embodiments of the present invention.

[0018] Figure 1 A schematic diagram of a dual-view facial identity recognition and authentication system provided by an embodiment.

[0019] Figure 2 Schematic diagram of the positions of the first camera, the second camera and the person waiting to pass in one embodiment.

[0020] Figure 3 Schematic diagram of the workflow of a dual-view face identification and authentication system in one embodiment.

[0021] Figure 4 Schematic diagram of a first process of a dual-view face identity recognition and authentication method in one embodiment.

[0022] Figure 5 Schematic diagram of the second process of the dual-view face identity recognition and authentication method in one embodiment.

[0023] Figure 6 Schematic diagram of the third process of the dual-view face identity recognition and authentication method in one embodiment.

[0024] Figure 7 Schematic diagram of the fourth process of the dual-view face identity recognition and authentication method in one embodiment.

[0025] Figure 8 Schematic diagram of the fifth process of the dual-view facial identity recognition and authentication method in one embodiment.

[0026] Fig. 9 Schematic diagram of the positions of the first camera, the ground and the people waiting to pass in one embodiment.

[0027] Among them, 10, dual-view facial identity recognition and authentication system, 110, gate body, 121, first adjustment mechanism, 122, first adjustment mechanism, 131, first camera, 132, second camera, 140, blocking member.

[0028] In the figures, structurally similar elements are denoted by the same reference numerals. DETAILED DESCRIPTION

[0029] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0030] The terms "first" and "second" in the present invention are used for descriptive purposes only and should not be understood as indicating or implying relative importance, and should not be construed as limiting the order of precedence. The terms "parallel" and "perpendicular" in the present invention are not limited to absolute parallelism or perpendicularity, and errors are allowed.

[0031] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0032] Facial identity recognition and authentication systems are often used in the field of access control, especially in foreign affairs, customs, rail transit, office buildings, or some occasions that require advance reservations. Taking office buildings as an example, users such as property departments usually pre-collect and store facial images of employees working in the building. When employees working in the building enter the building, the facial identity recognition and authentication system collects the employee's facial image and compares the collected facial image with the pre-stored facial image. When the two match, the employee is allowed to enter. However, there are inevitably differences in height among employees in a building. Taller employees need to squat to be collected, and shorter employees need to stand on tiptoe to be collected, resulting in an inefficient overall passage process.

[0033] The following is an embodiment of a dual-view face identification and authentication system provided by the present invention that can solve the above technical problems.

[0034] See also Figure 1 , Figure 1Schematic diagram of a dual-view facial identity recognition and authentication system provided by an embodiment of the present invention. The dual-view facial identity recognition and authentication system 10 includes a gate body 110, a first adjustment mechanism 121, a first camera 131, a second adjustment mechanism 122, a second camera 132, a blocking member 140 and a controller (not shown). The first camera 131 is arranged on the gate body 110 through the first adjustment mechanism 121, and the first adjustment mechanism 121 is used to adjust the height of the first camera 131; the second camera 132 is arranged on the gate body 110 through the second adjustment mechanism 122, and the second adjustment mechanism 122 is used to adjust the height of the second camera 132. The blocking member 140 is movably arranged on the gate body 110, and the blocking member 140 is used to block people waiting to pass. The first camera 131, the second camera 132 and the blocking member 140 are all connected to the controller. The controller is used to adjust the heights of the first camera 131 and the second camera 132 according to the height H of the person to pass when the first camera 131 and the second camera 132 capture the same facial image, and to compare the facial image captured by at least one of the first camera 131 and the second camera 132 with the pre-stored facial image. After the captured facial image matches the pre-stored facial image, the blocking member 140 is controlled to move to allow the person to pass.

[0035] The first camera 131 and the second camera 132 will adjust the height only after they have captured the same facial image, which can avoid the height adjustment caused by passers-by to a certain extent and reduce the energy consumption of the equipment. The height of the first camera 131 and the second camera 132 are adjusted according to the height H of the person to be passed, so as to facilitate the collection of facial images of people to be passed with different heights, and solve the problem in the prior art that the person to be passed needs to tiptoe or squat before his facial image can be collected, so as to realize senseless passage and greatly improve the passage experience. When the collected facial image matches the pre-stored facial image, the person to be passed is allowed to pass, and the entry of unauthorized persons is avoided, so as to achieve the purpose of security protection.

[0036] The gate body 110 mainly supports and provides installation positions for other components. In this embodiment, the gate body 110 includes a first sub-gate and a second sub-gate arranged at intervals, and a passage for people to pass through is formed between the first sub-gate and the second sub-gate. Of course, in other embodiments, the gate body 110 can be matched with a wall to form a passage, in which case only the first sub-gate or the second sub-gate is retained.

[0037] The first adjustment mechanism 121 is arranged on the first sub-gate, and the second adjustment mechanism 122 is arranged on the second sub-gate, so as to facilitate the collection of facial images of people waiting to pass through the passage. The first adjustment mechanism 121 and the second adjustment mechanism 122 can be of the same model or different models, as long as the height adjustment of the camera can be achieved. The adjustment method of the first adjustment mechanism 121 and the second adjustment mechanism 122 is the existing technology and will not be described in detail here.

[0038] The blocking member 140 mainly plays a blocking role to prevent unauthorized persons from illegally breaking in. The blocking member 140 can be a blocking rod or a blocking door, etc. In this embodiment, the blocking member 140 includes a first gate rotatably mounted on the first sub-gate and a second gate rotatably mounted on the second sub-gate.

[0039] Please also read Figure 2 , Figure 2 The figure is a schematic diagram of the positions of the first camera 131, the second camera 132 and the person to be passed in one embodiment. In the figure, the grid line area is both in the field of view of the first camera 131 and in the field of view of the second camera 132. The person to be passed is located in the grid line area, and the detection area of ​​the person to be passed includes the grid line area. The distance L between the first camera 131 and the second camera 132 ranges from 0.8m to 1.5m. When the distance L≧0.8m, the person to be passed feels more comfortable passing between the first camera 131 and the second camera 132, and it is convenient for the person to be passed to carry luggage through; when the distance L≦1.5m, the space occupied by the dual-view facial identity recognition and authentication system 10 is smaller, which is conducive to saving space.

[0040] A camera of a specific model usually has a specific field of view angle, and the size of the field of view angle determines the field of view of the camera. The larger the field of view angle, the larger the field of view. Optionally, the field of view angle θ of the first camera 131 and the second camera 132 are both in the range of 45°~75°, such as 50° or 60 degrees. The field of view angle θ of the first camera 131 and the second camera 132 is consistent, which is conducive to the unified setting of the camera. When the field of view angle θ≧75°, it is easy to capture irrelevant personnel in front of the channel; when the field of view angle θ≦45°, it is easy to fail to capture or miss the collection of people waiting to pass. Of course, in other embodiments, the field of view angle θ of one of the first camera 131 and the second camera 132 can be set in the range of 45°~75° as needed.

[0041] In one embodiment, the facial image of the person to be passed begins to be collected when the person to be passed is slightly away from the gate body 110, so that a relatively complete facial image can be collected before the person to be passed approaches the gate body 110. For this purpose, a predetermined deflection angle is set for the camera. Optionally, the value range of the predetermined deflection angle β of the first camera 131 toward the person to be passed and the value range of the predetermined deflection angle β of the second camera 132 toward the person to be passed are both 30°~50°, which is conducive to collecting images of the person to be passed within an appropriate range from the gate body 110, and also conducive to collecting images of the person to be passed within an appropriate range from the connection line of the first camera 131 and the second camera 132. The predetermined deflection angles β of the first camera 131 and the second camera 132 are consistent, which is conducive to unified camera settings. In some cases, according to the different needs of the user, images of people waiting to pass within the range of 1m~2m from the line connecting the first camera 131 and the second camera 132 can be collected, or images of people waiting to pass within the range of 0.5m~1.2m from the line connecting the first camera 131 and the second camera 132 can be collected, and the value of β can be flexibly adjusted accordingly.

[0042] For the convenience of description, the distance from the plane where the person to be passed is located to the line connecting the first camera 131 and the second camera 132 is represented by the distance D, wherein the plane where the person to be passed is parallel to the line connecting the first camera 131 and the second camera 132. Specifically, the plane where the person to be passed is located can be the plane where the nose tip of the person to be passed is located, or the plane where the eyes of the person to be passed are located, or the plane where the ears of the person to be passed are located. It can be flexibly selected according to the actual situation and is not limited here. The height H of the person to be passed can be calculated based on the distance D, and then the height of the first camera 131 and the second camera 132 can be adjusted based on the height H of the person to be passed.

[0043] Optionally, the distance D can be calculated in the following manner. Obtain the angle α1 with the first camera 131 as the endpoint, the line connecting the first camera 131 and the second camera 132 as one of the sides, and the line connecting the first camera 131 and the person to be passed as the other side; obtain the angle α2 with the second camera 132 as the endpoint, the line connecting the second camera 132 and the first camera 131 as one of the sides, and the line connecting the second camera 132 and the person to be passed as the other side; obtain the angle α3 with the person to be passed as the endpoint, the line connecting the person to be passed and the first camera 131 as one of the sides, and the line connecting the person to be passed and the second camera 132 as the other side; distance D=L·sinα1·sinα2 / sinα3. In fact, the first camera 131, the second camera 132 and the person waiting to pass are three endpoints, which can form a triangle, so the angle α1+angle α2+angle α3=180°, so after obtaining the angles of two of the angles, the angles of the remaining angles can be calculated. It should be noted that the distance D is not limited to this. In other embodiments, it can also be obtained using an infrared rangefinder or a range sensor.

[0044] Specifically, angle α1, angle α2 and angle α3 can be obtained in the following manner. Take the example that the predetermined deflection angle of the first camera 131 is consistent with the predetermined deflection angle of the second camera 132, and both are angle β. When the camera collects facial images, it will further deviate on the basis of the predetermined deflection angle to better collect facial images of people waiting to pass in front of the passage. The offset angle of the first camera 131 is recorded as angle θ1, and the offset angle of the second camera 132 is recorded as angle θ2, then angle α1=β+θ1, angle α2=β+θ2, angle α3=180°-α1-α2. When the camera is offset, angle θ1 and angle θ2 can be recorded by the camera or controller, and the acquisition difficulty is low.

[0045] It can be understood that the distance D is only one of the bases for calculating the height H. In this embodiment, the height H is also calculated based on the height h1 of the first camera 131 or the second camera 132 from the footrest surface. The footrest surface here refers to the plane stepped on by the feet of the person to be passed, such as the floor plate of a specific floor. When the gate is directly set on the ground, the footrest surface is the ground. The height h1 of the first camera 131 or the second camera 132 from the footrest surface is a fixed value after the dual-view facial identity recognition and authentication system 10 is installed, and it can be recorded and reused after only obtaining it once.

[0046] Furthermore, the angle θ3 between the line connecting the first camera 131 and the person to be passed and the horizontal plane where the first camera 131 is located is obtained, and the height H=h1+Dtanθ3; or the angle θ3 between the line connecting the second camera 132 and the person to be passed and the horizontal plane where the second camera is located is obtained, and the height H=h1+Dtanθ3. Among them, the angle θ3 can be calculated by the ratio of the pixels between the face position point and the horizontal center line of the image to the vertical pixels of the image. The face position point can be flexibly selected, such as the position of the nose tip, the eyeball position, the eyebrow position, etc. Of course, in other embodiments, the angle θ3 can also be measured by an angle measuring instrument, etc.

[0047] Considering that the property department is accustomed to pre-collecting facial information by collecting ID photos of employees, and the facial image in the ID photo is usually a frontal face image, in this embodiment, the facial image used for comparison is a frontal face image. It should be noted that the facial image used to calculate the height H can be a side face image.

[0048] When the face image captured by one of the first camera 131 and the second camera 132 is a frontal face image, any frontal face image is used to compare with the pre-stored face image. However, when the face images captured by the first camera 131 and the second camera 132 are both side face images, the side face images captured by the first camera 131 and the second camera 132 can be spliced ​​into a frontal face image, and the spliced ​​frontal face image is used to compare with the pre-stored face image. The image used for splicing can be the image with the most side face exposed in the images captured by the two cameras, so as to increase the probability of being able to splice a complete frontal face image.

[0049] Please also read Figure 3 , Figure 3 Schematic diagram of the workflow of a dual-view face identification and authentication system in one embodiment. In this embodiment, the distance between the first camera 131 and the second camera 132 is the distance L, and the height of the first camera 131 or the second camera 132 from the footrest surface is the height h1. When a face enters the acquisition range of the first camera 131 or the second camera 132, the first camera 131 and the second camera 132 perform image acquisition at a specific frequency. After the first camera 131 and the second camera 132 acquire the same face image, the controller determines that the person waiting to pass is ready to enter the passage.

[0050] Height H = height of camera + height difference between the person to be passed and camera = h1 + Dtanθ3, where angle θ3 is the angle between the line connecting the camera and the person to be passed and the horizontal plane where the camera is located, with the camera as the endpoint. The camera here can be either the first camera or the second camera. The dual-view facial identity recognition and authentication system 10 obtains angle α1, angle α2 and angle α3, thereby obtaining the distance D = L·sinα1·sinα2 / sinα3 from the plane where the person to be passed is located to the line connecting the first camera 131 and the second camera 132, then the height of the person to be passed H = h1 + Dtanθ3. The controller adjusts the height of the camera through the adjustment mechanism according to the height H of the person to be passed, so as to capture the frontal face image of the person to be passed without the person to be passed constantly adjusting his or her posture. If both the first camera 131 and the second camera 132 fail to capture a frontal face image, the side face image with the most exposed side face from the images captured by the first camera 131 and the second camera 132 is selected, and these two side face images are used to stitch together a complete frontal face image.

[0051] The front face image of the person to be passed is compared with the pre-stored face image. When the front face image matches the pre-stored face image, the controller controls the first gate and the second gate to open, and the person to be passed can pass through the passage. When the front face image does not match the pre-stored face image, the person to be passed is a non-permitted person, and the first gate and the second gate both remain in their normally closed state.

[0052] In order to solve the technical problem of poor traffic experience in the prior art, the present invention also provides an embodiment of a dual-view facial identity recognition and authentication method.

[0053] See also Figure 4 , Figure 4 Schematic diagram of a first process of a dual-view face identity recognition and authentication method in one embodiment.

[0054] Step 210, providing a first camera and a second camera that are spaced apart.

[0055] Step 220, image acquisition of the person to be passed: after the first camera and the second camera acquire the same facial image, the heights of the first camera and the second camera are adjusted according to the height H of the person to be passed, and image acquisition continues.

[0056] Step 230, face comparison: compare the face image captured by at least one of the first camera and the second camera with the pre-stored face image, and when the captured face image matches the pre-stored face image, allow the person to pass.

[0057] The height adjustment is performed only after the first camera and the second camera have captured the same facial image, which can avoid the height adjustment caused by passers-by to a certain extent and reduce the energy consumption of the equipment. The height adjustment of the first camera and the second camera is performed according to the height H of the person to be passed, so as to facilitate the collection of facial images of people to be passed with different heights. It can match a variety of passers-by and achieve efficient passage. It also solves the problem in the prior art that the person to be passed needs to tiptoe or squat before his facial image can be collected, and realizes senseless passage, which greatly improves the passage experience. After the collected facial image matches the pre-stored facial image, the person to be passed is allowed to pass, which can prevent unauthorized persons from entering and achieve the purpose of security protection.

[0058] Taking the application of this method in an office environment as an example, the height of an employee who passes between the first camera and the second camera first and the adjacent employee who passes later may be equal in time, so the height adjustment distance of the first camera can be zero, and the height adjustment distance of the second camera can also be zero.

[0059] When the first camera or the second camera collects a face image, any currently available target detection algorithm based on artificial intelligence or template matching can be used, such as YOLO, RetinaNet or Faster RCNN, etc. When determining whether the collected face image matches the pre-stored face image, a target matching algorithm such as an artificial intelligence algorithm or a feature point matching algorithm can be used.

[0060] See also Figure 5 , Figure 5 It is a second flow diagram of a dual-view facial identity recognition and authentication method in one embodiment. The height adjustment of the first camera and the second camera according to the height H of the person to be passed includes: step 221, obtaining the distance D from the plane where the person to be passed is located to the line connecting the first camera and the second camera, and the plane where the person to be passed is parallel to the line connecting the first camera and the second camera; step 222, calculating the height H of the person to be passed according to the distance D; and step 223, adjusting the height of the first camera and the second camera according to the height H of the person to be passed. Compared with directly obtaining the height H, first obtaining the distance D and then indirectly obtaining the height H is less difficult and more feasible. Although the person to be passed has a certain volume, the plane where the nose tip is located, the plane where the eyeballs are located, the plane where the eyebrows are located, etc. can be selected as the plane where the person to be passed is located.

[0061] See also Figure 6 , Figure 6The third flow diagram of the dual-view face identification authentication method in one embodiment. Obtaining the distance D from the plane where the person to be passed is located to the line connecting the first camera and the second camera includes: step 224, obtaining the distance L between the first camera and the second camera; step 225, obtaining the angle α1 with the first camera as the endpoint, the line connecting the first camera and the second camera as one of the sides, and the line connecting the first camera and the person to be passed as the other side; obtaining the angle α2 with the second camera as the endpoint, the line connecting the second camera and the first camera as one of the sides, and the line connecting the second camera and the person to be passed as the other side; obtaining the angle α3 with the person to be passed as the endpoint, the line connecting the person to be passed and the first camera as one of the sides, and the line connecting the person to be passed and the second camera as the other side; distance D=L·sinα1·sinα2 / sinα3. The distance L is a fixed distance, which can be reused after one measurement, increasing the convenience of calculating the height H. Optionally, the distance L has a value range of 0.8 m to 1.5 m, such as 0.8 m, 1 m or 1.5 m.

[0062] See also Figure 7 , Figure 7 The fourth flow diagram of the dual-view face identification authentication method in one embodiment. Step 241, provide a predetermined deflection angle β of the first camera, the predetermined deflection angle of the first camera takes the first camera as an endpoint, takes the line connecting the first camera and the second camera as one of the edges, and the other edge faces the detection area of ​​the person to be passed, the predetermined deflection angle β of the second camera is consistent with the predetermined deflection angle β of the first camera, the predetermined deflection angle of the second camera takes the second camera as an endpoint, takes the line connecting the second camera and the first camera as one of the edges, and the other edge faces the detection area of ​​the person to be passed; Step 242, obtain the offset angle θ1 of the first camera relative to the predetermined deflection angle toward the detection area of ​​the person to be passed; obtain the offset angle θ2 of the second camera relative to the predetermined deflection angle toward the detection area of ​​the person to be passed; angle α1=β+θ1, angle α2=β+θ2, angle α3=180°-α1-α2. The predetermined deflection angle β is pre-set and easy to obtain. Optionally, the predetermined deflection angle β is in the range of 30° to 50°, such as 30°, 45° or 50°. It is also relatively easy to obtain the offset angle θ1 and the offset angle θ2, and the overall acquisition difficulty of the angle α1, the angle α2 and the angle α3 is relatively low, and the feasibility of the solution is high. The offset angle θ1 and the offset angle θ2 can be directly derived from the camera without the need for other external components, and the structure is simple.

[0063] See also Figure 8 , Figure 82 is a fifth flow diagram of a dual-view face identification and authentication method in an embodiment. Step 222 includes: Step 2221, providing a height h1 between the first camera or the second camera and the footrest surface; Step 2222, calculating the height H of the person to be passed according to the height h1 and the distance D. The height h1 is a fixed distance and can be reused after being measured once, which increases the convenience of calculating the height H.

[0064] See also Fig. 9 , Fig. 9 Schematic diagram of an image captured by the first camera in an embodiment. Step 2222 is specifically: obtain the angle θ3 between the line connecting the first camera and the person to be passed and the horizontal plane where the first camera is located, with the first camera as the endpoint, and the height H=h1+Dtanθ3; or obtain the angle θ3 between the line connecting the second camera and the person to be passed and the horizontal plane where the second camera is located, with the second camera as the endpoint, and the height H=h1+Dtanθ3. It is easier to obtain the angle θ3, which reduces the difficulty of calculating the height H.

[0065] Continuing with the example of an office environment, height H = camera height + height difference between the person to be passed and the camera = h1 + Dtanθ3, where the angle θ3 can be calculated by the ratio of the pixels between the face position point and the horizontal center line of the image to the vertical pixels of the image, where the vertical direction is the same as the height direction of the person to be passed in the image, and the horizontal direction is perpendicular to the vertical direction. The face position point can be flexibly selected, such as the position of the nose tip, the position of the eyeball, the position of the center of the eyebrows, etc. Of course, in other embodiments, the angle θ3 can also be measured by an angle measuring instrument, etc.

[0066] In this embodiment, the face image used for comparison is a frontal face image. The property department is accustomed to obtaining the user's face information by collecting the user's frontal face photos in advance, and the frontal face image collected for comparison has higher comparability.

[0067] In this embodiment, when any face image captured by the first camera matches a pre-stored face image, the person waiting to pass is allowed to pass. In this way, the probability of successful passage can be increased. Similarly, when any face image captured by the second camera matches a pre-stored face image, the person waiting to pass can also be allowed to pass.

[0068] In the face comparison process, at least one of the first camera and the second camera is used to compare the face image with the pre-stored face image, including two situations. In the first situation, when the face image captured by one of the first camera and the second camera is a frontal face image, the frontal face image is used to compare with the pre-stored face image; in the second situation, when the face images captured by the first camera and the second camera are both side face images, the side face images captured by the first camera and the second camera are spliced ​​into a frontal face image, and the spliced ​​frontal face image is used to compare with the pre-stored face image. Splicing the side face images into the front face image increases the success rate of face comparison, and there is no need for the user to turn his head according to the prompt, which is conducive to further realizing senseless passage.

[0069] The technology for stitching two side face images into a front face image is to use seamless fusion technology, which can generally include image feature point detection, feature point matching and image fusion. Any existing feature point detection, feature point matching and image fusion methods can be used. For example, the feature point detection method can be SIFT (Scale-invariant feature transform) or SURF (Speeded-Up Robust Features), and the feature point matching and image fusion method can be RANSAC (Random Sample Consensus).

[0070] When walking, the person waiting to pass may shake his head to a certain extent, resulting in different degrees of exposure of the side face in the image captured by the camera. After detecting the person waiting to pass, the camera can be controlled to continuously capture facial images at a specific frequency, so in situation 2, the camera can capture multiple side face images. The image with the most exposed side face in the image captured by the first camera and the image with the most exposed side face in the image captured by the second camera can be spliced ​​into a front face image, making it more likely to splice a complete front face image, ensuring smooth passage of people who have obtained permission in advance.

[0071] The dual-view face identification and authentication method in the above embodiment of the present invention has at least the following advantages: The facial images of the people who may pass through are collected in advance and stored. The people who may pass through can approach the plane where the line connecting the first camera and the second camera is located along the perpendicular midline of the line connecting the first camera and the second camera. When the first camera and the second camera capture the same facial image, it can be determined that the people who may pass through are ready to pass through.

[0072] With the first camera, the second camera and the person waiting to pass as endpoints, a triangle can be connected. The line connecting the first camera and the second camera is one side of the triangle. Let the distance between the first camera and the second camera be the distance L. The three internal angles of the triangle are angle α1, angle α2 and angle α3. Among them, angle α1 takes the first camera as the endpoint, the line connecting the first camera and the second camera as one side, and the line connecting the first camera and the person waiting to pass as the other side; angle α2 takes the second camera as the endpoint, the line connecting the second camera and the first camera as one side, and the line connecting the second camera and the person waiting to pass as the other side; angle α3 takes the person waiting to pass as the endpoint, the line connecting the person waiting to pass and the first camera as one side, and the line connecting the person waiting to pass and the second camera as the other side. It can be concluded that the distance D from the plane where the person waiting to pass is located to the line connecting the first camera and the second camera is L·sinα1·sinα2 / sinα3.

[0073] Height H = camera height + height difference between the person to be passed and the camera = h1 + Dtanθ3, where angle θ3 is the angle between the camera as the endpoint, the line connecting the camera and the person to be passed and the horizontal plane where the camera is located. The camera here can be either the first camera or the second camera; h1 is the height of the camera that takes the image from the footrest surface. After the camera is installed, the height h1 can be measured. The height of the first camera and the second camera are adjusted according to the height H of the person to be passed, so as to facilitate the collection of facial images of people to be passed with different heights H, match the diverse passing people, achieve efficient and seamless passage, and enhance the passing experience.

[0074] A frontal face image captured by the first camera or the second camera is compared with a pre-stored face image. When the captured face image matches the pre-stored face image, the person waiting to pass is allowed to pass between the first camera and the second camera. When the face images captured by the first camera and the second camera are both side face images, considering that the person waiting to pass may turn his head subconsciously during the process of walking, the image with the most side face exposed in the image captured by the first camera and the image with the most side face exposed in the image captured by the second camera are spliced ​​into a frontal face image, and the frontal face image is used for face comparison.

[0075] To sum up, although the present invention has been disclosed as above with some embodiments, the protection scope of the present invention is not limited thereto, and any technician familiar with the technical field can make equivalent replacements or changes according to the concept of the technical solution of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A dual-view face identification and authentication method, characterized in that: include: Providing a first camera and a second camera arranged at intervals; Image acquisition of the person to be passed: After the first camera and the second camera acquire the same face image, the heights of the first camera and the second camera are adjusted according to the height H of the person to be passed, and image acquisition continues; Face matching: The face image captured by at least one of the first camera and the second camera is compared with the pre-stored face image. When the captured face image matches the pre-stored face image, the person waiting to pass is allowed to pass.

2. The dual-view face identification and authentication method according to claim 1, characterized in that: The step of adjusting the height of the first camera and the second camera according to the height H of the person to be passed includes: Obtain a distance D from a plane where the person to be passed is located to a line connecting the first camera and the second camera, wherein the plane where the person to be passed is located is parallel to the line connecting the first camera and the second camera; Calculate the height H of the person waiting to pass according to the distance D; The heights of the first camera and the second camera are adjusted according to the height H of the person to be passed.

3. The dual-view face identification and authentication method according to claim 2, characterized in that: The step of obtaining the distance D from the plane where the person to pass is located to the line connecting the first camera and the second camera includes: Obtain the distance L between the first camera and the second camera; Obtain an angle α1 with the first camera as the endpoint, the line connecting the first camera and the second camera as one of the sides, and the line connecting the first camera and the person to be passed as the other side; Obtain an angle α2 with the second camera as the endpoint, the line connecting the second camera and the first camera as one side, and the line connecting the second camera and the person to be passed as the other side; Obtain an angle α3 with the person waiting to pass as an endpoint, a line connecting the person waiting to pass and the first camera as one side, and a line connecting the person waiting to pass and the second camera as the other side; Distance D=L·sinα1·sinα2 / sinα3.

4. The dual-view face identification and authentication method according to claim 3, characterized in that: The angle α1, the angle α2 and the angle α3 are obtained in the following manner: A predetermined deflection angle β of the first camera is provided, wherein the predetermined deflection angle of the first camera takes the first camera as an endpoint, takes the line connecting the first camera and the second camera as one of its sides, and the other side faces the detection area for the persons to pass through, and the predetermined deflection angle β of the second camera is consistent with the predetermined deflection angle β of the first camera, wherein the predetermined deflection angle of the second camera takes the second camera as an endpoint, takes the line connecting the second camera and the first camera as one of its sides, and the other side faces the detection area for the persons to pass through; Obtaining a deviation angle θ1 of the first camera relative to a predetermined deflection angle toward a detection area for a person to pass; Obtaining a deviation angle θ2 of the second camera relative to a predetermined deflection angle toward a detection area for a person to pass; The included angle α1=β+θ1, the included angle α2=β+θ2, and the included angle α3=180°-α1-α2.

5. The dual-view face identification and authentication method according to claim 2, characterized in that: The calculation of the height H of the person to be passed according to the distance D includes: Providing a height h1 of the first camera or the second camera from the tread surface; The height H of the person waiting to pass is calculated based on the height h1 and the distance D.

6. The dual-view face identification and authentication method according to claim 5, characterized in that: The calculation of the height H of the person to be passed according to the height h1 and the distance D includes: Obtain the angle θ3 between the line connecting the first camera and the person to be passed and the horizontal plane where the first camera is located, with the first camera as the endpoint, and the height H=h1+Dtanθ3; or Obtain the angle θ3 between the line connecting the second camera and the person to be passed and the horizontal plane where the second camera is located, with the second camera as the endpoint, and the height H=h1+Dtanθ3.

7. The face recognition method according to claim 1, characterized in that: The face image used for comparison is a frontal face image.

8. The dual-view face identification and authentication method according to claim 1, characterized in that: Comparing the facial image captured by at least one of the first camera and the second camera with the pre-stored facial image includes: When the face image captured by one of the first camera and the second camera is a frontal face image, comparing the frontal face image with the pre-stored face image; When the facial images captured by the first camera and the second camera are both side face images, the side face images captured by the first camera and the second camera are spliced ​​into a front face image, and the spliced ​​front face image is compared with the pre-stored facial image.

9. The dual-view face identification and authentication method according to claim 8, characterized in that: When the facial images captured by the first camera and the second camera are both side face images, the image in which the side face is most exposed in the image captured by the first camera and the image in which the side face is most exposed in the image captured by the second camera are spliced ​​into a frontal face image.

10. A dual-view face identification and authentication system, characterized in that: include: A gate body, a first adjustment mechanism, a first camera, a second adjustment mechanism, a second camera, a blocking member and a controller, wherein the first camera is arranged on the gate body through the first adjustment mechanism, the first adjustment mechanism is used to adjust the height of the first camera, the second camera is arranged on the gate body through the second adjustment mechanism, the second adjustment mechanism is used to adjust the height of the second camera, the blocking member is movably arranged on the gate body, the blocking member is used to block people waiting to pass, the first camera, the second camera and the blocking member are all connected to the controller, and the controller is used to adjust the height of the first camera and the second camera according to the height H of the person waiting to pass when the first camera and the second camera capture the same facial image, and to compare the facial image captured by at least one of the first camera and the second camera with the pre-stored facial image, and after the captured facial image matches the pre-stored facial image, control the activity of the blocking member to allow the person waiting to pass.

11. The dual-view face identification and authentication system according to claim 10, characterized in that: The distance L between the first camera and the second camera ranges from 0.8m to 1.5m, the field of view angle θ of the first camera and the second camera ranges from 45° to 75°, and the predetermined deflection angle β of the first camera toward the person to pass and the predetermined deflection angle β of the second camera toward the person to pass range from 30° to 50°.

Images