CN114663527A - Camera self-calibration method under general motion - Google Patents
Camera self-calibration method under general motion Download PDFInfo
- Publication number
- CN114663527A CN114663527A CN202210284800.5A CN202210284800A CN114663527A CN 114663527 A CN114663527 A CN 114663527A CN 202210284800 A CN202210284800 A CN 202210284800A CN 114663527 A CN114663527 A CN 114663527A
- Authority
- CN
- China
- Prior art keywords
- camera
- calibration
- curve
- obtaining
- absolute
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/80—Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
- G06T7/248—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments involving reference images or patches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/66—Analysis of geometric attributes of image moments or centre of gravity
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- Multimedia (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
The invention discloses a camera self-calibration method under motion, which comprises seven steps. Compared with the prior art, the invention has the advantages that: the geometric relation between the Steiner curve and the absolute quadratic curve of the symmetrical part of the basic matrix under general motion is utilized, the complete calibration of five internal parameters of the camera and the calibration of six external parameters of the camera can be automatically realized, the operation of the calibration process is simple, only at least three pictures of general motion need to be shot in one calibration, the time of the calibration operation of the camera is saved, and the scheme has the advantages of simplicity, convenience in calculation and accurate result.
Description
Technical Field
The invention relates to the technical field of three-dimensional vision, in particular to a camera self-calibration method under general motion.
Background
In the field of three-dimensional vision, it is often necessary to accurately calibrate internal and external parameters of a camera so as to accurately perform operations such as subsequent photogrammetry, autonomous navigation based on vision, motion estimation, three-dimensional reconstruction and the like. However, in the method for completing camera self-calibration under general motion in actual calculation, the original technology neglects the inherent constraint of the basic matrix, and only part of camera parameters can be calculated by using the Kruppa equation; for five unknown camera parameters, there are still five possible solutions of the quintic power of two in the five quadratic equations. And the calculation accuracy is not accurate enough. Other methods attempt to simplify the Kruppa equation by eliminating the scale factors through some specific operations, however, there is still some ambiguity in the obtained camera auto-calibration constraints and the results are not accurate enough.
Disclosure of Invention
In order to solve the technical problems, the technical scheme provided by the invention is as follows: a camera self-calibration method under general motion comprises the following steps:
step one, a basic matrix is obtained through feature point correspondence;
step two, decomposing the basic matrix to obtain a conical quadratic curve Steiner curve Fs and a fixed point xa;
step three, obtaining general characteristic vector constraints of the absolute quadratic curve image and the Fs;
step four, obtaining initial estimation of three internal references and absolute secondary curve images of the camera by general feature vector constraint;
acquiring a circular ring point from the intersection point of the absolute quadratic curve image and Fs, and then acquiring initial solutions of two vanishing lines and two circle centers;
step six, establishing an objective function optimization double circle center;
and seventhly, obtaining the internal parameters of the camera by the optimal solution of the double circle centers.
Compared with the prior art, the invention has the advantages that: the geometric relation between the Steiner curve and the absolute quadratic curve of the symmetrical part of the basic matrix under general motion is utilized, the complete calibration of five internal parameters of the camera and the calibration of six external parameters of the camera can be automatically realized, the operation of the calibration process is simple, only at least three pictures of general motion need to be shot in one calibration, the time of the calibration operation of the camera is saved, and the scheme has the advantages of simplicity, convenience in calculation and accurate result.
Drawings
Fig. 1 is a schematic diagram of a second step in a camera self-calibration method under general motion.
Fig. 2 is a schematic diagram of step three in a camera self-calibration method under general motion.
Fig. 3 is a schematic diagram of step five of the camera self-calibration method under general motion.
Fig. 4 is a schematic diagram of step six of the camera self-calibration method under general motion.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings.
In the embodiment shown in fig. 1 to 4, the present invention provides a method for calibrating a camera under general motion, which includes the following steps:
step one, a basic matrix is obtained through feature point correspondence;
step two, decomposing the basic matrix to obtain a conical secondary curve Steiner curve Fs and a fixed point xa;
step three, obtaining general characteristic vector constraints of the absolute quadratic curve images and Fs;
step four, obtaining initial estimation of three internal references and absolute secondary curve images of the camera by general feature vector constraint;
acquiring a circular ring point from the intersection point of the absolute quadratic curve image and Fs, and then acquiring initial solutions of two vanishing lines and two circle centers;
step six, establishing an objective function optimization double circle center;
and seventhly, obtaining the internal parameters of the camera by the optimal solution with double circle centers.
Wherein:
1. at least three images with overlapping regions are taken of the scene to be measured. And selecting two adjacent image pairs, and acquiring high-precision feature points and matching by using a feature extraction and matching method such as SIFT. Based on the matching feature points in the image pair, the exact basis matrix F is estimated using the eight-point method or other methods, as well as the RANSAC method.
2. In the embodiment shown in FIG. 1, the fundamental matrix F can be decomposed into symmetrical parts Fs=(F+FT) /2 and asymmetric part Fa=(F-FT)/2. Wherein the symmetrical part FsIs a circleConic Steiner curve, asymmetric part FaIs an antisymmetric matrix, which can be written as Fa=[xa]×Thus point xaIs FaThe zero vector of (2). Meanwhile, the zero vector of the basic matrix F can obtain the polar point pair { e, e' }, which falls on the quadratic curve FsAbove, their connecting line laSatisfy about FsPole line quadrature constraint ofa=Fsxa. To summarize, in the geometric representation of the fundamental matrix F, the symmetrical part FsIs a conic section with F as asymmetrical partaZero vector of (2), i.e. point xa. The pair of poles { e, e' } falls on the quadratic curve FsAbove, they are the zero vectors of the fundamental matrix F, their connecting lines laSatisfy la ═ la=Fsxa。
3. In the embodiment shown in FIG. 2, let the absolute conic image be ω, then v⊥=ω*laIn the middle line laAnd point v⊥Is a pair of epipolar lines and poles with respect to the absolute conic section image omega. Connection point v⊥And point xaA fixed shaft l can be obtaineds. ω and the symmetrical part F of the basic matrixsForm omega*FsThe general eigenvector corresponding to the maximum eigenvalue is the point v1It is located on the line laThe other two general feature vectors are points v2And v3They are located atsOn the image of (a), the absolute conic image ω and the symmetric part F of the fundamental matrixsThe common eigenvector of (A) can form a common extreme triangle Deltav1v2v3. Wherein, with ω and FsGeneral eigenvector v corresponding to the largest eigenvalue1On the line laThe other two general feature vectors v2And v3On the axis lsAbove, and lsAnd laAbout ω quadrature, i.e. ls Tω*la=0。
4. Due to omega*FsGeneral eigenvector v corresponding to maximum eigenvalue1On the line laIn the above, the following constraints can be obtained,
three independent constraints are included. When the main shaft point (u) is known0,v0) Then, three unknown parameters, i.e., f, (v) can be recovered from the equation1x,v1y). Where f is the focal length of the natural camera, (v)1x,v1y) Is a generic feature vector v1The coordinates of (a). Using at least three images to obtain three image pairs, using and not limited to the above constraints, for camera parameters with known principal axis points, three camera parameters can be recovered, namely focal length fx,fyAnd a skew parameter s, and an initial estimate of the absolute conic image can be obtained
5. In the embodiment shown in FIG. 3, the image is due to an absolute quadratic curveAnd a symmetrical part F of a pair of image elementary matrices FsThere are two pairs of imaginary intersections, which are the circle points ii,jiAnd i is 1 and 2. Suppose FsIs the projection of a circle on a plane in space, and two vanishing lines l are corresponding to the ambiguity of the normal direction of the planehiAnd i is 1 and 2. Their intersection v1On the line laUpper, and two pairs of circle points ii,jiRespectively located on the two vanishing lines. Due to the vanishing line lhiAbout F by projection from the centre of a circlesWith polar line pole constraint, two circular projection can be recoveredAnd the center of the circleOn the axis lsUpper, ω and a pair of image momentsSymmetrical part F of matrix FsThere are two pairs of virtual intersections, which are circular points, located on the two vanishing lines brought by the plane normal ambiguity, respectively.
6. In the embodiment shown in fig. 4, the double center of the Fs is optimized: initial estimation at two centers of a circle respectivelyUniformly selecting a plurality of sampling points in the nearby areaFixed shaft image can be obtained by connecting a pair of sampling pointsFrom the center of a circleEpipolar pole constraint computation on Fs corresponds to two horizontal linesFurther in a horizontal lineTwo pairs of circular points are obtained from two pairs of imaginary intersections with FsThey satisfy
This constraint can be the calculation of an absolute quadratic curve image4 independent constraints are provided. Thus obtaining more circle points using at least three images and absolute conic image constraintsCalculated by the method in step 4Are points, respectivelyAnd calculating a relative absolute conic sectionPole of (2)Constructing an objective function, which may include, but is not limited to, the following constraints: 1) x is the number ofa,On the shaftThe above step (1); 2) dotOn-line laC, removing; 3) lsAnd laQuadrature with respect to ω; llAnd lrRegarding ω orthogonality, the objective function brought about by the above constraints is optimized to solve an optimal solution for the center oi (i ═ 1, 2).
7. Calculating a corresponding horizontal line l with respect to the epipolar pole constraint of Fs from the optimal solution of the center oi (i ═ 1,2)hi(ii) a Further on a horizontal line lhiTwo pairs of imaginary intersections with Fs obtain a circle point ii,jiI is 1, 2; they satisfy
It can provide 4 independent constraints for calculating the absolute quadratic curve image omega. Thus at least 3 images forming 3 image pairs will contain 6 pairs of circle points which provide 12 independent constraints to calculate the absolute quadratic curve image omega. The 5 internal parameters of the camera are obtained by Cholesky decomposition ω. The camera external parameters are thus obtainable from the decomposed essential matrix. Using more images will improve the accuracy of the camera self-calibration algorithm.
While there have been shown and described the fundamental principles and principal features of the invention and advantages thereof, it will be understood by those skilled in the art that the invention is not limited by the embodiments described above, which are given by way of illustration of the principles of the invention, but is susceptible to various changes and modifications without departing from the spirit and scope of the invention as defined by the appended claims. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (1)
1. A camera self-calibration method under general motion is characterized by comprising the following steps:
step one, a basic matrix is obtained through feature point correspondence;
step two, decomposing the basic matrix to obtain a conical quadratic curve Steiner curve Fs and a fixed point xa;
step three, obtaining general characteristic vector constraints of the absolute quadratic curve images and Fs;
step four, obtaining initial estimation of three internal references and absolute secondary curve images of the camera by general feature vector constraint;
step five, obtaining a circular point by the intersection point of the absolute secondary curve image and Fs, and then obtaining initial solutions of two vanishing lines and two circle centers;
step six, establishing an objective function optimization double circle center;
and seventhly, obtaining the internal parameters of the camera by the optimal solution of the double circle centers.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210284800.5A CN114663527A (en) | 2022-03-22 | 2022-03-22 | Camera self-calibration method under general motion |
PCT/CN2022/083071 WO2023178658A1 (en) | 2022-03-22 | 2022-03-25 | Camera self-calibration method under general motion |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210284800.5A CN114663527A (en) | 2022-03-22 | 2022-03-22 | Camera self-calibration method under general motion |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114663527A true CN114663527A (en) | 2022-06-24 |
Family
ID=82031880
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210284800.5A Pending CN114663527A (en) | 2022-03-22 | 2022-03-22 | Camera self-calibration method under general motion |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN114663527A (en) |
WO (1) | WO2023178658A1 (en) |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6437823B1 (en) * | 1999-04-30 | 2002-08-20 | Microsoft Corporation | Method and system for calibrating digital cameras |
CN104167001B (en) * | 2014-08-27 | 2017-02-15 | 大连理工大学 | Large-visual-field camera calibration method based on orthogonal compensation |
CN106530358A (en) * | 2016-12-15 | 2017-03-22 | 北京航空航天大学 | Method for calibrating PTZ camera by using only two scene images |
CN109064516B (en) * | 2018-06-28 | 2021-09-24 | 北京航空航天大学 | Camera self-calibration method based on absolute quadratic curve image |
-
2022
- 2022-03-22 CN CN202210284800.5A patent/CN114663527A/en active Pending
- 2022-03-25 WO PCT/CN2022/083071 patent/WO2023178658A1/en unknown
Also Published As
Publication number | Publication date |
---|---|
WO2023178658A1 (en) | 2023-09-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112102458A (en) | Single-lens three-dimensional image reconstruction method based on laser radar point cloud data assistance | |
CN108648240B (en) | Non-overlapping view field camera attitude calibration method based on point cloud feature map registration | |
CN106558080B (en) | Monocular camera external parameter online calibration method | |
CN109165680B (en) | Single-target object dictionary model improvement method in indoor scene based on visual SLAM | |
CN107833181B (en) | Three-dimensional panoramic image generation method based on zoom stereo vision | |
CN110264528B (en) | Rapid self-calibration method for binocular camera with fish-eye lens | |
CN110853075B (en) | Visual tracking positioning method based on dense point cloud and synthetic view | |
CN109754459B (en) | Method and system for constructing human body three-dimensional model | |
CN111612731B (en) | Measuring method, device, system and medium based on binocular microscopic vision | |
CN107818598B (en) | Three-dimensional point cloud map fusion method based on visual correction | |
WO2024045632A1 (en) | Binocular vision and imu-based underwater scene three-dimensional reconstruction method, and device | |
CN113361365B (en) | Positioning method, positioning device, positioning equipment and storage medium | |
CN111127401A (en) | Robot stereoscopic vision mechanical part detection method based on deep learning | |
CN113706381A (en) | Three-dimensional point cloud data splicing method and device | |
CN116051766A (en) | External planet surface environment reconstruction method based on nerve radiation field | |
CN113240597B (en) | Three-dimensional software image stabilizing method based on visual inertial information fusion | |
Wang et al. | Lrru: Long-short range recurrent updating networks for depth completion | |
CN110838146A (en) | Homonymy point matching method, system, device and medium for coplanar cross-ratio constraint | |
CN113393524A (en) | Target pose estimation method combining deep learning and contour point cloud reconstruction | |
CN110555880B (en) | Focal length unknown P6P camera pose estimation method | |
CN112288813A (en) | Pose estimation method based on multi-view vision measurement and laser point cloud map matching | |
CN114663527A (en) | Camera self-calibration method under general motion | |
CN114399547B (en) | Monocular SLAM robust initialization method based on multiframe | |
CN108595373B (en) | Uncontrolled DEM registration method | |
CN111339342A (en) | Three-dimensional model retrieval method based on angle ternary center loss |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |