A Fast Calibration Method for a Micro-Vision System with a Liquid Zoom Lens

Abstract

The present invention discloses a rapid calibration method for a micro-vision system with a liquid zoom lens, comprising the following steps: calculating the principal point and focal length under a single reference optical power; placing a periodic pattern calibration plate on a fixed platform; setting the lens input to the reference optical power and fixing the camera lens at a focused position; successively and equally spacing the focusing of the camera lens from the minimum optical power to the maximum optical power, and taking an image of the calibration plate each time the focusing is adjusted; using Fourier transform to extract the position of the feature points of the defocused calibration plate image during the focusing process; assuming that the principal point remains unchanged, the camera coordinate system is fixed, and the focal length and distortion parameters are in a polynomial relationship with the input optical power, and using the extracted feature point positions to calculate the camera pose and the coefficients of the polynomial function of the internal parameters; taking the obtained result as the initial value of the global optimization, performing global optimization for all input optical powers using all the obtained images, and obtaining the average reprojection error.

Description

Technical Field

[0001] The present invention relates to the field of computer vision technology, and more particularly, to a fast calibration method for a micro-vision system with a liquid zoom lens. Background Art

[0002] Compared with traditional fixed-focus lenses, the rapidly developing zoom lenses have greatly expanded the depth of field of optical systems. Due to their advantages such as fast zooming and easy miniaturization, they have been widely used in fields such as three-dimensional topography measurement, endoscopy, and microscope imaging in recent years. In these applications, the calibration of the camera system is a huge challenge because the camera system parameters change with the change of the focal length. Precise calibration results are the premise for the application of zoom lenses, but there is currently no general and mature calibration method for zoom lenses, especially in high magnification scenarios.

[0003] The simplest calibration method is to separately calibrate the internal parameters of the system at multiple focal lengths, and then use polynomial interpolation fitting to establish a polynomial function relationship between the focal length or the input value of the zoom lens and each internal parameter. This method has many problems. One is the huge workload, usually requiring taking hundreds of pictures. The second is that the depth of field of the camera system is small at high magnification, making it difficult to meet the multi-pose shooting requirements of the traditional two-dimensional calibration plate calibration method. The third is that after zooming, the calibration plate needs to be moved to the focusing position, and an external precision micro-stage is required, which is difficult to place auxiliary equipment in some scenarios. The above-mentioned disadvantages make it difficult to directly use the traditional calibration method in the calibration of micro-field zoom lenses.

[0004] An existing zoom lens calibration method proposes to use an electronic screen to design a complementary circular array target (CN202110941155.5), use the gray difference of the complementary circular array target for further processing to obtain the center coordinates, and obtain the camera internal parameters at multiple focal lengths according to the projection relationship, and then use polynomial interpolation to fit the function relationship between the internal parameters and the focal length. This method simplifies the process of extracting the center coordinates in a single pose, without the need to move the camera system or the calibration plate to the focusing position, but still requires the extraction process of multiple poses, with a large workload, and using an electronic screen to design the calibration pattern is not suitable for micro-field application scenarios. Summary of the Invention

[0005] The object of the present invention is to provide a fast calibration method for a micro-vision system with a liquid zoom lens: in the high magnification scenario of a micro-field, for any combination of focal lengths within a certain range of focal lengths, fix the camera system and the calibration plate, and take one out-of-focus calibration plate template image at each focal length, so as to solve the problems of large workload and cumbersome operation during the calibration of zoom lenses in a micro-field. Greatly reduce the calibration workload, without the need to move the calibration plate or the camera system during the zooming process, and without the need to use external auxiliary equipment, and can solve the problems existing in the calibration process of the above-mentioned calibration method in the micro-field zoom vision system.

[0006] The present invention is achieved by at least one of the following technical solutions.

[0007] A rapid calibration method for a micro-vision system with a liquid zoom lens, comprising the following steps:

[0008] Step 1, calculate the principal point and focal length under a single reference optical power;

[0009] Step 2, place the periodic pattern calibration plate on a fixed platform;

[0010] Step 3, set the input of the liquid zoom lens to the reference optical power, and fix the camera lens at a focused position;

[0011] Step 4, for N optical powers within the left and right intervals of the reference optical power, gradually and equally spaced focus the camera lens from the minimum optical power to the maximum optical power, and take an image of the calibration plate each time of focusing;

[0012] Step 5, use Fourier transform to extract the position of the feature points of the defocused calibration plate image during the focusing process;

[0013] Step 6, assuming that the principal point remains unchanged, the camera coordinate system is fixed, and the focal length and distortion parameters are in a polynomial relationship with the input optical power, use the position of the feature points extracted in Step 5 to calculate the camera pose and the coefficients of the internal parameter polynomial function;

[0014] Step 7, use the result obtained in Step 6 as the initial value of the global optimization, perform global optimization for all input optical powers using all the images obtained in Step 4, and obtain the average reprojection error.

[0015] Furthermore, use a checkerboard, a circular pattern calibration plate or a square periodic calibration plate to take multiple calibration plate images with pose changes, and use a single-focus calibration method to calibrate the principal point and focal length under the reference optical power.

[0016] Furthermore, use a square periodic calibration plate pattern calibration plate, and use the method described in Step 5 to extract the position of the feature points, and calibrate the principal point and focal length under the reference optical power.

[0017] Furthermore, in Step 2, use a square periodic pattern calibration plate, and place it at a certain position on the fixed platform, and keep the position and pose unchanged during the subsequent focusing and shooting process.

[0018] Furthermore, in Step 3, set the input of the liquid zoom lens to the reference optical power, adjust the position of the camera lens to the focused position, and fix it, and keep the position and pose of the camera unchanged during the subsequent focusing and shooting process.

[0019] Further, in the step 4, an optical power range is taken on each side of the reference optical power, and shooting starts from the minimum optical power. Focus adjustment is performed at equal-interval optical powers, and a calibration plate picture is taken at each optical power value. Among them, the picture at the reference optical power is a focused picture, and the pictures at other optical power values are defocused pictures.

[0020] Further, in the step 5, the captured pictures are transformed into the frequency domain by Fourier transform. Windows are respectively added to extract one fundamental frequency on each of the x and y axes, and then transformed back to the spatial domain by inverse Fourier transform to obtain complex graphs in two directions. The complex phase values of each point in the two graphs are respectively calculated, and the points with phase values of 2π in both graphs are extracted as the pixel coordinates of the center of the calibration plate grid, or the points with phase values of π / 2 and 3π / 2 are used as the pixel coordinates of the corner points of the calibration plate grid.

[0021] Further, in the step 6, the camera coordinate system is fixed. During the zooming process of the zoom lens, the focal length and distortion coefficient change, and the principal point remains stationary. The focal length and the input optical power value satisfy the following relationship:

[0022] f x (E j ) = k fx0 + k fx1 * E j + k fx2 * E j 2 + k fx3 * E j 3

[0023] f y (E j ) = k fy0 + k fy1 * E j + k fy2 * E j 2 + k fy3 * E j 3

[0024] In the formula, f x (E j ) and f y (E j ) are respectively the pixel length values in the x and y directions of the pixel coordinate system for the current focal length value, E j is the current optical power value input by the zoom lens drive, k fx0 and k fy0 are respectively the pixel length values in the x and y directions of the focal length value calibrated at the reference optical power value in step 1, k fx1 and k fx2 , k fx3 and kfy1 , k fy2 , k fy3 are the fitting coefficient values to be calibrated; the system internal parameter A(E j ) is expressed as:

[0025]

[0026] where A0, A1, A2, and A3 are constant coefficient matrices.

[0027] The distortion coefficient and the input optical power value satisfy the following relationship:

[0028]

[0029]

[0030]

[0031] In the formula, k 1,j is the radial distortion coefficient of the current input optical power value, p 1,j , p 2,j are the tangential distortion coefficients of the current input optical power value, a i , i = 1, 2,... 9 are the distortion model polynomial coefficients, f mean (E j ) is the average focal length pixel length of the current input optical power value, and satisfies the following relationship:

[0032] f mean (E j ) = [f x (E j ) + f y (E j )] / 2

[0033] Furthermore, in step 6, assuming that the world coordinate system is fixed on the calibration plate, the camera coordinate system is fixed, and the external parameter matrix remains unchanged, according to the relationship between the internal parameter coefficients and the input optical power value and the position of the feature points, the initial values of the external parameters and the internal parameter polynomial coefficients are calibrated using the coordinate transformation relationship and the radial arrangement constraint.

[0034] Furthermore, in step 7, using the calibration parameters in step 1 and the calibration parameters in step 6 as the input initial values of the bundle adjustment algorithm, the LM optimization algorithm is used to minimize the average reprojection error as the global optimization result, and the optimization objective function is as follows:

[0035]

[0036] where R and T represent the rotation matrix and the translation matrix respectively, A is the system internal parameter matrix, D is the distortion model polynomial coefficient, D = a i, where i = 1, 2,... 9, P d,ij is the pixel coordinates of the extracted feature points, P cal,ij is the distorted pixel coordinates obtained using the calculated parameters, P w,ij is the world coordinates of the feature points.

[0037] Compared with the existing technology, the beneficial effects of the present invention are as follows:

[0038] Quickly and simply calibrate the micro-view field zoom lens. During the calibration process, there is no need to move the calibration plate and the camera system, and there is no need to use additional auxiliary precision equipment to meet the focusing requirements during the focusing process. Only one calibration plate image needs to be taken at each focal length, which greatly reduces the calibration workload of the zoom lens. And it solves the problem that in the micro-view field with high magnification, the depth of field of the camera system is extremely small and cannot meet the requirements of multi-pose shooting of traditional calibration methods. This method solves the problems of large workload, cumbersome operation, and large calibration error during the calibration of the zoom lens in the micro-view field. Brief Description of the Drawings

[0039] Figure 1 is the theoretical model of the fast calibration method for the micro-view field zoom lens based on Fourier transform according to the embodiment of the present invention;

[0040] Figure 2 is the flow schematic diagram of the fast calibration method for the micro-view field zoom lens based on Fourier transform according to the embodiment of the present invention;

[0041] Figure 3 is the periodic pattern calibration plate according to the embodiment of the present invention;

[0042] Figure 4 is the flow schematic diagram of the feature point extraction of the defocused picture according to the embodiment of the present invention. Detailed Embodiments

[0043] The present invention will be further described below in conjunction with specific embodiments. The embodiments are only for explaining the technical concept and characteristics of the present invention, and their purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly, and cannot be used to limit the protection scope of the present invention. Any equivalent changes or modifications made according to the spirit of the present invention should be covered within the protection scope of the present invention.

[0044] As Figure 1 shown, it is the theoretical model of the present invention. Set the positions of each coordinate system and fix them. By inputting different optical power values, the liquid lens is zoomed to obtain the calibration plate patterns at different focal lengths.

[0045] As Figure 2 shown, a fast calibration method for a micro-vision system with a liquid zoom lens disclosed in this embodiment includes the following steps:

[0046] Step 1: Set the input of the liquid zoom lens to the reference optical power, capture multiple clear images of the checkerboard calibration plate in different poses, and use the single-focal length calibration method based on the traditional single-focal calibration idea to calculate the principal point and focal length under a single reference optical power, that is, obtain the initial value of A0.

[0047] Step 2: Fix the checkerboard calibration plate with a grid period as shown in Figure 3 on the fixed platform, and do not move the position of the calibration plate during the subsequent focusing and shooting process.

[0048] Step 3: Set the input of the liquid zoom lens to the reference optical power, and fix the camera lens at a focused position, and do not move the position of the camera system during the subsequent focusing and shooting process;

[0049] Step 4: For N optical powers within the left and right intervals of the reference optical power, gradually and equally spaced focus the camera lens from the minimum optical power to the maximum optical power, and capture an image of the calibration plate each time of focusing. Or start from the reference optical power and gradually increase or decrease, focus at equally spaced optical powers and capture a picture of the calibration plate at each optical power value, where the picture at the reference optical power is the focused picture, and the pictures at other optical power values are defocused pictures.

[0050] Step 5: As shown in Figure 4 , use the Fourier transform to convert all the captured images to the frequency domain, respectively window (such as Hanning window) to extract one fundamental frequency on each of the x and y axes, and then use the inverse Fourier transform to convert to the complex domain to extract the complex phase values. Take the points with phase values of 2π in both directions as the center points of the squares or the points where π / 2 and 3π / 2 are located as the corner points of the squares, and the coordinate of the feature points of the calibration plate image can be obtained.

[0051] Step 6: For the principal point and focal length extracted in Step 1, assume that the principal point remains unchanged during the focusing process, the camera coordinate system is fixed, the focal length and distortion parameters are in a polynomial relationship with the input optical power, and use the position of the feature points extracted in Step 5 to calculate the camera pose and the coefficients of the internal parameter polynomial function.

[0052] Estimate the external parameters using the radial alignment constraint, and use the least squares method to solve the unknown parameters R, t in the following formula x , t y .

[0053]

[0054]

[0055] In the formula, (x, y) is the coordinate of the feature point in the image coordinate system, (X c , Y c ) is the coordinate of the feature point in the camera coordinate system, (u, v) is the coordinate of the feature point in the pixel coordinate system, (Xw , Y w , Z w ) are the coordinates of the feature points in the world coordinate system. The subscripts ij represent the j-th feature point in the i-th image. r1 to r6 are the parameters arranged by rows in the rotation matrix, and t x , t y are the parameters in the translation matrix.

[0056] Estimate the internal parameters A1, A2, A3 and tz using the projection relationship, and the unknown parameters in the following formula can be solved by the least squares method.

[0057] [X c,ij ·E j X c,ij ·E 2 j X c,ij ·E 3 j u ij - u0][k fx1 k fx2 k fx3 t z

[0058] = (r7·X w,ij + r8·Y w,ij + r9·Z w,ij ) * (u ij - u0) - k fx0 ·X c,ij

[0059] [Y c,ij ·E j Y c,ij ·E 2 j Y c,ij ·E 3 j v ij - v0][k fy1 k fy2 k fy3 t z

[0060] = (r7·X w,ij + r8·Y w,ij + r9·Z w,ij ) * (v ij - v0) - k fy0 *Y c,ij

[0061] Using the relationship formula between the distortion parameter and the input optical power function, the distortion parameter can be solved by the following formula.

[0062] ​​

[0063]

[0064] where (u cal,ij , v cal,ij ) are the pixel coordinates of the feature points calculated using the calibrated parameter values,

[0065] (u ij , v ij ) are the pixel coordinates of the extracted feature points, k 1,j is the radial distortion coefficient of the current input optical power value, p 1,j , p 2,j are the tangential distortion coefficients of the current input optical power value, D = (a i , i = 1, 2,... 9) are the distortion model polynomial coefficients, f mean (E j ) is the average focal length pixel length of the current input optical power value, satisfying

[0066] the following relational expression:

[0067] f mean (E j ) = [f x (E j ) + f y (E j )] / 2

[0068] Step 7: Use the result obtained in Step 6 as the initial value for global optimization. Use all the images obtained in Step 4 to perform bundle adjustment for all input optical powers. Use the LM algorithm to perform global optimization on the following objective function and obtain the average reprojection error.

[0069]

[0070] where R and T respectively represent the rotation matrix and the translation matrix, A is the system internal parameter matrix, D is the distortion model polynomial coefficient, P d,ij are the pixel coordinates of the extracted feature points, P cal,ij are the distorted pixel coordinates obtained using the calculated parameters, P w,ij are the world coordinates of the feature points.

[0071] The present invention solves the problems of huge workload and cumbersome operation in calibrating a zoom lens in a micro field of view, eliminates the limitations brought about by the small depth of field and small working distance of the camera system at high magnification ratios, and can quickly and simply calibrate the parameters of the zoom system in the micro field of view, making the present invention feasible and practical. The calibration template and calibration method adopted simply and effectively solve the calibration problem of the zoom lens in the micro field of view. During the calibration process, there is no need to move the calibration plate and the camera system, and there is no need to use additional auxiliary precision equipment to meet the focusing requirements during the focusing process. Only one image of the calibration plate needs to be taken at each focal length, and there are no high requirements for the spatial size and the pose of the camera. All calibration parameters are globally optimized, making the final calibration result have high accuracy and robustness.

[0072] A rapid calibration method for a micro-vision system with a liquid zoom lens disclosed in this embodiment includes the following steps:

[0073] Step 1: Set the lens input to a reference optical power of 1.50 dpt, use a calibration plate with a grid period, and use a Fourier transform-based method to extract the positions of the feature points. Use a single-focal-length calibration method based on the traditional single-focus calibration idea to calculate the principal point and focal length at a single reference optical power, that is, obtain the initial value of A0.

[0074] Step 2: Fix the calibration plate with a grid period as shown in Figure 3 on a fixed platform, and do not move the position of the calibration plate during the subsequent focusing and shooting process.

[0075] Step 3: Set the lens input to a reference optical power of 1.50 dpt, and fix the camera lens at a focusing position, and do not move the position of the camera system during the subsequent focusing and shooting process;

[0076] Step 4: For the input optical power range of 1.10 dpt - 1.90 dpt, focus the liquid zoom lens at an optical power interval of 0.02 dpt. After each focusing, take an image of the calibration plate, and a total of 41 pictures are collected. Among them, the picture at the reference optical power is the focused picture, and the pictures at other optical power values are the defocused pictures.

[0077] Step 5: As shown in Figure 4 , use the Fourier transform to convert all the captured images into the frequency domain, respectively apply a window (such as a Hanning window) to extract one fundamental frequency on each of the x and y axes, and then use the inverse Fourier transform to convert to the complex domain to extract the complex phase value. Take the points with phase values of 2π in both directions as the center points of the grid or the points where π / 2 and 3π / 2 are located as the corner points of the grid, and the coordinate of the feature points of the calibration plate image can be obtained.

[0078] Step 6. For the principal point and focal length extracted in Step 1, assuming that the principal point remains unchanged during the focusing process, the camera coordinate system is fixed, and the focal length and distortion parameters have a polynomial relationship with the input optical power. Calculate the camera pose and the coefficients of the internal parameter polynomial function using the positions of the feature points extracted in Step 5. The focal length and the input optical power value satisfy the following relationship:

[0079] f x (E j )=k fx0 +k fx1 *E j +k fx2 *E j 2 +k fx3 *E j 3

[0080] f y (E j )=k fy0 +k fy1 *E j +k fy2 *E j 2 +k fy3 *E j 3

[0081] In the formula, f x and f y are the pixel length values of the current focal length in the x and y directions of the pixel coordinate system respectively, E j is the current optical power value input by the zoom lens drive, k fx0 and k fy0 are the pixel length values of the focal length calibrated by the reference optical power value in the x and y directions in Step 1 respectively, k fx1 , k fx2 , k fx3 , k fy1 , k fy2 , k fy3 are the fitting coefficient values to be calibrated. The internal parameters of the system are expressed as:

[0082]

[0083]

[0084] where A0, A1, A2, A3 are constant coefficient matrices.

[0085] The distortion coefficient and the input optical power value satisfy the following relationship:

[0086]

[0087]

[0088]

[0089] In the formula, k 1,j is the radial distortion coefficient of the current input optical power value, and p 1,j , p 2,j are the tangential distortion coefficients of the current input optical power value. D = (a i , i = 1, 2,... 9) are the coefficients of the distortion model polynomial, and f mean (E j ) is the average focal length pixel length of the current input optical power value, and satisfies the following relational expression:

[0090] f mean (E j ) = [f x (E j ) + f y (E j )] / 2

[0091] According to the above polynomial function relationship and the positions of the feature points extracted in step 5, use the coordinate transformation relationship and the Tsai two-step method to calibrate the external parameters and the coefficients of the above relational expression.

[0092] Step 7: Use the result obtained in step 6 as the initial value of the global optimization. Use all the images obtained in step 4 to perform beam adjustment for all input optical powers, and use the LM algorithm to globally optimize the following objective function and obtain the average reprojection error.

[0093]

[0094] where R and T respectively represent the rotation matrix and the translation matrix, A is the internal parameter matrix of the system, D is the coefficient of the distortion model polynomial, P d,ij is the pixel coordinate of the extracted feature point, P cal,ij is the distorted pixel coordinate obtained by using the calculated parameters, and P w,ij is the world coordinate of the feature point.

[0095] As a preferred embodiment, the input optical power interval can be set to 0.04 dpt, and 21 pictures are collected.

[0096] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations can be made according to the content of this specification. These embodiments are selected and specifically described in this specification in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can well understand and utilize the present invention. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. A rapid calibration method for a micro-vision system with a liquid zoom lens, characterized in that It includes the following steps: Step 1: Calculate the principal point and focal length under a single reference optical power; Step 2: Place the periodic pattern calibration plate on a fixed platform; Step 3: Set the input of the liquid zoom lens to the reference optical power, and fix the camera lens at a focused position; Step 4: For N optical powers within the range on both sides of the reference optical power, gradually and equally spacedly focus the camera lens from the minimum optical power to the maximum optical power, and capture an image of the calibration plate each time of focusing; Step 5: Use Fourier transform to extract the position of the feature points of the defocused calibration plate image during the focusing process; Step 6: Assume that the principal point remains unchanged, the camera coordinate system is fixed, the focal length and distortion parameters have a polynomial relationship with the input optical power, and use the position of the feature points extracted in Step 5 to calculate the camera pose and the coefficients of the internal parameter polynomial function; The camera coordinate system is fixed. During the zooming process of the zoom lens, the focal length and distortion coefficient change, the principal point remains fixed, and the focal length satisfies the following relational expression with the input optical power value: f x (E j ) = k fx0 + k fx1 * E j + k fx2 * E j 2 + k fx3 * E j 3 f y (E j )=k fy0 +k fy1 *E j +k fy2 *E j 2 +k fy3 *E j 3 where f x (E j ) and f y (E j ) are respectively the pixel length values of the current focal length value in the x and y directions of the pixel coordinate system, E j is the current optical power value input by the zoom lens drive, k fx0 and k fy0 are respectively the pixel length values of the focal length value calibrated by the reference optical power value in the x and y directions in step 1, k fx1 , k fx2 , k fx3 , k fy1 , k fy2 , k fy3 are the fitting coefficient values to be calibrated; the system internal parameter A(E j ) is expressed as: where A0, A1, A2, and A3 are constant coefficient matrices; The distortion coefficient satisfies the following relational expression with the input optical power value: where k 1,j is the radial distortion coefficient of the current input optical power value, p 1,j , p 2,j are the tangential distortion coefficients of the current input optical power value, a i , i = 1, 2, … 9 are the distortion model polynomial coefficients, f mean (E j ) is the average focal length pixel length of the current input optical power value, and satisfies the following relationship: f mean (E j ) = [f x (E j ) + f y (E j )] / 2 Step 7: Use the result obtained in Step 6 as the initial value of the global optimization, perform global optimization for all input optical powers using all the images obtained in Step 4, and obtain the average reprojection error; In Step 7, using the calibration parameters in Step 1 and the calibration parameters in Step 6 as the input initial values of the bundle adjustment algorithm, use the LM optimization algorithm to minimize the average reprojection error as the global optimization result, and the optimization objective function is as follows: where R and T represent the rotation matrix and the translation matrix respectively, A is the internal parameter matrix of the system, D is the polynomial coefficient of the distortion model, D = a i , i = 1, 2,... 9, P d,jm is the pixel coordinate of the extracted feature point, P cal,jm is the distorted pixel coordinate obtained by using the calculated parameters, P w,jm is the world coordinate of the feature point.

2. The rapid calibration method of a micro-vision system with a liquid zoom lens according to claim 1, characterized in that Use a checkerboard, circular pattern calibration plate or square periodic calibration plate to capture multiple calibration plate images with pose changes, and use the single-focus calibration method to calibrate the principal point and focal length under the reference optical power.

3. A rapid calibration method for a micro-vision system with a liquid zoom lens according to claim 1, characterized in that, Use a square periodic calibration plate pattern calibration plate, and use the method described in Step 5 to extract the position of the feature points, and calibrate the principal point and focal length under the reference optical power.

4. A rapid calibration method for a micro-vision system with a liquid zoom lens according to claim 1, characterized in that In Step 2, use a square periodic pattern calibration plate and place it at a certain position on the fixed platform, and keep the position and pose unchanged during the subsequent focusing and shooting process.

5. A rapid calibration method for a micro-vision system with a liquid zoom lens according to claim 1, characterized in that, In Step 3, set the input of the liquid zoom lens to the reference optical power, adjust the position of the camera lens to the focused position and fix it, and keep the camera position and pose unchanged during the subsequent focusing and shooting process.

6. A rapid calibration method for a micro-vision system with a liquid zoom lens according to claim 1, characterized in that, In Step 4, take a range of optical power on both sides of the reference optical power, start shooting from the minimum optical power, focus at equally spaced optical powers and capture a calibration plate picture at each optical power value, where the picture at the reference optical power is a focused picture, and the pictures at other optical power values are defocused pictures.

7. A rapid calibration method for a micro-vision system with a liquid zoom lens according to claim 1, characterized in that In Step 5, use Fourier transform to convert the captured pictures to the frequency domain, respectively window and extract one fundamental frequency on each of the x and y axes, and use inverse Fourier transform to convert to the spatial domain to obtain two complex pictures in two directions. Calculate the complex phase value of each point in the two pictures respectively, and extract the points with a phase value of 2π in both pictures as the pixel coordinates of the center of the calibration plate grid, or the points with π / 2 and 3π / 2 as the pixel coordinates of the corner points of the calibration plate grid.

8. A rapid calibration method for a micro-vision system with a liquid zoom lens according to claim 1, characterized in that, In step 6, it also includes setting the world coordinate system fixed on the calibration board, keeping the camera coordinate system fixed and the external parameter matrix unchanged, and calibrating the initial values of the external parameter and the internal parameter polynomial coefficients according to the relationship between the internal parameter coefficients and the input optical power value and the feature point positions, using the coordinate transformation relationship and the radial arrangement constraint.

Images