CN112465916A - RGBD binocular calibration method and system based on full-view-field plane calibration plate - Google Patents
RGBD binocular calibration method and system based on full-view-field plane calibration plate Download PDFInfo
- Publication number
- CN112465916A CN112465916A CN202011351521.3A CN202011351521A CN112465916A CN 112465916 A CN112465916 A CN 112465916A CN 202011351521 A CN202011351521 A CN 202011351521A CN 112465916 A CN112465916 A CN 112465916A
- Authority
- CN
- China
- Prior art keywords
- checkerboard
- matrix
- corner
- energy
- growth
- 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
- 238000000034 method Methods 0.000 title claims abstract description 46
- 238000001514 detection method Methods 0.000 claims abstract description 24
- 239000011159 matrix material Substances 0.000 claims description 147
- 230000036544 posture Effects 0.000 claims description 9
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000007792 addition Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 239000003550 marker Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
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
- G06T7/85—Stereo camera calibration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
The invention discloses an RGBD binocular calibration method based on a full-view plane calibration plate, which is characterized in that at least 6 images of the plane calibration plate under different poses of a camera to be calibrated are acquired, the plane calibration plate comprises a checkerboard and three marking dots, the checkerboard comprises a plurality of black and white unit cells, and the size of the plane calibration plate is far larger than the view field of the camera; adopting a growth-based checkerboard corner detection algorithm for the images, acquiring pixel coordinates of each checkerboard corner in each image, establishing a relative position relationship between the checkerboard corners, and establishing a one-to-one correspondence relationship between the pixel coordinates of the checkerboard corners and world coordinates; acquiring pixel coordinates of a marked dot in each image, establishing a relative position relation between the marked dot and a checkerboard corner point, and determining the origin and the direction of world coordinates; and calculating the internal reference, distortion and external reference of the camera to be calibrated and the relative rotation and translation of the binocular camera according to the Zhang calibration method. The invention improves the camera calibration efficiency and accuracy.
Description
Technical Field
The invention relates to the technical field of computer vision, in particular to an RGBD binocular calibration method and system based on a full-view plane calibration plate.
Background
In the using process of the RGBD camera, the RGB image and the depth image need to be registered, the process depends on the binocular calibration result of the RGB module and the depth module, and the more accurate the binocular calibration result is, the better the registration effect is. In the RGBD camera binocular calibration process, calibration object images in pairs and different poses need to be acquired. The camera calibration method based on the plane calibration plate is commonly used in terms of comprehensive consideration of equipment cost, calibration precision, calibration time consumption and the like. In the prior art, the size of a calibration object is mostly a plane calibration plate far smaller than the view field of a camera. On the one hand, images of different postures and multiple positions need to be collected to obtain a calibration result with higher accuracy on the basis of the plane calibration plate, so that the distribution sum of the characteristic points of the calibration plate in the images can cover the whole view field of the camera; on the other hand, most feature detection algorithms require consistent number and arrangement of feature points of images at each pose, so that one-to-one correspondence can be established. Therefore, the technical scheme needs to acquire a large number of images, and the number of the feature points of each image is consistent with the relative position relationship.
Disclosure of Invention
Based on the method, the RGBD binocular calibration method and the system based on the full-view-field plane calibration plate are provided, the plane calibration plate with the size far larger than the view field of a camera is adopted, the corresponding relation between pixel coordinates and world coordinates is established based on a grown checkerboard corner detection algorithm, the characteristic points of different postures can be distributed over the view field only by acquiring a small number of images, and the calibration efficiency and accuracy are improved.
In order to achieve the above object, the present invention provides an RGBD binocular calibration method based on a full view plane calibration plate, the method comprising:
s1, at least acquiring images of a plane calibration plate of 6 cameras to be calibrated under different poses, wherein the plane calibration plate comprises a checkerboard containing a plurality of black and white unit cells and three marking dots, and the size of the plane calibration plate is far larger than the field of view of the cameras;
s2, adopting a growth-based checkerboard corner detection algorithm to the images, obtaining the pixel coordinates of each checkerboard corner in each image, establishing the relative position relationship between the checkerboard corners, and establishing the one-to-one correspondence relationship between the pixel coordinates of the checkerboard corners and the world coordinates;
s3, acquiring the pixel coordinates of the marked dots in each image, establishing the relative position relationship between the marked dots and the checkerboard corner points, and determining the origin and the direction of world coordinates;
and S4, calculating the internal reference, distortion and external reference of the camera to be calibrated and the relative rotation and translation of the binocular camera according to the Zhang calibration method.
Preferably, the step S1 includes:
acquiring plane calibration plate images under postures of dead facing, pitching 15 degrees and deflecting 15 degrees at a position of a Z axis corresponding to the nearest depth of field of a camera to be calibrated, and acquiring 3 images in total;
and acquiring plane calibration plate images under the postures of dead, pitching of-15 degrees and deflecting of-15 degrees at the position of the Z axis corresponding to the central depth of field of the camera to be calibrated, and acquiring 3 images in total.
Preferably, the step S2 includes:
s201, growing leftwards by taking the leftmost three columns of the rectangular checkerboards as seed checkerboards to obtain left checkerboards;
s202, executing the same method as the step S201, respectively taking the rightmost three columns, the topmost three columns and the bottommost three columns of the rectangular checkerboards as seed checkerboards, and growing rightwards, upwards and downwards to obtain a right checkerboard, an upper checkerboard and a lower checkerboard;
s203, taking the uppermost 3 rows of the left checkerboard and the leftmost 3 columns of the upper checkerboard as a transverse seed checkerboard and a longitudinal seed checkerboard, and carrying out upper left region growth to obtain an upper left checkerboard;
s204, executing the same method as the step S203, taking the lowest 3 rows of the left checkerboard, the leftmost 3 columns of the lower checkerboard, the topmost 3 rows of the right checkerboard, the rightmost 3 columns of the upper checkerboard, the lowest 3 rows of the right checkerboard and the rightmost 3 columns of the lower checkerboard as transverse seed checkerboards and longitudinal seed checkerboards respectively, and performing growth of a left lower area, a right upper area and a right lower area to obtain a left lower checkerboard, a right upper checkerboard and a right lower checkerboard;
s205, growing leftwards and upwards by taking the leftmost three columns and the rightmost three columns of the left upper checkerboard as seed checkerboards respectively to obtain a left upper left checkerboard and a left upper checkerboard, and performing left upper regional growth by taking the left upper left checkerboard and the left upper left checkerboard as seed checkerboards to obtain a left upper checkerboard;
s206, executing the same method as the step S205 to obtain a left lower left checkerboard, a right upper right checkerboard, a right upper checkerboard, a right upper right checkerboard, a right lower right checkerboard and a right lower right checkerboard;
and S207, splicing the checkerboard grids to obtain extra-grown checkerboard grids.
Preferably, the step S201 specifically includes:
s2011, if the seeds grow leftwards, turning the seed checkerboard leftwards and rightwards, and if the seeds grow upwards, turning the seed checkerboard downwards;
s2012, recording the corner index which does not participate in the growth at present, the initial maximum structure energy and the current checkerboard corner number, obtaining the index of the growth in each row and column direction of the seed checkerboard, and calculating the structure energy of the seed checkerboard, if the structure energy meets the following conditions:
the structural energy < (the number of angular points of the current checkerboard + the maximum structural energy + 1)/(the number of angular points of the current checkerboard), if the growth is effective, the angular point index and the structural energy are recorded, the maximum structural energy is updated to the maximum value of the maximum structural energy and the calculated structural energy, the number of the angular points of the current checkerboard is added by one, and the current index in the angular point index is deleted;
s2013, constructing an index matrix, a second dimension energy matrix and a first dimension energy matrix according to the maximum value of the number of growing corners of each row, filling the corner indexes obtained in the step 2012 into the index matrix, and filling the obtained structural energy into the second dimension energy matrix;
s2014, constructing a first-dimension energy matrix, wherein the continuous length and the structural energy of the first-dimension energy matrix are taken as the basis;
s2015, deleting the columns of which all elements in the index matrix are zero, left-right inverting the index matrix, if the index matrix grows upwards, transposing the index matrix, and turning the index matrix up and down, if the index matrix grows downwards, transposing the index matrix.
Preferably, the step S2014 includes:
for each column of the index matrix from left to right, if the continuous length of a sequence formed by non-zero corner indexes is less than 3, adding indexes to filled elements in each row taking the corner indexes as starting points, writing corresponding position elements in the index matrix and the index matrix as 0, reducing the number of corner points of the current chessboard by one, and updating the maximum structure energy into the maximum value in the initial maximum structure energy and the second dimension energy matrix until all the corner indexes are traversed;
if the continuous length of a sequence formed by the corner indexes which are not 0 is more than or equal to 3 from left to right in each column of the index matrix, finding a first subsequence which meets the requirement that the continuous length of the sequence is 3 from the starting point of the sequence to the bottom, setting the rows with the corner indexes before the subsequence as the starting point to be zero, writing the corner indexes into the corner indexes which do not participate in growth, reducing the number of the current checkerboard corners by one every time of writing, setting the second dimension energy matrix at the corresponding position to be zero, and updating the maximum structure energy to be the maximum value among the initial maximum structure energy, the second dimension energy matrix and the first dimension energy matrix;
if the continuous length of a sequence formed by the corner indexes which are not 0 is more than or equal to 3 from left to right, finding a first subsequence which does not meet the condition sequence and has the continuous length of 3, setting the rows with the 2 nd and 3 rd corner indexes of the subsequence as the starting points to be zero, writing the corner indexes into the corner indexes which do not participate in growth, reducing the number of the current checkerboard corners by one every time the corner indexes are written, setting the second dimension energy matrix at the corresponding position to be zero, and updating the maximum structure energy to be the maximum value among the initial maximum structure energy, the second dimension energy matrix and the first dimension energy matrix.
Preferably, the step S203 includes:
s2031, if the left lower region grows, rotating the transverse seed checkerboard anticlockwise by 90 degrees, rotating the longitudinal seed checkerboard anticlockwise by 90 degrees, if the right upper region grows, turning the transverse seed checkerboard upside down, and turning the longitudinal seed checkerboard upside down;
s2032, recording the index of the corner points which do not participate in the growth at present, the initial maximum structure energy and the number of the corner points of the current chessboard, and constructing an index matrix;
s2033, setting the initial values of the effective growth height and the effective growth width to be zero, and adding one to the effective growth height if the angular point indexes of the leftmost three lines are not zero for each line of the longitudinal seed checkerboard from top to bottom; for each column of the transverse seed checkerboard from left to right, if the index of the corner point of the top three rows is not zero, the effective growth width is increased by one;
s2034, constructing an effective index matrix, an effective transverse structure energy matrix, an effective longitudinal structure energy matrix and a column source matrix;
s2035, assigning the front effective growth width column of the column source matrix as the front effective growth width column of the 2 nd and 3 rd rows of the transverse seed checkerboard;
s2036, marking the indexes of the rows corresponding to the first 3 columns of the longitudinal seed checkerboard as p1Row, p2Row and p3Row, carrying out the predicted growth in the Row direction by the p1Row, p2Row and p3Row for each Row of the effective index matrix, and calculating the structural energy of the effective index matrix once each growth;
s2037, calculating and predicting whether structural energy formed by the growth points and the corner points of the corresponding serial numbers of the corresponding columns of the column source matrix meets the requirement or not for each row of the effective index matrix, and if not, subtracting one from the width limiting column; otherwise, the value of the corresponding position of the effective transverse structural energy matrix is a row energy value, the value of the corresponding position of the effective longitudinal structural energy matrix is a column energy value, the corresponding position of the effective index matrix is an angular point index of a predicted angular point, a corresponding index in the angular point indexes which do not participate in growth at present is deleted, the number of the angular points of the chessboard grid is accumulated by one, and the maximum structural energy is updated to be the maximum value of the initial maximum structural energy, the effective transverse structural energy matrix and the effective longitudinal structural energy matrix;
s2038, if the effective growth width of each row of the effective index matrix is not zero, only the effective growth width column of the column source matrix is reserved, the 1 st row of the column source matrix is assigned as the 2 nd row of the column source matrix, and the 2 nd row of the column source matrix is assigned as the corresponding row and the front width limit column of the effective index matrix;
s2039, if the area grows in the upper left area, the index matrix is rotated 180 degrees anticlockwise, if the area grows in the lower left area, the index matrix is turned over left and right, and if the area grows in the upper right area, the index matrix is turned over up and down.
Preferably, the step S2 further includes:
taking the plane of the checkerboard as a plane with Z being 0 under a world coordinate system, and establishing the world coordinate and relative position relation of the checkerboard corner points;
and (3) taking the leftmost upper corner of the additionally grown checkerboard as the world coordinate origin, expressing the coordinates from each corner point to the origin as (x, y), and multiplying the (x, y) by the length of the cells of the checkerboard to obtain the world coordinates of each checkerboard corner point.
Preferably, the step S3 includes:
acquiring pixel coordinates of four vertexes of each unit lattice of the checkerboard;
taking the center of the circumscribed rectangle of the four vertexes of each cell as the center, and taking the half of the length and the half of the width of the circumscribed rectangle of the four vertexes of each cell as the length and the width to obtain the region of interest of each cell;
if the maximum gray scale in the region of interest is greater than 3 times of the minimum gray scale and the minimum gray scale is greater than 100, the center of a circle is marked in the cell;
the checkerboard angular point of the connecting line center of 1 pair of marking dots with the circle center distance as the diagonal length of the cell is taken as the world coordinate origin, the connecting line direction of 2 pairs of marking dots with the circle center distance as the side length of the cell is taken as the X direction and the Y direction, and the world coordinate of the checkerboard angular point is reestablished.
In order to achieve the above object, the present invention provides an RGBD binocular calibration system based on a full view plane calibration plate, the system comprising:
the camera calibration device is used for at least acquiring images of 6 planar calibration plates of a camera to be calibrated under different poses, and comprises a planar calibration plate, wherein the planar calibration plate comprises a checkerboard containing a plurality of black and white unit cells and three marking dots, and the size of the planar calibration plate is far larger than the field of view of the camera;
the angular point detection module is used for acquiring the pixel coordinates of each checkerboard angular point in each image by adopting a growth-based checkerboard angular point detection algorithm for the images, establishing the relative position relationship between the checkerboard angular points and establishing the one-to-one correspondence relationship between the pixel coordinates of the checkerboard angular points and world coordinates;
the marked dot detection module is used for acquiring the pixel coordinates of the marked dots in each image, establishing the relative position relationship between the marked dots and the checkerboard corner points, and determining the origin and the direction of world coordinates;
and the Zhang calibration module is used for calculating the internal reference, distortion and external reference of the camera to be calibrated and the relative rotation and translation of the binocular camera according to the Zhang calibration method.
Preferably, the three marking dots are respectively arranged in three different cells in a preset range of a central area of the plane calibration plate, the dots of the three marking dots are respectively positioned at the centers of the respective cells, the three marking dots are arranged in an L shape, the circle center distance of two pairs of marking dots is the side length of the cell, the circle center distance of the other pair of marking dots is the diagonal length of the cell, the marking dots in the white cell are black, and the marking dots in the black cell are white.
Compared with the prior art, the RGBD binocular calibration method and system based on the full-view plane calibration plate have the beneficial effects that: the camera calibration is carried out by adopting the plane calibration object with the size larger than the camera view field, the feature points of different postures can be distributed in the view field only by acquiring a small amount of images, the feature points of each image are not required to be consistent in quantity, the feature points of each image are not required to be in one-to-one correspondence, the calibration efficiency is improved, and the flexibility is higher; the characteristic points in the image of each pose are distributed in the market, the geometric constraint effect is better, the accuracy of the calculation result is better, and the assistance of a calibration device is reduced.
Drawings
Fig. 1 is a schematic flow chart of an RGBD binocular calibration method based on a full-field-of-view planar calibration plate according to an embodiment of the present invention.
Fig. 2 is a system diagram of an RGBD binocular calibration system based on a full field of view planar calibration plate according to an embodiment of the present invention.
Fig. 3 is a schematic diagram of a camera calibration apparatus according to an embodiment of the present invention.
Fig. 4 is a schematic view of a flat calibration plate according to an embodiment of the present invention.
Detailed Description
The present invention will be described in detail with reference to the specific embodiments shown in the drawings, which are not intended to limit the present invention, and structural, methodological, or functional changes made by those skilled in the art according to the specific embodiments are included in the scope of the present invention.
As shown in fig. 1, according to an embodiment of the present invention, the present invention provides an RGBD binocular calibration method based on a full view plane calibration plate, the method including:
s1, at least acquiring images of a plane calibration plate of 6 cameras to be calibrated under different poses, wherein the plane calibration plate comprises a checkerboard containing a plurality of black and white unit cells and three marking dots, and the size of the plane calibration plate is far larger than the field of view of the cameras;
s2, adopting a growth-based checkerboard corner detection algorithm to the images, obtaining the pixel coordinates of each checkerboard corner in each image, establishing the relative position relationship between the checkerboard corners, and establishing the one-to-one correspondence relationship between the pixel coordinates of the checkerboard corners and the world coordinates;
s3, acquiring the pixel coordinates of the marked dots in each image, establishing the relative position relationship between the marked dots and the checkerboard corner points, and determining the origin and the direction of world coordinates;
and S4, calculating the internal reference, distortion and external reference of the camera to be calibrated and the relative rotation and translation of the binocular camera according to the Zhang calibration method.
The size of the plane calibration plate is far larger than that of a camera view field, so that the camera view field only occupies one part of the plane calibration plate under each pose of a camera to be calibrated, namely, only characteristic points exist in the view field, and other objects or background interference does not exist. The calibration plate comprises a checkerboard containing a plurality of black and white unit cells and three marking dots. And acquiring and obtaining images of the plane calibration plate of the camera to be calibrated under different poses by using a camera calibration device, wherein the images are at least 6 images. The motion of the camera to be calibrated comprises a pose formed by three degrees of freedom in a range of Z-axis translation degree of freedom, a range of Y-axis rotation degree of freedom and a range of X-axis rotation degree of freedom. At least 6 images are acquired, each image containing a complete three marker dots. In a specific embodiment of the invention, the planar calibration plate images under the postures of dead facing, pitching 15 degrees and deflecting 15 degrees are collected at the position of the Z axis corresponding to the nearest depth of field of the camera to be calibrated, and 3 images are collected in total; and acquiring plane calibration plate images under the postures of dead, pitching of-15 degrees and deflecting of-15 degrees at the position of the Z axis corresponding to the central depth of field of the camera to be calibrated, and acquiring 3 images in total.
And establishing a corresponding relation between the characteristic point pixels and world coordinates. And according to the characteristic point form of the plane calibration plate, obtaining the pixel coordinates of the characteristic points by adopting a corresponding detection algorithm. The characteristic points of the checkerboard are the checkerboard angular points, and the characteristic points of the marked dots are the circle centers. And establishing a relative position relation of the feature points by combining the arrangement rules among the feature points, and corresponding the pixel coordinates of the feature points to the world coordinates one by one. And adopting a growth-based checkerboard corner detection algorithm for the images to acquire pixel coordinates of each checkerboard corner in each image, establishing a relative position relationship between the checkerboard corners, and establishing a one-to-one correspondence relationship between the pixel coordinates of the checkerboard corners and world coordinates. Acquiring pixel coordinates of all corner points of the checkerboard, direction vectors of all corner points, checkerboard corner point indexes of the grown rectangular checkerboard, the number of corner points of the rectangular checkerboard, the maximum structure energy of the checkerboard and corner point indexes not participating in growth. Based on this, additional growth is performed. Specifically, the method comprises the following steps:
s201, growing leftwards by taking the leftmost three columns of the rectangular checkerboards as seed checkerboards to obtain left checkerboards, and specifically comprising the following steps:
s2011, if the seeds grow leftwards, turning the seed checkerboard leftwards and rightwards, if the seeds grow upwards, turning the seed checkerboard downwards, turning the seed checkerboard leftwards and rightwards, and recording the height of the seed checkerboard as d 1; if the growth is rightward, the height of the seed checkerboard is marked as d 1; if the growth is upward, turning the seed checkerboard up and down, and recording the width of the seed checkerboard as d 1; if the growth is downward, the width of the seed checkerboard is marked as d 1;
s2012, recording the corner index which does not participate in the growth at present, the initial maximum structure energy and the current checkerboard corner number, obtaining the index of the growth in each row and column direction of the seed checkerboard, and calculating the structure energy of the seed checkerboard, if the structure energy meets the following conditions:
structure energy < (number of corner points of the current checkerboard + maximum structure energy + 1)/(number of corner points of the current checkerboard);
if the growth is effective, recording the angular point index and the structural energy, updating the maximum structural energy to be the maximum value of the maximum structural energy and the calculated structural energy, adding one to the number of the angular points of the current checkerboard, and deleting the current index in the angular point index; otherwise, the growth of the row is finished.
S2013, constructing an index matrix, a second dimension energy matrix and a first dimension energy matrix according to the maximum value d2 of the number of growing corners of each row, wherein the height of the index matrix is d1, the width of the index matrix is d2, and if d2 is 0, performing the step 13; otherwise, aligning the corner index obtained in the previous step to the left, filling the corner index into an index matrix, wherein the value of the filling which is not obtained is 0; and aligning the obtained structural energy to the left, filling the obtained structural energy into a second dimension energy matrix, wherein the value of the filling which is not obtained is 0. The first dimension energy matrix has a height d1-2 and a width d 2.
S2014, constructing a first-dimension energy matrix, wherein the continuous length and the structural energy of the first-dimension energy matrix are taken as the basis;
specifically, for each column of the index matrix from left to right, if the continuous length of the sequence formed by the non-zero corner indexes is less than 3, adding indexes to the filled elements in each row with the corner index as the starting point, writing the corresponding position elements in the index matrix and the index matrix as 0, reducing the number of the current checkerboard corners by one, and updating the maximum structure energy to the maximum value in the initial maximum structure energy and the second dimension energy matrix until all the corner indexes are traversed. If the continuous length of the sequence formed by the corner indexes which are not 0 is more than or equal to 3 from left to right, finding a first subsequence which meets the requirement that the continuous length of the sequence is 3 from the starting point of the sequence to the bottom, setting rows with the corner indexes before the subsequence as the starting point to be zero, writing the corner indexes into corner indexes which do not participate in growth, reducing the number of the current checkerboard corners by one every time of writing, setting the second dimension energy matrix at the corresponding position to be zero, updating the maximum structure energy to be the maximum value among the initial maximum structure energy, the second dimension energy matrix and the first dimension energy matrix, and repeatedly executing until all the corner indexes are traversed. If the continuous length of a sequence formed by the corner indexes which are not 0 is more than or equal to 3 from left to right, finding a first subsequence which does not meet the condition sequence and has the continuous length of 3, setting the rows with the 2 nd and 3 rd corner indexes of the subsequence as the starting points to be zero, writing the corner indexes into the corner indexes which do not participate in growth, reducing the number of the current checkerboard corners by one every time the corner indexes are written, setting the second dimension energy matrix at the corresponding position to be zero, and updating the maximum structure energy to be the maximum value among the initial maximum structure energy, the second dimension energy matrix and the first dimension energy matrix.
S2015, deleting columns of which all elements in the index matrix are 0, and turning the index matrix left and right; if the growth is upward, the index matrix is transposed and turned over up and down; if the growth is downward growth, the index matrix is transposed.
S202, executing the same method as the step S201, respectively taking the rightmost three columns, the topmost three columns and the bottommost three columns of the rectangular checkerboards as seed checkerboards, and growing rightwards, upwards and downwards to obtain a right checkerboard, an upper checkerboard and a lower checkerboard;
s203, taking the uppermost 3 rows of the left checkerboard and the leftmost 3 columns of the upper checkerboard as a transverse seed checkerboard and a longitudinal seed checkerboard, and carrying out upper left region growth to obtain an upper left checkerboard. The method specifically comprises the following steps:
s2031, if the growth is in the left lower area, rotating the transverse seed checkerboard 90 degrees anticlockwise, and rotating the longitudinal seed checkerboard 90 degrees anticlockwise; if the growth is in the upper right region, turning the transverse seed checkerboard up and down, and turning the longitudinal seed checkerboard up and down;
s2032, recording the corner index which does not participate in growth at present, the initial maximum structure energy and the current checkerboard corner number, and constructing an index matrix, wherein the height of the index matrix is the height of the transverse seed checkerboard, the width of the index matrix is the width of the longitudinal seed checkerboard, and the content, namely the corner index, is zero;
s2033, setting the initial value of effective growth height to be zero and the initial value of effective growth width to be zero, and for each line of the longitudinal seed checkerboard from top to bottom, if the angular point indexes of the leftmost three lines are not zero, adding the effective growth height, otherwise, not surveying the rest lines; for each column of the transverse seed checkerboard from left to right, if the index of the corner point of the top three rows is not zero, the effective growth width is increased by one, otherwise, the rest columns are not considered;
s2034, constructing an effective index matrix, an effective transverse structure energy matrix, an effective longitudinal structure energy matrix and a column source matrix, wherein the heights are all effective growth heights, the widths are all effective growth widths, the contents are all zero, the initial value of the specified width limit is the effective growth width, and the column source matrix is constructed, the height is 2, and the width is the effective growth width;
s2035, assigning the front effective growth width row of the effective growth width as the front effective growth width row of the 2 nd and 3 rd rows of the transverse seed checkerboard;
s2036, according to the same principle as the growth-based checkerboard corner detection method, for each Row of the effective index matrix, marking the indexes of the rows corresponding to the first 3 columns of the longitudinal seed checkerboard as p1Row, p2Row and p3Row, and marking the counter j as 1; and performing Row-direction predicted growth on each Row of the effective index matrix by using p1Row, p2Row and p3Row to obtain pPredict. And calculating the structural energy of the growth once, if the structural energy does not meet the requirement, reducing the width by 1, and executing a step S2037.
S2037, for each row of the effective index matrix: calculating whether structural energy formed by the predicted growing points and the angular points of the corresponding row corresponding to the serial numbers of the row source matrix meets the requirement or not, and if not, reducing the width limit row by one; otherwise, the value of the corresponding position of the effective transverse structural energy matrix is a row energy value, the value of the corresponding position of the effective longitudinal structural energy matrix is a column energy value, and the corresponding position of the effective index matrix is an angular point index of the prediction angular point;
s2038, if the effective growth width of each row of the effective index matrix is not zero, only the effective growth width column of the column source matrix is reserved, the 1 st row of the column source matrix is assigned as the 2 nd row of the column source matrix, and the 2 nd row of the column source matrix is assigned as the corresponding row and the front width limit column of the effective index matrix;
s2039, if the area grows in the upper left area, the index matrix is rotated by 180 degrees anticlockwise, and if the area grows in the lower left area, the index matrix is turned over left and right; and if the growth is in the upper right area, turning the index matrix up and down.
S204, the same method as the step S203 is carried out, the lowest 3 rows of the left checkerboard, the leftmost 3 columns of the lower checkerboard, the topmost 3 rows of the right checkerboard, the rightmost 3 columns of the upper checkerboard, the bottommost 3 rows of the right checkerboard and the rightmost 3 columns of the lower checkerboard are respectively used as transverse seed checkerboards and longitudinal seed checkerboards, and growth of a left lower area, a right upper area and a right lower area is carried out to obtain a left lower checkerboard, a right upper checkerboard and a right lower checkerboard;
s205, if the width and the height of the upper left checkerboard are both more than or equal to 3 and the index of the angular point is not zero, respectively growing the upper left checkerboard and the upper left checkerboard left and up by using the leftmost three columns and the rightmost three columns of the upper left checkerboard as seed checkerboards according to the same method as the step S201, and growing the upper left checkerboard and the upper left checkerboard left and up by using the upper left checkerboard and the upper left checkerboard left and up as seed checkerboards according to the same method as the step S203 to obtain the upper left checkerboard left and up; on the contrary, the height of the left upper left checkerboard is zero, the width of the left upper left checkerboard is zero, the height of the left upper left checkerboard is zero, and the width and the height of the left upper left checkerboard are both zero;
s206, obtaining a left lower left checkerboard, a right upper right checkerboard, a right upper checkerboard, a right upper right checkerboard, a right lower right checkerboard and a right lower right checkerboard by the same method as the step S205;
s207, splicing the checkerboard grids to obtain extra-grown checkerboard grids as shown in table 1:
| upper left upper chessboard grid | Left upper chessboard grid | 0 | Right upper chessboard grid | Right upper chessboard grid |
| Left upper left checkerboard | Left upper checkerboard | Go up chess board | Right upper checkerboard | Right upper right checkerboard |
| 0 | Left checkerboard | Initial checkerboard | Right checkerboard | 0 |
| Left lower left checkerboard | Left lower checkerboard | Lower checkerboard | Right lower checkerboard | Right lower right checkerboard |
| Left lower checkerboard | Left lower checkerboard | 0 | Right lower chess board grid | Right lower checkerboard |
Based on the additionally grown checkerboards, the up-down and left-right position relation between adjacent checkerboard angular points is obtained, the serial numbers of the up-down and left-right checkerboard angular points of each checkerboard angular point are obtained, the pixel coordinates of each checkerboard angular point are obtained, the up-down position corresponds to the Y axis of a world coordinate system, the left-right position corresponds to the X axis of the world coordinate system, and the world coordinates and the relative position relation of the checkerboard angular points are established. Knowing the length of each checkerboard unit grid, taking the plane where the checkerboard is located as a plane with Z being 0 in the world coordinate system, and establishing the world coordinate and relative position relation of the checkerboard corner points. Taking the leftmost corner of the additionally grown checkerboard as the world coordinate origin, the coordinates from each corner point to the origin can be expressed as (x, y), the meaning of which is the block distance from the origin, the unit is 1, and the world coordinates of each corner point can be obtained by multiplying (x, y) by the unit length of the checkerboard, and the unit is mm.
And acquiring the pixel coordinates of the marked dots in each image by adopting a detection algorithm, establishing a relative position relation between the marked dots and the checkerboard angular points, and determining the origin and the direction of the world coordinates. And according to the marking form of the marking dots of the plane calibration plate, obtaining the pixel coordinates of the marking dots by adopting a detection algorithm corresponding to the steps. And establishing the relative position relation between the marked round points and the checkerboard angular points by combining the topological rules of the positions of the marked round points and the checkerboard angular points as well as the topological rules of the positions of the marked round points and the marked round points. Specifically, pixel coordinates of four vertexes of each cell of the checkerboard are obtained according to pixel coordinates and relative position relations of all checkerboard angular points, the center of a circumscribed rectangle of the four vertexes of each cell is taken as a center, and the length half and the width half of the circumscribed rectangle of the four vertexes of each cell are taken as the length and the width, so that the region of interest of each cell is obtained. And if the maximum gray scale in the region of interest is more than 3 times of the minimum gray scale and the minimum gray scale is more than 100, determining that the center of the mark circle exists in the cell. The checkerboard angular point of the connecting line center of 1 pair of marking dots with the circle center distance as the diagonal length of the cell is taken as the world coordinate origin, the connecting line direction of 2 pairs of marking dots with the circle center distance as the side length of the cell is taken as the X direction and the Y direction, and the world coordinate of the checkerboard angular point is reestablished.
And calculating the internal reference, distortion and external reference of the camera to be calibrated and the relative rotation and translation of the two cameras according to the Zhang calibration method.
According to an embodiment of the present invention in fig. 2, the present invention provides an RGBD binocular calibration system based on a full view plane calibration plate, the system including:
the camera calibration device 20 is used for acquiring at least 6 images of a plane calibration plate of a camera to be calibrated under different poses, and comprises the plane calibration plate, wherein the plane calibration plate comprises a checkerboard containing a plurality of black and white grids and three marking dots, and the size of the plane calibration plate is far larger than the field of view of the camera;
the corner detection module 21 is configured to obtain pixel coordinates of each checkerboard corner in each image by using a growth-based checkerboard corner detection algorithm for the images, establish a relative position relationship between the checkerboard corners, and establish a one-to-one correspondence relationship between the pixel coordinates of the checkerboard corners and world coordinates;
the marked dot detection module 22 is used for acquiring the pixel coordinates of the marked dots in each image, establishing the relative position relationship between the marked dots and the checkerboard corner points, and determining the origin and the direction of world coordinates;
and the Zhang calibration module 23 is used for calculating the internal reference, distortion and external reference of the camera to be calibrated and the relative rotation and translation of the binocular camera according to the Zhang calibration method.
The camera calibration device comprises a plane calibration plate 30, a camera 31 to be calibrated and an adjusting device 32, as shown in fig. 3. The camera to be calibrated is fixed on the control device, and the adjusting device drives the camera to be calibrated to move together, so that the Z-axis translation, the Y-axis rotation and the X-axis rotation of the camera to be calibrated are realized. The adjusting device comprises an electric sliding table, an electric rotating table and an electric angular position table, the respective precision is not less than 0.1mm, 0.1 degree and 0.1 degree respectively, and the stroke is 2m, 360 degrees and +/-45 degrees. The planar calibration plate size is much larger than the camera field of view. As shown in fig. 4, the calibration board includes a checkerboard including a plurality of black and white unit cells and three marking dots, the unit cells are black squares and white squares which have the same size and are arranged longitudinally and transversely, the transverse and longitudinal adjacent checkerboard cells have different colors, and the unit cells in the diagonal direction have the same color; the corner points of the checkerboard, namely the intersection points of straight lines where the sides of the unit grids are located, are characteristic points. The three marking dots are respectively arranged in three different cells in a preset range of the central area of the plane calibration plate, the dots of the three marking dots are respectively positioned at the centers of the cells, the three marking dots are arranged in an L shape, the circle center distance of two pairs of marking dots is the side length of the checkerboard cells, the circle center distance of the other pair of marking dots is the diagonal length of the cells, the marking dots in the white cells are black, and the marking dots in the black cells are white. In one embodiment of the present invention, a checkerboard cell size of 50mm, 24 x 18 in total, is provided, with the marking dots having a diameter of 10 mm. The camera calibration device is used for at least acquiring 6 images of the plane calibration plate of the camera to be calibrated under different poses.
The corner detection module adopts a growth-based checkerboard corner detection algorithm to the images to obtain pixel coordinates of each checkerboard corner in each image, establishes a relative position relationship between the checkerboard corners, and establishes a one-to-one correspondence relationship between the pixel coordinates of the checkerboard corners and world coordinates. The marking dot detection module acquires the pixel coordinates of the marking dots in each image based on the same detection method, establishes the relative position relationship between the marking dots and the checkerboard corner points, and determines the origin and the direction of world coordinates. And the Zhang calibration module calculates the internal reference, distortion and external reference of the camera to be calibrated and the relative rotation and translation of the binocular camera according to the Zhang calibration method.
Although the preferred embodiments of the present invention have been disclosed for illustrative purposes, those skilled in the art will appreciate that various modifications, additions and substitutions are possible, without departing from the scope and spirit of the invention as disclosed in the accompanying claims.
Claims (10)
1. An RGBD binocular calibration method based on a full-view plane calibration plate is characterized by comprising the following steps:
s1, at least acquiring images of a plane calibration plate of 6 cameras to be calibrated under different poses, wherein the plane calibration plate comprises a checkerboard containing a plurality of black and white unit cells and three marking dots, and the size of the plane calibration plate is far larger than the field of view of the cameras;
s2, adopting a growth-based checkerboard corner detection algorithm to the images, obtaining the pixel coordinates of each checkerboard corner in each image, establishing the relative position relationship between the checkerboard corners, and establishing the one-to-one correspondence relationship between the pixel coordinates of the checkerboard corners and the world coordinates;
s3, acquiring the pixel coordinates of the marked dots in each image, establishing the relative position relationship between the marked dots and the checkerboard corner points, and determining the origin and the direction of world coordinates;
and S4, calculating the internal reference, distortion and external reference of the camera to be calibrated and the relative rotation and translation of the binocular camera according to the Zhang calibration method.
2. The RGBD binocular calibration method based on the full view plane calibration plate of claim 1, wherein the step S1 includes:
acquiring plane calibration plate images under postures of dead facing, pitching 15 degrees and deflecting 15 degrees at a position of a Z axis corresponding to the nearest depth of field of a camera to be calibrated, and acquiring 3 images in total;
and acquiring plane calibration plate images under the postures of dead, pitching of-15 degrees and deflecting of-15 degrees at the position of the Z axis corresponding to the central depth of field of the camera to be calibrated, and acquiring 3 images in total.
3. The RGBD binocular calibration method based on the full view plane calibration plate of claim 2, wherein the step S2 includes:
s201, growing leftwards by taking the leftmost three columns of the rectangular checkerboards as seed checkerboards to obtain left checkerboards;
s202, executing the same method as the step S201, respectively taking the rightmost three columns, the topmost three columns and the bottommost three columns of the rectangular checkerboards as seed checkerboards, and growing rightwards, upwards and downwards to obtain a right checkerboard, an upper checkerboard and a lower checkerboard;
s203, taking the uppermost 3 rows of the left checkerboard and the leftmost 3 columns of the upper checkerboard as a transverse seed checkerboard and a longitudinal seed checkerboard, and carrying out upper left region growth to obtain an upper left checkerboard;
s204, executing the same method as the step S203, taking the lowest 3 rows of the left checkerboard, the leftmost 3 columns of the lower checkerboard, the topmost 3 rows of the right checkerboard, the rightmost 3 columns of the upper checkerboard, the lowest 3 rows of the right checkerboard and the rightmost 3 columns of the lower checkerboard as transverse seed checkerboards and longitudinal seed checkerboards respectively, and performing growth of a left lower area, a right upper area and a right lower area to obtain a left lower checkerboard, a right upper checkerboard and a right lower checkerboard;
s205, growing leftwards and upwards by taking the leftmost three columns and the rightmost three columns of the left upper checkerboard as seed checkerboards respectively to obtain a left upper left checkerboard and a left upper checkerboard, and performing left upper regional growth by taking the left upper left checkerboard and the left upper left checkerboard as seed checkerboards to obtain a left upper checkerboard;
s206, executing the same method as the step S205 to obtain a left lower left checkerboard, a right upper right checkerboard, a right upper checkerboard, a right upper right checkerboard, a right lower right checkerboard and a right lower right checkerboard;
and S207, splicing the checkerboard grids to obtain extra-grown checkerboard grids.
4. The RGBD binocular calibration method based on the full view plane calibration plate according to claim 3, wherein the step S201 specifically includes:
s2011, if the seeds grow leftwards, turning the seed checkerboard leftwards and rightwards, and if the seeds grow upwards, turning the seed checkerboard downwards;
s2012, recording the corner index which does not participate in the growth at present, the initial maximum structure energy and the current checkerboard corner number, obtaining the index of the growth in each row and column direction of the seed checkerboard, and calculating the structure energy of the seed checkerboard, if the structure energy meets the following conditions:
the structural energy < (the number of angular points of the current checkerboard + the maximum structural energy + 1)/(the number of angular points of the current checkerboard), if the growth is effective, the angular point index and the structural energy are recorded, the maximum structural energy is updated to the maximum value of the maximum structural energy and the calculated structural energy, the number of the angular points of the current checkerboard is added by one, and the current index in the angular point index is deleted;
s2013, constructing an index matrix, a second dimension energy matrix and a first dimension energy matrix according to the maximum value of the number of growing corners of each row, filling the corner indexes obtained in the step 2012 into the index matrix, and filling the obtained structural energy into the second dimension energy matrix;
s2014, constructing a first-dimension energy matrix, and updating indexes, checkerboard indexes which do not participate in growth, the energy matrix and the maximum structure energy according to the continuous length and the structure energy of the first-dimension energy matrix;
s2015, deleting the columns of which all elements in the index matrix are zero, left-right inverting the index matrix, if the index matrix grows upwards, transposing the index matrix, and turning the index matrix up and down, if the index matrix grows downwards, transposing the index matrix.
5. The RGBD binocular calibration method based on the full field of view plane calibration plate according to claim 4, wherein the step S2014 includes:
for each column of the index matrix from left to right, if the continuous length of a sequence formed by non-zero corner indexes is less than 3, setting each row with the position of the corner index as a starting point as 0, if the sequence comprises the non-zero indexes, writing the indexes into the corner indexes, accumulating and adding one to the number of the corner points of the current chessboard once, setting the second dimension energy matrix of the corresponding position as 0, and updating the maximum structure energy to be the maximum value of the initial maximum structure energy and the maximum values of the second dimension energy matrix and the first dimension energy matrix;
if the continuous length of a sequence formed by the corner indexes which are not 0 is more than or equal to 3 from left to right in each column of the index matrix, finding a first subsequence which meets the requirement that the continuous length of the sequence is 3 from the starting point of the sequence downwards, setting rows with the corner indexes before the subsequence as the starting point to be zero, writing the corner indexes into corner indexes which do not participate in growth, reducing the number of the current checkerboard corners by one every time the corner indexes are written, setting a second dimension energy matrix at the corresponding position to be zero, and updating the maximum structure energy to be the maximum value among the initial maximum structure energy, the second dimension energy matrix and the first dimension energy matrix;
if the continuous length of a sequence formed by the corner indexes which are not 0 is more than or equal to 3 from left to right, finding a first subsequence which does not meet the condition sequence and has the continuous length of 3, setting the rows with the 2 nd and 3 rd corner indexes of the subsequence as the starting points to be zero, writing the corner indexes into the corner indexes which do not participate in growth, reducing the number of the current checkerboard corners by one every time the corner indexes are written, setting the second dimension energy matrix at the corresponding position to be zero, and updating the maximum structure energy to be the maximum value among the initial maximum structure energy, the second dimension energy matrix and the first dimension energy matrix.
6. The RGBD binocular calibration method based on the full view plane calibration plate according to claim 5, wherein the step S203 comprises:
s2031, if the left lower region grows, rotating the transverse seed checkerboard anticlockwise by 90 degrees, rotating the longitudinal seed checkerboard anticlockwise by 90 degrees, if the right upper region grows, turning the transverse seed checkerboard upside down, and turning the longitudinal seed checkerboard upside down;
s2032, recording the index of the corner points which do not participate in the growth at present, the initial maximum structure energy and the number of the corner points of the current chessboard, and constructing an index matrix;
s2033, setting initial values of the effective growth height and the effective growth width to be zero, adding one to the effective growth height for each line of the longitudinal seed checkerboard from top to bottom if the angular point indexes of the leftmost three lines are not zero, and adding one to the effective growth width for each column of the transverse seed checkerboard from left to right if the angular point indexes of the uppermost three lines are not zero;
s2034, constructing an effective index matrix, an effective transverse structure energy matrix, an effective longitudinal structure energy matrix and a column source matrix;
s2035, assigning the front effective growth width column of the column source matrix as the front effective growth width column of the 2 nd and 3 rd rows of the transverse seed checkerboard;
s2036, marking the indexes of the rows corresponding to the first 3 columns of the longitudinal seed checkerboard as p1Row, p2Row and p3Row, carrying out the predicted growth in the Row direction by the p1Row, p2Row and p3Row for each Row of the effective index matrix, and calculating the structural energy of the effective index matrix once each growth;
s2037, calculating and predicting whether structural energy formed by the growth points and the corner points of the corresponding serial numbers of the corresponding columns of the column source matrix meets the requirement or not for each row of the effective index matrix, and if not, subtracting one from the width limiting column; otherwise, the value of the corresponding position of the effective transverse structural energy matrix is a row energy value, the value of the corresponding position of the effective longitudinal structural energy matrix is a column energy value, the corresponding position of the effective index matrix is an angular point index of a predicted angular point, a corresponding index in the angular point indexes which do not participate in growth at present is deleted, the number of the angular points of the chessboard grid is accumulated by one, and the maximum structural energy is updated to be the maximum value of the initial maximum structural energy, the effective transverse structural energy matrix and the effective longitudinal structural energy matrix;
s2038, if the effective growth width of each row of the effective index matrix is not zero, only the effective growth width column of the column source matrix is reserved, the 1 st row of the column source matrix is assigned as the 2 nd row of the column source matrix, and the 2 nd row of the column source matrix is assigned as the corresponding row and the front width limit column of the effective index matrix;
s2039, if the area grows in the upper left area, the index matrix is rotated 180 degrees anticlockwise, if the area grows in the lower left area, the index matrix is turned over left and right, and if the area grows in the upper right area, the index matrix is turned over up and down.
7. The RGBD binocular calibration method based on the full field of view plane calibration plate according to claim 6, wherein the step S2 further comprises:
taking the plane of the checkerboard as a plane with Z being 0 under a world coordinate system, and establishing the world coordinate and relative position relation of the checkerboard corner points;
and (3) taking the leftmost upper corner of the additionally grown checkerboard as the world coordinate origin, expressing the coordinates from each corner point to the origin as (x, y), and multiplying the (x, y) by the length of the cells of the checkerboard to obtain the world coordinates of each checkerboard corner point.
8. The RGBD binocular scaling method based on the full field of view plane scaling plate of claim 7, wherein the step S3 comprises:
acquiring pixel coordinates of four vertexes of each unit lattice of the checkerboard;
taking the center of the circumscribed rectangle of the four vertexes of each cell as the center, and taking the half of the length and the half of the width of the circumscribed rectangle of the four vertexes of each cell as the length and the width to obtain the region of interest of each cell;
if the maximum gray scale in the region of interest is greater than 3 times of the minimum gray scale and the minimum gray scale is greater than 100, the chessboard is considered to have a marked circle center;
the checkerboard angular point of the connecting line center of 1 pair of marking dots with the circle center distance as the diagonal length of the cell is taken as the world coordinate origin, the connecting line direction of 2 pairs of marking dots with the circle center distance as the side length of the cell is taken as the X direction and the Y direction, and the world coordinate of the checkerboard angular point is reestablished.
9. An RGBD binocular calibration system based on a full-field plane calibration plate, the system comprising:
the camera calibration device is used for at least acquiring images of 6 planar calibration plates of a camera to be calibrated under different poses, and comprises a planar calibration plate, wherein the planar calibration plate comprises a checkerboard containing a plurality of black and white unit cells and three marking dots, and the size of the planar calibration plate is far larger than the field of view of the camera;
the angular point detection module is used for acquiring the pixel coordinates of each checkerboard angular point in each image by adopting a growth-based checkerboard angular point detection algorithm for the images, establishing the relative position relationship between the checkerboard angular points and establishing the one-to-one correspondence relationship between the pixel coordinates of the checkerboard angular points and world coordinates;
the marked dot detection module is used for acquiring the pixel coordinates of the marked dots in each image, establishing the relative position relationship between the marked dots and the checkerboard corner points, and determining the origin and the direction of world coordinates;
and the Zhang calibration module is used for calculating the internal reference, distortion and external reference of the camera to be calibrated and the relative rotation and translation of the binocular camera according to the Zhang calibration method.
10. The RGBD binocular calibration system based on the full view plane calibration plate of claim 9, wherein the three mark dots are respectively disposed in three different cells within a predetermined range of a central region of the plane calibration plate, the dots of the three mark dots are respectively located at the centers of the respective cells, the three mark dots are arranged in an L shape, wherein a distance between centers of two pairs of mark dots is a side length of the cell, a distance between centers of the other pair of mark dots is a diagonal length of the cell, the mark dots in a white cell are black, and the mark dots in a black cell are white.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011351521.3A CN112465916A (en) | 2020-11-27 | 2020-11-27 | RGBD binocular calibration method and system based on full-view-field plane calibration plate |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011351521.3A CN112465916A (en) | 2020-11-27 | 2020-11-27 | RGBD binocular calibration method and system based on full-view-field plane calibration plate |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112465916A true CN112465916A (en) | 2021-03-09 |
Family
ID=74808906
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011351521.3A Pending CN112465916A (en) | 2020-11-27 | 2020-11-27 | RGBD binocular calibration method and system based on full-view-field plane calibration plate |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112465916A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112991462A (en) * | 2021-03-15 | 2021-06-18 | 扬州大学 | Camera calibration method based on dot diagram |
| CN114862923A (en) * | 2022-07-06 | 2022-08-05 | 武汉市聚芯微电子有限责任公司 | Image registration method and device and storage medium |
| CN114923410A (en) * | 2022-05-09 | 2022-08-19 | 一汽解放汽车有限公司 | Longitudinal beam hole site online detection method and device |
Citations (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE102006055758A1 (en) * | 2006-11-25 | 2008-05-29 | Deutsches Zentrum für Luft- und Raumfahrt e.V. | Camera calibrating method for optical measuring system, involves marking one of cross-over points in checker pattern by position information, which is readable by image processing of images of pattern and has coding of orientation |
| US20120033852A1 (en) * | 2010-08-06 | 2012-02-09 | Kennedy Michael B | System and method to find the precise location of objects of interest in digital images |
| KR20180105875A (en) * | 2017-03-16 | 2018-10-01 | 한국전자통신연구원 | Camera calibration method using single image and apparatus therefor |
| CN108734743A (en) * | 2018-04-13 | 2018-11-02 | 深圳市商汤科技有限公司 | Method, apparatus, medium and electronic equipment for demarcating photographic device |
| CN109272555A (en) * | 2018-08-13 | 2019-01-25 | 长安大学 | A kind of external parameter of RGB-D camera obtains and scaling method |
| CN109285194A (en) * | 2018-09-29 | 2019-01-29 | 人加智能机器人技术（北京）有限公司 | Camera calibration plate and camera calibration collecting method |
| WO2019080229A1 (en) * | 2017-10-25 | 2019-05-02 | 南京阿凡达机器人科技有限公司 | Chess piece positioning method and system based on machine vision, storage medium, and robot |
| KR20190051463A (en) * | 2017-11-07 | 2019-05-15 | 현대모비스 주식회사 | Apparatus and method for detecting checkerboard corner point for camera calibration |
| CN110148174A (en) * | 2019-05-23 | 2019-08-20 | 北京阿丘机器人科技有限公司 | Scaling board, scaling board recognition methods and device |
| CN110223356A (en) * | 2019-06-10 | 2019-09-10 | 于兴虎 | A kind of monocular camera full automatic calibration method based on energy growth |
| CN110599548A (en) * | 2019-09-02 | 2019-12-20 | Oppo广东移动通信有限公司 | Camera calibration method and device, camera and computer readable storage medium |
| CN110660107A (en) * | 2019-08-23 | 2020-01-07 | 贝壳技术有限公司 | Plane calibration plate, calibration data acquisition method and system |
| CN110969668A (en) * | 2019-11-22 | 2020-04-07 | 大连理工大学 | Stereoscopic calibration algorithm of long-focus binocular camera |
| US20200151908A1 (en) * | 2017-06-30 | 2020-05-14 | Hangzhou Hikvision Digital Technology Co., Ltd. | Camera parameter calibration method, device, apparatus, and system |
| CN111179356A (en) * | 2019-12-25 | 2020-05-19 | 北京中科慧眼科技有限公司 | Binocular camera calibration method, device and system based on Aruco code and calibration board |
| CN111260731A (en) * | 2020-01-10 | 2020-06-09 | 大连理工大学 | Checkerboard sub-pixel level corner point self-adaptive detection method |
| CN111429532A (en) * | 2020-04-30 | 2020-07-17 | 南京大学 | Method for improving camera calibration accuracy by utilizing multi-plane calibration plate |
-
2020
- 2020-11-27 CN CN202011351521.3A patent/CN112465916A/en active Pending
Patent Citations (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE102006055758A1 (en) * | 2006-11-25 | 2008-05-29 | Deutsches Zentrum für Luft- und Raumfahrt e.V. | Camera calibrating method for optical measuring system, involves marking one of cross-over points in checker pattern by position information, which is readable by image processing of images of pattern and has coding of orientation |
| US20120033852A1 (en) * | 2010-08-06 | 2012-02-09 | Kennedy Michael B | System and method to find the precise location of objects of interest in digital images |
| KR20180105875A (en) * | 2017-03-16 | 2018-10-01 | 한국전자통신연구원 | Camera calibration method using single image and apparatus therefor |
| US20200151908A1 (en) * | 2017-06-30 | 2020-05-14 | Hangzhou Hikvision Digital Technology Co., Ltd. | Camera parameter calibration method, device, apparatus, and system |
| WO2019080229A1 (en) * | 2017-10-25 | 2019-05-02 | 南京阿凡达机器人科技有限公司 | Chess piece positioning method and system based on machine vision, storage medium, and robot |
| KR20190051463A (en) * | 2017-11-07 | 2019-05-15 | 현대모비스 주식회사 | Apparatus and method for detecting checkerboard corner point for camera calibration |
| CN108734743A (en) * | 2018-04-13 | 2018-11-02 | 深圳市商汤科技有限公司 | Method, apparatus, medium and electronic equipment for demarcating photographic device |
| CN109272555A (en) * | 2018-08-13 | 2019-01-25 | 长安大学 | A kind of external parameter of RGB-D camera obtains and scaling method |
| CN109285194A (en) * | 2018-09-29 | 2019-01-29 | 人加智能机器人技术（北京）有限公司 | Camera calibration plate and camera calibration collecting method |
| CN110148174A (en) * | 2019-05-23 | 2019-08-20 | 北京阿丘机器人科技有限公司 | Scaling board, scaling board recognition methods and device |
| CN110223356A (en) * | 2019-06-10 | 2019-09-10 | 于兴虎 | A kind of monocular camera full automatic calibration method based on energy growth |
| CN110660107A (en) * | 2019-08-23 | 2020-01-07 | 贝壳技术有限公司 | Plane calibration plate, calibration data acquisition method and system |
| CN110599548A (en) * | 2019-09-02 | 2019-12-20 | Oppo广东移动通信有限公司 | Camera calibration method and device, camera and computer readable storage medium |
| CN110969668A (en) * | 2019-11-22 | 2020-04-07 | 大连理工大学 | Stereoscopic calibration algorithm of long-focus binocular camera |
| CN111179356A (en) * | 2019-12-25 | 2020-05-19 | 北京中科慧眼科技有限公司 | Binocular camera calibration method, device and system based on Aruco code and calibration board |
| CN111260731A (en) * | 2020-01-10 | 2020-06-09 | 大连理工大学 | Checkerboard sub-pixel level corner point self-adaptive detection method |
| CN111429532A (en) * | 2020-04-30 | 2020-07-17 | 南京大学 | Method for improving camera calibration accuracy by utilizing multi-plane calibration plate |
Non-Patent Citations (2)
| Title |
|---|
| 张勤;贾庆轩;席宁;姜勇;: "基于双视点特征匹配的激光-相机***标定方法", 仪器仪表学报, vol. 33, no. 11, pages 2622 - 2628 * |
| 赖文敬;周延周;朱文卓;鲍鸿;: "一种基于棋盘格模板单目摄像机高精度标定方法", 广东工业大学学报, vol. 32, no. 03, pages 79 - 84 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112991462A (en) * | 2021-03-15 | 2021-06-18 | 扬州大学 | Camera calibration method based on dot diagram |
| CN114923410A (en) * | 2022-05-09 | 2022-08-19 | 一汽解放汽车有限公司 | Longitudinal beam hole site online detection method and device |
| CN114923410B (en) * | 2022-05-09 | 2024-05-14 | 一汽解放汽车有限公司 | On-line detection method and device for hole sites of longitudinal beams |
| CN114862923A (en) * | 2022-07-06 | 2022-08-05 | 武汉市聚芯微电子有限责任公司 | Image registration method and device and storage medium |
| CN114862923B (en) * | 2022-07-06 | 2022-09-09 | 武汉市聚芯微电子有限责任公司 | Image registration method and device and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112465916A (en) | RGBD binocular calibration method and system based on full-view-field plane calibration plate | |
| CN109443209B (en) | Line structured light system calibration method based on homography matrix | |
| CN111028350B (en) | Method for constructing grid map by using binocular stereo camera | |
| CN110443879B (en) | Perspective error compensation method based on neural network | |
| CN114018932B (en) | Pavement disease index measurement method based on rectangular calibration object | |
| CN106210716A (en) | A kind of panoramic video isodensity method of sampling and device | |
| CN107977996A (en) | Space target positioning method based on target calibrating and positioning model | |
| CN110298778B (en) | Traffic cell demographic method based on raster data and area ratio correction | |
| CN113052911A (en) | Calibration board, camera calibration method and device | |
| CN108234884A (en) | A kind of automatic focusing method of camera of view-based access control model conspicuousness | |
| CN112037284B (en) | Method for detecting center point of checkerboard lattice, method and system for calibrating distortion of camera | |
| CN109801242A (en) | A kind of grating image vectorization method based on the unrestrained water filling algorithm of improvement | |
| CN114723828B (en) | Multi-line laser scanning method and system based on binocular vision | |
| CN103761708A (en) | Image restoration method based on contour matching | |
| JP3218372B2 (en) | 3D graphic display method | |
| WO2021102948A1 (en) | Image processing method and device | |
| CN108460803B (en) | Checkerboard pattern-based AR sand table calibration model calculation method | |
| CN108320311B (en) | AR sand table calibration model calculation method capable of automatically correcting deviation | |
| CN108734148A (en) | A kind of public arena image information collecting unmanned aerial vehicle control system based on cloud computing | |
| CN114322865B (en) | Method and device for measuring semiconductor device and storage medium | |
| CN108182702B (en) | Real-time three-dimensional modeling method and system based on depth image acquisition equipment | |
| CN109829951A (en) | It is parallel to wait position detecting methods, device and automated driving system | |
| CN114998496A (en) | Orthoimage rapid generation method based on scene aerial photography image and sparse point cloud | |
| CN107507256A (en) | The real-time drawing method of high-volume ship target based on grid economy model in a kind of VTS system | |
| CN109472748B (en) | Image brightness correction method based on nanoparticle SEM image brightness extraction |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20210309 |