A structural light positioning method for intraoral three-dimensional reconstruction

Abstract

The present invention discloses a structural light calibration method for intra-oral three-dimensional reconstruction, comprising the following steps: step 1, initializing the optical system parameters of a three-dimensional scanner to obtain the initialization parameters; step 2, calculating the origin camera factor and the origin projector factor according to the initialization parameters; step 3, performing a calibration round according to the origin camera factor and the origin projector factor to obtain a calibration result; step 4, storing the calibration result.

Description

Technical Field

[0001] The present invention relates to the field of three-dimensional scanning technology, and in particular to a structural light positioning method for three-dimensional reconstruction in an oral cavity. Background Art

[0002] Traditional oral medical technology is mainly based on manual operation by doctors, which is often inefficient and causes discomfort to patients. Nowadays, many dental clinics have introduced advanced 3D scanners, and scanner calibration is a very critical step. Fast and accurate calibration methods are the prerequisite for achieving high-precision 3D reconstruction. The accuracy of the calibration results and the stability of the algorithm directly affect the accuracy of the results produced by the camera.

[0003] Nowadays, commonly used non-contact optical 3D scanning technologies include structured light, binocular vision, time-of-flight (TOF) and other technologies. The binocular vision system has a large amount of calculation and is very time-consuming; the TOF camera has a low resolution and is very expensive; while the structured light solution is mature, has a fast response speed, is low in cost, has high accuracy and resolution within a certain range, and is widely used in oral scanners. The principle of the structured light camera is to project specific coded stripes onto the surface of the object. The camera captures the projection pattern on the surface of the object and analyzes and decodes it. According to the matching relationship between the structured light image and the projection pattern, as well as the calibration parameters of the camera, the 3D coordinates of the object surface are calculated. After obtaining the point cloud, the reconstruction of the 3D model of the object can be completed through steps such as texture stitching and rendering.

[0004] In practical applications, there are many undesirable aspects in design and assembly that will introduce additional errors. The most obvious ones are deformation caused by the optical system and inaccurate positioning of components during assembly, which will cause errors in calibration and affect the accuracy of subsequent 3D point clouds and reconstructed dental models. Most existing calibration methods use checkerboard targets, which require positioning the camera multiple times at different distances and angles. The placement needs to cover all positions in the image and require as many accurate postures as possible. The corner point extraction accuracy is often insufficient, and it is also difficult to implement as a factory program. Summary of the invention

[0005] The present invention is made to solve the above-mentioned problem, and its purpose is to provide a structural light positioning method for three-dimensional reconstruction in the oral cavity.

[0006] The present invention provides a structural light calibration method for intra-oral three-dimensional reconstruction, which has the following characteristics and includes the following steps: Step 1, initializing the optical system parameters of the three-dimensional scanner to obtain the initialization parameters; Step 2, calculating the origin camera factor and the origin projector factor according to the initialization parameters; Step 3, performing a calibration round according to the origin camera factor and the origin projector factor to obtain the calibration result; Step 4, storing the calibration result.

[0007] The structured light positioning method for intraoral three-dimensional reconstruction provided by the present invention may also have the following characteristics: wherein, in step 1, the optical system parameters of the three-dimensional scanner are as shown in Table 1:

[0008] Table 1 Model parameters

[0009]

[0010]

[0011] Among them, the geometric parameters determined according to physical measurements are parameters P 11 , parameter P 12 And the parameter P 13 The geometric parameters determined by image measurement are parameter P9, parameter P 23 , parameter P 24 , parameters P1 to P6, parameter P 19 ~Parameter P 22 The initialization parameter is 0, the parameter P 23 is estimated based on the scale factor between the imager and the light source.

[0012] In the structural light positioning method for intraoral three-dimensional reconstruction provided by the present invention, it can also have the following characteristics: wherein, in step 2, the origin camera factor is the magnification between the three-dimensional target and the image, and the origin projector factor is the image center and the line angle. Specifically, it is divided into the following sub-steps: step 2-1, reading the image at position z=0; step 2-2, organizing the image, obtaining the tissue image; step 2-3, calculating the cross center; step 2-4, calculating the magnification between the three-dimensional target and the image according to the tissue image and the cross center; step 2-5, calculating the image center and the line angle according to the tissue image and the cross center.

[0013] In the structured light positioning method for intraoral three-dimensional reconstruction provided by the present invention, it can also have the following characteristics: wherein, in step 2-3, when calculating the cross center, the approximate center of the cross is searched on the tissue image, and then it is refined twice to determine the center of each cross in the x and y directions, and the main process is as follows: step 2-3-1, upsampling the image to normal size; step 2-3-2, finding all rough cross centers; step 2-3-3, refining the rough cross center. Step 2-4 includes the following sub-steps: step 2-4-1, calculating the distance between the two nearest cross centers; step 2-4-2, calculating the magnification between the three-dimensional target and the image. Step 2-5 includes the following sub-steps: step 2-5-1, upsampling the image to normal size; step 2-5-2, subtracting the cross line background from the image; step 2-5-3, detecting the peak value, converting the grayscale image peak value into a binary image peak value; step 2-5-4, detecting a horizontal line and a vertical line in the binary image; step 2-5-5, refining the horizontal line and the vertical line to obtain the image center and line angle.

[0014] The structural light positioning method for intra-oral three-dimensional reconstruction provided by the present invention may also have the following features: wherein, in step 2-3-2, the process of finding all rough cross centers is: moving the image of the cross to each pixel in the image, and calculating the correlation between each pair of pixels to obtain a cross-correlation graph, wherein the peak in the cross-correlation graph is the center of the cross within these two pixels.

[0015] In the structural light calibration method for intra-oral three-dimensional reconstruction provided by the present invention, it can also have the following characteristics: wherein, step 3 includes the following sub-steps: step 3-1, calculating the relevant information of the corresponding points based on the origin camera factor and the origin projector factor; step 3-2, detecting the stability of the workbench based on the relevant information of the corresponding points; step 3-3, calibration optimization: using the matrix to optimize the redundant calibration parameters and minimize the mean square error between the cross on the observed target and the cross at the predicted position; step 3-4, updating the calibration results to system parameters.

[0016] In the structural light calibration method for intraoral three-dimensional reconstruction provided by the present invention, it can also have the following features: wherein step 3-1 includes the following sub-steps: step 3-1-1, calculating the phase map according to the origin camera factor and the origin projector factor; step 3-1-2, finding the crosshairs with accurate positions according to the phase map; step 3-1-3, setting the two-dimensional image points and their phases according to the crosshairs; step 3-1-4, saving the two-dimensional image points and their phases, and saving the phase map on each plane. Among them, in step 3-1-1, the process of calculating the phase map is: there is a small angle between the column directions of the projector and the camera, so the coordinates on the camera sensor are calibrated by the following formula to compensate:

[0017]

[0018] For the three-dimensional coordinates of the object surface, suppose a point (u s , v s ) is projected onto a 3D point (x, y, z) on the target, and the camera captures the coordinates (u i , v i ), the reconstruction of the 3D points is as follows:

[0019] x=η i .τ i =-η i ·(d i -z) (2)

[0020] y=ξ i .τ i =-ξ i ·(d i -z) (3)

[0021]

[0022] Where:

[0023]

[0024]

[0025]

[0026]

[0027] At the same time, for the distortion of the lens module, a heuristic polynomial distortion correction is used at z:

[0028] z=z+k0xy+k1xy 2 +k2x 3 (9)

[0029] so

[0030] r=(x,y,z)=f(Θ,u i ,v i ,u s ) (10)

[0031] Θ represents all the parameters mentioned above. Since the surface of the calibration plate is not always parallel to the image plane, assuming that [R, T] is the transformation matrix from the scanner coordinate system to the target coordinate system, the following equation can be minimized to optimize the parameter Θ:

[0032]

[0033] Where r i is the ideal target 3D point with known position in the target system, (u i ,v i ,u s ) is the phase information extracted from the captured image.

[0034] In step 3-1-3, when setting the two-dimensional image points and their phases, the correspondence between the camera and the projector is first generated for each cross using local bilinear interpolation, and then the three-dimensional coordinates are generated for each cross. In step 3-2, if there are outliers in the sequence captured at a certain range of calibration positions, the point is unstable, and if more than 5 of the 9 corresponding points are unstable at one position, the position is unreliable.

[0035] In the structural light calibration method for intra-oral three-dimensional reconstruction provided by the present invention, it can also have the following features: wherein, step 3-3 includes the following sub-steps: step 3-3-1, calculate the position of the ideal cross: assuming that the calibration plate is parallel to the image sensor plane, calculate the target cross value on the XOY plane, and generate ideal three-dimensional targets at all different positions; step 3-3-2, start optimization and set the lookup table; step 3-3-3, solve the optimization problem.

[0036] In the structural light calibration method for intra-oral three-dimensional reconstruction provided by the present invention, it can also have the following characteristics: wherein, step 3-3-3 includes the following sub-steps: step 3-3-3-1, using the LM algorithm to calibrate parameters and remove outliers by iteration; step 3-3-3-2, using new parameters to calculate the calibrated measurement points; step 3-3-3-3, establishing a three-dimensional lookup table; step 3-3-3-4, establishing a Jacob matrix, and using the minimum mean square error method to calculate the lookup table results.

[0037] The structured light localization method for intraoral three-dimensional reconstruction provided by the present invention may also have the following features: wherein, in step 3-3-3-3, the three-dimensional lookup table is a model-free method, that is, the field of view space is divided into a set of cubes, and piecewise linear correction is applied to each point falling into the cube, as shown below:

[0038]

[0039] Where δ(k) indicates whether the vertex is in the cube containing r, φ(k) indicates the product of the three segment distances between vertex k and r, and β(k) is the correction vector of vertex k. This function is called a lookup table. The value of the lookup table can be estimated by minimizing the following formula:

[0040]

[0041] Where r is the target point transformed from the target coordinate system to the scanner coordinate system, and r' is the reconstructed point.

[0042] Functions and Effects of the Invention

[0043] According to the structural light calibration method for intraoral three-dimensional reconstruction involved in the present invention, the optical system parameters of the three-dimensional scanner are first initialized to obtain the initialization parameters; secondly, the origin camera factor and the origin projector factor are calculated according to the initialization parameters; then, according to the origin camera factor and the origin projector factor, a calibration round is performed to obtain the calibration result; finally, the calibration result is stored. The above process introduces more calibration parameters for the additional errors introduced by the actual design and assembly process, and optimizes the calibration results, which greatly improves the calibration accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 is a flow chart of a structural light positioning method for intraoral three-dimensional reconstruction in an embodiment of the present invention;

[0045] Figure 2 is a schematic diagram of an optical system of a three-dimensional scanner in an embodiment of the present invention;

[0046] Figure 3 is a flow chart of calculating the origin camera factor and the origin projector factor in an embodiment of the present invention;

[0047] Figure 4 is a schematic diagram of a cross target of a calibration plate in an embodiment of the present invention;

[0048] Figure 5 is a schematic diagram of a tissue image when z=0 in an embodiment of the present invention;

[0049] Figure 6 is a cross diagram of a thinned rough cross center in an embodiment of the present invention; and

[0050] Figure 7 It is a workflow diagram of a calibration round in an embodiment of the present invention. DETAILED DESCRIPTION

[0051] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the following embodiments and the accompanying drawings specifically illustrate a structural light positioning method for intra-oral three-dimensional reconstruction of the present invention.

[0052] In this embodiment, a structured light positioning method for intraoral three-dimensional reconstruction is provided.

[0053] Figure 1 4 is a flow chart of the structural cursor positioning method for intraoral three-dimensional reconstruction in this embodiment.

[0054] like Figure 1 As shown, the structured light positioning method for intraoral three-dimensional reconstruction involved in this embodiment includes the following steps:

[0055] Step S1, initializing the optical system parameters of the three-dimensional scanner to obtain initialization parameters.

[0056] Figure 2 FIG. 1 is a schematic diagram of the three-dimensional scanner optical system 100 in this embodiment. Figure 2 As shown, the 3D scanner optical system includes a camera system 10 and a projection system 20 . The camera system is composed of a camera sensor 11 and a camera lens 12 , and the projection system is composed of a DMD sensor 21 and a projection lens 22 .

[0057] Combination Figure 2 , the optical system parameters of the 3D scanner are shown in Table 1:

[0058] Table 1 Model parameters

[0059]

[0060] The device parameters are the number of pixels passing through the imager horizontally, the number of pixels passing through the imager vertically, and the number of pixels passing through the illuminator horizontally. In this embodiment, the current values ​​are [1024, 768, 480]. The geometric parameters determined by physical measurement are the parameters P 11 , parameter P 12 And the parameter P 13 ; The geometric parameters determined by image measurement are parameter P9, parameter P 23 , parameter P 24 ; Parameters P1 to P6, parameter P 19 ~Parameter P 22 The initialization parameter is 0; the parameter P 23 is estimated based on the scale factor between the imager and the light source.

[0061] Step S2, calculating the origin camera factor and the origin projector factor according to the initialization parameters.

[0062] Among them, the origin camera factor is the magnification between the 3D target and the image, and the origin projector factor is the image center and line angle.

[0063] Figure 3 This is the flow chart for calculating the origin camera factor and the origin projector factor in this paper.

[0064] like Figure 3 As shown in the figure, the process of calculating the origin camera factor and the origin projector factor is divided into the following sub-steps:

[0065] Step S2-1, reading the image at position z=0.

[0066] Step S2-2, tissue image, obtain tissue image.

[0067] Step S2-3, calculate the center of the cross: search for the approximate center of the cross on the uniformly illuminated image of the target, and then refine it twice to determine the center of each cross in the x and y directions with an accuracy of about 0.1 pixel.

[0068] Figure 4 Schematic diagram of the cross target of the calibration plate in this embodiment.

[0069] like Figure 4 As shown, a better calibration target is used in this embodiment, which is a white cross pattern with a cross line thickness of 0.2 mm, a cross line length of 1 mm, and a cross center distance of 2 mm.

[0070] Figure 5 It is a schematic diagram of the tissue image when z=0 in this embodiment.

[0071] like Figure 5 As shown, Figure 5 The first eight images (a)-(h) in the figure are centerline and multiline images, which are used to extract the phase map. The ninth image (i) is an image captured by the calibration plate with a projected vertical line and a horizontal line, which will be used to calculate the camera center and angle. The tenth image (j) is a crosshair image, which is used to calculate the magnification between the 3D target and the image.

[0072] When calculating the cross center, the approximate center of the cross is searched on the tissue image and then refined twice to determine the center of each cross in the x and y directions. The main process is as follows:

[0073] Step S2-3-1, upsampling the image to a normal size.

[0074] Step S2-3-2, find all the rough cross centers. In this embodiment, the center of the cross is searched without prior information. It uses a cross-correlation method written by PushkarApte, which moves the image of the cross to each pixel in the image and calculates the correlation between two pixels to obtain a cross-correlation graph. The peak in the cross-correlation graph is the center of the cross within these two pixels.

[0075] Step S2-3-3, refine the rough cross center.

[0076] Figure 6 : is a cross diagram of the center of the rough cross being refined in this embodiment.

[0077] like Figure 6 As shown, in this embodiment, a variation of the "Canny method" is used to find a more accurate center for each cross. First, determine Figure 6 The four regions shown in , one for each arm of the cross. For example, the region labeled "A" is a set of lines that cross horizontally through the lower arm of the cross. These lines are averaged to form a one-dimensional trajectory, which is convolved with a Gaussian function with a sigma of two pixels and then differentiated twice. The zero crossings of the second derivative represent the edges of the line, labeled a1 and a2 below. The average of these two edges gives the center of the bar. The above steps are then repeated for the top bar, and a line is drawn between the two points generated. The above steps are then repeated for the left and right bars, giving a second line, and the intersection of these two lines is the center of the cross. This method is robust to small rotations and defocus of the cross, and the center is accurate to about 0.1 pixels.

[0078] Step S2-4, calculating the magnification between the three-dimensional target and the image based on the tissue image and the cross center. The specific implementation includes the following sub-steps:

[0079] Step S2-4-1, calculate the distance between the two nearest cross centers.

[0080] Step S2-4-2, calculate the magnification ratio between the three-dimensional target and the image.

[0081] Step S2-5, calculating the image center and line angle according to the tissue image and the cross center. The specific implementation includes the following sub-steps:

[0082] Step S2-5-1, upsampling the image to a normal size.

[0083] Step S2-5-2, subtract the crosshair background from the image.

[0084] Step S2-5-3, detect the peak value and convert the grayscale image peak value into a binary image peak value.

[0085] Step S2-5-4, detect a horizontal line and a vertical line in the binary image.

[0086] Step S2-5-5, refine the horizontal line and the vertical line to obtain the image center and the line angle.

[0087] Step S3, performing a calibration round according to the origin camera factor and the origin projector factor to obtain a calibration result.

[0088] Figure 7 4 is a flowchart of the calibration round in this embodiment.

[0089] like Figure 7 As shown, the calibration round includes the following sub-steps:

[0090] Step S3-1, calculating relevant information of corresponding points according to the origin camera factor and the origin projector factor.

[0091] The specific implementation includes the following sub-steps:

[0092] Step S3-1-1, calculate the phase map according to the origin camera factor and the origin projector factor.

[0093] The process of calculating the phase diagram is:

[0094] like Figure 2 As shown, there is a slight angle between the column directions of the projector and the camera, so the coordinates on the camera sensor are calibrated to compensate for this using the following formula:

[0095]

[0096] For the three-dimensional coordinates of the object surface, suppose a point (u s , v s ) is projected onto a 3D point (x, y, z) on the target, and the camera captures the coordinates (u i , v i ), the reconstruction of the 3D points is as follows:

[0097] x=η i .τ i =-η i ·(d i -z) (2)

[0098] y=ξ i .τ i =-ξ i ·(d i -z) (3)

[0099]

[0100] Where:

[0101]

[0102]

[0103]

[0104]

[0105] At the same time, for the distortion of the lens module, a heuristic polynomial distortion correction is used at z:

[0106] z=z+k0xy+k1xy 2 +k2x 3 (9)

[0107] so

[0108] r=(x,y,z)=f(Θ,u i ,v i ,u s ) (10)

[0109] Θ represents all the parameters mentioned above. Since the surface of the calibration plate is not always parallel to the image plane, assuming that [R, T] is the transformation matrix from the scanner coordinate system to the target coordinate system, the following equation can be minimized to optimize the parameter Θ.

[0110]

[0111] Where r i is the ideal target 3D point with known position in the target system, (u i ,v i ,u s ) is the phase information extracted from the captured image.

[0112] Step S3-1-2, find the crosshairs with accurate position according to the phase diagram.

[0113] Step S3-1-3, set the two-dimensional image points and their phases according to the cross lines; when setting the two-dimensional image points and their phases, first use the local bilinear interpolation method to generate the corresponding relationship between the camera and the projector for each cross line, and then generate three-dimensional coordinates for each cross.

[0114] Step S3-1-4, save the two-dimensional image points and their phases, and save the phase map on each plane.

[0115] Step S3-2, based on the relevant information of the corresponding points, the stability of the workbench is detected; if there are outliers in the sequence captured at the -8 to +8 calibration positions, the point is unstable; if more than 5 of the 9 corresponding points are unstable at one position, the position is unreliable.

[0116] Step S3-3, calibration optimization: Because the target coordinate system and the world coordinate system do not overlap, in this embodiment, a matrix is ​​used to optimize these redundant calibration parameters by using a standard Levenberg-Marquardt procedure to minimize the mean square error between the observed cross on the target and the cross at the predicted position. The specific implementation includes the following sub-steps:

[0117] Step S3-3-1, calculate the position of the ideal cross: Assuming that the calibration plate is parallel to the image sensor plane, calculate the target cross value on the XOY plane to generate ideal three-dimensional targets at all different positions.

[0118] Step S3-3-2, start optimization and set the lookup table.

[0119] Step S3-3-3, solve the optimization problem. The specific implementation process includes the following sub-steps:

[0120] Step S3-3-3-1, use the LM algorithm to calibrate parameters and remove outliers through iteration.

[0121] Step S3-3-3-2, use the new parameters to calculate the calibrated measurement points.

[0122] Step S3-3-3-3, establish a three-dimensional lookup table; the three-dimensional lookup table is a model-free method, that is, the field of view space is divided into a set of cubes, and piecewise linear correction is applied to each point falling into the cube, as shown below:

[0123]

[0124] Where δ(k) indicates whether the vertex is in the cube containing r, φ(k) indicates the product of the three segment distances between vertex k and r, and β(k) is the correction vector of vertex k. This function is called a lookup table. The value of the lookup table can be estimated by minimizing the following formula:

[0125]

[0126] Where r is the target point transformed from the target coordinate system to the scanner coordinate system, and r' is the reconstructed point.

[0127] Step S3-3-3-4, establish the Jacob matrix and calculate the lookup table result using the minimum mean square error method.

[0128] Step S3-4, updating the calibration results to system parameters.

[0129] Step S4, storing the calibration results.

[0130] Functions and Effects of the Embodiments

[0131] According to the structural light calibration method for intraoral 3D reconstruction involved in this embodiment, the optical system parameters of the 3D scanner are first initialized to obtain the initialization parameters; secondly, the origin camera factor and the origin projector factor are calculated according to the initialization parameters; then, according to the origin camera factor and the origin projector factor, a calibration round is performed to obtain the calibration result; finally, the calibration result is stored. The above process introduces more calibration parameters for the additional errors introduced by the actual design and assembly process, and optimizes the calibration results, which greatly improves the calibration accuracy.

[0132] This embodiment takes into account the angle between the projector and the camera column direction and the problem that the calibration plate plane is not always parallel to the image plane, increases calibration parameters, compensates for the angle between the projector and the camera column direction, introduces a transformation matrix [R, T] to optimize redundant system parameters, and simultaneously performs nonlinear optimization on the calibration results to minimize the mean square error between the observed and predicted cross positions on the target. Finally, a 3D LUT is used as the final correction, which greatly improves the calibration accuracy.

[0133] The above-mentioned embodiments are preferred examples of the present invention and are not intended to limit the protection scope of the present invention.

Claims

1. A structural light positioning method for intraoral three-dimensional reconstruction, characterized in that: The following steps are involved: Step 1, initializing the optical system parameters of the three-dimensional scanner to obtain initialization parameters; Step 2, calculating the origin camera factor and the origin projector factor according to the initialization parameters; Step 3, performing a calibration round according to the origin camera factor and the origin projector factor to obtain a calibration result; Step 4, storing the calibration results; Wherein, step 3 includes the following sub-steps: Step 3-1, calculating relevant information of corresponding points according to the origin camera factor and the origin projector factor; Step 3-2, testing the stability of the workbench according to the relevant information of the corresponding points; Step 3-3, calibration optimization: use the matrix to optimize the redundant calibration parameters and minimize the mean square error between the cross on the observed target and the cross at the predicted position; Step 3-4, update the calibration results to system parameters; Wherein, the step 3-1 includes the following sub-steps: Step 3-1-1, calculating a phase map according to the origin camera factor and the origin projector factor; Step 3-1-2, finding a crosshair with a precise position according to the phase diagram; Step 3-1-3, setting the two-dimensional image points and their phases according to the cross lines; Step 3-1-4, saving the two-dimensional image points and their phases, and saving the phase map on each plane, Among them, in step 3-1-1, the process of calculating the phase diagram is: There is a slight angle between the projector and the camera, so the coordinates on the camera sensor are calibrated to compensate for this using the following formula: For the three-dimensional coordinates of the object surface, suppose a point (u s , v s ) is projected onto a 3D point (x, y, z) on the target, and the camera captures the coordinates (u i , v i ), the reconstruction of the 3D points is as follows: x=ηi.τi=-ηi·(d i -z) (2) y=ξi.τi=-ξi·(d i -z) (3) Where: At the same time, for the distortion of the lens module, a heuristic polynomial distortion correction is used at z: z=z+k0xy+k1xy 2 +k2x 3 (9) so r=(x,y,z)=f(Θ,u i ,v i ,u s ) (10) Θ represents all the parameters mentioned above, Since the surface of the calibration plate is not always parallel to the image plane, assuming that [R, T] is the transformation matrix from the scanner coordinate system to the target coordinate system, the following equation can be minimized to optimize the parameter Θ: Where r i is the ideal target 3D point with known position in the target system, (u i , v i ,u s ) is the phase information extracted from the captured image; In step 3-1-3, when setting the two-dimensional image points and their phases, firstly, a local bilinear interpolation method is used to generate a corresponding relationship between the camera and the projector for each cross line, and then a three-dimensional coordinate is generated for each cross line. In step 3-2, if there are outliers in the sequence captured at a certain range of calibration positions, the point is unstable. If more than 5 of the 9 corresponding points are unstable at a position, the position is unreliable.

2. The method for structural light localization for intraoral three-dimensional reconstruction according to claim 1, characterized in that: in, In step 1, the optical system parameters of the 3D scanner are shown in Table 1: Table 1 Model parameters Among them, the geometric parameters determined according to physical measurements are parameters P 11 , parameter P 12 And the parameter P 13 , The geometric parameters determined by image measurement are parameter P9, parameter P 23 , parameter P 24 , Parameters P1 to P6, Parameter P 19 ~Parameter P 22 The initialization parameter is 0, Parameter P 23 is estimated based on the scale factor between the imager and the light source.

3. The method for structural light localization for intraoral three-dimensional reconstruction according to claim 1, characterized in that: in, In step 2, the origin camera factor is the magnification between the three-dimensional object and the image, The origin projector factors are the image center and line angle, Step 2 is divided into the following sub-steps: Step 2-1, read the image at position z=0; Step 2-2, organizing the image to obtain a tissue image; Step 2-3, calculate the cross center; Step 2-4, calculating the magnification between the three-dimensional target and the image according to the tissue image and the cross center; Step 2-5, calculating the image center and the line angle according to the tissue image and the cross center.

4. The method for structural light localization for intraoral three-dimensional reconstruction according to claim 3, characterized in that: in, In step 2-3, when calculating the cross center, the approximate center of the cross is searched on the tissue image, and then refined twice to determine the center of each cross in the x and y directions. The main process is as follows: Step 2-3-1, upsample the image to normal size; Step 2-3-2, find all the rough cross centers; Step 2-3-3, refine the rough cross center, The steps 2-4 include the following sub-steps: Step 2-4-1, calculating the distance between the two nearest cross centers; Step 2-4-2, calculate the magnification between the three-dimensional target and the image, The steps 2-5 include the following sub-steps: Step 2-5-1, upsample the image to normal size; Step 2-5-2, subtract the crosshair background from the image; Step 2-5-3, detect the peak value, and convert the grayscale image peak value into a binary image peak value; Step 2-5-4, detecting a horizontal line and a vertical line in the binary image; Step 2-5-5, refine the horizontal line and the vertical line to obtain the image center and the line angle.

5. The method for structural light localization for intraoral three-dimensional reconstruction according to claim 4, characterized in that: in, In step 2-3-2, the process of finding all the rough cross centers is: moving the image of the cross to each pixel in the image, and calculating the correlation between the pixels in pairs to obtain a cross-correlation graph, wherein the peak in the cross-correlation graph is the cross center within these two pixels.

6. The method for structural light localization for intraoral three-dimensional reconstruction according to claim 1, characterized in that: The step 3-3 includes the following sub-steps: Step 3-3-1, calculate the position of the ideal cross: Assume that the calibration plate is parallel to the image sensor plane, calculate the target cross value on the XOY plane, and generate the ideal three-dimensional target at all different positions; Step 3-3-2, start optimization and set up the lookup table; Step 3-3-3, solve the optimization problem.

7. A method for structural light localization for intraoral three-dimensional reconstruction according to claim 6, characterized in that: Step 3-3-3 includes the following sub-steps: Step 3-3-3-1, use the LM algorithm to calibrate parameters and remove outliers through iteration; Step 3-3-3-2, use the new parameters to calculate the calibrated measurement points; Step 3-3-3-3, establish a three-dimensional lookup table; Step 3-3-3-4, establish the Jacob matrix and use the minimum mean square error method to calculate the lookup table results.

8. The method for structural light localization for intraoral three-dimensional reconstruction according to claim 7, characterized in that: In step 3-3-3-3, the three-dimensional lookup table is a model-free method, that is, the field of view space is divided into a set of cubes, and piecewise linear correction is applied to each point falling into the cube, as shown below: Where δ(k) indicates whether the vertex is in the cube containing r, φ(k) represents the product of the three segment distances between vertex k and r, and β(k) is the correction vector of vertex k. This function is called a lookup table. The values ​​of the lookup table can be estimated by minimizing the following equation: Where r is the target point transformed from the target coordinate system to the scanner coordinate system, and r' is the reconstructed point.

Images