Projection-guided calibration method based on stereo vision

Abstract

The present application provides a projection-guided calibration method based on stereo vision, which comprises projecting a first image onto a reference surface to obtain a second image, wherein the first image includes a plurality of first identification points; photographing the second image to obtain a third image, wherein the third image has a plurality of third identification points; traversing and solving the Euclidean distance between the first identification point and the third identification point to obtain a matching point set; inputting the matching point set into a mapping function to obtain a homography matrix; photographing a calibration plate to obtain a first calibration image; inputting the first calibration image and the homography matrix into an inversion function to obtain a second calibration image; monotonically calibrating the first calibration image to obtain a first extrinsic parameter matrix and a first distortion vector of a camera device; monotonically calibrating the second calibration image to obtain a second extrinsic parameter matrix and a second distortion vector of a projection device; inputting the first extrinsic parameter matrix and the second extrinsic parameter matrix into a conversion function to obtain a rotation matrix and a translation matrix between a projection device coordinate system and a camera device coordinate system.

Description

Technical Field

[0001] The present application relates to the field of assembly manufacturing technology, and in particular to a projection-guided calibration method based on stereoscopic vision. Background Art

[0002] Aerospace products, such as satellites, are usually highly complex and have many types of assembly parts. Currently, manual assembly is the main method of satellite assembly. In order to improve the difficulty of accessing paper documents, low manual assembly efficiency, and easy omission of installation requirements, engineering and technical personnel use augmented reality technology, visual and projection methods, and use projection devices to directly project graphics or text on the assembly object to guide workers' operations.

[0003] In the application process of this type of satellite assembly projection guidance system, it is necessary to calibrate the relative position of the "vision-projection" device composed of a camera device and a projection device, calculate the relative relationship between the coordinate system of the camera device and the coordinate system of the projection device, use the camera device to locate the actual assembly object, and then calculate the relative posture relationship between the projection device and the assembly object based on the calibration results, so as to accurately project the view of the parts to be assembled to the target position to achieve a good virtual-reality fusion effect. Therefore, the calibration of the "vision-projection" device is a prerequisite for realizing projection process guidance.

[0004] The current calibration method is generally a manual marking method, which uses artificial markers set up on site for calibration. It is necessary to stick and arrange two-dimensional or three-dimensional markers within the projection range of the projection device, and then solve the problem after measuring the relative position of the projection mark and the artificial marker. The disadvantage of this method is that it is necessary to arrange the markers according to the rules and perform measurements, which is time-consuming. Summary of the invention

[0005] In view of the above defects or deficiencies in the prior art, the present application aims to provide a projection-guided calibration method based on stereoscopic vision, comprising the following steps:

[0006] Setting a reference plane, wherein the reference plane is set in a common field of view of the camera device and the projection device, and the Z axes of the camera device coordinate system and the projection device coordinate system are both oriented toward the reference plane;

[0007] Projecting a first image onto the reference surface to obtain a second image, wherein the first image includes a plurality of first identification points;

[0008] photographing the second image to obtain a third image, wherein the third image has a plurality of third identification points;

[0009] Traversing and solving the Euclidean distance between the first identification point and the third identification point to obtain a matching point set, wherein the matching point set includes at least four pairs of the first identification point and the corresponding third identification point whose Euclidean distance is less than a set threshold;

[0010] Inputting the matching point set into a mapping function to obtain a homography matrix of the first image and the third image;

[0011] Setting a calibration plate under the field of view of the camera device;

[0012] photographing the calibration plate to obtain a first calibration image;

[0013] Inputting the first calibration image and the homography matrix into an inversion function to obtain a second calibration image, wherein the coordinate system of the second calibration image is the image coordinate system of the projection device;

[0014] The first calibration image is calibrated by a single object to obtain a first extrinsic parameter matrix and a first distortion vector of the camera device; the second calibration image is calibrated by a single object to obtain a second extrinsic parameter matrix and a second distortion vector of the projection device;

[0015] The first extrinsic parameter matrix and the second extrinsic parameter matrix are input into a conversion function to obtain a rotation matrix and a translation matrix between the projection device coordinate system and the camera device coordinate system.

[0016] According to the technical solution provided in the embodiment of the present application, after acquiring the first calibration image, the first calibration image and the first distortion vector are input into a correction function to perform distortion correction on the first calibration image. The correction function includes a radial correction function and a tangential correction function. The radial correction function is as follows:

[0017]

[0018] Among them, r 2 =x 2 +y 2 , k1, k2, k3 are radial distortion parameters.

[0019] The tangential correction function is as follows:

[0020]

[0021] Among them, p1, p2, p3 are tangential distortion parameters, (x', y') are the coordinates after radial distortion, and (x", y") are the coordinates after tangential correction.

[0022] According to the technical solution provided in the embodiment of the present application, after obtaining the second calibration image, the second calibration image and the second distortion vector are input into the correction function to perform distortion correction on the second calibration image.

[0023] According to the technical solution provided in the embodiment of the present application, one of the elements in the matching point set is (P pn , P cn ), P pn is the coordinate of the first identification point in the image coordinate system of the projection device, P cn is the coordinate of the third identification point in the camera coordinate system (2); set (P pn , P cn )=((X pn , Y pn , 1), (X cn , Y cn , 1));

[0024] The mapping function is shown in the following formula:

[0025]

[0026] The coordinates of the first identification points and the third identification points in the matching point set are input into formula (3) to obtain the homography matrix H. The homography matrix H is as follows:

[0027]

[0028] According to the technical solution provided in the embodiment of the present application, the first external parameter matrix includes the first translation matrix T of the camera device C and the first rotation matrix R C ; The second external parameter matrix includes the second translation matrix T of the projection device P and the second rotation matrix R P ; Set the marker point P in the world coordinate system W , the marking point P W The coordinate in the projection device coordinate system is X P , the coordinate in the camera coordinate system is X C ,

[0029] The conversion function is shown in the following formula:

[0030]

[0031] Among them, R C is the first rotation matrix, T C is the first translation matrix, R P is the second rotation matrix, TP is the second translation matrix;

[0032] Solving the conversion function, we obtain:

[0033] R=R C R P -1 X P

[0034] T=T C -R C R P -1 T C

[0035] Among them, R is the rotation matrix, and T is the translation matrix.

[0036] According to the technical solution provided in the embodiment of the present application, the calibration plate is photographed multiple times to obtain multiple first calibration images, thereby obtaining multiple second calibration images, and single-target calibration is performed on the multiple first calibration images and the second calibration images to obtain multiple rotation matrices and translation matrices. Then, the corresponding element values ​​in the multiple rotation matrices and the translation matrices are averaged to obtain the final rotation matrix and the translation matrix.

[0037] According to the technical solution provided in the embodiment of the present application, the inversion function is:

[0038] M=Hm (5)

[0039] Wherein, M is the coordinate of any point of the second calibration image, m is the coordinate of any point of the first calibration image, and H is the homography matrix.

[0040] In summary, the present application proposes a projection-guided calibration method based on stereoscopic vision, in which a first image is projected onto a reference surface by a projection device, the projection image is photographed by a camera device to obtain a third image, the identification points of the first image and the third image are traversed to obtain a matching point set, the matching point set is input into a mapping function, and the transformation relationship between the images in the image coordinate system of the camera device and the projection device, namely, the homography matrix, is obtained; the camera device photographs a calibration plate to obtain a first calibration image, and a second calibration image of the projection device is obtained through the homography matrix, and the external parameter matrix of the projection device of the camera device is obtained by performing single-target calibration on the first calibration image and the second calibration image, and the two external parameter matrices are input into the transformation function to obtain a rotation matrix and a translation matrix between the projection device coordinate system and the camera device coordinate system, thereby obtaining the relationship between the two coordinate systems; thereby, the camera device, the projection device and the two are calibrated; compared with the prior art, there is no need to set specific identification points, and the calibration speed is fast. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 A flowchart of the steps of the projection-guided calibration method based on stereoscopic vision provided in an embodiment of the present application;

[0042] Figure 2 A schematic diagram of the structure of a projection-guided calibration device for stereoscopic vision provided in an embodiment of the present application;

[0043] Figure 3 This is a schematic diagram of the structure of the device for calibrating the camera and projector of this application.

[0044] The text annotations in the figure represent:

[0045] 1. Camera device; 2. Camera device coordinate system; 3. Projection device; 4. Projection device coordinate system; 5. Reference surface; 6. Reference coordinate system; 7. Second image; 8. Calibration plate. DETAILED DESCRIPTION

[0046] The present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. It is also necessary to explain that, for ease of description, only the parts related to the invention are shown in the accompanying drawings.

[0047] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0048] As mentioned in the background technology, in order to solve the problems in the prior art, the present application proposes a projection-guided calibration method based on stereoscopic vision, comprising the following steps:

[0049] S101, set a reference plane 5, the reference plane 5 is set in the common field of view of the camera device 1 and the projection device 3, and the Z axes of the camera device coordinate system 2 and the projection device coordinate system 4 are both facing the reference plane 5; specifically, the camera device 1 is a color industrial camera, and the projection device 3 is a color projector. Before calibration, the camera and the projector are fixedly connected, and optionally, the camera is fixed on the top of the projector, so that when the camera and the projector move, the relative relationship between the projection device coordinate system 4 and the camera device coordinate system 2 does not change; Figure 2 As shown, the origin of the camera coordinate system 2 is the center of the camera, and the coordinate axes are X C , Y C , Z C The origin of the projection device coordinate system 4 is the center of the projector, and the coordinate axes are X P , Y P , Z P, the coordinate system where the reference surface 5 is located is the reference coordinate system 6, the origin of the reference coordinate system 6 is a corner of the reference surface 5, and the coordinate axes are X a , Y a , Z a , the corresponding coordinate axes of the projection device coordinate system 4 and the camera device coordinate system 2 are oriented in the same direction, and the Z axis is oriented toward the reference plane, but is not necessarily perpendicular to the reference plane 5;

[0050] S102, projecting a first image onto the reference surface to obtain a second image 7, wherein the first image includes a plurality of first identification points; wherein the coordinate system of the first image is the image coordinate system of the projection device 3, such as Figure 2 As shown, the second image 7 is a projected image, and there is a second identification point corresponding to the first identification point on the second image; the first identification point is a representative point in the first image, optionally, a corner point, an edge point, etc. in the first image. In some embodiments, whether it is the first identification point is determined by the gray gradient of the image plane. When the gray gradient of two adjacent points changes greatly, the intersection thereof is the first identification point;

[0051] S103, photographing the second image to obtain a third image, wherein the third image has a plurality of third identification points; the coordinate system of the third image is the image coordinate system of the camera device 1, the second image is photographed by the camera device 1 to obtain the third image, and the recognition method of the third identification points is the same as the recognition method of the first identification points;

[0052] S104, traverse and solve the Euclidean distance between the first identification point and the third identification point to obtain a matching point set, the matching point set including at least four pairs of the first identification points and the corresponding third identification points whose Euclidean distances are less than a set threshold; wherein, the matching point set is obtained by using a SURF (Speeded Up Robust Features) optimization algorithm, the first identification point and the third identification point have their own coordinates in the image coordinate system of the projection device 3 and the camera device 1, firstly, the feature vectors of the first identification point and the third identification point are calculated, and the Euclidean distance between each of the first identification point and each of the third identification point is calculated, the smaller the Euclidean distance is, the better the matching degree between the two is; set a set threshold of the Euclidean distance, and when it is determined that the Euclidean distance between the first identification point and the third identification point is less than the set threshold, the two identification points are determined to be matching points;

[0053] S105, inputting the matching point set into a mapping function to obtain a homography matrix of the first image and the third image;

[0054] Among them, one of the elements in the matching point set is (P pn , P cn ), P pn is the coordinate of the first identification point in the image coordinate system of the projection device 3, P cn is the coordinate of the third identification point in the image coordinate system of the camera device 1; set (P pn , P cn )=((X pn , Y pn , 1), (X cn , Y cn , 1));

[0055] The mapping function is shown in the following formula:

[0056]

[0057] The coordinates of the first identification points and the third identification points in the matching point set are input into formula (3) to obtain the homography matrix H. The homography matrix H is as follows:

[0058]

[0059] For the convenience of calculation, let h 33 is 1, then 8 values ​​need to be solved in H, then at least 4 pairs of matching points are required, and 8 equations are established to obtain the homography matrix H. In some embodiments, the matching point set is:

[0060]

[0061] Substitute the matching point set into formula (3) and solve for the conversion relationship:

[0062]

[0063] S106, setting a calibration plate 8 in the field of view of the camera device 1; wherein the coordinate system of the calibration plate 8 is the world coordinate system, and there is no requirement for the form of the calibration plate 8, as long as the intrinsic parameter matrix can be solved. Optionally, Figure 3 As shown, the calibration plate 8 is a checkerboard calibration plate 8, and the checkerboard calibration plate is composed of a plurality of black and white squares; the plane where the calibration plate 8 is located does not need to be perpendicular to the camera device coordinate system 2 and the projection device coordinate system 4,

[0064] S107, photographing the calibration plate 8 to obtain a first calibration image; wherein the coordinate system of the first calibration image is the image coordinate system of the camera device 1;

[0065] S108, inputting the first calibration image and the homography matrix into the inversion function to obtain a second calibration image, wherein the coordinate system of the second calibration image is the image coordinate system of the projection device 3; since the homography matrix represents the relationship between the two images in the image coordinate systems of the projection device 3 and the camera device 1, and the first calibration image and the second calibration image are also two images in the image coordinate systems of the projection device 3 and the camera device 1, the process of obtaining the second calibration image through the first calibration image and the homography matrix can be regarded as an inverse calculation process of obtaining the homography matrix;

[0066] Furthermore, the inversion function is:

[0067] M=Hm (5)

[0068] Wherein, M is the coordinate of any point of the second calibration image, m is the coordinate of any point of the first calibration image; H is the homography matrix; since the coordinate system of the first calibration image is the image coordinate system of the camera device 1, and its coordinates are two-dimensional coordinates, the two-dimensional coordinates are first converted into homogeneous coordinates, and then multiplied with the homography matrix to obtain the second calibration image in the image coordinate system of the projection device 3;

[0069] S109, monocularly calibrate the first calibration image to obtain a first extrinsic parameter matrix and a first distortion vector of the camera device 1; monocularly calibrate the second calibration image to obtain a second extrinsic parameter matrix and a second distortion vector of the projection device 3; wherein the first distortion vector includes a first radial distortion parameter and a first tangential distortion parameter, and the second distortion vector includes a second radial distortion parameter and a second tangential distortion parameter;

[0070] Furthermore, the first external parameter matrix includes a first translation matrix T of the camera device 1 C and the first rotation matrix R C The second external parameter matrix includes the second translation matrix T of the projection device 3 P and the second rotation matrix R P In some embodiments, the Zhang Zhengyou camera calibration method is used to obtain the intrinsic and extrinsic parameters of the camera and the projector, and the coordinates of a point on the plane of the calibration plate 8 are set to N, and the coordinates of the camera imaging plane are set to n. Substitute this coordinate into the following Zhang Zhengyou camera calibration formula:

[0071] S0n=K[R C ,T C ]N (6)

[0072] Wherein, S0 is the scale factor, K is the first internal parameter of the camera device 1, and R C is the first rotation matrix, TC is the first translation matrix; similarly, the second rotation matrix R of the projection device 3 is calculated by Zhang Zhengyou camera calibration method P and the second translation matrix T P ;

[0073] S110, inputting the first extrinsic parameter matrix and the second extrinsic parameter matrix into a conversion function to obtain a rotation matrix and a translation matrix between the projection device coordinate system 4 and the camera device coordinate system 2;

[0074] Among them, set the marker point P in the world coordinate system W , the marking point P W The coordinate in the projection device coordinate system 4 is X P , the coordinate in the camera coordinate system 2 is X C ,

[0075] The conversion function is shown in the following formula:

[0076]

[0077] Among them, R C is the first rotation matrix, T C is the first translation matrix, R P is the second rotation matrix, T P is the second translation matrix;

[0078] Eliminate Pw in the above formula and calculate:

[0079] X c =R c R p -1 X p +T c -R c R p -1 T c

[0080] The above formula can be expressed as:

[0081] X c =RX p +T

[0082] Wherein, R is the rotation matrix, and T is the translation matrix;

[0083] So we get R = R c R p -1 , T=T c -R c R p -1 Tc ; Thus, the conversion relationship between the projection device coordinate system 4 and the camera device coordinate system 2 is obtained; thus, the camera device 1, the projection device 3 and the two are calibrated; when in use, the image of the spare parts on the assembly plane is obtained by the camera device 1, and the conversion relationship between the projection device coordinate system 4 and the camera device coordinate system 2 obtained by this scheme is used to obtain the relationship between the parts on the assembly plane and the projection device coordinate system 4, so that the real object can be projected onto the assembly plane through the projection device 3, thereby achieving the purpose of guiding assembly.

[0084] Further, after acquiring the first calibration image, the first calibration image and the first distortion vector are input into a correction function to perform distortion correction on the first calibration image, the correction function includes a radial correction function and a tangential correction function, and the radial correction function is as follows:

[0085]

[0086] Among them, r 2 =x 2 +y 2 , k1, k2, k3 are radial distortion parameters.

[0087] The tangential correction function is as follows:

[0088]

[0089] Among them, p1, p2, p3 are tangential distortion parameters, (x', y') are coordinates after radial distortion, and (x", y") are coordinates after tangential correction; since the lens of the camera device 1 is not completely parallel to the image plane during the manufacturing process, tangential distortion is generated, and the light is more bent away from the center of the lens than near the center, resulting in radial distortion, so the camera device 1 needs to be corrected for distortion; the radial distortion parameter of the first calibration image is the first radial distortion parameter, and the tangential distortion parameter is the first tangential distortion parameter. After the radial correction function and the tangential correction function, the coordinate relationship between the corrected image pixel coordinates and the image pixel before correction is obtained.

[0090] Furthermore, after obtaining the second calibration image, the second calibration image and the second distortion vector are input into the correction function to perform distortion correction on the second calibration image; the distortion correction method of the second calibration image is the same as that of the first calibration image, the radial distortion parameter of the second calibration image is the second radial distortion parameter obtained in the single-target positioning, and the tangential distortion parameter thereof is the second tangential distortion parameter obtained in the single-target positioning; after performing distortion correction on the first calibration image and the second calibration image, a corrected first corrected image corresponding to the first calibration image and a second corrected image corresponding to the second calibration image are obtained; then single-target positioning is performed on the first corrected image and the second corrected image respectively to obtain a corrected first external parameter matrix and a second external parameter matrix, and the corrected first external parameter matrix and the second external parameter matrix are input into the conversion function to obtain the final rotation matrix and the translation matrix.

[0091] Further, the calibration plate 8 is photographed multiple times to obtain multiple first calibration images, thereby obtaining multiple second calibration images, and single-target calibration is performed on the multiple first calibration images and the second calibration images. After obtaining multiple rotation matrices and translation matrices, the corresponding element values ​​in the multiple rotation matrices and the translation matrices are averaged to obtain the final rotation matrix and the translation matrix; in order to improve the accuracy of single-target calibration, the first calibration image needs to be collected multiple times, and the calibration plate 8 does not need to be perpendicular to the Z axis of the projection device coordinate system 4 and the camera device coordinate system 2. In the process of collecting the first calibration images multiple times, the position and posture of the calibration plate 8 are adjusted in the world coordinate system, so that the camera device 1 has a better position and imaging quality when photographing the calibration plate 8; in this embodiment, 10 first calibration images are obtained and marked as Pc1~Pc10, and the coordinate system where the first calibration image is located is the image coordinate system of the camera device 1; through the homography matrix, the first calibration image is converted to the second calibration image in the image coordinate system of the projection device 3, which is set as P p1 ~P p10 ; By performing single-target calibration on the first calibration image and the second calibration image, the first external parameter matrix and the first translation matrix R corresponding to the 10 images of the calibration plate 8 are obtained. C1 / T C1 ~R C10 / T C10 , the second external parameter matrix and the second translation matrix R p1 / T p1 ~R p10 / T p10, put it into formula (4) to calculate 10 rotation matrices and translation matrices, and take the average value of the corresponding elements in the matrix to obtain the final rotation matrix and translation matrix.

[0092] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and its core ideas of this application. The above is only the preferred implementation method of this application. It should be pointed out that due to the limitations of textual expression and the objective existence of infinite specific structures, ordinary technicians in this technical field can make several improvements, modifications or changes without departing from the principles of the present invention, and can also combine the above technical features in an appropriate manner; these improvements, modifications, changes or combinations, or the direct application of the inventive concept and technical solution to other occasions without improvement, should be regarded as the scope of protection of this application.

Claims

1. A projection-guided calibration method based on stereo vision, characterized in that: The steps include: A reference plane (5) is set, wherein the reference plane (5) is set in a common field of view of the camera device (1) and the projection device (3), and the Z axes of the camera device coordinate system (2) and the projection device coordinate system (4) are both oriented toward the reference plane (5); Projecting a first image onto the reference surface (5) to obtain a second image (7), wherein the first image includes a plurality of first identification points; photographing the second image to obtain a third image, wherein the third image has a plurality of third identification points; Traversing and solving the Euclidean distance between the first identification point and the third identification point to obtain a matching point set, wherein the matching point set includes at least four pairs of the first identification point and the corresponding third identification point whose Euclidean distance is less than a set threshold; Inputting the matching point set into a mapping function to obtain a homography matrix of the first image and the third image; Setting a calibration plate (8) under the field of view of the camera device (1); photographing the calibration plate (8) to obtain a first calibration image; Inputting the first calibration image and the homography matrix into an inversion function to obtain a second calibration image, wherein the coordinate system of the second calibration image is the image coordinate system of the projection device (3); The first calibration image is calibrated by a single object to obtain a first extrinsic parameter matrix and a first distortion vector of the camera device (1); the second calibration image is calibrated by a single object to obtain a second extrinsic parameter matrix and a second distortion vector of the projection device (3); The first external parameter matrix and the second external parameter matrix are input into a conversion function to obtain a rotation matrix and a translation matrix between the projection device coordinate system (4) and the camera device coordinate system (2).

2. The projection-guided calibration method based on stereoscopic vision according to claim 1, characterized in that: After acquiring the first calibration image, the first calibration image and the first distortion vector are input into a correction function to perform distortion correction on the first calibration image. The correction function includes a radial correction function and a tangential correction function. The radial correction function is as follows: Among them, (x, y) is the coordinate of the image pixel point, r 2 =x 2 +y 2 , k1, k2, k3 are radial distortion parameters; The tangential correction function is as follows: Among them, p1, p2 are tangential distortion parameters, (x', y') are the coordinates after radial distortion, and (x", y") are the coordinates after tangential correction.

3. The projection-guided calibration method based on stereoscopic vision according to claim 2, characterized in that: After the second calibration image is obtained, the second calibration image and the second distortion vector are input into the correction function to perform distortion correction on the second calibration image.

4. The projection-guided calibration method based on stereoscopic vision according to claim 3, characterized in that: One of the elements in the matching point set is (P pn , P cn ), P pn is the coordinate of the first identification point in the image coordinate system of the projection device (3), cn is the coordinate of the third identification point in the image coordinate system of the camera device (1); setting (P pn , P cn )=((X pn , Y pn , 1), (X cn , Y cn , 1)), where (X pn , Y pn , 1) is the homogeneous coordinate form of the corresponding point on the imaging surface of the projection device (3), (X cn , Y cn , 1) is the homogeneous coordinate form of the corresponding point on the imaging surface of the camera device (1); The mapping function is shown in the following formula: The coordinates of at least four pairs of matching first identification points and third identification points in the matching point set are input into formula (3) to obtain the homography matrix H, which is as follows:

5. The projection-guided calibration method based on stereoscopic vision according to claim 4, characterized in that: The first external parameter matrix includes a first translation matrix T of the camera device (1) C and the first rotation matrix R C The second external parameter matrix includes the second translation matrix T of the projection device (3) P and the second rotation matrix R P ; Set the marker point P in the world coordinate system W , the marking point P W The coordinate in the projection device coordinate system (4) is X P , the coordinate in the camera coordinate system (2) is X C , The conversion function is shown in the following formula: Among them, R C is the first rotation matrix, T C is the first translation matrix, R P is the second rotation matrix, T P is the second translation matrix; Solving the conversion function, we obtain: R=R C R P -1 X P T=T C -R C R P -1 T C Among them, R is the rotation matrix, and T is the translation matrix.

6. The projection-guided calibration method based on stereoscopic vision according to claim 5, characterized in that: The calibration plate (8) is photographed multiple times to obtain multiple first calibration images, thereby obtaining multiple second calibration images, and single-target calibration is performed on the multiple first calibration images and the second calibration images to obtain multiple rotation matrices and translation matrices, and then the corresponding element values ​​in the multiple rotation matrices and the translation matrices are averaged to obtain the final rotation matrix and the translation matrix.

7. The guided calibration method based on stereoscopic vision according to claim 6, characterized in that: The inversion function is: M=Hm (5) Wherein, M is the coordinate of any point of the second calibration image, m is the coordinate of any point of the first calibration image, and H is the homography matrix.

Images