Monocular camera calibration effect evaluation method and system

Abstract

The invention relates to the technical field of image algorithms, and discloses a monocular camera calibration effect evaluation method and system, and the method comprises the steps: respectively shooting three groups of checkerboard calibration plates at different distances through a calibrated monocular camera, then carrying out the processing and analysis of a shot picture, and carrying out the evaluation of the calibration effect of a monocular camera through a digitally quantized evaluation method. And evaluating the distortion correction calibration effect of the monocular camera.

Description

Technical Field

[0001] The present invention relates to the field of image algorithm technology, and in particular to a method and system for evaluating the calibration effect of a monocular camera. Background Art

[0002] A monocular camera essentially captures a scene and leaves a projection on the imaging plane. In order to obtain a good imaging effect, a lens is generally added in front of the camera, which will have a new impact on the propagation of light during the imaging process. The distortion caused by the shape of the lens is called radial distortion. Radial distortion will make a straight line in the real world appear as a curve on the imaging plane, and the closer to the edge of the image, the more obvious the distortion. In addition to the radial distortion introduced by the shape of the lens, tangential distortion will be introduced because the lens and the imaging plane cannot be strictly parallel during the camera assembly process.

[0003] There are many camera distortion correction programs, such as Zhang Zhengyou calibration method, opencv, MATLAB built-in programs, etc., which can calculate the distortion correction parameters of monocular cameras. However, there is currently no method to evaluate the correction effect of monocular cameras. Usually, after completing the image distortion correction, there is a lack of an effective means to quantify the calibration effect of the monocular camera. Summary of the invention

[0004] The present invention provides a method and system for evaluating the calibration effect of a monocular camera, so as to solve the problem in the prior art that there is a lack of effective means for quantifying the calibration effect of a monocular camera.

[0005] In order to achieve the above object, the present invention is implemented by the following technical solutions:

[0006] In a first aspect, the present invention provides a method for evaluating the calibration effect of a monocular camera, comprising:

[0007] S1: Use the monocular camera that has completed distortion correction to collect images of the checkerboard calibration plate to obtain a checkerboard calibration plate image;

[0008] S2: extracting the checkerboard calibration plate corner points in the checkerboard calibration plate image;

[0009] S3: Divide the corner points of the checkerboard calibration plate into row corner points and column corner points, and calculate the slope mean square error according to the row corner points and the column corner points;

[0010] S4: Evaluate the calibration effect of the monocular camera distortion correction based on the slope mean square error.

[0011] Optionally, the S1 includes:

[0012] S11: Place an N*M checkerboard calibration plate at a first set position, a second set position, and a third set position, respectively; wherein N represents rows, M represents columns, and the number of rows N is less than the number of columns M; wherein the distance between the first set position and the monocular camera is less than the distance between the second set position and the monocular camera, and the distance between the second set position and the monocular camera is less than the distance between the third set position and the monocular camera.

[0013] The first setting position, the second setting position and the third setting position cover three common usage scenarios: near, medium and far. By analyzing the calibration results under the three common usage scenarios, it is possible to be more objective and comprehensive when evaluating the calibration effect.

[0014] S12: using the distortion-corrected monocular camera to acquire chessboard calibration plates respectively located at a first set position, a second set position, and a third set position to obtain three sets of chessboard calibration plate images.

[0015] Optionally, the S2 includes:

[0016] S21: performing a binarization transformation on the chessboard calibration plate image to convert the image from a color image to a black and white image;

[0017] S22: determining candidate corner points of the checkerboard calibration board according to the checkerboard calibration board image in the black and white image;

[0018] S23: Selecting a middle pixel point from the candidate points for the checkerboard calibration plate corner point as a checkerboard calibration plate corner point.

[0019] Since the candidate corner points of the checkerboard calibration plate are a set of points surrounding the true corner point, and the true corner point is in the middle of the set, the true corner point of the checkerboard calibration plate can be quickly found by selecting the middle pixel point.

[0020] Optionally, determining candidate corner points of the checkerboard calibration board in the checkerboard calibration board picture in the black-and-white image includes:

[0021] Calculate the average of the A*B pixel areas of the first target position, the second target position, the third target position, and the fourth target position of each pixel in the chessboard calibration board image, and then determine whether it is a candidate point for the corner point of the chessboard calibration board based on whether the colors of the diagonal areas of the corner point of the chessboard calibration board are consistent;

[0022] Among them, the first target position may be the upper left, the second target position may be the lower left, the third target position may be the upper right, and the fourth target position may be the lower right.

[0023] Optionally, the S3 includes:

[0024] S31: setting the pixel coordinates of the lower left corner of the chessboard calibration plate image to (0, 0), with changes to the right and upward being positive, and dividing the corner points of the chessboard calibration plate into C rows and D columns according to the horizontal and vertical coordinates;

[0025] S32: According to the row corner points and column corner points, respectively calculate the slopes between adjacent corner points in each row, and then obtain the mean square error of the slopes, and finally obtain the mean square error of the slopes of C row corner points;

[0026] S33: According to the row corner points and column corner points, respectively calculate the slopes between adjacent corner points in each column, and then obtain the mean square error of the slopes, and finally obtain the mean square errors of the slopes of D column corner points;

[0027] S34: deriving the parallelism of the C row corner points according to the mean square deviation of the slopes of the C row corner points; and deriving the parallelism of the D column corner points according to the mean square deviation of the slopes of the D column corner points.

[0028] Optionally, the S4 includes:

[0029] S41: averaging the C mean square errors of the slopes of the corner points in the rows and the D mean square errors of the slopes of the corner points in the columns calculated from the chessboard calibration plate image;

[0030] S42: Evaluate the distortion correction calibration effect of the monocular camera based on the flatness quantization correction effect according to the mean value of the mean square error of the chessboard calibration plate image.

[0031] In a second aspect, the present application provides a monocular camera calibration effect evaluation system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method described in the first aspect when executing the computer program.

[0032] Beneficial effects:

[0033] The present application provides a method for evaluating the calibration effect of a monocular camera. According to the unique image characteristics of the corner points of a chessboard calibration plate, the corner points are extracted, and the corner points are classified according to rows and columns. The corner points of the chessboard calibration plate are accurately extracted, and the parallelism between the diagonal points of the mean value of the slope mean square error is calculated, so as to perform a quantitative evaluation, and the correction effect is evaluated based on the calculated corrected parallelism. Through the evaluation method that can be digitally quantified, the distortion correction effect of the monocular camera is evaluated, and the calibration effect of the monocular camera is obtained, thereby evaluating the camera correction effect.

[0034] In a further solution, the corner points are extracted according to the unique image characteristics of the corner points of the chessboard calibration plate, and the corner points are classified according to rows and columns to facilitate the subsequent calculation of the mean square error of the slopes between the corner points in the rows and columns. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a flow chart of a method for evaluating the calibration effect of a monocular camera according to a preferred embodiment of the present invention;

[0036] Figure 2 The calibrated chessboard calibration plate of the preferred embodiment of the present invention;

[0037] Figure 3 The candidate points of the checkerboard calibration plate of the preferred embodiment of the present invention;

[0038] Figure 4 This is the extraction of corner points of a chessboard calibration plate according to a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0039] The technical solution of the present invention is described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0040] Unless otherwise defined, the technical terms or scientific terms used in the present invention shall have the usual meanings understood by persons with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. Similarly, words such as "one" or "one" do not indicate quantity restrictions, but indicate the existence of at least one. Words such as "connect" or "connected" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the object being described changes, the relative positional relationship also changes accordingly.

[0041] See also Figure 1 , the present application provides a method for evaluating the calibration effect of a monocular camera, comprising:

[0042] S1: Use the monocular camera that has completed distortion correction to collect images of the checkerboard calibration plate to obtain a checkerboard calibration plate image;

[0043] S2: extracting the checkerboard calibration plate corner points in the checkerboard calibration plate image;

[0044] S3: Divide the corner points of the checkerboard calibration plate into row corner points and column corner points, and calculate the slope mean square error according to the row corner points and the column corner points;

[0045] S4: Evaluate the calibration effect of the monocular camera distortion correction based on the slope mean square error.

[0046] The above-mentioned monocular camera calibration effect evaluation method proposes a monocular camera calibration effect evaluation method to solve the problem that there is a lack of effective means to quantify the monocular camera calibration effect. This method extracts the corner points according to the unique image characteristics of the corner points of the chessboard calibration plate, and classifies the corner points according to rows and columns, accurately extracts the corner points of the chessboard calibration plate, calculates the mean of the slope mean square error and the parallelism between the corner points, so as to perform quantitative evaluation, and evaluates the correction effect according to the calculated corrected parallelism. Through the digitally quantifiable evaluation method, the distortion correction effect of the monocular camera is evaluated, and the calibration effect of the monocular camera is obtained, thereby evaluating the camera correction effect.

[0047] The steps of the above method are described in detail below with a complete embodiment.

[0048] 1. Use the distortion-corrected monocular camera to capture images of the checkerboard calibration plate

[0049] Step 1: Place an N*M checkerboard calibration plate at a first set position, a second set position, and a third set position, respectively; wherein N represents a row and M represents a column; wherein a distance between the first set position and the monocular camera is smaller than a distance between the second set position and the monocular camera, and a distance between the second set position and the monocular camera is smaller than a distance between the third set position and the monocular camera.

[0050] Among them, the N*M checkerboard calibration plate can be a 6*7 checkerboard calibration plate, and the length of each grid is 4 cm; this is only an example and not a limitation. In other feasible implementations, checkerboard calibration plates of other sizes can also be used, and the length of each grid can also be adjusted appropriately.

[0051] The first set position may be 0.5 m away from the monocular camera, the second set position may be 1 m away from the monocular camera, and the third set position may be 2 m away from the monocular camera; this is only an example and not a limitation. In other feasible implementations, other distances may be possible.

[0052] Step 2: Use the distortion-corrected monocular camera to collect the checkerboard calibration plates at the first set position, the second set position, and the third set position to obtain three sets of checkerboard calibration plate images, such as Figure 2 shown.

[0053] 2. Extract the corner points of the checkerboard calibration board in the checkerboard calibration board image

[0054] Step 1: Binarize the checkerboard calibration plate image to convert the image from a color image to a black and white image.

[0055] Step 2: Determine candidate corner points of the checkerboard calibration board according to the checkerboard calibration board image in the black and white image.

[0056] The average of the four A*B pixel areas of the first target position, the second target position, the third target position, and the fourth target position of each pixel of the chessboard calibration board screen is calculated. The A*B pixel area can be an 8*8 pixel area. Then, whether the diagonal area of ​​the corner point of the chessboard calibration board has the same color is determined to determine whether it is a candidate point for the corner point of the chessboard calibration board, such as Figure 3 As shown;

[0057] Step 3: Select the middle pixel point from the candidate points of the checkerboard calibration plate corner point as the checkerboard calibration plate corner point, such as Figure 4 shown.

[0058] 3. Divide the corner points of the chessboard calibration plate into row corner points and column corner points, and calculate the mean square error of the slopes of the row corner points and the column corner points

[0059] Step 1: Set the pixel coordinates of the lower left corner of the three groups of checkerboard calibration plate images to (0, 0), and change to the right and upward as positive, and divide the corner points of the checkerboard calibration plate into C rows and D columns according to the horizontal and vertical coordinates.

[0060] Among them, the C rows and D columns can be 6 rows and 7 columns; this is only an example and not a limitation. In other feasible implementations, other numbers of rows and columns can be used.

[0061] The chessboard calibration plate image is an image taken when the monocular camera is placed upright. If the monocular camera rotates, the pixel coordinates should rotate with the monocular camera.

[0062] Step 2: According to the row corner points and column corner points, the slopes between adjacent corner points in each row are calculated respectively, and then the mean square error of the slopes is calculated, and finally the mean square error of the slopes of the six row corner points is obtained.

[0063] Step 3: According to the row corner points and column corner points, the slopes between adjacent corner points in each column are calculated respectively, and then the mean square error of the slopes is calculated, and finally the mean square error of the slopes of the seven column corner points is obtained.

[0064] Step 4: Based on the slope mean square error, the parallelism of the row corner points and the column corner points are obtained respectively.

[0065] Among them, it can be seen from the distortion type of the monocular camera that when the distortion is correctly corrected, the closer the slopes calculated between adjacent corner points in the same row are, the closer the line connecting the corner points is to a straight line, and the smaller the parallelism is. Similarly, the corner points in the same column also have this property. Based on the principle that a straight line in the real world after correction will still appear as a straight line on the imaging plane, the straight lines that should have been parallel in the corrected image are detected.

[0066] 4. Evaluating the calibration effect of the monocular camera distortion correction based on the parallelism

[0067] Step 1: Calculate the average of the mean square errors of the slopes of the six row corner points and the mean square errors of the slopes of the seven column corner points calculated from the chessboard calibration plate image.

[0068] Step 2: According to the mean of the mean square error of the chessboard calibration plate image, the distortion correction calibration effect of the monocular camera is evaluated based on the flatness quantification correction effect.

[0069] In one example, the same monocular camera is calibrated 8 times, and the calibration precision is improved successively. Then, the calibration effect is evaluated using the method proposed in this application, and the mean of the calculated mean square error is shown in Table 1.

[0070] Table 1 Quantitative evaluation of 8 calibration effects

[0071]

[0072] It can be seen from Table 1 that with the increase of precision, the result calculated by this application gradually decreases, indicating that the monocular camera calibration effect evaluation method proposed in this patent can effectively perform quantitative evaluation on the camera calibration effect.

[0073] The embodiment of the present application also provides a monocular camera calibration effect evaluation system, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the above method when executing the computer program. The monocular camera calibration effect evaluation system can implement various embodiments of the above method and achieve the same beneficial effects, which will not be described in detail here.

[0074] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A method for evaluating the calibration effect of a monocular camera, characterized in that: include: S1: Use the monocular camera that has completed distortion correction to collect images of the checkerboard calibration plate to obtain a checkerboard calibration plate image; S2: extracting the checkerboard calibration plate corner points in the checkerboard calibration plate image; S3: Divide the corner points of the chessboard calibration plate into row corner points and column corner points, and calculate parallelism according to the row corner points and the column corner points; S4: Evaluate the calibration effect of the monocular camera distortion correction based on the parallelism.

2. The method for evaluating the calibration effect of a monocular camera according to claim 1, wherein: The S1 includes: S11: placing an N*M checkerboard calibration plate at a first set position, a second set position, and a third set position, respectively; wherein N represents a row, M represents a column, and the number of rows N is less than the number of columns M; wherein the distance between the first set position and the monocular camera is less than the distance between the second set position and the monocular camera, and the distance between the second set position and the monocular camera is less than the distance between the third set position and the monocular camera; S12: using the distortion-corrected monocular camera to acquire chessboard calibration plates respectively located at a first set position, a second set position, and a third set position to obtain three sets of chessboard calibration plate images.

3. The method for evaluating the calibration effect of a monocular camera according to claim 1, characterized in that: The S2 includes: S21: performing a binarization transformation on the chessboard calibration plate image to convert the image from a color image to a black and white image; S22: determining candidate corner points of the checkerboard calibration board according to the checkerboard calibration board image in the black and white image; S23: Selecting a middle pixel point from the candidate points for the checkerboard calibration plate corner point as a checkerboard calibration plate corner point.

4. The method for evaluating the calibration effect of a monocular camera according to claim 3, characterized in that: The S22 includes: Calculate the average of the A*B pixel areas of the first target position, the second target position, the third target position, and the fourth target position of each pixel in the chessboard calibration board image, and then determine whether it is a candidate point for the corner point of the chessboard calibration board based on whether the colors of the diagonal areas of the corner point of the chessboard calibration board are consistent.

5. The method for evaluating the calibration effect of a monocular camera according to claim 1, characterized in that: The S3 includes: S31: setting the pixel coordinates of the lower left corner of the chessboard calibration plate image to (0, 0), with changes to the right and upward being positive, and dividing the corner points of the chessboard calibration plate into C rows and D columns according to the horizontal and vertical coordinates; S32: According to the row corner points and column corner points, respectively calculate the slopes between adjacent corner points in each row, and then obtain the mean square error of the slopes, and finally obtain the mean square error of the slopes of C row corner points; S33: According to the row corner points and column corner points, respectively calculate the slopes between adjacent corner points in each column, and then obtain the mean square error of the slopes, and finally obtain the mean square errors of the slopes of D column corner points; S34: deriving the parallelism of the C row corner points according to the mean square deviation of the slopes of the C row corner points; and deriving the parallelism of the D column corner points according to the mean square deviation of the slopes of the D column corner points.

6. The method for evaluating the calibration effect of a monocular camera according to claim 5, characterized in that: The S4 includes: S41: averaging the C mean square errors of the slopes of the corner points in the rows and the D mean square errors of the slopes of the corner points in the columns calculated from the chessboard calibration plate image; S42: Evaluate the distortion correction calibration effect of the monocular camera based on the flatness quantization correction effect according to the mean value of the mean square error of the chessboard calibration plate image.

7. A monocular camera calibration effect evaluation system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method described in any one of claims 1 to 6 are implemented.

Images