Internal reference calibration method of catadioptric camera based on principal line property

Abstract

The invention discloses an internal reference calibration method of a catadioptric camera based on principal line properties, and belongs to the technical field of computer vision and machine vision calibration, and the method comprises the following steps: S1, obtaining four antipodal image points in each image; s2, obtaining two groups of orthogonal vanishing points in each image; s3, constructing a main line based on the vanishing points and estimating a camera main point; s4, screening bad images; and S5, determining the focal length of the camera. According to the internal reference calibration method of the catadioptric camera based on the main line property, the problems that an existing internal reference calibration method of the catadioptric camera depends on a mirror surface geometric model, a complex calibration structure or high-precision feature detection, the robustness is insufficient, and operation is tedious are solved.

Description

An Intrinsic Parameter Calibration Method for a Catalytic Reflection Camera Based on Principal Line Properties Technical Field

[0001] This invention belongs to the field of computer vision and machine vision calibration technology, specifically relating to an intrinsic parameter calibration method for catadioptric cameras based on principal axis properties. Background Technology

[0002] In fields such as computer vision, panoramic imaging, 3D measurement, and industrial inspection, camera calibration is a crucial and fundamental step connecting 3D physical space with a 2D image plane. Its core objective is to accurately estimate intrinsic parameters such as principal point position and focal length. Accurate intrinsic parameters are a prerequisite for ensuring the accuracy of subsequent tasks such as attitude estimation, 3D reconstruction, and panoramic stitching. If the calibration parameters are deviated, it will directly lead to problems such as target localization distortion, reconstructed structural distortion, and stitching failure.

[0003] However, existing technologies have the following shortcomings: the surface parameters of the mirror (such as the curvature of parabolic or hyperboloid surfaces) need to be obtained in advance, but there are errors in the mirror processing, and inaccurate parameters will directly reduce the calibration accuracy; a complex three-dimensional calibration device is required, which is costly and has a complicated operation process, making it difficult to deploy quickly in industrial settings; the traditional checkerboard corner extraction method is used, but there are no obvious corners on the quadratic curves in the catadioptric images, making corner extraction difficult and lacking robustness; and the acquired distorted / noisy images are not filtered, which can interfere with the calibration results, resulting in insufficient noise resistance and stability.

[0004] Therefore, a new method is urgently needed. Summary of the Invention

[0005] The purpose of this invention is to provide an intrinsic parameter calibration method for catadioptric cameras based on principal properties. This method uses multi-dimensional technical means to avoid interference from mirror processing errors to ensure calibration accuracy, significantly reduce equipment costs and operational complexity, achieve rapid on-site deployment, adapt to the catadioptric imaging rules to improve feature extraction robustness, and enhance the stability and noise resistance of calibration results.

[0006] To achieve the above objectives, this invention provides an intrinsic parameter calibration method for a catadioptric camera based on principal line properties, comprising the following steps: S1, placing the planar calibration plate in front of the catadioptric camera, changing the camera's orientation to acquire multiple images, performing edge extraction and fitting on the four quadratic curve arcs formed by the projection of parallel straight lines in each image, solving for the intersection points of each set of quadratic curves, and obtaining four antipodal image points; S2, receiving the four antipodal image points of each image output from S1; connecting the four antipodal image points in a preset order to construct a quadrilateral, obtaining the first set of orthogonal vanishing points and determining the vanishing line by finding the intersection points of corresponding sides of the quadrilateral, and then finding the intersection points of the quadrilateral's diagonal with the vanishing line to obtain the second set of orthogonal vanishing points; S3, receiving... S2 outputs two sets of orthogonal vanishing points for each image; constructs principal line equations based on each set of orthogonal vanishing points, collects multiple principal lines from multiple images, and estimates camera principal points using the least squares solution method; S4 receives the camera principal points and principal line parameters for each image output by S3; calculates the geometric distance from the camera principal points to each principal line, removes images corresponding to principal lines whose distances are greater than a preset threshold, and updates principal point estimates based on the retained images; S5 receives the updated camera principal points output by S4 and the two sets of orthogonal vanishing points corresponding to the images retained after filtering; given the updated camera principal points, simplifies the camera intrinsic parameter matrix, constructs a system of equations using the absolute quadratic curve algebraic constraints satisfied by the orthogonal vanishing points, and solves to obtain the camera focal length parameters.

[0007] Preferably, in S1, the arc of the quadratic curve is fitted using the least squares method or other fitting methods, and the complete equation of the quadratic curve is expressed as: In the formula, These are the coefficients of a quadratic curve.

[0008] Preferably, in S2, the preset connection order of the four antipodal image points is based on the spatial relationship of the straight lines on the calibration board, corresponding to the connection order of the square formed by the top points on the unit sphere; the vanishing line is determined by the cross product operation of the first set of orthogonal vanishing points, and the expression is: In the formula, These are the first set of orthogonal vanishing points; It is the vanishing line of the image plane.

[0009] Preferably, the principal line equation is constructed based on each group of orthogonal vanishing points, specifically: the direction of the principal line is determined by the coordinate difference of the same group of orthogonal vanishing points, expressed as: ; In the formula, Here are the coordinates of the first vanishing point. The coordinates of the second vanishing point are given; the principal intercept is obtained by jointly calculating another set of orthogonal vanishing points, and the final principal equation is in the form of: In the formula, Let these be the coordinates of the camera's principal point. The main line intercept.

[0010] Preferably, in S3, the specific method for solving the least squares problem is as follows: collect n principal lines from n images and construct a matrix. sum matrix , is represented as: ; In the formula, For the first The parameters of the main line, The principal points of the camera are solved using least squares, using the following formula: .

[0011] Preferably, in S4, the formula for calculating the geometric distance from the principal point to the principal line is: In the formula, For the first The distance from the main line to the main point.

[0012] Preferably, in S5, the simplified form of the camera intrinsic parameter matrix is: In the formula, Let be the camera focal length parameter to be solved.

[0013] Preferably, in S5, the specific expression for the algebraic constraint of the absolute conic section is: ; In the formula, There are two sets of orthogonal vanishing points. Internal parameter matrix The transpose inverse matrix, Internal parameter matrix The inverse matrix; the system of linear equations constructed based on the constraints is: In the formula, These are the coordinates of the first set of orthogonal vanishing points.

[0014] Therefore, the present invention adopts the above-mentioned intrinsic parameter calibration method for catadioptric cameras based on principal line properties. Compared with the prior art, the technical solution of the present invention has the following beneficial effects: (1) It adopts the logical chain of "quadratic curve-opposite image point-principal line constraint", which does not require the technical means of obtaining the surface parameters of the reflector in advance. This overcomes the technical problems of the prior art relying on the geometric model of the mirror surface and the influence of mirror surface processing error on the calibration accuracy, thereby achieving the technical effect of avoiding interference from mirror surface processing error and ensuring calibration accuracy; (2) It adopts the technical means of using only a planar calibration plate containing two sets of orthogonal parallel lines, which overcomes the technical problems of the prior art relying on complex three-dimensional calibration objects, high equipment cost, cumbersome operation and difficulty in on-site deployment, thereby achieving a significant reduction in equipment cost. (3) By abandoning the traditional corner detection and using the technique of quadratic curve fitting and solving the metatarsal image point, the technical problem that the traditional corner detection is not applicable and the feature extraction is not robust due to the fact that the spatial straight line projection of the catadioptric camera is a quadratic curve is overcome, thereby achieving the technical effect of adapting to the catadioptric imaging law and improving the robustness of feature extraction; (4) By using the technique of removing distorted / noisy images through the "geometric distance threshold from the principal point to the principal line", the technical problem that the existing technology lacks a bad data screening mechanism, distorted / noisy images interfere with the calibration results, and the stability and noise resistance are insufficient, thereby achieving the technical effect of improving the stability and noise resistance of the calibration results.

[0015] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0016] Figure 1 is a flowchart of an embodiment of the intrinsic parameter calibration method for a catadioptric camera based on principal line properties according to the present invention; Figure 2 is a schematic diagram of parallel straight lines forming a quadratic curve under catadioptric imaging according to an embodiment of the intrinsic parameter calibration method for a catadioptric camera based on principal line properties according to the present invention; Figure 3 is a schematic diagram of orthogonal parallel straight lines forming a square at their top points on a unit sphere according to an embodiment of the intrinsic parameter calibration method for a catadioptric camera based on principal line properties according to the present invention; Figure 4 is a schematic diagram of quadratic curve fitting and vanishing point solving according to an embodiment of the intrinsic parameter calibration method for a catadioptric camera based on principal line properties according to the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Unless otherwise defined, the technical or scientific terms used in the present invention should have the ordinary meaning understood by those skilled in the art.

[0018] As shown in Figures 1-4, this embodiment provides an intrinsic parameter calibration method for a catadioptric camera based on principal properties. It should be understood that the specific parameters, models, and protocols mentioned in this embodiment are merely examples to help those skilled in the art understand the present invention, and are not intended to limit the present invention.

[0019] This invention discloses an intrinsic parameter calibration method for a catadioptric camera based on principal line properties, comprising the following steps: S1. Placing the planar calibration plate in front of the catadioptric camera, changing the camera's pose to acquire multiple images, and performing edge extraction and fitting on the four quadratic curve arcs formed by the projection of parallel straight lines in each image, solving for the intersection points of each set of quadratic curves, and obtaining four antipodal image points. Specifically, in a catadioptric camera, a spatial straight line, after being reflected back through the light path of a parabolic mirror and a lens, will form a quadratic curve arc on the image plane, usually a part of an elliptical arc. Two sets of mutually orthogonal parallel straight lines on the calibration plate are placed in front of the camera, and the camera's pose is changed to acquire multiple images. In each image, the first set of parallel straight lines generates two quadratic curve arcs, and the second set of parallel straight lines generates another two quadratic curve arcs. Edge extraction is performed on the four quadratic curve arcs obtained by the projection of parallel straight lines in each image, and the general quadratic curve equation is fitted using the least squares method or other fitting methods. In the formula, The coefficients of the quadratic curves are given. For two quadratic curves belonging to the same set of parallel lines, solving for their intersection point yields the projection of the paired points of these lines onto the image plane in the unit sphere. Since the calibration plate contains a set of orthogonal parallel lines, four paired image points can be obtained for each image. S2. Receive four antipodal image points from each image output by S1; connect the four antipodal image points in a preset order to construct a quadrilateral; obtain the first set of orthogonal vanishing points and determine the vanishing line by finding the intersection points of corresponding sides of the quadrilateral; then find the intersection points of the diagonals of the quadrilateral and the vanishing line to obtain the second set of orthogonal vanishing points. Specifically, under the unit sphere model of the catadioptric camera, the top points of two sets of orthogonal parallel lines form a square on the unit sphere, with their diagonal directions remaining orthogonal, and this orthogonal structure is maintained after projection. The "direction" of the two sets of orthogonal lines on the unit sphere is mapped to two pairs of top points on the sphere, wherein the first set of parallel lines corresponds to a pair of top points on the sphere. The second set of parallel lines corresponds to another pair of top points on the sphere. In the unit sphere model, these four points necessarily form a square structure on a sphere. Therefore, although the lines of a catadioptric camera are not straight lines, by pairing the top points, the vanishing point structure that is completely consistent with that of a pinhole camera can be recovered. This invention utilizes this property to transform the problem of solving the vanishing point of a catadioptric camera into pinhole camera geometry. By pairing the four top points in the image according to the connection order of the unit sphere square, four straight lines are constructed, represented as: ; ; ; In the formula, , , , These are the sides of the quadrilateral; the first group of orthogonal vanishing points is represented as: ; In the formula, These are the first set of orthogonal vanishing points; the vanishing line is determined by the first set of vanishing points, and is represented as: In the formula, Let be the vanishing line of the image plane; then take the diagonal of the antipodal point, represented as: ; In the formula, , These are the diagonals of the quadrilateral; thus, we obtain two sets of vanishing points in orthogonal directions. S3. Receive two sets of orthogonal vanishing points for each image output from S2; construct the principal line equation based on each set of orthogonal vanishing points, collect multiple principal lines from multiple images, and estimate the camera principal points using the least squares solution method; in this step, let the recovered principal line in a certain image be: In the formula, Let be the homogeneous coefficient vector of the line; the direction of the principal line is determined by the coordinate difference of the same set of orthogonal vanishing points, expressed as: ; In the formula, Here are the coordinates of the first vanishing point. The coordinates of the second vanishing point are given; the principal intercept is obtained by jointly calculating another set of orthogonal vanishing points, and the principal equation is in the form of: ; In the formula, Let these be the coordinates of the camera's principal point. Main line intercept; ( 、( The second set of orthogonal vanishing points; collect n principal lines from n images and construct a matrix. sum matrix , is represented as: ; In the formula, For the first The parameters of the main line, The principal points of the camera are solved using least squares, using the following formula: In the formula, S4: Receive the camera principal point coordinates and principal line parameters of each image output from S3; calculate the geometric distance from the camera principal point to each principal line, discard images corresponding to principal lines whose distance is greater than a preset threshold, and update the principal point estimation based on the retained images; in this step, the formula for calculating the geometric distance from the principal point to the principal line is: In the formula, For the first The distance from the main line to the principal point; if If the value exceeds the threshold, the image is removed and the principal point is re-estimated to improve robustness; S5: Receive the updated camera principal points output from S4 and the two sets of orthogonal vanishing points corresponding to the retained images after filtering; Given the updated camera principal points, simplify the camera intrinsic parameter matrix, construct a system of equations using the algebraic constraints of the absolute quadratic curves satisfied by the orthogonal vanishing points, and solve to obtain the camera focal length parameters; In this step, the simplified form of the camera intrinsic parameter matrix is: In the formula, Let be the camera focal length parameter to be solved; the specific expression for the algebraic constraint of the absolute conic section is: ; In the formula, There are two sets of orthogonal vanishing points. Internal parameter matrix The transpose inverse matrix, Internal parameter matrix The inverse matrix; the system of linear equations constructed based on the constraints is: In the formula, These are the coordinates of the first set of orthogonal vanishing points.

[0020] Therefore, the present invention adopts the above-mentioned intrinsic parameter calibration method of catadioptric camera based on principal properties. This method uses multi-dimensional technical means to avoid interference from mirror processing errors to ensure calibration accuracy, significantly reduce equipment costs and operational complexity, achieve rapid on-site deployment, adapt to the catadioptric imaging law to improve feature extraction robustness, and enhance the stability and noise resistance of calibration results.

[0021] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0022] 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 preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for calibrating the intrinsic parameters of a catadioptric camera based on principal axis properties, characterized in that, Includes the following steps: S1. Place the planar calibration plate in front of the catadioptric camera, change the camera's orientation to acquire multiple images, and perform edge extraction and fitting on the four quadratic curve arcs formed by the projection of parallel straight lines in each image. Solve for the intersection points of each set of quadratic curves to obtain four antipodal image points; S2. Receive the four antipodal image points of each image output from S1; connect the four antipodal image points in a preset order to construct a quadrilateral, obtain the first set of orthogonal vanishing points and determine the vanishing line by finding the intersection points of corresponding sides of the quadrilateral, and then find the intersection points of the quadrilateral's diagonal with the vanishing line to obtain the second set of orthogonal vanishing points. S3. Receive two sets of orthogonal vanishing points for each image output from S2; construct principal line equations based on each set of orthogonal vanishing points, collect multiple principal lines from multiple images, and estimate camera principal points using the least squares solution method; S4. Receive camera principal points and principal line parameters for each image output from S3; calculate the geometric distance from the camera principal point to each principal line, discard images corresponding to principal lines whose distance is greater than a preset threshold, and update principal point estimation based on the retained images; S5. Receive the updated camera principal points output from S4 and the two sets of orthogonal vanishing points corresponding to the filtered and retained images; Given the updated camera principal points, simplify the camera intrinsic parameter matrix, construct a system of equations using the absolute quadratic curve algebraic constraints satisfied by the orthogonal vanishing points, and solve for the camera focal length parameters.

2. The intrinsic parameter calibration method for a catadioptric camera based on principal axis properties according to claim 1, characterized in that, In S1, the quadratic curve arc is fitted using the least squares method or other fitting methods, and the complete equation of the quadratic curve is expressed as: In the formula, These are the coefficients of a quadratic curve.

3. The intrinsic parameter calibration method for a catadioptric camera based on principal axis properties according to claim 1, characterized in that, In S2, the preset connection order of the four antipodal image points is based on the spatial relationship of the straight lines on the calibration board, corresponding to the connection order of the square formed by the top points on the unit view sphere; the vanishing line is determined by the cross product operation of the first set of orthogonal vanishing points, and the expression is: In the formula, These are the first set of orthogonal vanishing points; It is the vanishing line of the image plane.

4. The intrinsic parameter calibration method for a catadioptric camera based on principal axis properties according to claim 1, characterized in that, In S3, the principal line equation is constructed based on each set of orthogonal vanishing points. Specifically, the direction of the principal line is determined by the coordinate difference of the same set of orthogonal vanishing points, expressed as follows: ； In the formula, Here are the coordinates of the first vanishing point. The coordinates of the second vanishing point are given; the principal intercept is obtained by jointly calculating another set of orthogonal vanishing points, and the final principal equation is in the form of: In the formula, Let these be the coordinates of the camera's principal point. The main line intercept.

5. The intrinsic parameter calibration method for a catadioptric camera based on principal axis properties according to claim 1, characterized in that, In S3, the specific method for solving the least squares problem is as follows: collect the n principal lines of n images and construct a matrix. sum matrix , is represented as: ； In the formula, For the first The parameters of the main line, The principal points of the camera are solved using least squares, using the following formula: 。 6. The intrinsic parameter calibration method for a catadioptric camera based on principal axis properties according to claim 1, characterized in that, In S4, the formula for calculating the geometric distance from the principal point to the principal line is: In the formula, For the first The distance from the main line to the main point.

7. The intrinsic parameter calibration method for a catadioptric camera based on principal axis properties according to claim 1, characterized in that, In S5, the simplified form of the camera intrinsic parameter matrix is: In the formula, Let be the camera focal length parameter to be solved.

8. The intrinsic parameter calibration method for a catadioptric camera based on principal axis properties according to claim 1, characterized in that, In S5, the specific expression for the algebraic constraint of the absolute conic section is: ； In the formula, There are two sets of orthogonal vanishing points. Intrinsic parameter matrix The transpose inverse matrix, Intrinsic parameter matrix The inverse matrix; the system of linear equations constructed based on the constraints is: In the formula, These are the coordinates of the first set of orthogonal vanishing points.

9. A computer device, characterized in that, include: A processor configured to be coupled to memory, read and execute instructions and / or program code in the memory to perform the method as described in any one of claims 1-8.

10. A computer-readable medium, characterized in that, The computer-readable medium stores computer program code that, when executed on a computer, causes the computer to perform the method as described in any one of claims 1-8.

Images