Feature point position detection method and electronic device

Abstract

The present application provides a feature point position detection method and an electronic device. The method includes: obtaining a plurality of first relative positions of a plurality of feature points on a specific object relative to a first image capturing component; obtaining a plurality of second relative positions of the plurality of feature points on the specific object relative to a second image capturing component; and in response to determining that the first image capturing component is unreliable, estimating a current three-dimensional position of each feature point based on a historical three-dimensional position of each feature point and the plurality of second relative positions. Thus, a user in front of a 3D display can not see a three-dimensional image with severe 3D crosstalk due to an unreliable image capturing component.

Description

Technical Field

[0001] This invention relates to an image processing mechanism, and more particularly to a method and electronic device for detecting feature point locations. Background Technology

[0002] Current glasses-free 3D displays first place the pixels for the left and right eyes at their corresponding pixel positions on the display panel. Then, the liquid crystal within the 3D lens controls the light path, projecting the images for the left and right eyes into the correct eyes. Because focusing is required for both eyes, the 3D lens typically has a curved design, ensuring the image for the left (right) eye is focused and projected into the left (right) eye. However, due to limitations in the light refraction path, some light rays may be projected into the wrong eye. In other words, the image for the left (right) eye may mistakenly appear in the right (left) eye; this phenomenon is called 3D crosstalk.

[0003] Generally, glasses-free 3D displays are equipped with eye-tracking systems to provide corresponding images for each eye after obtaining the user's eye position. Currently, the most common eye-tracking methods utilize binocular cameras for facial recognition and triangulation to determine the positions of the two eyes. However, in some cases, facial recognition using binocular cameras may fail to accurately measure the eye positions due to insufficient facial feature points, potentially affecting the quality of subsequent 3D image rendering. Summary of the Invention

[0004] In view of this, the present invention provides a feature point location detection method and electronic device, which can be used to solve the above-mentioned technical problems.

[0005] The present invention provides a feature point position detection method, suitable for an electronic device including a first imaging component and a second imaging component, comprising: acquiring a plurality of first relative positions of a plurality of feature points on a specific object relative to the first imaging component; acquiring a plurality of second relative positions of the plurality of feature points on the specific object relative to the second imaging component; and, in response to determining that the first imaging component is unreliable, estimating a current three-dimensional position of each feature point based on a historical three-dimensional position of each feature point and the plurality of second relative positions.

[0006] The present invention provides an electronic device including a first imaging component, a second imaging component, and a processor. The processor is coupled to the first imaging component and the second imaging component and configured to perform: acquiring a plurality of first relative positions of a plurality of feature points on a specific object relative to the first imaging component; acquiring a plurality of second relative positions of the plurality of feature points on the specific object relative to the second imaging component; and, in response to determining that the first imaging component is unreliable, estimating a current three-dimensional position of each feature point based on a historical three-dimensional position of each feature point and the plurality of second relative positions. Attached Figure Description

[0007] The accompanying drawings are included to further illustrate the invention, and are incorporated in and constitute a part of this specification. The drawings illustrate embodiments of the invention and, together with the description, serve to explain the principles of the invention.

[0008] Figure 1 This is a schematic diagram of an electronic device according to an embodiment of the present invention;

[0009] Figure 2 This is a flowchart illustrating a feature point location detection method according to an embodiment of the present invention;

[0010] Figure 3 This is a schematic diagram of facial feature points according to an embodiment of the present invention;

[0011] Figure 4 This is a schematic diagram illustrating the estimation of the current three-dimensional position of each feature point according to an embodiment of the present invention;

[0012] Figure 5 This is an application scenario diagram illustrating the determination of the current three-dimensional position of each feature point according to an embodiment of the present invention. Detailed Implementation

[0013] Reference will now be made in detail to exemplary embodiments of the invention, examples of which are illustrated in the accompanying drawings. Wherever possible, the same component reference numerals are used in the drawings and description to denote the same or similar parts.

[0014] Please refer to Figure 1 This is a schematic diagram of an electronic device according to embodiments of the present invention. In different embodiments, the electronic device 100 may be implemented as various intelligent devices and / or computer devices. In some embodiments, the electronic device 100 may be implemented as an eye-tracking device. In one embodiment, the electronic device 100 may be externally connected to a 3D display (e.g., a glasses-free 3D display) to provide relevant eye-tracking information to the 3D display. In another embodiment, the electronic device 100 may also be implemented as a 3D display with eye-tracking functionality.

[0015] After obtaining eye-tracking information, the electronic device 100, which is implemented as a 3D display, can adjust the display content accordingly, so that users viewing the 3D display can enjoy the display content of the 3D display with lower 3D crosstalk.

[0016] exist Figure 1 In the embodiment, electronic device 100 includes image-capturing components 101 and 102 and a processor 104. In different embodiments, electronic device 100 may also include more image-capturing components coupled to processor 104, and is not limited to these components. Figure 1 The implementation method shown.

[0017] In different embodiments, the first image-capturing component 101 and the second image-capturing component 102 may be, for example, any image-capturing device having a charge-coupled device (CCD) lens or a complementary metal-oxide-semiconductor (CMOS) lens, but are not limited thereto. In some embodiments, the first image-capturing component 101 and the second image-capturing component 102 may be integrated into a binocular camera on the electronic device 100, but are not limited thereto.

[0018] The processor 104 is coupled to the first image-capturing component 101 and the second image-capturing component 102, and may be a general-purpose processor, a special-purpose processor, a conventional processor, a digital signal processor, multiple microprocessors, one or more microprocessors incorporating a digital signal processor core, a controller, a microcontroller, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), any other type of integrated circuit, a state machine, a processor based on an advanced reduced instruction set machine (ARM), and the like.

[0019] In an embodiment of the present invention, the processor 104 accesses relevant modules and program code to implement the eye-tracking method proposed in the present invention, the details of which are described below.

[0020] Please refer to Figure 2 This is a flowchart illustrating a feature point location detection method according to an embodiment of the present invention. The method of this embodiment can be derived from... Figure 1 The electronic device 100 performs the following: Figure 1 Component description shown Figure 2 Details of each step.

[0021] First, in step S210, the processor 104 obtains multiple first relative positions of multiple feature points on a specific object relative to the first imaging component 101. For ease of explanation, it is assumed below that the specific object under consideration is a face, and the multiple feature points on the specific object are, for example, multiple facial feature points located on this face, but are not limited to this.

[0022] In one embodiment, the processor 104 may control the first imaging component 101 to capture a first image of a specific object under consideration. The processor 104 may then identify feature points on the specific object in this first image and determine a plurality of first relative positions of these feature points relative to the first imaging component 101.

[0023] Please refer to Figure 3 This is a schematic diagram of facial feature points according to an embodiment of the present invention. Figure 3 In this context, assuming the processor 104, after the first imaging component 101 captures a first image of the specific object under consideration (i.e., a face), finds on the first image... Figure 3 The multiple feature points shown are described. In one embodiment, the processor 104 can identify the multiple feature points shown in the first image based on any known face recognition algorithm, and accordingly obtain multiple first relative positions of these feature points relative to the first imaging component 101.

[0024] In one embodiment, the first relative position corresponding to each feature point can be characterized, for example, as a unit vector corresponding to each feature point. Figure 3 Taking feature point 0 (hereinafter referred to as feature point 0) as an example, after finding feature point 0, the processor 104 can generate a corresponding unit vector. This unit vector is a vector with a length of 1 that points to feature point 0, originating from the three-dimensional position of the first imaging component 101 (i.e., the position of the first imaging component 101 in three-dimensional space). Figure 3 Taking feature point number 1 (hereinafter referred to as feature point 1) as an example, after finding feature point 1, the processor 104 can generate a corresponding unit vector. This unit vector is a vector with a length of 1 and a starting point at the three-dimensional position of the first imaging component 101, pointing to feature point 1.

[0025] Based on the above principles, processor 104 can obtain Figure 3 After identifying each feature point, find the corresponding unit vector for each feature point.

[0026] In one embodiment, after identifying multiple feature points in the first image, the processor 104 can further determine whether the first imaging component 101 is reliable. In one embodiment, the processor 104 can determine whether the number of feature points in the first image is lower than a preset threshold. If so, this indicates that there may be too few feature points in the first image, and therefore the information obtained by the first imaging component 101 may not be suitable for subsequent judgment. Therefore, the processor 104 can accordingly determine that the first imaging component 101 is unreliable.

[0027] On the other hand, if the number of feature points in the first image is not lower than a preset threshold, this means that there are enough feature points in the first image, and therefore the information obtained by the first imaging component 101 is suitable for subsequent judgment. Therefore, the processor 104 can determine that the first imaging component 101 is reliable, but it is not limited to this.

[0028] Additionally, in step S220, the processor 104 obtains multiple second relative positions of the plurality of feature points on the specific object relative to the second imaging component 102. In one embodiment, the processor 104 may control the second imaging component 102 to capture a second image of the specific object under consideration. Then, the processor 104 may identify feature points on the specific object in this second image and determine the multiple second relative positions of these feature points relative to the second imaging component 102.

[0029] Similar to Figure 3 The concept is that after the processor 104 finds multiple feature points based on the second image, it can correspondingly find the unit vector of each feature point as the second relative position corresponding to each feature point. Further details can be found in [reference needed]. Figure 3 The explanation will not be repeated here.

[0030] Furthermore, in one embodiment, after identifying multiple feature points in the second image, the processor 104 can also determine whether the second imaging component 102 is reliable. In one embodiment, the processor 104 can determine whether the number of feature points in the second image is lower than a preset threshold. If so, this indicates that there may be too few feature points in the second image, and therefore the information obtained by the second imaging component 102 may not be suitable for subsequent judgment. Therefore, the processor 104 can accordingly determine that the second imaging component 102 is unreliable.

[0031] On the other hand, if the number of feature points in the second image is not lower than a preset threshold, this means that there are enough feature points in the second image, and therefore the information obtained by the second imaging component 102 is suitable for subsequent judgment. Therefore, the processor 104 can determine that the second imaging component 102 is reliable, but it is not limited to this.

[0032] In some embodiments, if the processor 104 determines at a certain time that both the first imaging component 101 and the second imaging component 102 are reliable, the processor 104 can perform feature matching and bundle adjustment based on the first relative position of the feature points corresponding to the first imaging component 101 and the second relative position of the feature points corresponding to the second imaging component 102. This allows the processor to determine the current three-dimensional position of each feature point on a specific object. Related details can be found in relevant literature on bundle adjustment (e.g., "Chen, Yu & Chen, Yisong & Wang, Guoping. (2019). Bundle Adjustment Revisited."), and will not be elaborated further here.

[0033] In other embodiments, if either the first imaging component 101 or the second imaging component 102 is determined to be unreliable, the processor 104 may estimate the current three-dimensional position of each feature point based on the one determined to be reliable and the historical three-dimensional positions of each feature point on the specific object. For ease of explanation, it is assumed below that the first imaging component 101 is the one determined to be unreliable, but this is only used as an example and is not intended to limit the possible implementations of the present invention.

[0034] Therefore, in step S230, in response to the determination that the first imaging component 101 is unreliable, the processor 104 estimates the current three-dimensional position of each feature point based on the historical three-dimensional position of each feature point and the plurality of second relative positions.

[0035] In some embodiments, the historical 3D position of each feature point is, for example, the current 3D position of each feature point at a certain time point obtained by previous estimation / detection. For instance, assuming that the processor 104 determines that the first imaging component 101 is unreliable at time point t (t is an index value), the processor 104 may, for example, take the current 3D position of each feature point at time point tk (k is a positive integer) as the historical 3D position considered at time point t, but it is not limited to this.

[0036] In one embodiment, the processor 104 obtains a first distance between feature points based on their historical 3D positions. Then, the processor 104 estimates a second distance between the second imaging component 102 and each feature point based on the unit vector corresponding to each feature point and the first distance between the feature points. Next, the processor 104 estimates the current 3D position of each feature point based on the 3D position of the second imaging component 102 and the second distance corresponding to each feature point. To make the above concepts easier to understand, the following further explains… Figure 4 Further explanation is needed.

[0037] Please refer to Figure 4This is a schematic diagram illustrating the estimation of the current three-dimensional position of each feature point according to an embodiment of the present invention. Figure 4 In the above, it is assumed that the second imaging component 102 has a three-dimensional position O at time point t, and the processor 104 finds feature points A, B, and C based on the second image at time point t. As previously mentioned, after finding feature points A, B, and C, the processor 104 can correspondingly find the unit vectors corresponding to each feature point A, B, and C as the second relative positions corresponding to feature points A, B, and C.

[0038] exist Figure 4 In the image, the second relative position between feature point A and the second imaging component 102 can be characterized as a unit vector. For example, it is a vector with a starting point at a three-dimensional position O, a length of 1, and pointing to feature point A. The second relative position between feature point B and the second imaging component 102 can be characterized as a unit vector. For example, it can be a vector with a starting point at a three-dimensional position O, a length of 1, and pointing to feature point B. Furthermore, the second relative position between feature point C and the second imaging component 102 can be represented as a unit vector. For example, it is a vector that starts at a three-dimensional position O, has a length of 1, and points to a feature point C.

[0039] In an embodiment of the present invention, it is assumed that the relative positions of feature points A, B, and C are constant between the t-th time point and the tk-th time point.

[0040] In this case, the processor 104 may, for example, obtain a first distance c between feature points A and B based on the historical three-dimensional positions of feature points A and B, obtain a first distance b between feature points A and C based on the historical three-dimensional positions of feature points A and C, and obtain a first distance a between feature points B and C based on the historical three-dimensional positions of feature points B and C.

[0041] In addition, Figure 4 In this scenario, processor 104 can determine the direction in which feature points A, B, and C are located at three-dimensional position O (which can be determined by unit vectors). (It is known), but the second distance x between the three-dimensional position O and feature point A, the second distance y between the three-dimensional position O and feature point B, and the second distance z between the three-dimensional position O and feature point C are not yet known.

[0042] To obtain the second distances x, y, z, processor 104 can be based on Figure 4 The geometric relationships shown are used to establish multiple relationships that can be used to calculate the second distances x, y, and z.

[0043] In one embodiment, the processor 104 may be based on unit vectors. Establish multiple relationships between the first distance a, b, c and the second distance x, y, z, and estimate the second distance x, y, z based on these relationships.

[0044] In one embodiment, the processor 104 may establish the following relation based on the law of cosines: and Due to unit vector Since the first distances a, b, and c are all known, the processor 104 can obtain the second distances x, y, and z by solving the above relationships (which can be regarded as simultaneous equations), but it is not limited to this.

[0045] After obtaining the second distances x, y, and z, the processor 104 can determine the current three-dimensional positions of feature points A, B, and C. Specifically, the processor 104 can be located at the corresponding unit vector. The position of feature point A at time t is defined as the position along the direction of the feature point A and at a second distance x from the 3D position O. Additionally, processor 104 can be located at the corresponding unit vector. The position of feature point B at time t is defined as the position along the direction of the feature point B and at a second distance y from the 3D position O. Similarly, processor 104 can be located at the corresponding unit vector. The position of feature point C at time t is defined as the position of feature point C in the direction of the feature point and at a second distance z from the three-dimensional position O.

[0046] In another embodiment, assuming the processor 104 determines that the second imaging component 102 is unreliable at time point t, the processor 104 may, for example, take the current 3D position of each feature point at time point tk as the historical 3D position considered at time point t. Then, the processor 104 may obtain a first distance between the feature points based on the historical 3D positions of each feature point. Next, the processor 104 estimates a second distance between the first imaging component 101 and each feature point based on the unit vector corresponding to each feature point and the first distance between the feature points. Finally, the processor 104 estimates the current 3D position of each feature point based on the 3D position of the first imaging component 101 and the second distance corresponding to each feature point.

[0047] Specifically, processor 104 can still be based on Figure 4 The relevant instructions are used to estimate the current three-dimensional position of each feature point, but in the previous embodiment, the three-dimensional position of the second imaging component 102 was used as... Figure 4 The three-dimensional position O in the image is used. However, if the second imaging component 102 is determined to be unreliable, the processor 104 needs to use the three-dimensional position of the first imaging component 101 as the position. Figure 4The three-dimensional position O is determined and used for subsequent estimation. Details can be found in the teachings of the previous embodiments and will not be repeated here.

[0048] Please refer to Figure 5 This is an application scenario diagram illustrating the determination of the current three-dimensional position of each feature point according to an embodiment of the present invention. In this embodiment of the present invention, the operation of the processor 104 obtaining the current three-dimensional position of each feature point when both the first imaging component 101 and the second imaging component 102 are reliable can be referred to as a first beam adjustment mechanism. Furthermore, the operation of the processor 104 obtaining the current three-dimensional position of each feature point when the first imaging component 101 is unreliable can be referred to as a second beam adjustment mechanism, and the operation of the processor 104 obtaining the current three-dimensional position of each feature point when the second imaging component 102 is unreliable can be referred to as a third beam adjustment mechanism.

[0049] exist Figure 5 In this scenario, at each time point, the processor 104 may execute the first beam adjustment mechanism 511, the second beam adjustment mechanism 521, and the third beam adjustment mechanism 531 before determining whether the first imaging component 101 and / or the second imaging component 102 are reliable, in order to obtain the current three-dimensional position of each feature point corresponding to the first beam adjustment mechanism 511 (hereinafter referred to as the first result 512), the current three-dimensional position of each feature point corresponding to the second beam adjustment mechanism 521 (hereinafter referred to as the second result 522), and the current three-dimensional position of each feature point corresponding to the third beam adjustment mechanism 531 (hereinafter referred to as the third result 532).

[0050] That is, before determining whether the first imaging component 101 and / or the second imaging component 102 are reliable, the processor 104 may first perform feature matching and bundle adjustment based on the first relative position of the feature points corresponding to the first imaging component 101 and the second relative position of the feature points corresponding to the second imaging component 102 to find the current three-dimensional position of each feature point as the first result 512. In addition, the processor 104 may also use the three-dimensional position of the second imaging component 102 as... Figure 4 In the case of three-dimensional position O, based on Figure 4 The mechanism obtains the current three-dimensional position of each feature point as the second result 522. Furthermore, the processor 104 can also use the three-dimensional position of the first imaging component 101 as... Figure 4 In the case of three-dimensional position O, based on Figure 4 The mechanism obtains the current three-dimensional position of each feature point as the third result 532.

[0051] Subsequently, in step S500, the processor 104 may adaptively select the first, second, or third result as the final result based on whether the first imaging component 101 and / or the second imaging component 102 are reliable.

[0052] In one embodiment, assuming the processor 104 determines at time t that both the first imaging component 101 and the second imaging component 102 are reliable, the processor 104 may select the first result 512 in step S501 to determine the current three-dimensional position of each feature point (or, in other words, discard the second and third results). Alternatively, assuming the processor 104 determines at time t that the first imaging component 101 is unreliable, the processor 104 may select the second result 522 in step S502 to determine the current three-dimensional position of each feature point (or, in other words, discard the first result 512 and the third result 532). Furthermore, assuming the processor 104 determines at time t that the second imaging component 102 is unreliable, the processor 104 may select the third result 532 in step S503 to determine the current three-dimensional position of each feature point (or, in other words, discard the first result 512 and the second result 522).

[0053] In other words, the processor 104 can execute the first beam adjustment mechanism 511, the second beam adjustment mechanism 521, and the third beam adjustment mechanism 531 at each time point, and then adaptively determine the current three-dimensional position of each feature point based on the first result 512, the second result 522, or the third result 532.

[0054] In one embodiment, after determining the current three-dimensional position of each feature point at time t, the processor 104 can further process the current three-dimensional position of each feature point at time t based on the concept of a Kalman filter (e.g., a linear Kalman filter). For example, the processor 104 can input the current three-dimensional positions of each feature point individually obtained from time tm to time t into a Kalman filter (e.g., a linear Kalman filter) so that the Kalman filter can correct the current three-dimensional position of each feature point at time t, but this is not limited to this. Related details can be found in relevant literature on Kalman filters, and will not be elaborated here.

[0055] In one embodiment, after the processor 104 acquires multiple eye feature points of the eyes on a human face according to the above teachings, it can determine the three-dimensional display content of the 3D display based on these eye feature points. For example, the processor 104 can activate the lenticular lens on the 3D display and adjust the pixel positions on the 3D display. Details related to this can be found in existing literature on 3D rendering, and will not be elaborated further here. This ensures that a user in front of the 3D display will not see a three-dimensional image with severe 3D crosstalk due to an unreliable imaging component.

[0056] In the embodiments of the present invention, although the above description uses two imaging components (i.e. Figure 1 The first imaging component 101 and the second imaging component 102) and three feature points (i.e., Figure 3 Taking feature points A, B, and C as an example, the concept of the present invention can also be applied to scenarios with more imaging components and more feature points in other embodiments, and is not limited to the above implementation methods.

[0057] Furthermore, although the above embodiments are illustrated using a 3D display as an example, the concepts of the embodiments of the present invention can be applied to any mechanism for detecting the three-dimensional position of feature points, and are not limited to 3D displays.

[0058] In summary, the embodiments of the present invention can first obtain the relative positions of multiple feature points on a specific object with respect to each imaging component, and when it is determined that a certain imaging component is unreliable, estimate the current three-dimensional position of each feature point based on its historical three-dimensional position and the relative position of another reliable imaging component. This ensures that users in front of a 3D display will not see a 3D image with severe 3D crosstalk due to the unreliability of a certain imaging component.

[0059] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A feature point position detection method, suitable for an electronic device including a first imaging component and a second imaging component, characterized in that, The method includes: Obtain multiple first relative positions of multiple feature points on a specific object relative to the first imaging component; Obtain the multiple second relative positions of the plurality of feature points on the specific object relative to the second imaging component; and In response to the determination that the first imaging component is unreliable, the step of estimating the current three-dimensional position of each feature point based on the historical three-dimensional position of each feature point and the plurality of second relative positions, wherein the plurality of second relative positions includes a unit vector corresponding to each feature point, and the step of estimating the current three-dimensional position of each feature point based on the historical three-dimensional position of each feature point and the plurality of second relative positions includes: The first distance between the multiple feature points is obtained based on the historical three-dimensional position of each feature point. Based on the unit vector corresponding to each of the feature points and the first distance between the plurality of feature points, the second distance between the second imaging component and each of the feature points is estimated; The current three-dimensional position of each feature point is estimated based on the three-dimensional position of the second imaging component and the second distance corresponding to each feature point.

2. The method of claim 1, wherein the step of obtaining the plurality of first relative positions of the plurality of feature points on the specific object relative to the plurality of first imaging components comprises: The first imaging component captures a first image of the specific object; The plurality of feature points are identified in the first image, and the plurality of feature points are determined relative to the plurality of first relative positions of the first imaging component.

3. The method according to claim 2, comprising: The first imaging component is deemed unreliable if the number of the plurality of feature points in the first image is lower than a preset threshold. as well as The first imaging component is deemed reliable if the number of the plurality of feature points in the first image is not less than the preset threshold.

4. The method according to claim 1, further comprising: In response to the determination that both the first imaging component and the second imaging component are reliable, the current three-dimensional position of each feature point is estimated based on the plurality of first relative positions and the plurality of second relative positions.

5. The method according to claim 1, wherein the plurality of feature points includes a first feature point, a second feature point, and a third feature point, the second imaging component has a first unit vector, a second unit vector, and a third unit vector respectively corresponding to the first feature point, the second feature point, and the third feature point, and the step of estimating the second distance between the second imaging component and each of the feature points based on the unit vectors corresponding to each of the feature points and the first distance between the plurality of feature points includes: Multiple relationships are established based on the first unit vector, the second unit vector, the third unit vector, the first distance between the first feature point and the second feature point, the first distance between the second feature point and the third feature point, the second distance between the second imaging component and the first feature point, the second distance between the second imaging component and the second feature point, and the second distance between the second imaging component and the third feature point. Based on the multiple relationships, estimate the second distance between the second imaging component and the first feature point, the second distance between the second imaging component and the second feature point, and the second distance between the second imaging component and the third feature point.

6. The method according to claim 5, wherein the plurality of relations comprises: in The first unit vector corresponding to the first feature point. The second unit vector corresponding to the second feature point. Let a be the first unit vector corresponding to the third feature point, b be the first distance between the second feature point and the third feature point, c be the first distance between the first feature point and the second feature point, x be the second distance between the second imaging component and the first feature point, y be the second distance between the second imaging component and the second feature point, and z be the second distance between the second imaging component and the third feature point.

7. The method according to claim 5, wherein the plurality of second relative positions are obtained at the t-th time point, the historical three-dimensional position of each of the feature points is obtained at the tk-th time point, t is an index value, k is a positive integer, and the relative positions of the first feature point, the second feature point and the third feature point to each other are constant between the t-th time point and the tk-th time point.

8. The method of claim 1, wherein the electronic device is a three-dimensional display, and the first image-capturing component and the second image-capturing component belong to a dual-pupil camera on the three-dimensional display.

9. The method of claim 8, wherein the specific object is a face, and after the step of estimating the current three-dimensional position of each of the feature points based on the historical three-dimensional positions of each of the feature points and the plurality of second relative positions, further comprising: Obtain multiple eye feature points corresponding to the eyes on the face; as well as The 3D display content of the 3D display is determined based on the multiple eye feature points.

10. An electronic device, characterized in that, include: First image-capturing component; Second imaging component; as well as A processor, coupled to the first image-capturing component and the second image-capturing component, and configured to execute: Obtain multiple first relative positions of multiple feature points on a specific object relative to the first imaging component; Obtain the multiple second relative positions of the multiple feature points on the specific object relative to the second imaging component; as well as In response to the determination that the first imaging component is unreliable, the current three-dimensional position of each feature point is estimated based on the historical three-dimensional position of each feature point and the plurality of second relative positions, wherein the plurality of second relative positions includes a unit vector corresponding to each feature point, and the estimation of the current three-dimensional position of each feature point based on the historical three-dimensional position of each feature point and the plurality of second relative positions includes: The first distance between the multiple feature points is obtained based on the historical three-dimensional position of each feature point. Based on the unit vector corresponding to each of the feature points and the first distance between the plurality of feature points, the second distance between the second imaging component and each of the feature points is estimated; The current three-dimensional position of each feature point is estimated based on the three-dimensional position of the second imaging component and the second distance corresponding to each feature point.

Images