A microlens array bottom drawing generation method based on process distortion adaptive compensation

Abstract

This invention discloses a method for generating microlens array base maps based on adaptive compensation for process distortion. The method involves collecting thin film distortion parameters, identifying a circular 90° rectangular array, and then enabling independent adjustment of horizontal and vertical parameters. The compensation size is calculated based on the distortion coefficients, and pixel conversion is performed. An asymmetric mesh matrix adapted to the deformation is generated with differentiated step sizes. Image noise reduction and enhancement are achieved through multi-mode parallax sampling and other steps. High-definition base maps are output using multi-threaded parallel rendering of tiles and asynchronous lossless compression. Imaging defect data is collected during mass production, and the compensation coefficients are iteratively corrected to complete closed-loop parameter optimization. This invention overcomes the shortcomings of traditional methods, such as uniform horizontal and vertical parameters and difficulty in adapting to anisotropic thin film deformation. It eliminates imaging stretching distortion, moiré patterns, and viewing angle breaks. It can adapt to fluctuations in production processes, significantly improving the uniformity and clarity of batch microlens imaging, and is suitable for the industrial fabrication of high-precision base maps for naked-eye 3D optical thin films.

Description

Technical Field

[0001] This invention relates to the field of imaging thin film design and manufacturing technology, and in particular to a method for generating microlens array base maps based on adaptive compensation for process distortion. Background Technology

[0002] Microlens array grating films, including structures such as fly-eye lenses and linear parallax gratings, are core optical components in the fields of glasses-free stereoscopic displays, light field imaging, and optical decoration, and are widely used in the new display and consumer electronics industries. In their precision manufacturing process, the accuracy of the base image generation directly determines the arrangement morphology and optical effect of the microlens array, and is a crucial prerequisite for ensuring the continuity of the finished product's field of view and reducing optical distortion and scrap rates.

[0003] Currently, mainstream fabrication processes for microlens films include UV embossing, hot roll calendering, and laser direct writing. During the stretching and shrinking process, significant anisotropic process distortion occurs, causing the designed standard circular microlens to become an elliptical structure, resulting in inconsistencies in horizontal and vertical geometric parameters. Existing pattern generation software is developed based on the isotropic ideal assumption, using a single set of lens diameter and array spacing parameters in both directions. This fails to match the asymmetric deformation caused by the process, leading to a mismatch between the generated standard symmetrical pattern and the actual formed lens array, easily causing defects such as optical distortion and field-of-view jumps.

[0004] Existing base map generation technologies have several inherent limitations. First, their geometric models are rigid, supporting only fixed scalar parameter configurations. They cannot characterize or compensate for lens elliptic distortion, and can only adapt to ideal, distortion-free conditions, resulting in poor mass production adaptability. Second, their parallax modes are limited, supporting only linear parallax sampling with fixed offsets. They cannot achieve the 360° full parallax light field encoding required by fly-eye lenses, making it difficult to meet the production needs of high-end global stereoscopic lighting products. Third, their image quality optimization capabilities are insufficient. They often use simple random noise to optimize the image, failing to effectively eliminate far-field low-frequency moiré patterns and contour stripes, resulting in poor imaging effects.

[0005] Meanwhile, ultra-high resolution basemap rendering suffers from severe performance bottlenecks. High-end thin film production requires basemap resolutions of up to 10160 DPI and canvas sizes of 50000×50000μm, resulting in an extremely large number of pixels. Traditional single-threaded rendering methods not only take more than 30 minutes, making them unsuitable for mass production pace, but also frequently encounter problems such as memory overflows and rendering anomalies, leading to poor production stability.

[0006] In summary, existing technologies suffer from drawbacks such as lack of process distortion compensation, limited parallax encoding modes, low rendering efficiency, and poor imaging quality, failing to meet the demands for high-precision, high-efficiency, and automated mass production of microlens array base maps. Therefore, there is an urgent need to propose a microlens array base map generation method based on adaptive process distortion compensation to address these technical problems. Summary of the Invention

[0007] The purpose of this invention is to provide a microlens array base map generation method based on adaptive compensation for process distortion, so as to solve the problems of imaging distortion, multiple image defects and poor image quality consistency caused by process distortion in the prior art. Through adaptive compensation for process distortion, multi-mode parallax sampling, adaptive contour clipping and closed-loop parameter iteration mechanism, the image enhancement of the microlens array base map is realized, thereby improving imaging accuracy and image quality.

[0008] To achieve the above objectives, the present invention provides a method for generating microlens array base maps based on adaptive compensation for process distortion, comprising: S1. Obtain the process distortion parameters of the microlens film, wherein the process distortion parameters include at least the longitudinal stretching ratio; S2. Based on the process distortion parameters, determine the pixelation diameter and pixelation array spacing of the circular microlens in the horizontal and vertical directions, respectively. S3. Based on the pixelated diameter and pixelated array spacing, pre-calculate the lens center point array in rectangular array mode, wherein the horizontal arrangement step size and the vertical arrangement step size are independent of each other. S4. For the center point of each lens, image slices of matching pixel size are extracted through multi-mode parallax sampling, and adaptive cropping optimization is performed according to the process distortion parameters to suppress imaging distortion and image noise, thereby generating a high-fidelity microlens array enhancement base map.

[0009] Preferably, S1 specifically includes: S11. Collect the lateral shrinkage rate, target DPI, array type, and lens shape parameters of the microlens film; S12. When the lens shape is detected to be circular and the array type is a 90° rectangular array, the independent adjustment mode of horizontal and vertical geometric parameters is activated to provide parameter support for subsequent image distortion correction and image quality enhancement.

[0010] Preferably, S2 specifically includes: S21. Obtain the nominal diameter and nominal array spacing of the microlens array; S22. Based on the preset longitudinal stretching rate and transverse shrinkage rate, calculate the transverse physical diameter, longitudinal physical diameter, transverse physical spacing and longitudinal physical spacing after process distortion compensation respectively. S23. Based on the target DPI, perform pixel-discrete conversion on each physical dimension to obtain the corresponding horizontal and vertical pixel diameters and pixel array spacing, thereby achieving accurate mapping of physical distortion parameters to image pixel parameters.

[0011] The preferred S3 specifically includes: S31. For a 90° rectangular array structure, the horizontal pixelated array spacing is used as the horizontal arrangement step size, and the vertical pixelated array spacing is used as the vertical arrangement step size. S32. Based on the independent horizontal and vertical arrangement step size, an asymmetric rectangular grid is pre-calculated to adapt to the anisotropic process distortion in the thin film forming process.

[0012] Preferably, S4 specifically includes: S41. According to the preset fixed window ratio, the central region of the foreground image is inverted and sampled to generate an initial lens image slice; S42. An adaptive irregular cutting path is used instead of a fixed circular cutting path. The lateral and longitudinal dimension parameters of the irregular cutting path are matched with the lateral and longitudinal pixelation radii of the microlens, respectively, to form an asymmetric cutting profile that adapts to the anisotropic process distortion of the thin film. S43. Configure a sampling window that matches the horizontal and vertical dimensions of the asymmetric cropping outline to ensure that the cropped image slices are free from geometric stretching distortion and improve the regularity of the base image.

[0013] Preferably, the image slicing sampling process of S4 supports multiple types of disparity sampling logic, including linear disparity sampling mode and fly-eye full disparity sampling mode; Specifically, the fly-eye total parallax sampling mode includes: S44. Calculate the normalized offset of the center point of each lens relative to the center of the field of view; S45. Combining the preset maximum offset and the virtual depth parameter, the initial disparity offset of image sampling is obtained; S46. Radial phase noise is introduced to correct the initial parallax offset, suppress low-frequency contour stripe defects in the image, and optimize the overall image uniformity.

[0014] Preferably, S4 also includes: S47. Perform high-frequency dithering processing on the sampled image slices using a preset order of Bayer matrix; S48. Based on the radial normalized distance, the jitter intensity is adaptively and dynamically adjusted to suppress far-field moiré interference noise, thereby achieving image detail optimization and image quality enhancement.

[0015] Preferably, it also includes an ultra-high resolution parallel rendering step, specifically including: S51. Divide the overall rendering canvas into several independent rectangular tile areas; S52: The rendering of each tile area is completed in parallel through multi-threading, and the tile image data is cached by memory pool reuse mechanism to reduce the peak memory usage and ensure the stability of high-resolution image rendering. S53: Image conversion and compression storage are completed asynchronously through an independent background thread, enabling non-blocking rendering of the main thread and efficient output of high-definition enhanced base maps.

[0016] Preferably, S53 specifically includes: S531. Convert the rendered image into a palette image with preset hue levels to optimize the image color levels and visual effects. S532. A high-precision base map file is generated using a preset lossless compression algorithm. By fixing the white index value, grayscale recognition deviation during image processing is eliminated, ensuring the accuracy consistency of the enhanced image.

[0017] Preferably, the method further includes a closed-loop image enhancement step for iterative optimization of image quality, specifically including: S61. Collect the actual imaging parameters and geometric dimensions of the microlens array after the film is formed, and accurately obtain image defect data such as imaging distortion, texture deviation, and image noise. S62. Based on actual imaging defect deviations, the distortion compensation coefficients of longitudinal stretching rate and transverse shrinkage rate are optimized in reverse iteration, and the base map generation parameters are dynamically corrected. S63. Based on the iteratively optimized distortion compensation coefficient, update the process parameters for generating the base map to complete the iterative generation of the microlens array base map.

[0018] The beneficial effects of this invention are: 1. Based on the process distortion parameters of microlens films, this invention enables independent control of horizontal and vertical geometric parameters and asymmetric array arrangement, adaptive contour cutting, and targeted compensation for anisotropic distortion in film forming. It corrects image stretching and distortion caused by process distortion from the source, and effectively achieves image enhancement of microlens imaging. 2. The present invention is equipped with multi-mode parallax sampling, radial phase correction and adaptive jitter optimization mechanism, which can effectively suppress various image defects such as imaging moiré patterns, contour stripes and image unevenness, refine image details, greatly improve the imaging clarity and uniformity of microlens array, and have a significant image enhancement effect. 3. This invention adopts a tile-based multi-threaded parallel rendering and lossless image processing mechanism, which ensures the efficiency and accuracy of high-definition base map rendering while avoiding image processing deviations and stably outputting high-quality enhanced base maps. 4. This invention sets up a closed-loop parameter iterative optimization mechanism, which can dynamically correct process parameters according to the actual forming and imaging deviation, continuously optimize the image enhancement effect, solve the problem of poor image quality consistency in mass production of traditional technology, and is suitable for large-scale industrial production applications. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0020] Figure 1 This is a flowchart of the method of the present invention.

[0021] Figure 2 This is a schematic diagram of the rectangular array X / Y independent step size dot matrix of the present invention.

[0022] Figure 3 This is a schematic diagram of the sampling slice of the elliptical lens of the present invention.

[0023] Figure 4 This is a diagram showing the coordinate relationship of fly-eye total parallax sampling in this invention.

[0024] Figure 5 This is a diagram of the 24-thread tile segmentation and memory pool reuse architecture of the present invention.

[0025] Figure 6 This is a schematic diagram illustrating the principle of superimposing the high-frequency jitter Bayer matrix and phase noise in this invention.

[0026] Figure 7 This is the logic diagram for the 256-color palette conversion of the present invention. Detailed Implementation

[0027] The embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, but should not be construed as limiting the present invention.

[0028] like Figures 1 to 7 As shown, to address the anisotropic process distortion problem caused by lateral shrinkage and longitudinal stretching during the microlens film forming process, this invention first calculates the physical compensation dimensions and array spacing of the microlens in the lateral and longitudinal directions, and then uses the device's DPI to achieve precise mapping between physical dimensions and pixel dimensions, constructing an asymmetric grid that matches the process distortion characteristics. Simultaneously, it abandons the traditional fixed circular cutting method and adopts an irregularly shaped elliptical cutting contour adapted to the film deformation characteristics, ensuring that the lens shape in the prepared base image precisely matches the deformation characteristics of the actual formed lens. This fundamentally improves image defects caused by process distortion, such as geometric stretching distortion, pixel misalignment, and edge blurring, achieving basic enhancement of microlens images.

[0029] The microlens base map generation method of this invention is based on the acquisition of process distortion parameters. Step S1 primarily involves acquiring various process distortion parameters of the microlens film, with core parameters including at least the longitudinal stretching rate. During actual production, the transverse shrinkage rate, target DPI, lens array type, and lens shape parameters are simultaneously acquired. The system identifies the lens and array structure in real time. When the lens is determined to be a circular structure arranged in a 90° rectangular array, the system automatically activates the independent control mode for the transverse and longitudinal geometric parameters, removing the limitation of uniformly binding transverse and longitudinal parameters in traditional processes. This provides the basic operating conditions for subsequent differential size calculations and array arrangement design.

[0030] Specifically, step S1 includes two refinement sub-steps, S11 and S12: S11. Collect the lateral shrinkage rate, target output DPI, array type and lens shape parameters of the microlens film, and take the longitudinal stretching rate as the core distortion parameter to comprehensively cover the key influencing factors of anisotropic deformation during the film forming process, and ensure the data integrity and accuracy of subsequent distortion compensation calculations.

[0031] S12. After parameter acquisition is completed, the system identifies and determines the lens structure and array arrangement in real time. When the lens is detected to be circular and the array to be a 90° rectangular array, the independent adjustment mode of horizontal and vertical geometric parameters is automatically activated. This mode enables independent configuration and calculation of horizontal and vertical parameters, completely breaking the limitations of the traditional unified and fixed process parameters, and providing functional support for subsequent differential size conversion and asymmetric array arrangement.

[0032] After completing the process parameter acquisition and control mode switching, proceed to step S2. Based on the real distortion parameters acquired in S1, calculate the independent pixel diameter and pixel array spacing of the circular microlens in the horizontal and vertical directions. By completing parameter retrieval, physical size compensation, and pixel conversion step by step, achieve precise differential adaptation of the horizontal and vertical dimensions.

[0033] Specifically, step S2 includes three refinement sub-steps: S21, S22, and S23. S21. Retrieve the factory reference parameters of the microlens array to accurately obtain the nominal diameter of the microlens and the nominal array spacing, and use them as the original reference data for subsequent distortion compensation calculations.

[0034] S22. Combining the longitudinal stretching rate and transverse shrinkage rate collected in S1, the transverse shrinkage and longitudinal stretching of the film are calculated independently to obtain the transverse physical diameter, longitudinal physical diameter, transverse physical spacing and longitudinal physical spacing after process distortion compensation. This changes the traditional uniform size compensation mode and accurately matches the anisotropic deformation law of the film.

[0035] S23. Based on the device's target DPI parameters, perform pixel-based discrete conversion on all compensated horizontal and vertical physical dimensions, accurately converting the physical scale into the device's rendering pixel scale, and finally obtaining mutually independent horizontal pixel diameter, vertical pixel diameter, horizontal pixel array spacing, and vertical pixel array spacing, thus completing the accurate conversion from physical size to pixel size.

[0036] This invention, through dimensional differential size compensation and precise pixelation calculation, fully adapts to the deformation characteristics of the film's lateral shrinkage and longitudinal stretching, effectively offsetting the dimensional deviations generated during the film forming process, significantly improving the fit between the microlens design size and the actual formed size, and reducing geometric imaging distortion problems from the source.

[0037] After completing the pixel size conversion, proceed to step S3. Based on the independent pixelation parameters obtained in S2, independently set the horizontal and vertical arrangement step size in the rectangular array mode. Complete the precise pre-arrangement of the microlens center point array through the refinement sub-steps, and build an asymmetric array structure adapted to process distortion.

[0038] Specifically, step S3 includes two refinement sub-steps, S31 and S32: S31. For the 90° rectangular array structure, continuing the differentiated design approach, the horizontal pixel array spacing is set as the horizontal direction arrangement step size, and the vertical pixel array spacing is set as the vertical direction arrangement step size, ensuring that the horizontal and vertical arrangement steps are independent of each other and do not constrain each other.

[0039] S32. Based on mutually independent horizontal and vertical arrangement step sizes, the center coordinates of all microlenses are calculated in batches to generate an asymmetric rectangular grid, replacing the traditional regular and uniform symmetric grid, so that the array arrangement structure can accurately adapt to the anisotropic process distortion of thin film molding.

[0040] Traditional base map generation processes use uniform lens dimensions and array spacing. After film molding and stretching, the geometry of the microlenses undergoes passive deformation, easily leading to defects such as image stretching misalignment, lens edge blurring, and array offset. This solution collects real process distortion data through S1, performs differentiated size compensation calculations through S2, and then constructs an asymmetric grid matrix adapted to the deformation through S3. This ensures that the base map array arrangement closely matches the actual deformation characteristics of the film, effectively improving various imaging distortions caused by traditional uniform grids, significantly enhancing array arrangement accuracy and global imaging uniformity, and achieving fundamental image enhancement.

[0041] After the center point array of the lenses is arranged, the core image enhancement step S4 is entered. This step mainly focuses on performing precise image sampling, adaptive contour cropping, parallax correction, and multi-level noise reduction optimization for each lens center point. Through multiple refinement sub-steps from S41 to S48, the image quality of microlens imaging is comprehensively upgraded.

[0042] S41. According to the preset fixed window ratio, the central area of ​​the foreground image is cropped and inverted for sampling to quickly generate a regular initial lens image slice, providing standard basic material for subsequent cropping optimization, parallax correction and noise reduction processing.

[0043] S42. Abandoning the industry-standard fixed circular cutting method, an adaptive irregular cutting path is adopted, so that the horizontal and vertical dimensions of the cutting path match the horizontal and vertical pixelation radii of the microlens, forming an asymmetric cutting contour that adapts to the anisotropic distortion of the film.

[0044] S43. Strictly refer to the horizontal and vertical proportions of the asymmetrical cropping contour to configure a dedicated sampling window, ensuring that the sampling range accurately matches the lens deformation contour and avoiding sampling redundancy and loss of detail.

[0045] This step supports two working modes: linear parallax sampling and fly-eye full parallax sampling. Linear parallax sampling is suitable for 90° rectangular arrays, while fly-eye full parallax sampling is specifically designed for 60° cellular arrays and 45° rhomboid arrays. When using fly-eye full parallax sampling mode, precise parallax correction is performed through S44, S45, and S46: S44. Calculate the normalized offset of the center point of each lens relative to the center of the field of view to accurately distinguish the differences in the field of view of lenses at different positions. S45. Combining the preset maximum offset and the virtual depth parameter, the initial disparity offset of image sampling is obtained; S46. Radial phase noise is introduced to correct the initial disparity offset, eliminating systematic imaging bias caused by fixed parameter sampling.

[0046] After completing the parallax sampling correction, further refinement and noise reduction optimization are performed using S47 and S48: S47. Perform high-frequency dithering of the Bayer matrix of a preset order on all sampled image slices to remove subtle noise in the image and achieve basic detail noise reduction; S48. Based on the radial normalized distance of the lens relative to the center of the field of view, the jitter intensity is adaptively and dynamically adjusted to achieve precise noise reduction in different regions and with differentiation.

[0047] Traditional fixed sampling modes are prone to problems such as image brightness and darkness layering, moiré interference in long-distance observation, and viewfinder discontinuity when viewed from multiple angles. This step distinguishes between two parallax sampling logics: linear parallax adapts to a 90° rectangular array, while fly-eye full parallax adapts to a 60° honeycomb array and a 45° rhomboid array. Combined with irregular cropping to adapt to lens deformation characteristics and a radial adaptive jitter noise reduction mechanism, it can effectively correct global sampling offset, suppress periodic noise, and eliminate low-frequency interference fringes while fully preserving image details and textures. This achieves continuous and uniform 360° full-view imaging, significantly improving the continuity, uniformity, and clarity of microlens imaging.

[0048] After image slicing optimization and enhancement, the process proceeds to the ultra-high resolution base map rendering and output stage. Through steps S51, S52, S53, and their refinement sub-steps, an efficient and stable parallel rendering and lossless output architecture is built to meet the batch generation requirements of industrial-grade ultra-high precision base maps. The entire process of this invention also relies on a supporting software system for automated operation, with built-in modules for complete parameter configuration, image rendering, and format export.

[0049] S51. The ultra-large rendering canvas is divided into regions, dividing the complete canvas into multiple independent rectangular tile regions, distributing the computing power pressure, and solving the problems of high overall rendering difficulty and poor stability of large canvases.

[0050] S52: It starts a multi-threaded parallel rendering architecture, allocates independent threads to synchronously complete the rendering work of each tile area, and uses a memory pool reuse mechanism to cache tile image data, effectively reducing device memory usage and avoiding problems such as resource redundancy and rendering stuttering.

[0051] S53 adopts a decoupled rendering and storage model, using an independent background thread to asynchronously complete image conversion and compression storage, achieving non-blocking rendering on the main thread. This specifically includes two levels of refinement operations: S531 converts the rendered image into a palette image with preset tone levels. The system has a built-in CMYK four-color management channel to unify the global image tone standard and adapt to different color output requirements of optical thin film printing and photolithography. S532 uses a preset lossless compression algorithm to generate high-precision base map files. At the same time, by fixing the white index value, it eliminates grayscale recognition deviation in the image processing process and ensures the accuracy of the output image quality. The software also supports exporting GDS file format for lithography, which can directly output standardized pattern files to interface with the microlens lithography plate making process, eliminating the intermediate format conversion step and avoiding data distortion.

[0052] Traditional single-threaded rendering of ultra-large microlens base maps commonly suffers from rendering stuttering, memory overflow, and local pixel distortion. Furthermore, the uncompressed raw image files are enormous, hindering storage and transmission for mass production. This step employs tile-based parallel rendering technology, significantly improving rendering efficiency and device stability. Combined with palette calibration and lossless compression algorithms, it drastically reduces file size while fully preserving the image enhancement effects and detail accuracy from the initial rendering, outputting base maps in the required production format. This balances imaging accuracy with the output efficiency of industrial-scale mass production. The accompanying software is compatible with CMYK four-color output and photolithography GDS pattern saving, adapting to the processing needs of both printing and photolithography production lines, thus expanding the range of process adaptability.

[0053] During the mass production of thin films, the characteristics of raw materials and the operating conditions of equipment can cause slight dynamic fluctuations. If fixed parameters are used to generate the base map for a long time, it will be impossible to adapt to real-time process deviations, which can easily lead to inconsistent imaging quality between different batches of products. To address this, after the base map is finalized and put into mass production, this invention establishes a molding feedback closed-loop optimization system. Through the S61, S62, and S63 closed-loop sub-steps, the base map parameters are dynamically and iteratively optimized to continuously ensure the imaging consistency of batch products and achieve long-term image enhancement at the mass production stage.

[0054] S61. After the film molding mass production is completed, the actual geometric dimensions and global imaging parameters of the finished microlens array are collected to accurately capture various defect data such as imaging distortion, texture deviation, and image noise generated during the molding process, and to complete the accurate detection of process deviation and image quality defects.

[0055] S62. Based on the measured imaging defects and process deviation data, the longitudinal stretching rate and transverse shrinkage rate distortion compensation coefficients in S1 are optimized in reverse iteration, and all core process parameters generated by the base map are dynamically corrected to adapt to the real-time fluctuations of production conditions.

[0056] S63. Based on the iteratively optimized distortion compensation coefficient, update the entire set of base map generation process parameters, and re-execute the entire process from S1 to S5 to complete the iterative upgrade and optimized generation of the microlens array base map.

[0057] This closed-loop optimization system can adapt to various working condition fluctuations in the production process in real time, dynamically correct distortion compensation parameters, continuously correct imaging defects, and completely solve the problems of poor adaptability of traditional fixed parameter base maps and uneven batch image quality. It can effectively stabilize the imaging quality of mass-produced products and reduce the production defect rate.

[0058] Based on the complete execution of all technical steps S1 to S6 above, the following example illustrates the complete implementation process of the present invention in conjunction with a set of specific industrial conditions. All parameters are only used to explain the technical solution of the present invention and do not constitute a limitation on the scope of protection of the present invention.

[0059] Execute S1: Select a circular microlens with a 90° rectangular array structure, acquire the target output DPI=10160, the measured longitudinal tensile rate of the film is 4.4%, matching the corresponding transverse shrinkage rate; input the nominal transverse diameter of the lens is 125.0μm, the nominal longitudinal diameter is 130.5μm, the nominal transverse spacing of the array is 130.0μm, and the longitudinal spacing is 135.2μm; the system recognizes that the lens shape and array type meet the preset conditions and automatically starts the independent adjustment mode of transverse and longitudinal parameters.

[0060] Execute S2: Calculate the physical dimensions after distortion compensation by combining the collected stretching and shrinkage coefficients, and complete the pixel conversion based on the target DPI. Finally, the lens has a horizontal pixel size of 50px and a vertical pixel size of 52px, and the array has a horizontal pixel spacing of 52px and a vertical pixel spacing of 54px.

[0061] Execute S3: Set 52px as the horizontal layout step size and 54px as the vertical layout step size. The system will automatically generate an asymmetric rectangular grid that adapts to the longitudinal stretching characteristics of the film.

[0062] Execute S4: Set 35% of the foreground image area as a uniform sampling window, and extract a 52×54 pixel elliptical image slice. Use irregular ellipse cropping instead of traditional standard circle cropping. This example uses a fly-eye full parallax sampling mode adapted to a 60° honeycomb / 45° rhombus array, configures the imaging field of view to 60° and virtual depth to 0.5, calculates the field of view offset for each lens and uses radial phase noise to correct the sampling error; adaptively adjust the Bayer jitter intensity according to the radial distance of each lens to complete the slice global noise reduction optimization.

[0063] Execute S5: Set the physical canvas size to 50000×50000μm, corresponding to a pixel canvas of approximately 20000×20000 pixels. The system automatically divides the complete canvas into 48 independent tiles, and uses 24 threads to complete the rendering of all tiles in parallel. After the tiles are stitched together, an asynchronous background thread converts the image into a graded color palette image, enables CMYK four-color mode to calibrate the image colors, and uses LZW lossless compression to complete conventional image storage (such as TIFF). The software also supports exporting GDS layout files specifically for photolithography, which can be directly used for microlens array photolithography. After compression, the size of the conventional image file is only 15% of the original ARGB image.

[0064] S6: After the film is mass-produced and sampled, the actual longitudinal stretching rate of the lens is measured to be 4.8%, which deviates from the initial parameters entered in S1. Slight longitudinal stretching distortion appears at the edges of the finished image. The system automatically iterates backward to optimize the longitudinal stretching compensation coefficient, and simultaneously updates all related parameters such as S2 size conversion, S3 dot matrix arrangement, and S4 sampling and cropping. The S1 to S5 processes are then completely re-run to generate a new base map. After preparing the film using the new base map, the lens geometric imaging deviation is completely eliminated, and the finished product is free of defects such as moiré patterns and viewing angle breaks. The entire batch of products has uniform imaging brightness and stable and consistent image quality.

[0065] The above-disclosed embodiments are merely one or more preferred embodiments of this application and should not be construed as limiting the scope of this application. Those skilled in the art can understand that all or part of the processes for implementing the above embodiments and equivalent changes made in accordance with the claims of this application still fall within the scope of this application.

Claims

1. A method for generating a microlens array base map based on adaptive compensation for process distortion, characterized in that, include: S1. Obtain the process distortion parameters of the microlens film, wherein the process distortion parameters include at least the longitudinal stretching ratio; S2. Based on the process distortion parameters, determine the pixelation diameter and pixelation array spacing of the circular microlens in the horizontal and vertical directions, respectively. S3. Based on the pixelated diameter and pixelated array spacing, pre-calculate the lens center point array in rectangular array mode, wherein the horizontal arrangement step size and the vertical arrangement step size are independent of each other. S4. For the center point of each lens, image slices of matching pixel size are extracted through multi-mode parallax sampling, and adaptive cropping optimization is performed according to the process distortion parameters to suppress imaging distortion and image noise, thereby generating a high-fidelity microlens array enhancement base map.

2. The method for generating a microlens array base map based on adaptive compensation for process distortion as described in claim 1, characterized in that, S1 specifically includes: S11. Collect the lateral shrinkage rate, target DPI, array type, and lens shape parameters of the microlens film; S12. When the lens shape is detected to be circular and the array type is a 90° rectangular array, the independent adjustment mode of horizontal and vertical geometric parameters is activated to provide parameter support for subsequent image distortion correction and image quality enhancement.

3. The method for generating a microlens array base map based on adaptive compensation for process distortion as described in claim 1, characterized in that, S2 specifically includes: S21. Obtain the nominal diameter and nominal array spacing of the microlens array; S22. Based on the preset longitudinal stretching rate and transverse shrinkage rate, calculate the transverse physical diameter, longitudinal physical diameter, transverse physical spacing and longitudinal physical spacing after process distortion compensation respectively. S23. Based on the target DPI, perform pixel-discrete conversion on each physical dimension to obtain the corresponding horizontal and vertical pixel diameters and pixel array spacing, thereby achieving accurate mapping of physical distortion parameters to image pixel parameters.

4. The method for generating a microlens array base map based on adaptive compensation for process distortion as described in claim 1, characterized in that, S3 specifically includes: S31. For a 90° rectangular array structure, the horizontal pixelated array spacing is used as the horizontal arrangement step size, and the vertical pixelated array spacing is used as the vertical arrangement step size. S32. Based on the independent horizontal and vertical arrangement step size, an asymmetric rectangular grid is pre-calculated to adapt to the anisotropic process distortion in the thin film forming process.

5. The method for generating a microlens array base map based on adaptive compensation for process distortion as described in claim 1, characterized in that, S4 specifically includes: S41. According to the preset fixed window ratio, the central region of the foreground image is inverted and sampled to generate an initial lens image slice; S42. An adaptive irregular cutting path is used instead of a fixed circular cutting path. The lateral and longitudinal dimension parameters of the irregular cutting path are matched with the lateral and longitudinal pixelation radii of the microlens, respectively, to form an asymmetric cutting profile that adapts to the anisotropic process distortion of the thin film. S43. Configure a sampling window that matches the horizontal and vertical dimensions of the asymmetric cropping outline to ensure that the cropped image slices are free from geometric stretching distortion and improve the regularity of the base image.

6. The method for generating a microlens array base map based on adaptive compensation for process distortion as described in claim 1, characterized in that, S4's image slicing sampling process supports multiple types of disparity sampling logic, including linear disparity sampling mode and fly-eye full disparity sampling mode. Specifically, the fly-eye total parallax sampling mode includes: S44. Calculate the normalized offset of the center point of each lens relative to the center of the field of view; S45. Combining the preset maximum offset and the virtual depth parameter, the initial disparity offset of image sampling is obtained; S46. Radial phase noise is introduced to correct the initial parallax offset, suppress low-frequency contour stripe defects in the image, and optimize the overall image uniformity.

7. The method for generating a microlens array base map based on adaptive compensation for process distortion as described in claim 6, characterized in that, S4 also includes: S47. Perform high-frequency dithering processing on the sampled image slices using a preset order of Bayer matrix; S48. Based on the radial normalized distance, the jitter intensity is adaptively and dynamically adjusted to suppress far-field moiré interference noise, thereby achieving image detail optimization and image quality enhancement.

8. The method for generating a microlens array base map based on adaptive compensation for process distortion as described in claim 1, characterized in that, It also includes ultra-high resolution parallel rendering steps, specifically including: S51. Divide the overall rendering canvas into several independent rectangular tile areas; S52: The rendering of each tile area is completed in parallel through multi-threading, and the tile image data is cached by memory pool reuse mechanism to reduce the peak memory usage and ensure the stability of high-resolution image rendering. S53: Image conversion and compression storage are completed asynchronously through an independent background thread, enabling non-blocking rendering of the main thread and efficient output of high-definition enhanced base maps.

9. The method for generating a microlens array base map based on adaptive compensation for process distortion as described in claim 8, characterized in that, S53 specifically includes: S531. Convert the rendered image into a palette image with preset hue levels to optimize the image color levels and visual effects. S532. A high-precision base map file is generated using a preset lossless compression algorithm. By fixing the white index value, grayscale recognition deviation during image processing is eliminated, ensuring the accuracy consistency of the enhanced image.

10. The method for generating a microlens array base map based on adaptive compensation for process distortion as described in claim 1, characterized in that, It also includes a closed-loop image enhancement step for iterative optimization of image quality, specifically including: S61. Collect the actual imaging parameters and geometric dimensions of the microlens array after film forming to accurately obtain image defect data such as imaging distortion, texture deviation, and image noise. S62. Based on actual imaging defect deviations, the distortion compensation coefficients of longitudinal stretching rate and transverse shrinkage rate are optimized in reverse iteration, and the base map generation parameters are dynamically corrected. S63. Based on the iteratively optimized distortion compensation coefficient, update the process parameters for generating the base map to complete the iterative generation of the microlens array base map.

Images