A test body for 3D face recognition performance test

Abstract

The utility model discloses a kind of for 3D face recognition performance test test body, ear analog body, human eye analog body, malar bone analog body, cheek analog body, nose analog body and lip analog body are respectively equipped with stereoscopic structure on surface concrete, simulate real face feature, improve stereoscopic face simulation effect, and the movement and telescopic change of each analog body on surface concrete can be mutually matched to form the 3D face test sample of multiple distribution forms, can effectively test the depth collection, three-dimensional modeling, three-dimensional recognition ability and face recognition accuracy of 3D face recognition equipment, improve test result accuracy and reliability, can be used for the depth recognition training of 3D face recognition equipment, each analog body is respectively configured into different stereoscopic shape, can also test the depth collection, three-dimensional modeling, three-dimensional recognition ability of 3D face recognition equipment to circular arc body, sphere, cuboid, conical body, ellipsoid and have height difference ladder, to improve test result accuracy and reliability.

Description

Technical Field

[0001] This utility model relates to the field of facial recognition technology, specifically to a test body for 3D facial recognition performance testing. Background Technology

[0002] 3D face recognition devices are categorized based on the imaging principle of their cameras, including 3D structured light, Time-of-Flight (TOF) laser ranging (where the illumination source typically uses square wave pulse modulation, and distance is calculated based on the time difference between pulse emission and reception), and binocular stereo vision. They process 3D face data such as point cloud data, voxel meshes, triangular meshes, and multi-view images. Compared to 2D face data, this data contains an additional dimension of depth information, making it more complete, three-dimensional, and better able to express individual facial features from various angles, thus effectively improving the accuracy of face recognition in 3D face recognition devices.

[0003] To test the performance of facial recognition devices, existing technologies typically use two-dimensional images or simplified facial models. For example, patent document CN 108235769 A discloses a performance testing device for facial recognition devices, comprising a testing mechanism consisting of a horizontal movement mechanism and an angle testing fixture, and a head model mounted on the testing mechanism. The head model is positioned in front of the facial recognition device under test, with adjustable distance and angle, capable of simulating several positions. However, the facial features of these head models, such as contours, facial feature positions, and facial feature depth, remain fixed. Even if the overall position of the head model can be adjusted using the horizontal movement mechanism and angle testing fixture, a single head model can only serve as a single facial sample, failing to provide the large amount of point cloud data required for depth acquisition, 3D modeling, and 3D recognition of different faces in the process of testing the accuracy of 3D facial recognition devices.

[0004] Therefore, providing a testing device with good simulation effect and capable of providing a variety of 3D face data test samples to effectively test the face recognition accuracy of 3D face recognition devices and improve the reliability of test results has become an urgent problem to be solved in this field. Utility Model Content

[0005] In view of the shortcomings of the existing technology, the purpose of this utility model is to provide a test body for 3D face recognition performance testing that has good simulation effect and accurate and reliable test results.

[0006] To achieve the above objectives, the present invention provides a test body for 3D face recognition performance testing, comprising a face body, wherein an ear simulator is provided at a position corresponding to the human ear, an eye simulator is provided at a position corresponding to the human eye, a cheekbone simulator is provided at a position corresponding to the human cheek, a cheek simulator is provided at a position corresponding to the human nose, and a lip simulator is provided at a position corresponding to the human lips.

[0007] The ear simulator, eye simulator, cheekbone simulator, cheek simulator, nose simulator, and lip simulator are respectively arranged in a three-dimensional structure on the surface body, and form a distribution height difference between them;

[0008] The ear simulator, eye simulator, cheek simulator, nose simulator, and lip simulator can move and stretch on the surface to form 3D face test samples with several different distributions.

[0009] Furthermore, the surface is specifically configured as a frustum distribution structure, forming a first surface and a second surface, which are smoothly connected by an arc-shaped circumferential surface.

[0010] Furthermore, the ear simulator is disposed on both sides of the first surface, and the eye simulator, cheekbone simulator, cheek simulator, nose simulator, and lip simulator are respectively disposed on the second surface.

[0011] Furthermore, the ear simulator is composed of a first layer of arc-shaped bodies and a second layer of arc-shaped bodies with adjustable spacing, and the ear simulator can rotate around the surface.

[0012] Furthermore, the human eye simulator has an outwardly protruding hemispherical structure and is composed of several hemispherical shells of increasing diameter coaxially stacked.

[0013] Furthermore, the cheekbone simulator is distributed below the human eye simulator and on both sides of the nose simulator, and the cheekbone simulator is distributed in a progressively higher manner from the inside to the outside of the face.

[0014] Furthermore, the cheekbone simulator is configured as a gradually increasing stepped structure.

[0015] Furthermore, the cheek simulator has a cuboid structure, the nose simulator has an oblique trivertebral shape, and the height of the nose simulator gradually increases from the top to the bottom of the facet, and it can extend and retract in the height direction.

[0016] Furthermore, the nose simulator is specifically connected to the surface via a flexible telescopic folding component.

[0017] Furthermore, the telescopic folding component is composed of several nested elastic folding segments, with folds formed between adjacent elastic folding segments.

[0018] Furthermore, the lip simulator has a semi-elliptical structure and is composed of a first arc and a second arc, which can extend and retract and move relative to each other.

[0019] The test body provided by this utility model for 3D face recognition performance testing has three-dimensional structures on its surface, including ear simulators, eye simulators, cheekbone simulators, cheek simulators, nose simulators, and lip simulators, which simulate real human facial features, reflect the contours and shapes of the face, thereby improving the three-dimensional face simulation effect. Furthermore, the movement and expansion of each simulator on the surface can cooperate to form 3D face test samples with various distribution forms, which can effectively test the depth acquisition, three-dimensional modeling, three-dimensional recognition capabilities, and face recognition accuracy of 3D face recognition devices, improve the accuracy and reliability of test results, and can be used for depth recognition training of 3D face recognition devices.

[0020] Furthermore, each simulation object is configured with a different three-dimensional shape, which can also test the 3D face recognition device's ability to acquire depth, 3D model, and 3D recognize arcs, spheres, cuboids, cones, ellipsoids, and objects with elevation differences, thereby improving the accuracy and reliability of the test results. Attached Figure Description

[0021] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0022] Figure 1 This is a front view schematic diagram of the test object used for 3D face recognition performance testing in this utility model;

[0023] Figure 2 This is a side view of the test object used for 3D face recognition performance testing in this utility model;

[0024] Figure 3 This is a bottom view of the test object used for 3D face recognition performance testing in this utility model;

[0025] Figure 4 and Figure 5 This is a schematic diagram of the structure of the ear simulator in this utility model;

[0026] Figure 6 This is a schematic diagram of the structure of the human eye simulator in this utility model;

[0027] Figure 7 This is a schematic diagram of the unfolded state of the nose simulator in this utility model;

[0028] Figure 8 This is a schematic diagram of the contracted state of the nose simulator in this utility model;

[0029] Figure 9 and Figure 10 This is a schematic diagram of the structure of the lip simulator in this utility model;

[0030] Figure 11 This is a schematic diagram of the structure of the cheekbone simulator in this utility model.

[0031] Figure label:

[0032] 1. Surface details; 11. First surface; 12. Second surface; 121. Mounting groove; 13. Peripheral surface;

[0033] 2. Ear simulator; 21. First layer arc-shaped body; 211. Damping pivot; 22. Second layer arc-shaped body; 221. Flexible telescopic component; 222. Arc-shaped body component;

[0034] 3. Human eye simulator; 31. Hemispherical shell; 311. First hemispherical shell; 312. Outer hemispherical shell;

[0035] 4. Nose simulator;

[0036] 5. Lip simulator; 51. First arc; 52. Second arc; 53. Vertical slide rail; 54. First telescopic support rod; 55. Second telescopic support rod; 56. Elastic support layer;

[0037] 6. Cheekbone simulator; 61. First height step; 62. Second height step; 63. Third height step;

[0038] 7. Cheek simulator; 8. Telescopic folding component; 81. Elastic folding section; 82. Wrinkle. Detailed Implementation

[0039] To make the technical means, creative features, objectives and effects of this utility model easier to understand, the following description, in conjunction with specific illustrations, further elaborates on this utility model.

[0040] See Figures 1 to 3 The image shown is an example of a test body provided by this utility model for 3D face recognition performance testing.

[0041] As shown in the figure, the test object used for 3D face recognition performance testing in this example mainly includes a face body 1. On the face body 1, an ear simulator 2 is set at the position corresponding to the human ear, an eye simulator 3 is set at the position corresponding to the human eye, a nose simulator 4 is set at the position corresponding to the human nose, a lip simulator 5 is set at the position corresponding to the human mouth, a cheekbone simulator 6 is set at the position corresponding to the human cheekbone, and a cheek simulator 7 is set at the position corresponding to the human cheek. Each simulator is set on the face body 1 in a three-dimensional structure and forms a height difference between them to simulate real human facial features, reflect the facial contour and facial features, thereby improving the simulation effect.

[0042] Furthermore, the ear simulator 2, eye simulator 3, nose simulator 4, lip simulator 5, and cheek simulator 7 can move and adjust their height on the face body 1 to form several distribution forms of 3D face test samples. This can effectively test the depth acquisition, 3D modeling, 3D recognition capabilities, and face recognition accuracy of 3D face recognition devices, improve the accuracy and reliability of test results, and can be used for depth recognition training of 3D face recognition devices.

[0043] Combination Figures 1 to 3 The surface body 1 is configured as a frustum distribution structure with an elliptical cross section, forming a first surface 11 and a second surface 12 respectively. The first surface 11 and the second surface 12 are smoothly connected by an arc-shaped circumferential surface 13, so that the surface body 1 as a whole forms an arc-shaped three-dimensional structure to simulate the curvature of the human head and improve the simulation effect.

[0044] At the same time, the first surface 11 and the second surface 12 of the surface 1 form a height difference, which can detect the recognition accuracy of the 3D face recognition device for depth data.

[0045] Combination Figures 1 to 3 In conjunction with the face body 1, ear simulators 2 are respectively set on both sides of the first surface 11 of the face body 1 to simulate human ears. Eye simulators 3, nose simulators 4, lip simulators 5, cheekbone simulators 6 and cheek simulators 7 are respectively set in three-dimensional structure on the second surface 12 of the face body 1, so that the ear simulators 2, eye simulators 3, nose simulators 4, lip simulators 5, cheekbone simulators 6 and cheek simulators 7 form a distribution height difference. While simulating the structure of human facial features, it can detect the depth recognition accuracy of 3D face recognition equipment for different parts.

[0046] As an example, the nose simulator 4 is located at the longitudinal center of the second surface 12. The eye simulator 3, cheekbone simulator 6, and cheek simulator 7 each have two components, which are respectively located on both sides of the second surface 12 with the longitudinal direction of the nose simulator 4 as the dividing line. The lip simulator 5 is located below the nose simulator 4.

[0047] Combination Figures 1 to 3 Specifically, the ear simulator 2 is generally arc-shaped and is composed of a first arc-shaped body 21 and a second arc-shaped body 22. The surface area of ​​the first arc-shaped body 21 is larger than that of the second arc-shaped body 22, and the second arc-shaped body 22 is placed on the first arc-shaped body 21. This allows the first arc-shaped body 21 to simulate the outer auricle of the human ear, and the second arc-shaped body 22 to simulate the inner auricle of the human ear, thus realistically simulating the human ear. At the same time, the first arc-shaped body 21 and the second arc-shaped body 22 form a height difference distribution, which can effectively test the recognition performance of the 3D face recognition device.

[0048] Combination Figure 4 and Figure 5 Furthermore, the two ends of the second arc-shaped body 22 are respectively connected by a flexible telescopic component 221 and an arc-shaped body component 222. The flexible telescopic component 221 is made of a corrugated tube, with one end connected to the first arc-shaped body 21 and the other end connected to the first end of the arc-shaped body component 222. Meanwhile, the second end of the arc-shaped body component 222 is slidably connected to the first arc-shaped body 21 through a slider. Correspondingly, the first arc-shaped body 21 is provided with a track adapted to the slider, thereby stably setting the second arc-shaped body 22 on the first arc-shaped body 21.

[0049] In this way, the arc-shaped component 222 of the second arc-shaped body 22 can move on the first arc-shaped body 21 and drive the flexible telescopic component 221 to extend and retract synchronously, thereby adjusting the mating distance between the second arc-shaped body 22 and the first arc-shaped body 21 to simulate the distribution structure of the outer and inner auricles of different human ears.

[0050] As an example, the arc-shaped component 222 moves away from the surface component 1 on the first arc-shaped component 21, which can drive the flexible telescopic component 221 to extend synchronously, thereby reducing the mating distance between the second arc-shaped component 22 and the first arc-shaped component 21, simulating the human ear part where the distance between the outer and inner auricles is relatively small.

[0051] Conversely, the arc-shaped component 222 moves on the surface component 1 above the first arc-shaped component 21, which can drive the flexible telescopic component 221 to shorten synchronously, thereby increasing the mating distance between the second arc-shaped component 22 and the first arc-shaped component 21, simulating the human ear part where the outer and inner auricles are distributed with a large distance.

[0052] Therefore, the first layer of arc-shaped body 21 and the second layer of arc-shaped body 22 can work together to simulate different parts of the human ear. The flexible telescopic component 221 can be extended and retracted to adapt to the movement and adjustment of the second layer of arc-shaped body 22, so as to ensure that the second layer of arc-shaped body 22 can form a smooth and continuous arc surface in any adjustment position, thus ensuring the continuity and integrity of the second layer of arc-shaped body 22, thereby more realistically simulating the structure of the human ear auricle and improving the simulation effect.

[0053] Combination Figure 4 and Figure 5 Furthermore, the first layer of arc-shaped body 21 is hinged to the surface body 1, so that the first layer of arc-shaped body 21 can rotate around the surface body 1 and drive the second layer of arc-shaped body 22 to rotate synchronously, so as to adjust the distribution angle between the ear simulation body 2 and the surface body 1, thereby simulating the distribution shape of the human ear part, such as abduction or attachment.

[0054] Preferably, the first arc-shaped body 21 is hinged to the surface body 1 via a damping pivot 211, so that when the first arc-shaped body 21 is pushed, it can rotate relative to the surface body 1 via the damping pivot 211 to adjust the distribution angle between the ear simulator 2 and the surface body 1. At the same time, the damping structure (such as friction plate, hydraulic oil or spring) inside the damping pivot 211 can generate controllable resistance to balance the centrifugal force or external load when the first arc-shaped body 21 rotates, so that the first arc-shaped body 2 can rotate to any angle and remain suspended stably to form human ear parts with various distribution patterns.

[0055] Here, the damping shaft 211 is a conventional technical means in this field, and will not be described in detail here.

[0056] Combination Figure 5 As an example, the first layer of arc-shaped body 21 rotates away from the surface body 1 via the damping shaft 211, which can drive the second layer of arc-shaped body 22 to rotate synchronously, thereby increasing the distribution angle between the ear simulator 2 and the surface body 1, so that the ear simulator 2 moves away from the second surface 1 of the surface body 1. The damping shaft 211 can maintain the rotation angle of the ear simulator 2, thereby simulating the attached form of the human ear, such as the face ear.

[0057] Combination Figure 4 Conversely, the first layer of arc-shaped body 21 rotates towards the surface body 1 via the damping shaft 211, which can drive the second layer of arc-shaped body 22 to rotate synchronously, thereby reducing the distribution angle between the ear simulator 2 and the surface body 1. When the ear simulator 2 and the surface body 1 are distributed at right angles, the damping shaft 211 can maintain the rotation angle of the ear simulator 2, thereby simulating the outward-facing shape of the human ear, such as protruding ears.

[0058] The resulting ear simulator 2 can simulate various shapes of human ears, thus providing a variety of test samples for 3D face recognition devices and testing the ability of 3D face recognition devices to recognize curved objects.

[0059] Combination Figures 1 to 3 Furthermore, the human eye simulator 3 is a hemispherical structure protruding outwards and protruding on the second surface 12 at the position corresponding to the human eye, so as to simulate the shape of the human eye.

[0060] Combination Figure 6 The human eye simulator 3 is composed of several hemispherical shells 31 with increasing diameters, which are coaxially stacked. The first hemispherical shell 311 with the smallest diameter can be used as the innermost component of the human eye simulator 3 and is set on the surface body 1. The outer hemispherical shells 312 with progressively increasing diameters are sequentially placed outside the first hemispherical shell 311, thus forming a human eye simulator 3 with gradually increasing diameters from the inside to the outside and stacked together.

[0061] In this way, by sequentially removing the outer hemispherical shell 312, the inner hemispherical shell with a smaller diameter can be exposed, so as to adjust the diameter and height of the human eye simulator 3, and test the 3D face recognition device's performance in recognizing different eye features and the depth of the hemisphere.

[0062] Combining 1 to Figure 3 Furthermore, the nose simulator 4 is shaped like a trivertebra and protrudes from the second surface 12 at the position corresponding to the human nose. The height of the nose simulator 4 gradually increases from the top to the bottom of the surface 1 to simulate the height distribution structure of the human nasal bridge.

[0063] Furthermore, the nose simulator 4 is configured to extend and retract in the height direction to change the height of the nose simulator 4 and simulate different bridge heights of the nose.

[0064] Combination Figure 7 and Figure 8 Specifically, the nose simulator 4 is connected to the surface body 1 through a flexible telescopic folding component 8. One end of the telescopic folding component 8 is connected to the nose simulator 4, and the other end is connected to the surface body 1. The telescopic folding component 8 is composed of several nested elastic folding segments 81.

[0065] Furthermore, along the nose simulator 4 to the surface 1, the cross-section of each layer of elastic folding segment 81 is enlarged proportionally, and folds 82 are formed between adjacent elastic folding segments 81, so that the upper layer of elastic folding segment 81 can be embedded into the lower layer of elastic folding segment 81 through the folds 82, so as to realize the nesting of several elastic folding segments 81.

[0066] As an example, the telescopic folding component 8 can be made of materials such as silicone, PE (thermoplastic elastomer), and TPR (thermoplastic rubber), which have good resilience and can withstand repeated folding and stretching without breaking or permanent deformation.

[0067] Combination Figure 7 Therefore, when the nose simulator 4 is pulled away from the surface 1, the telescopic folding component 8 will be pulled out synchronously, causing the folds 82 between adjacent elastic folding sections 81 to open up and simultaneously increasing the height of the nose simulator 4. When all the folds 82 between the elastic folding sections 81 are fully opened, the telescopic folding component 8 is fully opened, and the height of the nose simulator 4 reaches its maximum.

[0068] Combination Figure 8 Correspondingly, when the nose simulator 4 is pushed towards the surface 1, the telescopic folding component 8 will contract synchronously, so that the folds 82 between adjacent elastic folding segments 81 are embedded into the lower elastic folding segment 81 based on the thrust, so as to fold the adjacent elastic folding segments 81 and simultaneously reduce the height of the nose simulator 4. When all the elastic folding segments 81 are nested in layers, the telescopic folding component 8 is fully contracted, and the height of the nose simulator 4 reaches its minimum.

[0069] Here, the height of the telescopic folding component 8 is not limited. Preferably, the height of the telescopic folding component 8 can cover different nose bridge heights.

[0070] Furthermore, by controlling the distance between the nose simulator 4 and the surface body 1, the number and degree of unfolding of the folds 82 between the elastic folding segments 81 can be controlled simultaneously, thereby adjusting the height of the nose simulator 4 to test the recognition performance of the 3D face recognition device for different nose features and triangular pyramids.

[0071] Combination Figures 1 to 3 Correspondingly, the lip simulator 5 is a semi-elliptical structure protruding from the second surface 12 at the position corresponding to the human lips, so as to simulate the shape of the human lips.

[0072] Combination Figure 9 and Figure 10 The lip simulator 5 is composed of a first arc body 51 and a second arc body 52 that are relatively distributed. The first arc body 51 and the second arc body 52 can respectively simulate the upper lip and lower lip of a human body. The lip simulator 5 also includes an adjustment component that can drive the first arc body 51 and the second arc body 52 to extend and retract, as well as move relative to each other, in order to simulate the depth and thickness of different human lips and form a variety of test samples.

[0073] Correspondingly, the second surface 12 of the surface body 1 is provided with a mounting groove 121 corresponding to the lip simulation body 5, so that the adjustment component can be set in the mounting groove 121 and cooperate with the first arc body 51 and the second arc body 52 protruding on the second surface 12 to drive the first arc body 51 and the second arc body 52 to move.

[0074] Specifically, the adjustment assembly includes a vertical slide rail 53, a first telescopic support rod 54, and a second telescopic support rod 55. The vertical slide rail 53 is built into the mounting groove 121 along the height direction. One end of the first telescopic support rod 54 extends out of the mounting groove 121 and connects to the bottom surface of the first arc body 51, and the other end is connected to the vertical slide rail 53 through a slider. Correspondingly, one end of the second telescopic support rod 55 extends out of the mounting groove 121 and connects to the bottom surface of the second arc body 52, and the other end is connected to the vertical slide rail 53 through a slider.

[0075] Preferably, the first telescopic support rod 54 and the second telescopic support rod 55 are each composed of a telescopic sleeve, which includes several layers of interlocking pipes, so that the first telescopic support rod 54 and the second telescopic support rod 55 can achieve the interlocking and extension of the pipes under the action of external force, so as to achieve telescopic movement.

[0076] In this way, pushing the tube or pulling the first arc body 51 and the second arc body 52 can drive the first telescopic support rod 54 and the second telescopic support rod 55 to extend and retract synchronously, so as to adjust the distance between the first arc body 51 and the second arc body 52 and the second surface 12, thereby adjusting the depth of the lip simulator 5.

[0077] Furthermore, the first telescopic support rod 54 can move along the vertical slide rail 53 via a slider, and drive the first arc body 51 to move synchronously to adjust the height of the first arc body 51. Correspondingly, the second telescopic support rod 55 can move along the vertical slide rail 53 via a slider, and drive the second arc body 52 to move synchronously to adjust the height of the second arc body 52, and realize the relative movement of the first arc body 51 and the second arc body 52, thereby adjusting the distance between the first arc body 51 and the second arc body 52, and realizing the thickness adjustment of the lip simulator 5.

[0078] Combination Figure 10 When the first arc 51 and the second arc 52 are relatively far apart, there is a gap between them, which will expose the mounting groove 121 and the internal adjustment components, affecting the overall appearance of the surface 1 and causing test errors in the 3D face recognition device. In order to ensure the accuracy of the test results, the adjustment components also include an elastic support layer 56. The elastic support layer 56 can cooperate with the first arc 51 and the second arc 52 to achieve seamless relative movement between the first arc 51 and the second arc 52.

[0079] Specifically, the elastic support layer 56 connects the inner sides of the first arc body 51 and the second arc body 52 respectively. Preferably, the elastic support layer 56 is made of existing memory foam, spandex cloth or silicone foam, so that the elastic support layer 56 can expand and contract synchronously when the first arc body 51 and the second arc body 52 move relative to each other, so as to fill the gap between the first arc body 51 and the second arc body 52, ensuring that the lip simulation body 5 can be stably adjusted in thickness and has a full shape, thereby improving the simulation effect.

[0080] Combination Figure 10As an example, when the first telescopic support rod 54 and the second telescopic support rod 55 respectively drive the first arc body 51 and the second arc body 52 away from each other along the vertical slide rail 53, the distance between the first arc body 51 and the second arc body 52 increases. The elastic support layer 56 extends under its own elastic force to support the first telescopic support rod 54 and the second telescopic support rod 55, and fill the gap between the first arc body 51 and the second arc body 52, so as to increase the thickness of the lip simulation body 5 and ensure the structural integrity and full shape of the lip simulation body 5.

[0081] Combination Figure 9 Conversely, when the first telescopic support rod 54 and the second telescopic support rod 55 respectively drive the first arc body 51 and the second arc body 52 to move closer to each other along the vertical slide rail 53, the distance between the first arc body 51 and the second arc body 52 decreases, which will simultaneously compress the elastic support layer 56 to reduce the thickness of the lip simulator 5.

[0082] Here, it is necessary to ensure that the movement stroke of the first arc body 51 and the second arc body 52 is adapted to the extreme compression and extension state of the elastic support layer 56, so as to ensure stable adjustment of the thickness of the lip simulator 5.

[0083] Combination Figure 9 Meanwhile, the overall length of the placement groove 121 is adapted to the moving stroke of the first arc body 51 and the second arc body 52, ensuring that when the first arc body 51 and the second arc body 52 approach each other to the closest distance and reach the limit compression state of the elastic support layer 56, the edges of the first arc body 51 and the second arc body 52 can respectively cover the placement groove 121, ensuring that the placement groove 121 will not be exposed, will not affect the appearance of the surface body 1, and will not cause testing errors of the 3D face recognition device.

[0084] The resulting lip simulator 5 can be adjusted in depth and thickness to simulate different shapes of human lips and ensure the appearance integrity of the face body 1, providing a variety of test samples for 3D face recognition devices, thereby improving the accuracy of test results.

[0085] Meanwhile, the hemispherical eye simulator 3 and the semi-elliptical lip simulator 5 each have an arc-shaped surface, with the highest and lowest points on the arc-shaped surface, which can be used to test the recognition accuracy of 3D face recognition devices for different depths of the arc-shaped surface.

[0086] Combination Figures 1 to 2Furthermore, the second surface 12 of the facepiece 1 forms a cheekbone simulator 6 at the position corresponding to the human cheekbone. The cheekbone simulator 6 is distributed below the human eye simulator 3 and located on both sides of the nose simulator 4. The cheekbone simulator 6 and the second surface 12 form an integrated structure. The cheekbone simulator 6 is distributed in a progressively higher manner from the inside to the outside of the facepiece 1 to simulate the cheekbone structure of the real human body and make the facepiece 1 more three-dimensional and improve the simulation effect.

[0087] Combination Figure 11 To accurately test the depth recognition performance of the 3D face recognition device, the cheekbone simulator 6 was configured with a gradually increasing stepped structure, forming several steps with increasing height. Each step has a fixed and known height difference from the previous step, thus forming a precise and controllable stepped three-dimensional contour from low to high on the cheekbone simulator 6 as a whole, in order to test the depth recognition accuracy of the 3D face recognition device for steps of different heights.

[0088] Combination Figure 11 As an example, the cheekbone simulator 6 is formed with a first height step 61, a second height step 62 and a third height step 63, and the height of the first height step 61, the second height step 62 and the third height step 63 increases uniformly from 0.1cm to 2cm to simulate the height of the cheekbone of a real human body. At the same time, it is used to test the depth recognition accuracy of the 3D face recognition device on the first height step 61, the second height step 62 and the third height step 63 respectively.

[0089] Furthermore, the first height step 61, the second height step 62, and the third height step 63 are each equipped with identifiers, such as different numbers or colors, to distinguish different height steps and height changes, thereby enabling the depth recognition results of the 3D face recognition device to be quickly compared with the corresponding height steps, improving the accuracy of the test results.

[0090] Combination Figures 1 to 3 The surface 1 also has a cheek simulator 7 at the position corresponding to the human cheek. The cheek simulator 7 is a cuboid structure protruding on the second surface 12 to simulate the human cheek and is also used to test the depth recognition accuracy of the 3D face recognition device for the cuboid structure.

[0091] Furthermore, the cheek simulator 7 is also connected to the face body 1 via a telescopic folding component 8. One end of the telescopic folding component 8 is connected to the cheek simulator 7, and the other end is connected to the face body 1, so that the cheek simulator 7 can extend and retract in the height direction via the telescopic folding component 8 to change the height of the nose simulator 4 and simulate different cheek heights.

[0092] As an example, when the cheek simulator 7 is pulled away from the face body 1, the telescopic folding component 8 will be pulled out simultaneously, causing the folds 82 between adjacent elastic folding sections 81 to open up and simultaneously increasing the height of the cheek simulator 7. When the folds 82 between all elastic folding sections 81 are fully opened, the telescopic folding component 8 is fully opened, and the height of the cheek simulator 7 reaches its maximum.

[0093] Correspondingly, when the cheek simulator 7 is pushed towards the face body 1, the telescopic folding component 8 will contract synchronously, so that the folds 82 between adjacent elastic folding segments 81 are embedded into the lower elastic folding segment 81 based on the thrust, so as to fold the adjacent elastic folding segments 81 and simultaneously reduce the height of the cheek simulator 7. When all elastic folding segments 81 are nested in layers, the telescopic folding component 8 is fully contracted, and the height of the cheek simulator 7 reaches its minimum, so as to simulate different cheek heights and provide a variety of test samples for 3D face recognition devices.

[0094] Combination Figures 1 to 3 The ear simulator 2, eye simulator 3, nose simulator 4, lip simulator 5, cheekbone simulator 6, and cheek simulator 7 are respectively set in three-dimensional structure on the surface body 1, and form a distribution height difference with each other, which can effectively simulate real facial features, improve the simulation effect, and enable the 3D face recognition device to perform depth recognition tests on different parts and different three-dimensional structures of the face, thereby improving the accuracy of the test results.

[0095] Furthermore, each simulated body can move and adjust its height on surface 1 to form 3D face test samples with various distributions. This can effectively test the depth acquisition, 3D modeling, 3D recognition capabilities and face recognition accuracy of 3D face recognition devices, improve the accuracy and reliability of test results, and can be used for depth recognition training of 3D face recognition devices.

[0096] In practical applications, the distribution of one or more simulators on surface 1 can be adjusted according to testing requirements. The adjustment of each simulator will change the height difference between the simulator and other simulators, thereby forming a new 3D face test sample. Therefore, the coordinated adjustment of multiple simulators will form a sufficient number of samples to enrich the 3D face test samples and improve the accuracy of test results.

[0097] For example, when the heights of the nose simulator 4 and the eye simulator 3 are changed, the height differences between the nose simulator 4 and the eye simulator 3 and the ear simulator 2, lip simulator 5, cheekbone simulator 6 and cheek simulator 7 also change accordingly, so as to form a variety of test samples that can be used to test the generalization ability and depth recognition accuracy of 3D face recognition devices for different facial features, thereby improving the accuracy of test results.

[0098] Furthermore, each simulation object is configured with a different three-dimensional shape, which can also test the 3D face recognition device's ability to acquire depth, 3D model, and 3D recognize arcs, spheres, cuboids, cones, ellipsoids, and objects with elevation differences, thereby improving the accuracy and reliability of the test results.

[0099] The test body provided by this utility model for 3D face recognition performance testing has three-dimensional structures on the surface body 1, including an ear simulator 2, an eye simulator 3, a nose simulator 4, a lip simulator 5, a cheekbone simulator 6, and a cheek simulator 7, which simulate real human facial features, reflect the facial contour and shape of facial features, thereby improving the three-dimensional face simulation effect. Furthermore, the movement and height changes of each simulator on the surface body can provide 3D face test samples with various distribution forms, which can effectively test the depth acquisition, three-dimensional modeling, three-dimensional recognition capabilities and face recognition accuracy of 3D face recognition devices, improve the accuracy and reliability of test results, and can be used for depth recognition training of 3D face recognition devices.

[0100] The foregoing has shown and described the basic principles, main features, and advantages of this utility model. Those skilled in the art should understand that this utility model is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of this utility model. Various changes and modifications can be made to this utility model without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claims. The scope of protection of this utility model is defined by the appended claims and their equivalents.

Claims

1. A test body for 3D face recognition performance testing, characterized in that, The surface includes a facet, on which an ear simulator is provided at the position corresponding to the human ear, an eye simulator is provided at the position corresponding to the human eye, a cheekbone simulator is provided at the position corresponding to the human cheek, a cheek simulator is provided at the position corresponding to the human nose, and a lip simulator is provided at the position corresponding to the human lips. The ear simulator, eye simulator, cheekbone simulator, cheek simulator, nose simulator, and lip simulator are respectively arranged in a three-dimensional structure on the surface body, and form a distribution height difference between them; The ear simulator, eye simulator, cheek simulator, nose simulator, and lip simulator can move and stretch on the surface to form 3D face test samples with several different distributions.

2. The test body for 3D face recognition performance testing according to claim 1, characterized in that, The surface is specifically configured as a frustum distribution structure, forming a first surface and a second surface, which are smoothly connected by an arc-shaped circumferential surface.

3. The test body for 3D face recognition performance testing according to claim 2, characterized in that, The ear simulator is disposed on both sides of the first surface, and the eye simulator, cheekbone simulator, cheek simulator, nose simulator, and lip simulator are disposed on the second surface.

4. The test body for 3D face recognition performance testing according to claim 1, characterized in that, The ear simulator is composed of a first layer of arc-shaped bodies and a second layer of arc-shaped bodies with adjustable spacing, and the ear simulator can rotate around the surface.

5. The test body for 3D face recognition performance testing according to claim 1, characterized in that, The human eye simulator has an outwardly protruding hemispherical structure and is composed of several hemispherical shells of increasing diameter coaxially stacked.

6. The test body for 3D face recognition performance testing according to claim 1, characterized in that, The cheekbone simulators are distributed below the human eye simulators and on both sides of the nose simulators, with the cheekbone simulators arranged in a progressively higher manner from the inside to the outside of the face.

7. The test body for 3D face recognition performance testing according to claim 6, characterized in that, The cheekbone simulator is configured in a gradually increasing stepped structure.

8. The test body for 3D face recognition performance testing according to claim 1, characterized in that, The cheek simulator has a cuboid structure, the nose simulator has an oblique trivertebral shape, and the height of the nose simulator gradually increases from the top to the bottom of the face, and it can extend and retract in the height direction.

9. The test body for 3D face recognition performance testing according to claim 8, characterized in that, The nose simulator is specifically connected to the surface via a flexible, telescopic, and folding component.

10. The test body for 3D face recognition performance testing according to claim 9, characterized in that, The telescopic folding component is composed of several nested elastic folding segments, with folds formed between adjacent elastic folding segments.

11. The test body for 3D face recognition performance testing according to claim 1, characterized in that, The lip simulator has a semi-elliptical structure and is composed of a first arc and a second arc. The first and second arcs can extend and retract and can move relative to each other.

Images