Large-view-field light field microscopic image distortion correction method and system

Abstract

The invention discloses a large-view-field light field microscopic image distortion correction method and system, and belongs to the technical field of computer vision, and the method comprises the steps: obtaining a calibration image, and extracting a center coordinate of a central micro lens as a feature point; constructing an ideal circle center grid and an actual circle center grid according to the feature points, and obtaining grid coordinates; the two-dimensional nonlinear distortion model comprises an external parameter matrix, an internal parameter matrix, radial distortion and tangential distortion, and elements in the external parameter matrix comprise image rotation angle parameters and translation parameters; elements in the internal reference matrix comprise scaling parameters and optical center position parameters; substituting the normalized ideal circle center grid coordinates and the actual circle center grid coordinates into the two-dimensional nonlinear distortion model, and solving to obtain distortion parameters of the light field microscope; and performing correction and remapping based on the obtained distortion parameters. According to the method and the system, the distortion correction precision of the large-view-field light field microscopic image is improved.

Description

Technical Field

[0001] This invention relates to the field of computer vision technology, and more specifically to a method and system for correcting distortion in large field-of-view light field microscopic images. Background Technology

[0002] Light field microscopy is an imaging technique based on microlens arrays. By introducing a microlens array between the microscope objective and the imaging sensor, it can simultaneously record the spatial and angular information of a sample in a single exposure, thereby achieving three-dimensional volume reconstruction and high-speed dynamic imaging. Because light field microscopy can acquire multi-view information in a single acquisition, it is widely used in fields requiring high temporal resolution, such as neuroscience, live-cell imaging, and developmental biology.

[0003] However, in large-field light field microscopy imaging, significant geometric distortions often exist in the acquired light field images due to factors such as the design limitations of the optical system itself, aberrations of the microscope objectives, geometric deviations of the microlens array, and projection distortion introduced by the large-size sensor. These distortions are relatively insignificant under small-field conditions, but are significantly amplified in large-field imaging, manifesting as inconsistent sampling angles of microlenses at different spatial locations, deformation of the reconstructed three-dimensional voxels, and structural distortion in edge regions.

[0004] Traditional distortion correction methods are primarily based on two-dimensional planar projection models. They typically involve acquiring calibration plate images in single-view microscopy and then using polynomial fitting or perspective transformation to correct global distortion. However, the imaging principle of light field microscopy differs from ordinary microscopy; the images simultaneously encode spatial and angular information. Correction methods based solely on planar geometry cannot effectively handle the complex distortions introduced by multi-view sampling. Furthermore, under large field-of-view conditions, a single global distortion model struggles to account for the differences between central and peripheral regions, easily leading to over-correction or under-correction in local areas, thus reducing the accuracy and resolvability of subsequent 3D reconstruction.

[0005] Therefore, in order to address the problem of spatial non-uniform distortion caused by non-ideal factors such as optical systems, microlens arrays and sensors in large field-view light field microscopy, there is an urgent need for a distortion correction method that can combine the characteristics of light field imaging and take into account the features of multiple viewpoints and multiple regions, so as to improve the quality of light field reconstruction and the spatial accuracy of microscopic images. Summary of the Invention

[0006] In view of the above problems, the present invention is proposed to provide a method and system for correcting distortion of large field-of-view light field microscopic images that overcomes or at least partially solves the above problems.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, embodiments of the present invention provide a method for correcting distortion in large field-of-view light field microscopic images, comprising the following steps: S1. Obtain the calibration image and extract the coordinates of the center of the central microlens as feature points; S2. Construct ideal and actual circle center grids based on feature points, and obtain the grid coordinates of the ideal and actual circle center grids respectively; S3. Establish a two-dimensional nonlinear distortion model including an extrinsic parameter matrix, an intrinsic parameter matrix, radial distortion, and tangential distortion, wherein the elements in the extrinsic parameter matrix include image rotation angle parameters and translation parameters; the elements in the intrinsic parameter matrix include scaling parameters and optical center position parameters. S4. Substitute the normalized ideal center grid coordinates and the actual center grid coordinates into the two-dimensional nonlinear distortion model, and solve for the distortion parameters of the light field microscope. S5. Based on the obtained distortion parameters, the distortion correction of the original light field microscope image to be corrected is realized.

[0008] Furthermore, step S1 specifically includes: Obtain a blank two-dimensional light field map as a calibration image; Based on the obtained two-dimensional light field map, the coordinates of the center of the microlens of the microlens array are obtained and used as feature points of the calibration image.

[0009] Further, obtaining a blank two-dimensional light field map specifically includes: Under white light extended light source illumination, a two-dimensional light field map with blank, uniform illumination was obtained by taking a picture using a light field microscope.

[0010] Furthermore, based on the obtained two-dimensional light field map, the coordinates of the center of the central microlens of the microlens array are obtained, specifically including: Based on the spatial period N of the microlens array in the image coordinate system, a template of size N*N matching the diameter of the microlens is generated, and the template contains a circular region of size N. Extract a local image patch of size 2N*2N from the geometric center region of the two-dimensional light field map; The generated template is cross-correlated with the pixel values ​​of the central local image block using normalized methods to obtain the coordinates of the center of the microlens array. These coordinates can be denoted as: .

[0011] Furthermore, in S2, constructing the ideal center grid and obtaining the grid coordinates of the ideal center grid specifically includes: Based on the spatial period N of the microlens array in the image coordinate system, the spacing step between adjacent centers in the ideal circle grid is determined, wherein the spacing step is an integer multiple of the spatial period N; Coordinates of the center of the central microlens Starting from the point, the points are expanded outwards according to the interval step to obtain discrete two-dimensional points with a step size step as the ideal center grid. The coordinate index of each grid position in the ideal center grid is calculated according to the coordinates of the center of the central microlens.

[0012] Furthermore, in S2, constructing the actual center grid and obtaining the grid coordinates of the actual center grid specifically includes: Use the center coordinates of the central microlens as the center starting point and record its grid index; create a distCoor actual center grid coordinate mapping to store the actual center coordinates corresponding to each grid index. Simultaneously, create a queue and add the center starting grid index to the queue; Create a visited collection and add the starting grid index to the visited collection. While the queue is not empty, the following loop is executed: a. Retrieve a grid node from the queue and use it as the current node; b. Obtain the actual center coordinates of the circle corresponding to the current node; c. Traverse the four adjacent directions of the current node, and check whether the adjacent nodes are within the grid range and not in the visited set; if the conditions are not met, skip the adjacent node; If the conditions are met, proceed with the following steps: Based on the actual center coordinates of the current node and the theoretical distance step, predict the center coordinates of the adjacent nodes. Using the center coordinates of adjacent nodes as the center, extract a local image patch of size 2N×2N from the original image; The actual center coordinates of adjacent nodes in the local image block are obtained using the template matching method. Store the actual center coordinates of the adjacent nodes into the corresponding adjacent node index position in the actual center grid coordinate mapping distCoor; Mark the adjacent node as visited and add it to the visited set `visited`. At the same time, add the adjacent node to the queue. When the queue is empty, the search ends, and the actual center grid coordinates that correspond one-to-one with the ideal center grid coordinates are finally obtained.

[0013] Furthermore, a two-dimensional nonlinear distortion model is established, including extrinsic parameter matrices, intrinsic parameter matrices, radial distortion, and tangential distortion, specifically including the following model: Extrinsic parameter matrix:

[0014] Intrinsic parameter matrix:

[0015] Matrix transformation:

[0016] Radial distortion:

[0017]

[0018] Tangential distortion:

[0019]

[0020] in, Represents the extrinsic parameter matrix. This represents the intrinsic parameter matrix, where x and y are the standard horizontal and vertical coordinates. , The changed x and y coordinates Represents radial distortion. This represents intermediate variables in the process of obtaining radial distortion, where x_dist and y_dist are the x and y coordinates of the distortion. Indicates the image rotation angle. Indicates horizontal translation of the image. Indicates vertical translation of the image. This represents the horizontal scaling value of the image. This represents the vertical scaling value of the image. Represents the x-coordinate of the optical center. Represents the ordinate of the optical center. Both represent radial distortion coefficients. Both represent radial distortion coefficients.

[0021] Furthermore, the normalized ideal center grid coordinates and the actual center grid coordinates are substituted into the two-dimensional nonlinear distortion model, and the distortion parameters of the light field microscope are solved, specifically including: Normalize the coordinates of the ideal center grid and the actual center grid respectively:

[0022] in, Represents the x-coordinate of the ideal circle center grid. Represents the ordinate of the ideal circle center grid. This represents the x-coordinate of the actual circle center grid. This represents the ordinate of the actual circle center grid. This represents the normalized x-coordinate of the ideal circle center. This represents the normalized ordinate of the ideal circle's center. This represents the normalized x-coordinate of the actual circle center. This represents the normalized ordinate of the actual center of the circle. Normalized ideal circle center grid coordinates and As standard coordinates, the normalized actual circle center grid coordinates and As the distortion coordinates, they are substituted into the distortion model `distort_model`, and the distortion parameters `params` are obtained by least-squares fitting. Specifically, they can be expressed as the following formula:

[0023] in, These represent the height and width of the ideal circular grid, respectively.

[0024] Secondly, embodiments of the present invention provide a large field-of-view light field microscopic image distortion correction system, including a computer system, wherein the computer system is capable of implementing the large field-of-view light field microscopic image distortion correction method as described in any one of the first aspects of the present invention when it is running.

[0025] As can be seen from the above technical solution, compared with the prior art, the present invention discloses a method and system for correcting distortion of large field-of-view light field microscopic images, which has the following beneficial effects: In the process of large field-of-view light field microscopic image distortion correction, the present invention fully considers rotation, translation, scaling, radial and tangential distortion, and can achieve sub-pixel level correction under large field-of-view conditions, thereby improving the distortion correction accuracy of large field-of-view light field microscopic images.

[0026] This invention employs a combination of gridded template matching and breadth-first search for mesh creation, avoiding edge region distortion caused by a single global model. This invention eliminates the need for manual target calibration and automatically estimates distortion parameters based on the original white field image; This invention is applicable to the angular sampling characteristics of light field microscopes, which helps to improve the spatial accuracy of subsequent 3D reconstruction. Attached Figure Description

[0027] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0028] Figure 1This is a schematic diagram of the overall process of the distortion correction method provided in the embodiments of the present invention.

[0029] Figure 2 This is a schematic diagram of the template matching method provided in an embodiment of the present invention. Detailed Implementation

[0030] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0031] This invention addresses the geometric distortion problem in images acquired by light field microscopy under large field-of-view conditions caused by factors such as optical system non-ideality and sensor projection distortion. It discloses a method for correcting image distortion in large field-of-view light field microscopy images. This method can simultaneously estimate the extrinsic, intrinsic, and higher-order distortion parameters of the optical system, achieving accurate geometric correction of the light field image and improving the accuracy of subsequent volume reconstruction and multi-view registration. This method is based on grid calibration, and the overall steps are described in [reference needed]. Figure 1 Specifically, it includes the following steps: Capture data and determine calibration images and parameter initialization: Illumination is achieved using a white light extended source, without a sample, to capture a two-dimensional light field image as a calibration image. In one specific embodiment, the dimensions of the two-dimensional light field image are recorded as H*W, and each microlens in the microlens array corresponds to N*N pixels.

[0032] Center point detection: Based on the known period of the microlens array being N pixels, the spatial period of the microlens array in the image coordinate system generates a circular region template that matches the diameter of the microlenses. The overall size of the generated circular region template is N*N, and the circular region template contains a circular region with a diameter of N. The values ​​inside the circular region are 1, and the values ​​outside are 0. A local block with a size of 2N*2N is extracted from the geometric center region of the two-dimensional light field map; the coordinates of the center of the microlens array are obtained by matching this local block using a circular region template matching method. : This is used for subsequent coordinate normalization and global distortion calibration; Specifically, the cross-correlation of pixel values ​​between the circular template image and the local block image is calculated, and the position of the maximum cross-correlation value is taken as the center coordinate of the central microlens; for example... Figure 2 The template circle ( Figure 2The size of the red circle in the middle is the same as the size of the target circle. The template circle slides within the target area. The product of the two circles is maximized only when the template circle and the target circle coincide, and the target position (coordinates of the center of the central microlens circle) can be determined.

[0033] Constructing the ideal center grid and the actual center grid: First, in the step of constructing the ideal center grid, the ideal center grid is a discrete two-dimensional grid centered at the coordinates centerPT of the central microlens with a step size of step. Based on the image size H and the distance step between adjacent center points, the sizes of the ideal center grid and the actual center grid are determined. Where step is an integer multiple of N, and the specific formula for calculating step is as follows: ; After determining the distance between adjacent centers (step) and the center coordinates (centerPT) of the central microlens in the ideal center grid, the ideal center grid coordinates (gtCoor) are generated, where the size of the ideal center grid is... .

[0034] Specifically, the row and column indices of the ideal circle center grid coordinates gtCoor are defined as follows:

[0035] ; Where i represents the row number of the grid (vertical direction) and j represents the column number (horizontal direction).

[0036] Coordinates of the center of the central microlens The grid index is represented as: ; For each grid position of the ideal circle center grid coordinates Their position coordinates are represented as follows: ; .

[0037] Secondly, in the step of constructing the actual center mesh, breadth-first search can be used. In this step, a breadth-first search structure is constructed starting from the central microlens, expanding layer by layer according to the adjacency relationships in all directions. At each step, local image patches are extracted and template matching is performed, thereby sequentially determining the actual center of the central microlens in different local regions, obtaining the actual center mesh coordinates (distCoor). The size of the actual center mesh is... The row and column indices of the actual center grid coordinates (distCoor) in the actual center grid are defined as follows:

[0038]

[0039] Where i represents the row number of the grid (vertical direction) and j represents the column number (horizontal direction).

[0040] The actual center grid is also centered on the coordinates of the center microlens, centerPT. The grid index is: .

[0041] In practice, the actual center grid coordinates can be obtained by constructing a breadth-first search structure starting from the central microlens. The specific steps are described in detail below.

[0042] First, initialize the queue and access set, taking the center coordinates of the central microlens (centerPT) as the starting point and recording its row and column indices. It also creates an actual center grid coordinate mapping distCoor to store the actual center coordinates corresponding to each grid index; and creates a queue and adds the starting grid index to the queue.

[0043] Next, create a visited set and add the starting grid index to the visited set.

[0044] Then define the adjacency relationship for any node. Its adjacent nodes are defined as adjacent grid nodes in the four directions of up, down, left, and right with indices of (i-1, j), (i+1, j), (i, j-1), and (i, j+1).

[0045] Next, we proceed with a layer-by-layer expansion search: When the queue is not empty, execute the following loop: a. Take a grid node (cur_i, cur_j) from the queue.

[0046] b. Obtain the actual center coordinates (current_center) of the node. The actual center coordinates are represented as follows: current_center = distCoor[(cur_i, cur_j)]. c. Traverse the four adjacent directions of the current node: First, calculate the grid index of the adjacent node. The grid index of the adjacent node is obtained using the following formula: (nei_i, nei_j) = (cur_i + di, cur_j + dj), where (di, dj) are the direction offsets, and the direction offsets are (-1, 0), (1, 0), (0, -1), (0, 1) in sequence.

[0047] Then, check whether the adjacent node is within the grid range and not in the visited set visited. If the condition is not met, skip this adjacent node. To check whether the adjacent node is within the grid range, it is to determine whether the adjacent node meets the following conditions: 0 <= nei_i < rows and 0 <= nei_j < cols, where rows and cols represent the maximum row and column ranges of the grid respectively.

[0048] If the adjacent node is valid and unvisited, perform the following operations: i. Predict the center coordinates of the adjacent node according to the actual center coordinates of the current node and the theoretical spacing step. The center coordinates of the adjacent node predicted_center can be obtained through the following expression: predicted_center = (current_center[0] + dj * step, current_center[1]+di * step). ii. Extract a local image patch of size 2N×2N in the original image centered at the center coordinates predicted_center of the adjacent node. iii. Use a circular template matching method, such as the correlation matching method, to accurately locate the center in this local image patch to obtain the actual center coordinates actual_center of the adjacent node. iv. Store the actual center coordinates actual_center of the adjacent node at the corresponding adjacent node index position in distCoor: distCoor[(nei_i, nei_j)] = actual_center. v. Mark the adjacent node as visited: visited.add((nei_i, nei_j)). vi. Add the adjacent node to the queue: queue.put((nei_i, nei_j)).

[0049] When the queue is empty, the search ends. At this time, the visited set visited contains all valid grid nodes (i.e., rows×cols nodes). If all grid nodes in the microlens array are valid, then the size of visited should be equal to the total number of grid nodes.

[0050] Finally, obtain the actual center grid coordinates distCoor that correspond one-to-one with the ideal center grid coordinates gtCoor.

[0051] This invention employs a breadth-first search (BFS) strategy, starting from the known center of the microlens array and progressively expanding outwards to locate the actual center coordinates of the entire array. Utilizing the spatial continuity of the located centers, the precise positions of adjacent microlenses are determined sequentially through local template matching, ultimately achieving efficient and accurate positioning of the actual center.

[0052] Establish a distortion model: The distortion model constructed in this invention is a two-dimensional nonlinear distortion model, distort_model, which can be represented as:

[0053] Where x and y are standard coordinates, and x_dist and y_dist are distorted coordinates. Distortion parameters, distortion parameters Includes: rotation Translation ( ), scaling ), optical center position ( Radial distortion coefficient ( ) and tangential distortion coefficient ( The initial value of the distortion parameter is set to 0. The distortion parameter is obtained by using the ideal center grid coordinates and the actual center grid coordinates, and the original light field microscope image to be corrected is then corrected.

[0054] Specifically, in this invention, the constructed two-dimensional nonlinear distortion model `distort_model` includes extrinsic parameter matrices, intrinsic parameter matrices, radial distortion, and tangential distortion. First, the intrinsic and extrinsic parameter matrices are calculated, and then the distortion is calculated. Extrinsic parameter matrix:

[0055] Intrinsic parameter matrix:

[0056]

[0057] Radial distortion:

[0058]

[0059] Tangential distortion:

[0060]

[0061] The distortion parameters are solved based on the obtained ideal and actual center grid coordinates: First, the ideal center grid coordinates and the actual center grid coordinates are normalized. Specifically:

[0062] gtCoor is the ideal center grid coordinate defined above, with three dimensions. , representing the x and y coordinates of the row, column, and point respectively; gtCoor[:,:,0] means taking the first value of the third dimension, that is, the x coordinate of each point.

[0063] Then, the normalized ideal circle center grid coordinates are... , and actual circle center grid coordinates , Substituting these values ​​into the distortion model `distort_model` (i.e., using the ideal center grid coordinates and the actual center grid coordinates as x, y and x_dist, y_dist respectively), the distortion parameters `params` are obtained using least-squares fitting. The specific formula is as follows:

[0064] Distortion correction and remapping: Within the entire image area, construct an ideal coordinate system based on a pixel-level grid: ; For each ideal coordinate, its corresponding distorted coordinate in the distorted image is calculated based on the distorted parameters params obtained from the fitting:

[0065] Global distortion correction is achieved on the original image using bilinear interpolation.

[0066] Based on the same inventive concept, this invention also provides a large field-of-view light field microscope image distortion correction system, including a computer system. When the system is executed, it can implement the distortion correction method of this invention. Therefore, the implementation of this system can refer to the implementation of the aforementioned method, and repeated details will not be described again.

[0067] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0068] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for large field of view light field microscopic image distortion correction, the method comprising: The method comprises the following steps: S1, acquiring a calibration image and extracting the center microlens circle center coordinates as feature points; S2, constructing an ideal circle center grid and an actual circle center grid according to the feature points, and acquiring grid coordinates of the ideal circle center grid and the actual circle center grid respectively; S3, establishing a two-dimensional nonlinear distortion model comprising an external parameter matrix, an internal parameter matrix, a radial distortion and a tangential distortion, wherein the elements in the external parameter matrix comprise an image rotation angle parameter and a translation parameter; the elements in the internal parameter matrix comprise a scaling parameter and a light center position parameter; S4, bringing the normalized ideal circle center grid coordinates and the actual circle center grid coordinates into the two-dimensional nonlinear distortion model, and solving to obtain distortion parameters of the light field microscope; S5, based on the obtained distortion parameters, realizing distortion correction of the original light field microscope image to be corrected.

2. The method of claim 1, wherein, S1 specifically comprises: acquiring a blank two-dimensional light field image as a calibration image; based on the obtained two-dimensional light field image, obtaining the center microlens circle center coordinates of the microlens array as the feature points of the calibration image.

3. The method of claim 2, wherein, Acquiring a blank two-dimensional light field image specifically comprises: under the illumination condition of a white light extended light source, using a light field microscope to take a picture to obtain a blank and uniformly illuminated two-dimensional light field image.

4. The method of claim 2, wherein, Based on the obtained two-dimensional light field image, the center microlens circle center coordinates of the microlens array are obtained, specifically comprising: based on the spatial period N of the microlens array in the image coordinate system, a template with a size of N*N matching the diameter of the microlens is generated, and the template contains a circular region directly with N; a 2N*2N center local image block is cut in the geometric center region of the two-dimensional light field image; the generated template and the center local image block are subjected to pixel value normalization cross-correlation calculation to obtain the center microlens circle center coordinates of the microlens array.

5. The method of claim 1, wherein, In S2, the grid coordinates of the ideal circle center grid are obtained, specifically comprising: based on the spatial period N of the microlens array in the image coordinate system, the interval step between adjacent circle centers in the ideal circle center grid is determined, and the interval step step is an integer multiple of the spatial period N; taking the center microlens circle center coordinates as the starting point and expanding according to the interval step step, a discrete two-dimensional point with a step of step is obtained as the ideal circle center grid, and the coordinate index of each grid position in the ideal circle center grid is calculated according to the center microlens circle center coordinates.

6. The method of claim 1, wherein, In S2, the grid coordinates of the actual circle center grid are obtained, specifically comprising: taking the center microlens circle center coordinates as the center starting point, recording the grid index; creating an actual circle center grid coordinate mapping distCoor for storing the actual circle center coordinates corresponding to each grid index; an queue queue is created at the same time, and the center starting point grid index is added to the queue queue; create a visited set visited and add the starting grid index to the visited set visited when the queue queue is not empty, the following loop is executed: a. taking a grid node from the queue queue as a current node; b. obtaining the actual circle center coordinates corresponding to the current node; c. traversing the four adjacent directions of the current node, checking whether the adjacent nodes are within the grid range and not in the visited set visited; if the conditions are not met, skip the adjacent nodes; If the conditions are met, proceed with the following steps: Based on the actual center coordinates of the current node and the theoretical distance step, predict the center coordinates of the adjacent nodes. Using the center coordinates of adjacent nodes as the center, extract a local image patch of size 2N×2N from the original image; The actual center coordinates of adjacent nodes in the local image block are obtained using the template matching method. Store the actual center coordinates of the adjacent nodes into the corresponding adjacent node index position in the actual center grid coordinate mapping distCoor; The adjacent nodes are marked as visited and added to the visited set, and the adjacent nodes are also added to the queue. When the queue is empty, the search ends, and the actual center grid coordinates that correspond one-to-one with the ideal center grid coordinates are finally obtained.

7. The method of claim 1, wherein, A two-dimensional nonlinear distortion model is established, including extrinsic parameter matrices, intrinsic parameter matrices, radial distortion, and tangential distortion. Specifically, the model includes the following: Extrinsic parameter matrix: Intrinsic parameter matrix: Matrix transformation: Radial distortion: Tangential distortion: wherein, denotes an extrinsic matrix, denotes an intrinsic matrix, x, y are standard horizontal and vertical coordinates, , are changed horizontal and vertical coordinates, denotes a radial distortion variable, denotes an intermediate variable in the process of obtaining the radial distortion variable, x_dist, y_dist are distorted horizontal and vertical coordinates, denotes an image rotation angle, denotes an image horizontal translation, denotes an image vertical translation, denotes an image horizontal scaling value, denotes an image vertical scaling value, denotes an optical center horizontal coordinate, denotes an optical center vertical coordinate, both denote a radial distortion coefficient, both denote a radial distortion coefficient.

8. The method of claim 1, wherein, The normalized ideal center grid coordinates and the actual center grid coordinates are substituted into the two-dimensional nonlinear distortion model, and the distortion parameters of the light field microscope are obtained by solving the model. Specifically, the distortion parameters include: Normalize the coordinates of the ideal center grid and the actual center grid, respectively; The normalized ideal center grid coordinates are used as standard coordinates, and the normalized actual center grid coordinates are used as distortion coordinates. These are substituted into the two-dimensional nonlinear distortion model constructed by S3, and the distortion parameters are obtained by least squares fitting.

9. A large field of view light field microscopic image distortion correction system, characterized in that, The system includes a computer system that, when running, can implement the large field-of-view light field microscopic image distortion correction method according to any one of claims 1-8.

Images