A multi-light-source pixel-level splicing method for DLP surface exposure 3D printing

Abstract

A multi-source pixel-level stitching method for DLP surface exposure 3D printing involves intelligent control, image acquisition, and image processing technologies. It includes the following steps: Image acquisition: Projector A is fixed in a horizontal position. By controlling projector B to deflect at different angles in different directions, the overlapping area between the two projection light sources is acquired at equal time intervals. Image analysis: The acquired overlapping area image is preprocessed to determine if the pixels are aligned. If not aligned, the projected image of projector B is extracted from the overlapping image, and the pixel contours are extracted. The centroid coordinates are used to describe different discrete points, and finally, curves are drawn based on the discrete points to obtain a relationship expression. Projector calibration: The calculated average value is compared with a standard value. If they are not equal, projector B is calibrated according to a fitting function. This invention can print large-size products while maintaining consistent printing accuracy.

Description

Technical Field

[0001] This invention relates to intelligent control, image acquisition, and image processing technologies. Specifically, it involves acquiring and analyzing images of the overlapping areas of two projection light sources on an exposure surface at different directional deflection positions. By combining the light source distribution characteristics of the digital micromirror device and the pixel coordinate distribution of the projection light source, the spatial position of the pixels in the captured image is analyzed and corrected, thereby achieving pixel-level stitching of multiple DLP light sources. Background Technology

[0002] 3D printing, also known as additive manufacturing, is a type of rapid prototyping technology. It's a technique that uses digital model files as a basis and employs powdered metals or plastics and other bondable materials to construct objects layer by layer. This technology has applications in the medical industry, aerospace, mold making, and other fields.

[0003] The 3D printing technology industry is constantly expanding into new 3D printing technology routes and implementation methods. Based on differences in materials and printing methods, these can be broadly categorized into three main technology types: photopolymerization molding, powder material molding, and extrusion molding. Photopolymerization molding based on liquid photosensitive resin materials includes stereolithography (SLA) and polymer jetting (PolyJet) technologies. TM 3D printing technologies based on powder materials include selective laser sintering (SLS) and selective laser melting (SLM), while those based on extrusion include fused deposition modeling (FDM) and fused extrusion modeling (MEM). Among these technologies, photopolymerization 3D printing based on liquid photosensitive resin materials has attracted widespread attention due to its high forming accuracy and fast printing speed. Comparative studies have shown that surface-exposure technology has a simpler hardware system structure and lower cost, giving it a significant cost-performance advantage over similar photopolymerization technologies. Therefore, surface-exposure technology (DLP) has become a major research focus in current 3D continuous printing photopolymerization technology.

[0004] In existing photopolymerization technologies, the printing accuracy of printed products decreases as the size of the printed product increases. Dynamic multi-light source combinations for large-size printing cannot maintain the original printing accuracy. This paper proposes a vision-based static multi-light source combination scheme. By analyzing the overlapping images generated by two projection light sources at the midpoint of the focal plane, the deflected projector is corrected to achieve pixel-level stitching of multiple light sources, thus achieving the goal of printing large-size models while maintaining accuracy. Summary of the Invention

[0005] This invention provides a multi-light source pixel-level stitching method for DLP surface exposure 3D printing, which ensures that the original forming accuracy remains unchanged when printing large-size models.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0007] A multi-light source pixel-level stitching method for DLP surface exposure 3D printing includes the following steps:

[0008] Step 100: Place projectors A and B. Projector A has been leveled and placed in a horizontal position. Fix projector B on the five-axis combined control platform. Then, focus the light source of projector B separately so that its focal plane is exactly on the upper surface of the resin tank. Then, place an industrial camera above the resin tank, positioning it in the exact center above the two projectors. Adjust the camera position so that its focal plane coincides with the focal plane of the projection light source.

[0009] Step 200: Control projector B to deflect at different angles in three different directions, X, Y and Z. Projectors A and B project their own different images. Use a camera to capture the overlapping area of ​​the two projection light sources of projectors A and B on the focal plane in this state, and store the images at equal intervals on the PC.

[0010] Step 300: Analyze the acquired overlapping area image; first, determine whether the pixels in the image are aligned. If they are not aligned, extract the overlapping image to obtain the projected images of projectors A and B respectively; analyze the changes in the centroid coordinates of the pixels in the image projected by projector B under different directional deflection angles, and fit a curve to the discrete points based on this change to obtain their relationship expression.

[0011] Step 400: Analyze the pixel parameter values ​​of the projected image extracted from projector B under different deflection angles, substitute the values ​​into the fitting curve to estimate the direction and angle of projector B deflection, and determine whether projector B has deflected; if deflection occurs, the projector B is controlled and corrected in all directions through a five-axis combination device, so that the pixels on the overlapping area of ​​the projection light source of projector A in the horizontal position are stitched together.

[0012] The process includes placing projectors A and B, with projector A already leveled and positioned horizontally, fixing projector B on a five-axis integrated control platform, focusing the light source of projector B individually so that its focal plane is exactly on the upper surface of the resin tank, and then placing an industrial camera above the resin tank, positioning it precisely above the two projectors, and adjusting the camera position so that its focal plane coincides with the focal plane of the projection light source. Step 100 includes:

[0013] The printing equipment consists of three units: an optical system unit, a resin tank unit, and a camera unit. The optical system unit includes a projector mounting bracket and a projector. The projector mounting bracket is a motorized five-axis control platform composed of a motorized corner stage and a motorized translation stage, capable of rotating in the X, Y, and Z directions. It is equipped with a stepper motor and an RS232 interface, allowing for automated control of projector B via a host computer. Projector A is already in a horizontal position. Based on the printing platform settings, the distance from the projection light source to the resin tank is determined; in the experiment, it is 340mm. Projector B is then focused to ensure that the resin tank plane is its focal plane, allowing for a well-focused and clear projected image.

[0014] The camera unit consists of a camera and a movement axis. The movement axis controls the camera's movement in three directions: X, Y, and Z. The camera uses CMOS imaging and lacks a focusing function; the lens must be manually moved directly to the focusing plane of the resin tank. Due to the small size of its built-in chip, the CMOS camera can only capture images of an area of ​​approximately 5mm x 5mm. Because the projector's light intensity is too strong, the camera is overexposed. This requires manual reduction of the light intensity, either by projecting a low-grayscale slice of image or reducing the projector's brightness, and by adding a filter to the camera. Fine-tuning the camera vertically is necessary until the clearest image can be captured.

[0015] The step 200, which involves controlling projector B to deflect at different angles in three different directions (X, Y, and Z), and projecting different images from projectors A and B respectively, and using a camera to capture the overlapping area of ​​the two projection light sources (projectors A and B) on the focal plane in this state, and storing the images at equal intervals on a PC, includes:

[0016] First, ensure projector A remains horizontal. To distinguish between different projection light sources and ensure clear capture of the overlapping area between the two light sources, projectors A and B each project different monochromatic light images. Once the projected light output stabilizes and stops flickering, control projector B to deflect at different angles in different directions. In this state, capture multi-source images of the overlapping area between the two light sources; this overlapping area is defined as the camera capture area. Set an interval time, acquire target images at equal intervals, and then continue deflecting. Store images within ±10° of deflection in each direction. Repeat the above steps, recording the deflection images in the X, Y, and Z directions.

[0017] The process of analyzing the acquired overlapping region image includes: first, determining whether the pixels in the image are aligned; if not, extracting the overlapping image to obtain the projected images from projectors A and B respectively; analyzing the changes in the centroid coordinates of the pixels in the image projected by projector B under different directional deflection angles, and fitting a curve to the discrete points based on this change to obtain the relationship expression step 300, which includes:

[0018] The acquired multi-source images are color images. First, they are binarized to remove noise, preserving the original image and determining pixel alignment. If misaligned, the monochromatic light images projected by projectors A and B are extracted from the overlapping images. The extracted image from projector B is then denoised, edges are smoothed, and the contour of each pixel is extracted, with its centroid coordinates calculated. A set of discrete points is obtained using these centroid coordinates. Adjacent pixels are connected, and the changes in pixel spacing and slope under different deflection angles are calculated. The overall image is then analyzed, and the average of each set of parameters is calculated. Based on the obtained data, a fitting curve of different deflection angles versus pixel parameter changes is plotted.

[0019] The step 400 involves analyzing the pixel parameter values ​​of the projected image extracted from projector B under different deflection angles, substituting the values ​​into a fitted curve to estimate the direction and angle of projector B's deflection, and determining whether projector B has deflected. If deflection occurs, the step 400 involves using a five-axis combination device to achieve omnidirectional control and correction of projector B, so that the pixels on the overlapping area of ​​the projection light source of projector A at the same horizontal position are stitched together.

[0020] The average pixel spacing and slope of the projected image from projector B, extracted from the overlapping region image, are calculated. The calculated average value is then compared to a standard value. If they do not match, it indicates that projector B has undergone angular deflection. Substituting the values ​​into the fitted curve allows estimation of the direction and angle of projector B's deflection, which is then corrected based on the calculated values. After each fine-tuning of projector B, the camera re-acquires images, performs image analysis and projector correction, repeating this process until the calculated values ​​match the standard values, achieving pixel-level stitching of multiple light sources.

[0021] The multi-light source pixel-level stitching method based on surface exposure of the present invention has the following advantages:

[0022] 1) Increase the size of the printed model. Since the projection area of ​​a single projection light source is limited, when the projection area of ​​a single light source is smaller than the size of the model to be printed, multiple projectors need to be manually combined and spliced. Designing a multi-light source combination and splicing scheme increases the size of the printed model.

[0023] 2) Ensure the printing accuracy of the model remains constant. The accuracy of the printed model decreases as the size of the printed model increases. Dynamic light source combinations for large-size printing cannot maintain the original printing accuracy. A static multi-light source combination scheme is designed. By collecting and analyzing the overlapping area between two projection light sources, the deflected projector is corrected to achieve pixel-level splicing of multiple light sources, thus maintaining consistent printing accuracy when printing large-size models.

[0024] 3) It is universally applicable to multiple light sources. Depending on the type of printing light source used in different surface exposure 3D printing methods, the method proposed in this invention can be used for image acquisition, analysis, and correction to achieve pixel-level precise splicing of multiple light sources, thereby ensuring that the original forming accuracy remains unchanged when printing large-size models. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the process of this invention.

[0026] Figure 2 (a) and (b) are the projection images captured by the camera from projectors A and B, respectively.

[0027] Figure 3 (a) Diagram showing the initial horizontal positions of projectors A and B. Figure 3 (b), (c), and (d) are the rotation positions of projector B along the X+, Z+, and Y+ axes, respectively.

[0028] Figure 4 It is a map of the overlapping area of ​​two projected light sources captured by the camera.

[0029] Figure 5 (a) is a pixel outline map of the image. Figure 5 (b) is used to calculate the spacing between pixels in each row. Figure 5 (c) is used to calculate the spacing between pixels in each column. Figure 5 (d) is used to calculate the slope of each row of pixels.

[0030] Figure 6 (a), (b), and (c) are curves showing the relationship between pixel size and angle change under X, Y, and Z direction deflection, respectively.

[0031] Figure 7 (a) and (b) are wrench products printed using single and dual projection light sources, respectively.

[0032] Figure 8 It refers to the setting of test points at different locations on the product.

[0033] Figure 9 It is the roughness value measured at a certain test point on the product. Detailed Implementation

[0034] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be noted that, unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

[0035] Surface exposure 3D printing using DLP as the light source is called DLP 3D printing technology. This printing method can be divided into top exposure printing and bottom exposure printing methods depending on the position of the light source. This invention will use bottom exposure as an example. DLP technology was first developed by Texas Instruments (TI). It mainly involves projecting an image through a projector, forming a two-dimensional pattern in a resin bath solution. Upon irradiation, a polymerization reaction occurs, causing the layer to solidify. A moving platform controls the layer-by-layer solidification, ultimately forming a three-dimensional entity. The hardware system mainly consists of three parts: a camera unit, an optical system unit, and a resin bath unit. The camera moves to the projector's focusing surface, and its position remains unchanged during image capture. The optical system unit includes the projector and a projector mounting bracket. The projector is the projection light source, and the projector mounting bracket is a five-axis combined system that can control the deflection of the projector B at different angles in various directions, achieving precise pixel stitching in the overlapping area between the two projection light sources. The resin bath unit includes a resin bath, liquid photosensitive resin, fluorinated oil, and a flow system.

[0036] To increase the size of printed products, existing research mainly focuses on processing sliced ​​images to achieve image-level stitching. However, in actual printing, even minute deflections imperceptible to the human eye can significantly impact the stitched parts. Research on the impact of projector position changes on pixels, based on the principles of projectors, is virtually nonexistent. The method described in this invention uses a static multi-light source combination to increase the size of the printed model while maintaining its original forming accuracy, achieving pixel-level precise stitching from multiple light sources.

[0037] The core of DLP projection technology is the Digital Micromirror Device (DMD) chip, which contains millions of closely packed micromirrors. Each micromirror represents a pixel. Ignoring the gaps between the micromirrors, the pixels in the projected image can be considered to be arranged in a matrix. The camera uses CMOS imaging and lacks a focusing function. The lens must be manually moved to the focusing plane of the resin tank to control the camera and capture the image.

[0038] This invention proposes a multi-light source pixel-level stitching method for DLP surface exposure 3D printing, the flowchart of which is shown below. Figure 1 As shown. Includes:

[0039] Step 100: Place projectors A and B. Projector A has been leveled and placed in a horizontal position. Fix projector B on the five-axis combined control platform. Then, focus the light source of projector B separately so that its focal plane is exactly on the upper surface of the resin tank. Then, place an industrial camera above the resin tank, positioning it in the exact center above the two projectors. Adjust the camera position so that its focal plane coincides with the focal plane of the projection light source.

[0040] Optionally, step 100 includes the following sub-steps:

[0041] Sub-step 110: Fix projector A in a horizontal position and place projector B on the five-axis combination control platform. Turn on projectors A and B. Place the two projection light sources below the printing platform, at a distance of 300-400mm from the resin tank. Set the brightness of both projectors to 50 and simultaneously project different monochromatic light images onto the resin tank; when the projected images no longer flicker, it indicates that the light from the projection light sources is stable. Then, focus projector B, using the first layer in the actual printing process, i.e., the bottom of the resin tank, as the focus plane. Place a piece of white paper on this plane and judge whether projector B is focused by the clarity of the projected pattern on the white paper. If it is not focused, turn the focus ring to the clear position to complete the initial setup. The above focus is judged by observing whether the edges of the projected pattern are clear.

[0042] Sub-step 120: Move the camera position using a three-axis mechanism so that its focal plane coincides with the focal planes of the two projection light sources. Fix the camera on a three-axis mechanical system directly above and between the two projection light sources. Control the rotation of the three axes to move the camera in different directions. The three-axis mechanical system includes three axes: X, Y, and Z. It controls the camera to move left and right, and forward and backward in the X and Y directions to reach the specified image capture area; it can also move along the Z axis to fine-tune the camera's vertical movement, ensuring that its focal plane coincides with the focal plane of the projection light source, thus capturing a bright and clear target image.

[0043] Step 200: Control projector B to deflect at different angles in three different directions, X, Y and Z. Projectors A and B project their own different images. Use a camera to capture the overlapping area of ​​the two projection light sources of projectors A and B on the focal plane in this state, and store the images at equal intervals on the PC.

[0044] Optionally, step 200 includes the following sub-steps:

[0045] Sub-step 210: Determine the test image and use an area scan camera to capture the overlapping area between the two projected images. In order to distinguish between two different projection light sources and simultaneously capture clear pixels on the image, projectors A and B select two different monochromatic light projections. Figure 2 To utilize the corresponding images acquired by the hardware system, each grid represents a pixel.

[0046] Sub-step 220: Projector A is fixed in a horizontal position. Projector B can be deflected at different angles in different directions. After projector B is placed on the projector stand, it may deflect at various angles due to human error. To study the impact of projector B's deflection in different directions on the pixels in the overlapping area of ​​the two projection light sources, a coordinate system is established with projector B as the origin. The overlapping areas of the two projection light sources are obtained under the deflection states in the X, Y, and Z directions, as shown below. Figure 3 As shown in (b), 3(c), and 3(d), Figure 3 (a) shows the initial fixed position of projector B. Projector B is deflected at a frequency of 1° each time.

[0047] Sub-step 230: Capture the overlapping area between the two projection light sources, such as... Figure 4As shown. The test image signal is transmitted to projectors A and B via the host computer. Both projectors output light simultaneously, ensuring the projected image is stably output and clearly displayed on the focusing surface. The overlapping area between the centers of the two projection light sources at the bottom of the resin tank is selected as the image capture area. The resin tank is then removed, and the camera is moved to the image capture area, with minor vertical adjustments to ensure a clear and bright image. Projector A remains horizontal throughout, while projector B is fixed to a five-axis control unit. An interval is set, and the target image is acquired at equal intervals before deflection continues. The target image within ±10° of deflection is stored. The above steps are repeated to record deflection images at different angles in the X, Y, and Z directions.

[0048] Step 300: Analyze the acquired overlapping area image; first, determine whether the pixels in the image are aligned. If they are not aligned, extract the overlapping image to obtain the projected images of projectors A and B respectively; analyze the changes in the centroid coordinates of the pixels in the image projected by projector B under different directional deflection angles, and fit a curve to the discrete points based on this change to obtain their relationship expression.

[0049] Optionally, step 300 includes the following sub-steps:

[0050] Sub-step 310: Preprocess the acquired image. The image captured by the camera is an overlapping image of monochromatic red light and monochromatic blue light projected by projectors A and B respectively, and is a color image. First, the image is preprocessed for denoising, including filtering, erosion, and dilation. Then, color conversion is performed, and finally, the luminance component is extracted, with histogram equalization point enhancement selected. The processed image can be used to preliminarily determine whether the two projectors are pixel-aligned. If not, the R and B components are extracted from the overlapping image to obtain the monochromatic light images projected by each projector separately. Then, the image projected by projector B is analyzed, including denoising, edge smoothing, and pixel contour extraction, such as... Figure 5 As shown in (a), each sub-contour in the image represents a pixel of the projected light source. All coordinate points of each sub-contour are detected, its centroid coordinates are calculated, and all centroid coordinates are stored in an array to complete the preprocessing of the acquired image.

[0051] Sub-step 320: Perform a levelness analysis on the deflected image in the X and Y directions. The distance between two bright pixels is represented by the distance between the centroid coordinates of the two pixel contours in each row and column of the image, such as... Figure 5 (b) and Figure 5As shown in (c). Taking any contour point as the center, calculate the distance between the centroid coordinates of each row and column in the image, and then calculate the average value from the entire image to obtain the average pixel spacing value between all rows and columns. Repeat the above steps to obtain the average pixel row spacing (R-avg) and (C-avg) after deflection at different deflection angles in the X and Y directions, respectively.

[0052] To avoid the influence of different focal lengths during printing, the pixel size relationship at the origin position is set to a standard state, including standard row spacing (SV-R) and standard column spacing (SV-C). The relationship between the difference between the average pixel spacing and the standard spacing (R-avg-D, C-avg-D) at different deflection angles and the deflection angle is analyzed. Based on the obtained discrete points, a curve fitting is performed to obtain the pixel relationship curve. The X-axis curve is shown below. Figure 6 As shown in (a), the relational expression is (1); the Y-axis curve is as follows: Figure 6 As shown in (b), its corresponding relational expression is (2), as follows:

[0053] F1(θ x )=0.0001*θ x 3 -0.00042*θ x 2 +0.1188*θ x (1)

[0054] F2(θ y ) = -8 * 10 -6 *θ y 3 -0.0035*θ y 2 +0.2684*θ y (2)

[0055] Where the angle θ x and θ y These represent the tilt angles along the X and Y axes, respectively.

[0056] By comparing whether the R-avg-D and C-avg-D values ​​are 0, it can be preliminarily determined whether the projector B has deflected on the X and Y axes; if they are not 0, the values ​​are substituted into the above-mentioned fitting curve to estimate the direction and angle of the projector B's deflection. The estimated angle value is then input into the automatic control system to first complete the position correction in the X and Y directions.

[0057] Sub-step 330: Perform a levelness analysis on the deflected image in the Z direction. From the perspective of the overall image, the centroid coordinates can be hierarchically classified according to rows, such as... Figure 5As shown in (d), the complete pixels within each red box belong to the same level. Select all discrete coordinate points within each level, list the corresponding X and Y coordinate values, insert a scatter plot, generate a chart, and then add a trend line in the chart elements. Set the trend format to a linear function to obtain the function expression formula: y0 = kOx + bO. Similarly, the fitting function for each level can be obtained: y1 = k1x + b1, y2 = k2x + b2, ... It is found that the more scatter points the curve covers, the better the fitting effect. The value of k in the fitting function of each level represents the slope of the trend line. By calculating the average of all k values, the slope of the pixels in the entire image at this deflection angle can be obtained: k = (kO + k1 + ... + km-2 + km-1) / m. Based on the obtained discrete points, a curve fitting is performed to obtain the pixel relationship curve. The Z-axis curve is shown below. Figure 6 As shown in (c), the relational expression is as shown in (3):

[0058] F3(θ z )=0.00863*θ z (3)

[0059] Where the angle θ Z This represents the tilt angle along the Z-axis.

[0060] Step 400: Analyze the pixel parameter values ​​of the projected image extracted from projector B under different deflection angles, substitute the values ​​into the fitting curve to estimate the direction and angle of projector B deflection, and determine whether projector B has deflected; if deflection occurs, the projector B is controlled and corrected in all directions through a five-axis combination device, so that the pixels on the overlapping area of ​​the projection light source of projector A in the horizontal position are stitched together.

[0061] Optionally, step 400 includes the following sub-steps:

[0062] Sub-step 410: Before the actual printing process begins, place projector A horizontally, and then fix projector B on the five-axis combination control console. First, calculate the average pixel row spacing, column spacing, and average slope between pixels in the pixel coordinates of the projected image of projector B extracted from the overlapping image in this state.

[0063] Sub-step 420: Based on the calculated average pixel spacing, calculate the difference between it and the standard value. If the difference is not 0, it means that the X and Y axes are not horizontal. Substitute this value into the standard value. Figure 6 By using curve (a) or (b), the deflection direction and angle along that axis can be obtained. Projector B is then calibrated according to the obtained angle value. After calibration, the camera recaptures an image, calculates the average pixel spacing on that image, and compares it with the standard value until it matches the standard value. At this point, the deflection correction of projector B in the X and Y directions is complete.

[0064] Sub-step 430: Compare the calculated average slope between pixels with the standard slope. In a horizontal state, the standard slope between pixels is 0. If the average slope is not 0, it means that the projector B has deflected in the Z direction. Substitute the average slope value into... Figure 6 By finding the curve (c), the deflection direction and angle along that axis can be obtained. The projector B is then calibrated according to the obtained angle value. After calibration, the camera captures an image again, and the average pixel slope on that image is calculated until the value is 0. At this point, the deflection correction of the projector B in the Z direction is complete.

[0065] Print test

[0066] To more clearly observe the printing quality of the spliced ​​parts, the sliced ​​images of the overlapping parts are processed into grayscale, making the spliced ​​parts identifiable in the printed product. Figure 7 (a) shows a wrench product printed using a single light source, with dimensions of 110mm*40mm. The structure is complete without missing details, the surface is smooth, and the final formed structure meets the expected requirements; printing was successful. A wrench product printed using two light sources is shown below. Figure 7 As shown in (b), the dimensions are 214mm*78mm. The product has a complete structure, a smooth surface, and no impurities on the edges.

[0067] To further illustrate the experimental results, a roughness meter was used to measure the roughness of identical models printed using both methods to compare their accuracy. For each printed product, five test points were selected for roughness measurement: test1 for the non-overlapping surface, test2 for the non-overlapping side surface, test3 for the overlapping surface, test4 for the left side of the overlapping surface, and test5 for the right side of the overlapping surface. Figure 8 As shown. In the experiment, a TR110 surface roughness meter was selected. The average roughness Ra within the sampling length was used to represent the roughness of the test point on the printed product. Figure 9 As shown in Table 1 below, the test results demonstrate that the roughness of single-source printing and pixel-aligned multi-source printing models are not significantly different, proving that the proposed scheme can maintain the accuracy of large-size printed models.

[0068]

[0069] Table 1

Claims

1. A multi-light source pixel-level stitching method for DLP surface exposure 3D printing, characterized in that, Includes the following steps:

100. Place projectors A and B. Projector A has been leveled and placed in a horizontal position. Fix projector B on the five-axis combined control platform. Then, focus the light source of projector B separately so that its focal plane is exactly on the upper surface of the resin tank. Then, place an industrial camera above the resin tank, positioning it in the exact center above the two projectors. Adjust the camera position so that its focal plane coincides with the focal plane of the projection light source.

200. Control projector B to deflect at different angles in three different directions X, Y and Z. Projectors A and B project their own different images. Use a camera to capture the overlapping area of ​​the two projection light sources A and B on the focal plane in this state, and store the images at equal intervals on the PC.

300. Analyze the acquired overlapping area images; first, determine whether the pixels in the image are aligned. If they are not aligned, extract the overlapping images to obtain the projected images of projectors A and B respectively; analyze the changes in the centroid coordinates of the pixels in the image projected by projector B under different directional deflection angles, and fit a curve to the discrete points based on this change to obtain their relationship expression.

400. Analyze the pixel parameter values ​​of the projected image extracted from the projector B under different deflection angles, substitute the values ​​into the fitting curve to estimate the direction and angle of deflection of the projector B, and determine whether the projector B has deflected. If a deflection occurs, the projector B is controlled and corrected in all directions through a five-axis combination device, so that the pixels on the overlapping area of ​​the projection light source of the projector A in the horizontal position are stitched together. Step 200 includes: Projector B is controlled to deflect at different angles in different directions. After projector B is placed on the projection stand, in order to study the effect of its deflection angle in different directions on the pixels on the overlapping image, a coordinate system is established with projector B as the origin, and the overlapping area images of projector B in the three deflection directions of X, Y and Z are obtained respectively. Projector B is controlled to move 1° each time. If the change in focal length causes the image to become unclear, the focus ring is moved to adjust the focus. Each test image is projected onto the resin tank using a projector, ensuring the projected image is clearly displayed on the focusing surface. A center position is selected on the focusing interface, precisely capturing the overlapping portion between the two projection light sources. This area serves as the image capture region. The focusing surface is marked, the resin tank is removed, and the camera is moved to the image capture region, with minor adjustments made vertically to ensure image clarity. An interval is set, and the camera acquires target images at equal intervals before continuing to deflect. Images within ±10° of deflection are stored. The above steps are repeated to record deflection images in three directions. Step 300 includes: Image preprocessing is performed; the captured image is the overlapping area of ​​two projection light sources at the middle position of the focal plane. First, the image is preprocessed, then the projected image of projector B is extracted from the overlapping image, and then it is denoised, the edges are smoothed, and the pixel contours are extracted; each sub-contour on the image represents a pixel of the projection light source; all coordinate points of each sub-contour are detected, its centroid coordinates are calculated, and all centroid coordinates are stored in an array to complete the preprocessing of the overlapping image; Perform a levelness analysis on the image deflected in the X and Y directions; represent the distance between two bright pixels by calculating the distance between the centroid coordinates of two contours in each row and column of the image; calculate the distance between the centroid coordinates of each row and column in the image with any contour point as the center, and then calculate the average value from the whole image to obtain the average distance value between pixels in all rows and columns; repeat the above steps to obtain the pixel distance value after deflection at different deflection angles in the X and Y directions; Set the pixel size at the origin position to a standard value, including standard row spacing and standard column spacing; analyze the relationship between the difference between the average spacing of pixels on the image and the standard spacing and the angle under different deflection angles to obtain the fitting curve; A levelness analysis was performed on the image deflected in the Z direction. Analyzing the overall image, pixel coordinates were categorized by row. Complete pixels within each row belonged to the same level. All discrete coordinate points within each level were selected, and their corresponding X and Y coordinate values ​​were listed. A scatter plot was inserted, and a chart was generated. A trend line was added to the chart elements, and the trend format was set to a linear function, resulting in the formula y0 = k0x + b0. This process was repeated for each level, yielding fitting functions y1 = k1x + b1, y2 = k2x + b2, ... The more scatter points the curve covered, the better the fitting effect. The value of k in the fitting function at each level represented the slope of the trend line. By calculating the average of all k values, the slope of the pixels in the entire image at this deflection angle was obtained: k = (k0 + k1 + ... + km-2 + km-1) / m. Based on these discrete points, a curve fitting was performed to obtain the pixel relationship change curve.

2. The method according to claim 1, characterized in that, Step 100 includes: First, fix projector A in a horizontal position, then place projector B on the five-axis combination control device and adjust the focus of projector B; place the two projection light sources below the printing platform, at a distance of 300-400mm from the resin tank; use CMOS imaging for the camera, and move the lens directly to the focusing plane of the resin tank.

3. The method according to claim 1, characterized in that, Step 400 includes: Before the actual printing process begins, projector A is placed horizontally, and then projector B is fixed on the five-axis combination control console. The average pixel row spacing, column spacing, and average slope between pixels in the pixel outline of the projected image of projector B extracted from the overlapping image in this state should be calculated first. Based on the calculated average pixel spacing, the difference between it and the standard value is calculated. If the difference is not 0, it means that the projector B is not in a horizontal position in the X and Y axis directions. The value is substituted into the fitting curve to obtain the angle of deflection in this direction. The projector B is corrected according to this angle value. After calibration, the camera captures the image again and analyzes it. The average pixel spacing of the pixels in the image is calculated and compared with the standard value until it is consistent with the standard value, and the calibration is completed. The calculated average slope between pixels is compared with the standard slope. If they are inconsistent, it means that the projector B has deflected in the Z direction. In the horizontal state, the standard slope between pixels is 0. The average slope value is substituted into the fitted curve to obtain the deflection angle in this direction. The projector B is then corrected according to this angle. After calibration, the camera captures the image again and analyzes it. The average slope value of the pixels in the image is calculated and compared with the standard value until it is consistent with the standard value, thus completing the calibration.

Images