Image digitization method, system and device and storage medium

Abstract

The invention discloses an image digitization method, system and device and a storage medium, and relates to the field of data processing. The method comprises the following steps: acquiring spectral data, three-dimensional shape data and material reflectivity parameters of a target image; based on the three-dimensional shape data, analyzing surface deformation characteristics of the target image, and generating a geometric distortion distribution diagram; generating geometric correction parameters according to the surface deformation features, and reconstructing pixel positions of the geometric distortion distribution diagram to obtain reconstructed image data; and mapping the spectral data to a preset color space, matching a pre-established color database, and adjusting the color component of the reconstructed image data to obtain corrected image data. By implementing the technical scheme provided by the invention, a set of traditional Chinese painting digitization technical system considering high-precision acquisition, intelligent correction and efficient management can be developed.

Description

Technical Field

[0001] The present application relates to the technical field of data processing, and in particular to a method, system, device and storage medium for image digitization. Background Art

[0002] Classical Chinese paintings, as a core part of China's cultural heritage, possess irreplaceable artistic value and historical significance. However, due to limitations in traditional preservation methods and the fragility of physical materials (such as rice paper and silk fabrics, which age and fade easily), many of these treasures face the risk of irreversible damage. With the advent of the digital age, the use of technology to achieve high-precision preservation, research, and dissemination of Chinese paintings has become a pressing need in the field of cultural preservation.

[0003] While multispectral imaging and high-resolution scanning technologies have been gradually applied to cultural heritage in recent years, their adaptability to the unique materials and artistic expressions of traditional Chinese painting remains insufficient. For example, existing color management relies heavily on universal standards and lacks specific calibration for the characteristics of traditional mineral pigments and ink blending. Furthermore, geometric correction algorithms need to improve their accuracy in handling complex non-rigid deformations.

[0004] Therefore, developing a digital technology system for traditional Chinese painting that takes into account high-precision acquisition, intelligent correction and efficient management has become a key direction to break through industry bottlenecks. Summary of the Invention

[0005] In order to develop a traditional Chinese painting digitization technology system that takes into account high-precision acquisition, intelligent correction and efficient management, this application provides a method, system, device and medium for image digitization.

[0006] In a first aspect of the present application, a method for digitizing an image is provided, which adopts the following technical solution: Obtain spectral data, three-dimensional topography data and material reflectance parameters of the target image; Analyzing the surface deformation characteristics of the target image based on the three-dimensional topography data to generate a geometric distortion distribution map; generating geometric correction parameters according to the surface deformation characteristics, and reconstructing pixel positions of the geometric distortion distribution map to obtain reconstructed image data; The spectral data is mapped to a preset color space, matched with a pre-established color database, and the color components of the reconstructed image data are adjusted to obtain corrected image data.

[0007] By adopting the above technical solutions, ultra-high-definition digital imaging with a resolution of hundreds of millions of pixels is achieved, which can accurately capture micro details such as rice paper fibers and silk cloth textures; a geometric correction algorithm based on feature point detection is used to effectively eliminate deformation errors such as wrinkles and bends, so that the geometric accuracy error rate between the corrected image and the original Chinese painting is greatly reduced; through color space conversion and self-built Chinese painting color gamut database, not only can the original color be accurately restored, but faded areas can also be intelligently enhanced; a set of Chinese painting digitalization technology system is developed that takes into account high-precision acquisition, intelligent correction and efficient management.

[0008] Optionally, the acquiring of spectral data, three-dimensional topography data, and material reflectance parameters of the target image includes: Scanning the target image to obtain the spectral data of each band and the three-dimensional coordinate data of the surface, and constructing the three-dimensional topography data of the target image according to the three-dimensional coordinate data; The target image is illuminated by a light source, the incident angle and polarization direction of the light source are adjusted, and the spectral distribution of the reflected light is detected by a spectrometer to obtain the material reflectivity parameters of the material surface.

[0009] By adopting the above technical solution, by scanning to obtain spectral data and three-dimensional coordinate data of each band, and combining multispectral information with three-dimensional morphology information, the three-dimensional brushstrokes, material texture and other details of the image can be accurately restored, making up for the limitations of traditional two-dimensional digitization technology. By adjusting the incident angle and polarization direction of the light source and using a spectrometer to detect the distribution of reflected light, the reflective characteristics of Chinese painting materials under different lighting conditions can be dynamically captured. This parameter provides a physical basis for the color restoration algorithm, significantly improving the color correction accuracy of faded areas or complex pigment areas. The combination of three-dimensional morphology data and material reflectivity parameters can analyze the impact of surface deformation of Chinese paintings on color performance, thereby simultaneously optimizing color compensation during the geometric correction stage and further improving the overall restoration consistency.

[0010] Optionally, analyzing the surface deformation characteristics of the target image based on the three-dimensional topography data to generate a geometric distortion distribution map specifically includes: Dividing the surface of the target image into a preset number of grid units according to a preset size; Comparing the three-dimensional surface topography data of the grid unit with the corresponding preset plane model to obtain the deformation offset of the grid unit; Marking the grid cells whose deformation offset exceeds a preset threshold as geometric distortion areas; Boundary features of all the geometric distortion areas are extracted to generate the geometric distortion distribution map.

[0011] By employing the aforementioned technical solution, the target image surface is gridded and analyzed unit by unit, enabling refined detection and visualization of geometric distortion on the surface of traditional Chinese paintings. Specifically, gridding enables detection accuracy to submillimeter levels, making it particularly suitable for capturing subtle deformations such as rice paper wrinkles and silk fabric stretching. The threshold determination mechanism for deformation offset intelligently distinguishes between natural material textures and real structural deformations, avoiding overcorrection that damages the original features. The resulting geometric distortion distribution map visually annotates the degree of distortion in each area through heat maps or contour lines. This not only provides precise spatial parameter guidance for subsequent correction algorithms, but also allows for reverse tracing of physical damage locations, assisting in decision-making regarding cultural relic restoration.

[0012] Optionally, generating geometric correction parameters according to the surface deformation features and reconstructing pixel positions of the geometric distortion distribution map to obtain reconstructed image data specifically includes: Extracting edge feature points and texture key points in the target image to generate an initial geometric deformation mapping relationship; Constructing a physical deformation model based on the material properties of the target image, wherein the physical deformation model includes a material elastic modulus, an environmental humidity parameter, and a force deformation curve; Iteratively adjusting the initial geometric deformation mapping relationship according to the physical deformation model to generate the geometric correction parameters; The pixels of the geometric distortion distribution map are reconstructed according to an interpolation algorithm and the geometric correction parameters to obtain reconstructed image data.

[0013] By adopting the above technical solution, firstly, based on the dual extraction of edge feature points and texture key points, the deformation characteristics of the overlapping areas of brush strokes and material textures can be accurately located, effectively improving the accuracy of the initial deformation mapping relationship; secondly, a physical deformation model including parameters such as material elastic modulus and ambient humidity is introduced, so that the correction process not only considers the deformation performance at the image level, but also reflects the real physical deformation laws of different materials such as rice paper and silk cloth, enhancing the ability to interpret complex deformations such as wrinkles and bends; thirdly, through the iterative adjustment of the initial mapping relationship through the physical model, the correction parameters can be dynamically optimized, significantly reducing the cumulative error of a single correction; finally, an adaptive interpolation algorithm is used for pixel reconstruction, which eliminates geometric distortion while fully preserving the sharp edges of the brush strokes and the micro-texture characteristics of the material.

[0014] Optionally, mapping the spectral data to a preset color space, matching a pre-established color database, and adjusting the color components of the reconstructed image data to obtain corrected image data specifically includes: Converting the color representation of the spectral reflectance data into a preset color space to extract the chromaticity component of the target image; Obtaining the material type and lighting conditions of the target image; Retrieving a standard color sample that matches the material type and lighting conditions from a pre-established color database, and obtaining a corresponding standard chromaticity component; Comparing the chromaticity component of the target image with the standard chromaticity component, and generating a color adjustment parameter according to the comparison result; Based on the color adjustment parameter, the chroma component of the target image is corrected so that the difference between the corrected chroma component and the standard chroma component is less than a preset threshold; The corrected color components are output to form corrected image data.

[0015] By employing the above technical solution, the color representation of spectral reflectance data is converted to a preset color space and the chromaticity components of the target image are extracted, ensuring accurate extraction and representation of color information. This color space conversion fully utilizes the characteristics of the color space, laying the foundation for subsequent color matching and correction. The material type and lighting conditions of the target image are determined, and matching standard color samples are retrieved from a pre-established color database, making color correction more accurate to the actual situation. Color representation varies under different materials and lighting conditions, and this technical solution fully accounts for these factors, thereby improving the accuracy of color correction. The chromaticity components of the target image are compared with the standard chromaticity components, and color adjustment parameters are generated based on the comparison results, ensuring the scientific and reasonable color adjustment. In this way, the chromaticity components of each pixel can be finely adjusted, ensuring that the difference between the corrected chromaticity components and the standard chromaticity components is within a preset threshold, thereby achieving high color restoration.

[0016] Optionally, comparing the chromaticity component of the target image with the standard chromaticity component and generating the color adjustment parameter according to the comparison result specifically includes: Calculating a color difference vector between the chroma component of the target image and the standard chroma component, wherein the color difference vector includes a brightness difference and a chroma difference; Based on the color difference vector, a color adjustment matrix is ​​generated by a preset color correction model, wherein the color correction model includes linear transformation coefficients and nonlinear compensation parameters; The weight of the color adjustment matrix is ​​adjusted according to the material reflectance parameter of the target image to generate a color adjustment parameter.

[0017] By employing this technical solution, the difference between the target image's chromaticity components and the standard's chromaticity components is calculated, accurately quantifying the difference between them. This quantification method ensures scientific and accurate color adjustment, making the digitized image's colors closer to the true colors of the original Chinese painting. Based on the color difference vectors, a color adjustment matrix is ​​generated using a preset color correction model. This model not only includes linear transformation coefficients but also incorporates nonlinear compensation parameters. This extends the color correction process beyond simple linear adjustments and enables it to handle more complex color variations, particularly when dealing with gradient colors and special pigments commonly found in Chinese painting. Taking into account the diversity of image materials, the weights of the color adjustment matrix are dynamically adjusted based on the target image's material reflectance parameters. This feature ensures that the color adjustment parameters can be personalized and optimized based on the characteristics of different materials, further improving the accuracy of color reproduction. This is particularly effective for aging Chinese paintings whose materials have changed over time, enabling a better restoration of their original colors.

[0018] Optionally, the method further includes: Decomposing the corrected image data into a data set comprising at least two levels; Encoding the first level data in the data set using a transformation-based compression algorithm and encoding the second level data using a sparse feature-based compression algorithm to form a layered compressed data stream; generating a unique identifier based on the layered compressed data stream, wherein the unique identifier includes a data feature summary, a timestamp, and processing device information; The unique identifier is bound to the layered compressed data stream and stored in a storage network.

[0019] By employing the above technical solution, the rectified image data is decomposed into at least two levels of data sets. Transform-based compression algorithms and sparse feature-based compression algorithms are applied to the data characteristics of each level, respectively. This hierarchical compression strategy significantly reduces data volume while maintaining image quality. Hierarchical compression not only reduces data volume but also optimizes data processing. The first-level data retains the image's primary visual information, suitable for quick browsing and preliminary analysis; the second-level data contains more detailed feature information, suitable for in-depth research and high-precision applications. This hierarchical processing approach allows users to load different levels of data based on their specific needs, improving data access speed and flexibility. By generating a unique identifier for the layered compressed data stream and binding it to the data for storage, the uniqueness and traceability of each image data set are ensured. The unique identifier includes a data feature summary, a timestamp, and processing device information. This information not only helps verify data integrity and authenticity, but also records the data generation and processing history, facilitating subsequent audits and accountability tracking. Storing the uniquely identified layered compressed data stream on a distributed storage network enables distributed redundant data storage, enhancing data security and reliability.

[0020] In a second aspect of the present application, a system for digitizing an image is provided, comprising: Data acquisition module, used to obtain spectral data, three-dimensional shape data and material reflectance parameters of the target image; a feature extraction module, configured to analyze the surface deformation features of the target image based on the three-dimensional shape data and generate a geometric distortion distribution map; a geometric correction module, configured to generate geometric correction parameters according to the surface deformation characteristics, reconstruct pixel positions of the geometric distortion distribution map, and obtain reconstructed image data; The spectral color correction module is used to map the spectral data to a preset color space, match it with a pre-established color database, and adjust the color components of the reconstructed image data to obtain corrected image data.

[0021] In the third aspect of the present application, an electronic device is provided, including a processor, a memory, a user interface and a network interface, the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device performs any of the methods described above.

[0022] In a fourth aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores instructions. When the instructions are executed, any one of the methods described above is executed.

[0023] In summary, one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: By optimizing acquisition parameters, digital images of traditional Chinese paintings with resolutions reaching hundreds of millions of pixels have been captured. This not only clearly displays the subtle brushstrokes, lines, and color gradations of traditional Chinese paintings, but also captures material details such as the fiber structure of rice paper and the warp and weft patterns of silk cloth, providing high-quality image data for the study and appreciation of traditional Chinese paintings. Utilizing feature point detection and matching algorithms and image correction algorithms, geometric distortion caused by wrinkles, bends, and other deformations during the acquisition process is effectively eliminated. The corrected images are highly consistent in shape and size with the originals, ensuring geometric accuracy and overcoming the difficulties associated with traditional techniques in correcting geometric distortion. Leveraging multispectral acquisition technology and color correction algorithms, combined with a traditional Chinese painting color database, this technology accurately reproduces the true colors of traditional Chinese paintings. Especially for aged and faded traditional Chinese paintings, the digitized images exhibit vibrant colors and distinct gradations, closely resembling the originals and even improving color stability and consistency. Advanced image compression algorithms reduce the data size of high-definition digitized images while maintaining excellent image quality, significantly improving data storage and transmission efficiency. The supporting data management system supports the efficient management and sharing of large-scale Chinese painting data, facilitates users to query, retrieve and analyze, and provides strong technical support for the digital protection, research and dissemination of classic Chinese paintings. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 This is a schematic diagram of the system architecture of a method or system for image digitization disclosed in an embodiment of the present application; Figure 2 This is a flow chart of a method for image digitization disclosed in an embodiment of the present application; Figure 3 This is a module diagram of an image digitization system disclosed in an embodiment of the present application; Figure 4 This is a structural diagram of an electronic device disclosed in an embodiment of the present application.

[0025] Explanation of the accompanying drawings: 301, data acquisition module; 302, feature extraction module; 303, geometric correction module; 304, spectral color correction module; 305, data compression and storage module; 401, processor; 402, communication bus; 403, user interface; 404, network interface; 405, memory. DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments.

[0027] In the description of the embodiments of this application, words such as "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "for example" or "for instance" in the embodiments of this application should not be construed as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "for example" or "for instance" is intended to present the relevant concepts in a concrete manner.

[0028] In the description of the embodiments of the present application, the term "multiple" means two or more. For example, multiple systems refer to two or more systems, and multiple screen terminals refer to two or more screen terminals. In addition, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized.

[0029] like Figure 1 As shown, system architecture 100 may include terminal devices 101, 102, 103, a network 104, and a cloud server 105. Network 104 is used to provide a medium for communication links between terminal devices 101, 102, 103 and cloud server 105. Network 104 may include various connection types, such as wired or wireless communication links or fiber optic cables.

[0030] Users can use terminal devices 101, 102, and 103 to interact with cloud server 105 via network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, and 103, such as model training applications, video recognition applications, web browser applications, social platform software, etc.

[0031] Terminal devices 101, 102, and 103 can be hardware or software. When terminal devices 101, 102, and 103 are hardware, they can be various electronic devices with display screens, including but not limited to smartphones, tablet computers, e-book readers, MP3 (Moving Picture Experts Group Audio Layer III, Moving Picture Experts Group Audio Layer 3) players, MP4 (Moving Picture Experts Group Audio Layer IV, Moving Picture Experts Group Audio Layer 4) players, laptop computers, and desktop computers, etc. When terminal devices 101, 102, and 103 are software, they can be installed in the electronic devices listed above. They can be implemented as multiple software or software modules (for example, multiple software or software modules used to provide distributed services), or they can be implemented as a single software or software module. No specific limitation is made here.

[0032] When terminals 101, 102, and 103 are hardware, they may also be equipped with a video capture device. The video capture device may be any device capable of capturing video, such as a camera, a sensor, and the like. Users can use the video capture device on terminals 101, 102, and 103 to capture video.

[0033] The cloud server 105 may be a server that provides various services, such as a backend server that processes data displayed on the terminal devices 101, 102, and 103. The backend server may analyze and process the received data, and may feed back the processing results (e.g., recognition results) to the terminal device.

[0034] It should be noted that the cloud server can be either hardware or software. When the cloud server is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When the cloud server is software, it can be implemented as multiple software programs or software modules (for example, multiple software programs or software modules used to provide distributed services), or as a single software program or software module. This is not specifically limited here.

[0035] It should be understood that Figure 1 The number of terminal devices, networks, and cloud servers described above is merely illustrative. Any number of terminal devices, networks, and cloud servers may be provided as needed. In particular, if the target data does not need to be acquired remotely, the system architecture may not include a network, but may instead include only terminal devices or cloud servers.

[0036] This embodiment discloses a method for digitizing an image. Figure 2This is a flow chart of a method for image digitization disclosed in an embodiment of the present application, such as Figure 2 As shown, the method includes the following steps: S201, obtaining spectral data, three-dimensional topography data and material reflectance parameters of a target image; Specifically, a professional large-format digital scanner equipped with a high-resolution CCD (charge-coupled device) sensor and high-quality optical lenses is used to capture the finest details of traditional Chinese paintings. The scanner must possess multispectral acquisition capabilities, capable of scanning the target image in multiple wavelength bands (such as visible light and near-infrared) to obtain spectral data in each band. This facilitates subsequent color restoration and material analysis. During the scanning process, the three-dimensional coordinate data of the target image surface is simultaneously recorded. Laser triangulation or structured light scanning technology precisely measures the spatial position of each point on the painting surface, constructing three-dimensional topographic data of the target image. This data can reflect the physical form of the painting, such as wrinkles and undulations. To obtain material reflectance parameters, a specific light source (such as an LED) is used to illuminate the target image. The incident angle and polarization direction of the light source are adjusted to ensure uniform illumination across different areas of the painting surface. A high-precision spectrometer is then used to measure the spectral distribution of the reflected light, thereby determining the reflectance parameters of the material surface at different wavelengths. These parameters are crucial for subsequent color correction and material property analysis. The acquired spectral data, 3D topographic data, and material reflectance parameters are integrated to ensure temporal and spatial consistency across the datasets. Preliminary data preprocessing, such as denoising and calibration, is performed to improve the accuracy of subsequent processing. Finally, the processed data is stored in a dedicated database for further analysis and application.

[0037] Optionally, the acquiring of spectral data, three-dimensional topography data, and material reflectance parameters of the target image includes: Scanning the target image to obtain the spectral data of each band and the three-dimensional coordinate data of the surface, and constructing the three-dimensional topography data of the target image according to the three-dimensional coordinate data; The target image is illuminated by a light source, the incident angle and polarization direction of the light source are adjusted, and the spectral distribution of the reflected light is detected by a spectrometer to obtain the material reflectivity parameters of the material surface.

[0038] Specifically, first, a hyperspectral imager (such as SPECIM IQ or Headwall Photonics series) is used as the core of spectral acquisition. Its spectroscopic system decomposes the incident light into dozens to hundreds of continuous bands (such as the 400-900nm range, with an interval of 10nm), and acquires data frame by frame / line by line in a planar or linear scanning manner, accurately capturing the characteristic absorption spectrum of the azurite pigment in the 520nm band in the traditional Chinese painting "A Thousand Miles of Rivers and Mountains"; synchronously start a three-dimensional scanner (such as Artec Eva structured light scanner or Faro Focus laser scanner) to obtain the XYZ coordinates of the surface discrete point cloud by projecting a coded grating or measuring the flight time of the laser pulse, with an accuracy of up to 0.1mm, which is sufficient to record the micron-level fluctuations of the rust layer of bronze artifacts; then, use a feature point matching algorithm (such as SIFT / SURF) to align the two-dimensional coordinates of the spectral data with the three-dimensional point cloud, and then use point cloud processing software (such as Geomagic) to align the two-dimensional coordinates of the spectral data with the three-dimensional point cloud. Wrap) to perform denoising, simplification and triangulation, and finally map the hyperspectral data to the three-dimensional model in the form of pseudo-color texture, forming a digital replica that combines the chemical information of mineral pigments (such as the reflectivity of ochre at 580nm) and physical morphological characteristics (such as the texture of pine needle strokes), realizing non-contact, multimodal data fusion throughout the process.

[0039] Furthermore, an adjustable-angle LED light source array or laser is used as the illumination source, and its incident angle is controlled by an electric rotation stage or a manual fine-tuning bracket, for example, it can be gradually adjusted from 0° to 75° in steps of 15°. At the same time, the polarization direction is precisely controlled by an electric rotating polarizer (such as Thorlabs LPVIS100), which can be set to four gears of 0°, 45°, 90°, and 135° or a continuous rotation mode; the reflected light is received by a fiber optic spectrometer (such as Ocean Insight Maya2000 Pro), and its probe maintains a fixed angle (such as 45°) with the light source-sample normal through a collimating mirror, and is 5-10 cm away from the sample surface. During detection, it is first calibrated with a standard white board (such as Labsphere Spectralon), and then the sample reflected light intensity is measured. For example, when detecting mineral pigments in Chinese paintings, the light source is irradiated with an incident angle of 45° and a polarization direction of 90°. The spectrometer collects a reflected light intensity of 12.6 mW / sr nm in the 550 nm band. Combined with the white board data of 15.8 mW / sr nm, the formula The calculated reflectivity for this band is 79.7%. The following characterizes the reflected light intensity (unit: mW / sr·nm) of the target material (such as the pigment layer of a traditional Chinese painting) at a specific wavelength as measured by the spectrometer: I_sample. I_white represents the reflected light intensity of a standard white plate (such as a polytetrafluoroethylene calibration plate, with a theoretical reflectivity of 100%) under the same conditions. ρ represents the reflectivity of the target material at this wavelength (as a percentage). Finally, through multi-angle scanning and polarization modulation, a parameterized model of the material's reflectivity versus wavelength, incident angle, and polarization state is established.

[0040] S202, analyzing the surface deformation characteristics of the target image based on the three-dimensional topography data to generate a geometric distortion distribution map; Specifically, based on the three-dimensional morphological data of the target image, a basic reference system for deformation analysis is established by aligning it with the original lossless state data or a reference morphological model constructed by mirror replication and statistical averaging; then, a multi-scale geometric analysis method is used to quantitatively extract surface deformation features, including calculating the normal offset and curvature change rate of each vertex to capture local deformation, and identifying global deformation features such as overall tilt and periodic fluctuations through principal component analysis and fast Fourier transform; then, the deformation is quantified and classified, and the deformed areas are marked as different degrees of severity by setting deformation thresholds, and independent deformation events are identified using cluster analysis; finally, a geometric distortion distribution map is generated, and the deformation features are mapped into visual elements such as color and brightness, and the visualization is enhanced by combining the three-dimensional coordinate system and scale. At the same time, the deformation traceability analysis can be optionally extended, and the material decay information is associated with spectral data to comprehensively evaluate the surface deformation condition of the target image.

[0041] Optionally, analyzing the surface deformation characteristics of the target image based on the three-dimensional topography data to generate a geometric distortion distribution map includes: Dividing the surface of the target image into a preset number of grid units according to a preset size; Comparing the three-dimensional surface topography data of the grid unit with the corresponding preset plane model to obtain the deformation offset of the grid unit; Marking the grid cells whose deformation offset exceeds a preset threshold as geometric distortion areas; Boundary features of all the geometric distortion areas are extracted to generate the geometric distortion distribution map.

[0042] Specifically, the original 3D morphology data is preprocessed using a point cloud processing library (such as Open3D or PCL), including eliminating discrete noise points through radius filtering or statistical filtering, and using voxel downsampling technology to balance data accuracy and computational efficiency. Subsequently, the grid size is dynamically determined according to the target material characteristics. A tiny grid of 0.5mm×0.5mm is used for flexible materials (such as rice paper) to capture micron-level morphology changes, while a grid of 2mm×2mm is set for rigid materials (such as ceramics) to take into account macroscopic deformation detection. The grid density of mixed material areas (such as the mounting layer of calligraphy and painting) is dynamically adjusted according to the risk level. Next, the point cloud is projected onto the XY plane through the octree algorithm for spatial division to generate a uniform cubic grid, and RANSAC (random sampling consensus, RANdom SAmple The local plane of each grid is fitted using the CNN Consensus algorithm to construct the initial grid surface. The pyvista library is used to visually verify the grid model to ensure that the grid density of key areas (such as the inscription area of ​​traditional Chinese paintings) meets the analysis requirements. To adapt to morphological mutations, the standard deviation of the point cloud curvature within the grid is calculated to identify the mutation area, and the quadtree subdivision strategy is used to recursively subdivide the high-curvature grid until the curvature threshold is met or the maximum subdivision level is reached. At the same time, the grid boundaries of the morphological edges (such as pigment cracks) are aligned to avoid feature truncation. Finally, the grid uniformity is evaluated by calculating the grid area variation coefficient, and the grid coverage of key morphological features (such as mineral pigment particles) is verified to ensure that more than 95% of the features are fully covered.

[0043] Furthermore, a reference plane model is constructed. If the original lossless three-dimensional data exists, it is directly used as a comparison benchmark. If there is no original data, a reference plane needs to be generated based on the current morphology. The specific methods include: for flexible materials (such as rice paper), local neighborhood fitting is used. With each grid as the center, the undistorted area within the surrounding 3×3 grid range is selected, and the local best fit plane (BFP, Best Fit Plane) is fitted using the RANSAC algorithm as a benchmark. This algorithm iteratively randomly samples a set of points and fits a plane model, and selects the model with the largest number of internal points as the optimal result. For rigid materials (such as ceramics), a global benchmark is used, and the main direction plane of the overall morphology is extracted as a reference through PCA (Principal Component Analysis). PCA determines the main direction of morphology change by calculating the eigenvector of the data covariance matrix. In the deformation calculation stage, for each point cloud data in the grid, its vertical distance to the reference plane is calculated. The specific method is as follows: for the fitted plane equation, the degree of deformation is quantified by calculating the shortest distance from the point to the plane. To reduce noise interference, the median of the distance of all points in the grid is statistically calculated as the deformation offset δ of the grid. The median is more resistant to the influence of outliers than the mean.

[0044] Furthermore, deformation thresholds are set based on specific application scenarios and accuracy requirements. For the field of cultural relics protection (such as restoration of ancient calligraphy and painting), the deformation threshold of flexible materials (rice paper) is set to 50μm, and the threshold of rigid materials (ceramics) is set to 200μm. For mixed material areas (such as calligraphy and painting mounting layers), a dynamic threshold is used, which is adaptively adjusted within the range of 30-150μm according to the complexity of local morphology (such as the standard deviation of neighborhood grid curvature). Subsequently, all grid cells are traversed, and the deformation offset δ of each grid is compared with the preset threshold. If δ exceeds the threshold, the grid is marked as a geometric distortion area. In order to reduce noise interference, a sliding window mechanism is adopted. Only when more than 50% of the grids in a continuous 3×3 neighborhood simultaneously meet δ>threshold , finally confirming the distortion mark to avoid false positives of isolated outliers. A confidence weighting mechanism is introduced for mixed-material areas. For high-risk areas such as the junction of mounting layers, the deformation threshold is lowered by 30% and the neighborhood judgment weight is increased to ensure that subtle deformations (such as 20μm-level pigment layer warping) are not missed. Finally, the marked areas are highlighted through 3D visualization rendering (e.g., using the PyVista library), and the degree of deformation is encoded using color mapping (e.g., red indicates severe distortion (δ>300μm), yellow indicates moderate distortion (100μm<δ≤300μm), and blue indicates mild distortion (50μm<δ≤100μm). A distortion heat map is generated and superimposed on the original topography model, providing an intuitive basis for subsequent restoration decisions. This method achieves a complete closed loop from threshold setting to distortion marking, ensuring the accuracy and scene adaptability of geometric distortion detection.

[0045] Furthermore, a boundary detection algorithm based on normal mutation is used to identify the contour of the distorted area. For each grid unit marked as distorted, its normal angle with the neighboring grid is calculated. When the normal angle of the flexible material (such as rice paper) area exceeds 15° and that of the rigid material (such as ceramics) exceeds 30°, the grid is determined to be a candidate boundary point. Subsequently, the discrete boundary points are contour fitted by the Alpha Shape algorithm. The α parameter of the flexible material area is set to 0.8 times the grid size to capture the slightly deformed boundary, and the α parameter of the rigid material area is set to 1.5 times the grid size to smooth the contour noise. A dynamic α parameter is used for the mixed material area (such as the calligraphy and painting mounting layer), and it is adaptively adjusted within the range of 0.5-2 times the grid size according to the local morphological complexity (such as the standard deviation of the neighborhood grid curvature). To address the boundary break problem, a contour closure strategy is introduced. When a short boundary with a length less than 3 times the grid size is detected, it is aligned with the neighboring boundary through morphological expansion (setting a 3×3 grid expansion kernel). Connections are made to ensure the continuity of deformed areas (such as pigment cracks); finally, a geometric distortion distribution map is generated, using a dual-mode visualization scheme: in three-dimensional space, the boundary contours are rendered as translucent red pipelines using the PyVista library and superimposed on the original morphology model. The pipeline thickness is proportional to the deformation offset δ (for every 50μm increase in δ, the pipeline diameter increases by 0.2mm); in two-dimensional projection, a deformation heat map is generated through orthographic projection, using Jet color scale mapping (blue to red corresponds to δ from 0 to the maximum value), and a 50% transparency texture mask is superimposed on the flexible material area to enhance the contrast.

[0046] S203, generating geometric correction parameters according to the surface deformation characteristics, and reconstructing pixel positions of the geometric distortion distribution map to obtain reconstructed image data; Specifically, the thin plate spline (TPS) algorithm is first used to establish a global displacement field based on the gridded deformation data, and a continuous deformation map is generated with the boundary feature points of the distorted area as the control points. The deformation amplitude is constrained by introducing a regularization term. The deformation energy weight λ=0.1 is set in the cultural relics protection scene to balance the feature retention and correction strength. The λ value is dynamically adjusted to 0.05 in the mixed material area. Then, the reverse texture mapping strategy is implemented. The corresponding position of the input image is inversely calculated for each pixel coordinate of the output image through the deformation field. The target intensity is calculated by weighting 16 neighborhood pixels in combination with the cubic convolution interpolation algorithm to ensure sub-pixel reconstruction accuracy. In order to maintain boundary consistency, the geometric distortion contour is converted into a stretched contour. The Grangian constraint condition forces the boundary pixel displacement deviation to not exceed 5%, and the geodesic constraint is used to maintain the directionality of linear features. At the same time, a four-layer image pyramid is constructed to achieve multi-resolution optimization. The bottom layer captures the macro deformation trend, and the top layer refines the micro texture. Finally, the fidelity is evaluated by the structural similarity index (SSIM), requiring SSIM ≥ 0.95 for flexible materials and SSIM ≥ 0.85 for rigid materials, or local phase congruence (LPC) is used to verify texture continuity, ensuring that key morphological features (such as inscriptions and mineral patterns) remain clearly distinguishable in both millimeter-level macro correction and micron-level micro restoration.

[0047] Optionally, generating geometric correction parameters according to the surface deformation features and reconstructing pixel positions of the geometric distortion distribution map to obtain reconstructed image data further includes: Extracting edge feature points and texture key points in the target image to generate an initial geometric deformation mapping relationship; Constructing a physical deformation model based on the material properties of the target image, wherein the physical deformation model includes a material elastic modulus, an environmental humidity parameter, and a force deformation curve; Iteratively adjusting the initial geometric deformation mapping relationship according to the physical deformation model to generate the geometric correction parameters; The pixels of the geometric distortion distribution map are reconstructed according to an interpolation algorithm and the geometric correction parameters to obtain reconstructed image data.

[0048] Specifically, the Canny edge detection algorithm is first used to extract edge feature points of the target image. Significant edges are screened through double thresholding, and non-maximum suppression is combined to ensure single-pixel edge width. Texture keypoints are then detected using the SIFT algorithm. Extreme points are located within a 16×16 pixel region using a Difference of Gaussian pyramid, and a 128-dimensional feature descriptor is calculated to enhance rotation and scale invariance. Based on the detected feature points, a triangulation algorithm is used to construct an initial geometric deformation mapping relationship. The feature point correspondences between the source and target images are mapped into a triangular mesh, and the local deformation of each triangle is described using affine transformation parameters. To ensure mapping continuity, a thin plate spline energy function is introduced to globally optimize the triangular mesh, balancing local deformation with global smoothness. Ultimately, an initial geometric deformation mapping field from the original shape to the target image is generated.

[0049] Furthermore, the system queries a material database based on the material type of the target image to obtain mechanical parameters such as elastic modulus and Poisson's ratio. For example, the elastic modulus of rice paper is typically in the range of 0.5-2 GPa, while that of ceramics can reach 100-300 GPa. An environment-material interaction model is then constructed, incorporating ambient humidity and temperature. The diffusion equation describes the effect of moisture migration on material deformation, and the stress-strain relationship is established in conjunction with a viscoelastic model. Furthermore, a library of force-deformation curves is established, using experimental data to fit the stress-strain responses of different materials under uniaxial tension and bending loads. For example, the yield stress of rice paper at 50% humidity is approximately 2 MPa. Ultimately, a physical deformation model is generated that incorporates material properties, environmental parameters, and mechanical responses.

[0050] Furthermore, the initial geometric deformation mapping relationship is used as an initial guess, combined with a physical deformation model to calculate the theoretical deformation variables. For example, finite element analysis is used to solve the thin plate bending equation. The residual between the theoretical deformation variables and the initial mapped deformation variables is then calculated, and the Levenberg-Marquardt algorithm is used for iterative optimization to update the deformation parameters. In each iteration, the optimization weights are dynamically adjusted based on the material properties, increasing the weight of the bending stiffness term for flexible materials and the tensile stiffness term for rigid materials. The iteration is terminated when the residual decreases by less than a preset threshold or the maximum number of iterations is reached. Ultimately, geometric correction parameters that meet physical laws are generated.

[0051] Furthermore, the geometric correction parameters are converted into a pixel displacement field, and a bilinear interpolation algorithm is used to calculate the target position of each pixel. To avoid interpolation artifacts, cubic convolution is used for areas where the deformation gradient exceeds a threshold, and the target intensity is calculated using the weighted values ​​of the surrounding 16 pixels. A hybrid strategy of forward and reverse mapping is then implemented, with reverse mapping used to ensure pixel coverage in densely deformed areas and forward mapping used to improve computational efficiency in flat areas. Finally, the corrected image data is generated, and the reconstruction quality is evaluated using a structural similarity index. Key morphological features are verified to ensure feature geometric accuracy errors of less than 2μm.

[0052] S204 : Mapping the spectral data to a preset color space, matching it with a pre-established color database, and adjusting the color components of the reconstructed image data to obtain corrected image data.

[0053] Specifically, the original spectral data is first subjected to wavelet threshold denoising and normalization processing to eliminate noise and unify the dimension. Then, the target color space is selected according to the application scenario, such as the CIE 1976 L commonly used in the field of cultural relics protection. a b* space, the spectral data is converted into tristimulus values ​​XYZ through the International Commission on Illumination standard observer function, and then linearly transformed to the target color space; then, using a pre-established color database containing standard color card data and material-specific color data, an improved K-nearest neighbor algorithm is used to search for nearest neighbor samples in the color space, and a material weight coefficient is introduced to adapt to different material characteristics; color correction parameters are calculated based on the matching results, and a piecewise linear transformation is implemented on the RGB channels. A soft correction is used for low-saturation areas, an enhanced correction is used for high-saturation areas, and a guided filtering algorithm is used to maintain edge smoothness. Local histogram matching is performed on key color areas; finally, the fidelity is evaluated by calculating the average color difference between the corrected image and the standard color card to ensure color authenticity and visual appeal. If the color difference exceeds the standard, the conversion parameters are optimized through a simulated annealing algorithm until the accuracy requirements are met, thus completing the complete process from spectral data to color correction.

[0054] Optionally, mapping the spectral data to a preset color space, matching a pre-established color database, and adjusting the color components of the reconstructed image data to obtain corrected image data further includes: Converting the color representation of the spectral reflectance data into a preset color space to extract the chromaticity component of the target image; Obtaining the material type and lighting conditions of the target image; Retrieving a standard color sample that matches the material type and lighting conditions from a pre-established color database, and obtaining a corresponding standard chromaticity component; Comparing the chromaticity component of the target image with the standard chromaticity component, and generating a color adjustment parameter according to the comparison result; Based on the color adjustment parameter, the chroma component of the target image is corrected so that the difference between the corrected chroma component and the standard chroma component is less than a preset threshold; The corrected color components are output to form corrected image data.

[0055] Specifically, the raw spectral reflectance data is first preprocessed, using a moving average filter to remove high-frequency noise and retain key color features. The spectral data is then converted to a preset color space (such as the CIE 1976 Lab space) based on the International Commission on Illumination (CIE) standard colorimetry system. The tristimulus values ​​(X, Y, and Z) are calculated using a standard illuminant (such as D65) and a standard observer function (such as a 2° field of view), and then mapped to the chromaticity coordinates (L, a, and b) of the target color space. L represents lightness, and a and b* represent the red, green, and yellow, blue chromaticity components, respectively. This completes the conversion from the physical spectrum to the color perceived by the human eye, providing a benchmark for subsequent color correction.

[0056] Furthermore, image classification algorithms (such as a material recognition model based on a convolutional neural network) are first used to analyze the target image content. Combining texture features (such as the fiber structure of rice paper or the reflective surface of ceramic glaze) with color distribution, the material type (such as rice paper, mineral pigment, silk, etc.) is automatically determined. Simultaneously, image metadata is analyzed or ambient light sensors are used to obtain lighting condition parameters, including light source type (such as LED, natural light), color temperature (such as 5000K), and light intensity (such as 500 lux). If metadata is missing, color temperature estimation algorithms (such as automatic white balance based on the grayscale world assumption) are used to infer lighting conditions, providing key screening criteria for color database retrieval.

[0057] Furthermore, the pre-established color database contains standardized color card data (such as X-Rite ColorChecker24 color card) and material-specific color data. Each sample records the CIE L color of a specific material (such as rice paper) under a specific light (such as D65 light source). a b* value and spectral reflectance curve. During the search, the sub-database (e.g., the cultural relic material library) is first filtered based on the target image's material type. Lighting condition parameters (e.g., a color temperature of 5000K) are then used to further narrow the scope, ultimately selecting the standard sample that most closely matches the target image's shooting environment. If multiple candidate samples exist, the material's aging level (e.g., unaged / mildly aged / heavily aged rice paper) is prioritized to ensure physical consistency of color correction.

[0058] Furthermore, the chromaticity components (L, a, b) of the target image are compared with those of the retrieved standard sample. The color difference is calculated (e.g., using the CIEDE2000 color difference formula) and the sources of color difference are decomposed (e.g., lightness deviation, hue shift, and saturation difference). Adjustment priority is set based on the material type: for flexible materials like rice paper, lightness (L) is prioritized to maintain layering; for rigid materials like mineral pigments, hue (a, b) is prioritized to restore true color. When generating adjustment parameters, a piecewise linear mapping strategy is used, limiting the adjustment range (±5%) for low-saturation areas (e.g., rice paper base) and allowing larger adjustments (±15%) for high-saturation areas (e.g., mineral pigments) to avoid color distortion.

[0059] Furthermore, based on the generated adjustment parameters, a nonlinear transformation is applied to the target image's RGB channels. Gamma correction is applied to the lightness channel (L) (the γ value is dynamically adjusted based on the material's reflectance curve, set at 2.2 for rice paper and 1.8 for ceramic). For the chromaticity channels (a and b*), a directional adjustment is performed using the Hue-Saturation-Value (HSV) model, approximating standard colors by rotating the hue circle (angular deviation ≤ ±5°) and stretching the saturation axis (scale factor 0.8-1.2). Mask-guided local adjustments are applied to mixed-material areas (such as the interface between mounting layers) to ensure color corrections for different materials do not interfere with each other while maintaining overall visual consistency.

[0060] Furthermore, the adjusted CIE L a The b* chromaticity component is converted back to a device-dependent color space (such as sRGB), and gamma decoding (following the IEC 61966-2-1 standard) is used to restore linear RGB values. Color gamut mapping (such as perceptual quantization) is then performed to ensure accurate color reproduction on the display device. Finally, the corrected image data is generated and verified through both visual inspection and instrumental measurement. Visual inspection focuses on the naturalness and legibility of key color areas (such as inscriptions and mineral patterns). Instrumental measurement calculates the average color difference (ΔE) between the corrected image and a standard color chart, requiring ΔE ≤ 2.0 for cultural relics preservation and ΔE ≤ 1.5 for commercial printing, ensuring that the color correction meets the requirements of the application scenario.

[0061] Optionally, the method further includes: Decomposing the corrected image data into a data set comprising at least two levels; Encoding the first level data in the data set using a transformation-based compression algorithm and encoding the second level data using a sparse feature-based compression algorithm to form a layered compressed data stream; generating a unique identifier based on the layered compressed data stream, wherein the unique identifier includes a data feature summary, a timestamp, and processing device information; The unique identifier is bound to the layered compressed data stream and stored in a storage network.

[0062] Specifically, the rectified image is first subjected to a wavelet transform (e.g., using the Daubechies 9 / 7 wavelet basis) to decompose the image into a low-frequency approximation component and a high-frequency detail component. The low-frequency component contains information about the overall brightness and main structure of the image, serving as the first-level data. The high-frequency component, containing detailed features such as texture and edges, is further subdivided into horizontal, vertical, and diagonal sub-bands based on spatial frequency, serving as the second-level data. Based on the characteristics of cultural relic images, key areas (such as inscriptions and mineral patterns) can be marked, and the image blocks containing them are elevated to independent levels, forming a multi-level structure of "global layer + detail layer + key area layer" to ensure that important information is prioritized during compression and transmission.

[0063] Furthermore, the first-level data, such as low-frequency approximation components, is compressed using discrete cosine transform (DCT) or modified wavelet coding algorithms (such as the JPEG 2000 standard). Transform coefficients are non-uniformly quantized using a quantization matrix, employing fine-grained quantization (quantization step size ≤ 2) for low-frequency coefficients and coarse-grained quantization (quantization step size ≥ 8) for high-frequency coefficients. This approach removes redundancy while preserving important visual information. Entropy coding (such as arithmetic coding) is then implemented, leveraging symbol probability statistics to further compress the data, ensuring a manageable compression ratio (typically ≤ 20:1) for the first-level data and preventing loss of critical structural information. Second-level data, such as high-frequency detail components, is encoded using a compressed sensing framework. First, a measurement matrix (such as a Gaussian random matrix) is designed to project the original signal into a low-dimensional space, resulting in a measurement value count of only 10%-25% of the original data. These measurements are then sparsely represented using a sparse basis (such as a discrete wavelet transform basis), and the original signal is reconstructed using the orthogonal matching pursuit (OMP) algorithm. Based on the texture characteristics of cultural relic images, an overcomplete dictionary (such as the K-SVD algorithm) can be trained to improve the sparsity of the sparse representation (sparseness ≥ 0.8), thereby achieving high compression (compression ratio can reach over 50:1) while ensuring the quality of detail restoration.

[0064] Furthermore, a data feature summary is extracted and a secure hash algorithm (SHA-256) is applied to the compressed data at each level to generate a fixed-length digest sequence. The processing timestamp (accurate to the millisecond level) and processing device information (such as device ID and operating system version) are then recorded. These elements are concatenated in the format of "digest + timestamp + device ID" and encoded using Base64 to generate a unique identifier string. To enhance tamper resistance, a digital signature technology (such as RSA-2048) can be used to sign the unique identifier and ensure reliable data traceability.

[0065] Furthermore, the generated unique identifier is first embedded as metadata into the header field of the compressed data stream, and the data packet is encapsulated using Extensible Markup Language (XML) to form a standardized data unit containing data fingerprint, hierarchical encoding content and associated metadata; then the storage strategy is dynamically selected according to the importance of the data, and a distributed storage system (such as the Ceph file system) is used for key cultural relic data (such as high-precision scanning data of unique cultural relics), the number of replicas is set to no less than 3 and cross-rack storage is implemented, and the WORM (write once read many) mode is enabled to ensure that the data cannot be tampered with. For ordinary scientific research data (such as intermediate results of experimental image processing), an erasure code storage scheme (such as Reed-Solomon (10,4) code) is used, and the data is sharded into 10 data blocks and 4 check blocks to tolerate up to 4 node failures, reducing storage overhead while ensuring availability; finally, the encapsulated data packet is written to the distributed storage pool through the storage network interface (such as iSCSI or NFS protocol), completing the full process closed loop from data processing to persistent storage, ensuring data security, traceability and efficient access.

[0066] This embodiment also discloses a system for digitizing images. Figure 3 This is a module diagram of an image digitization system disclosed in an embodiment of the present application, such as Figure 3 As shown, the system includes: The data acquisition module 301 is used to obtain the spectral data, three-dimensional shape data and material reflectance parameters of the target image; A feature extraction module 302 is configured to analyze the surface deformation features of the target image based on the three-dimensional shape data and generate a geometric distortion distribution map; A geometric correction module 303 is configured to generate geometric correction parameters according to the surface deformation characteristics, reconstruct pixel positions of the geometric distortion distribution map, and obtain reconstructed image data; The spectral color correction module 304 is configured to map the spectral data to a preset color space, match the spectral data to a pre-established color database, and adjust the color components of the reconstructed image data to obtain corrected image data.

[0067] Optionally, the data acquisition module 301 is specifically configured to: Scanning the target image to obtain the spectral data of each band and the three-dimensional coordinate data of the surface, and constructing the three-dimensional topography data of the target image according to the three-dimensional coordinate data; The target image is illuminated by a light source, the incident angle and polarization direction of the light source are adjusted, and the spectral distribution of the reflected light is detected by a spectrometer to obtain the material reflectivity parameters of the material surface.

[0068] Optionally, the feature extraction module 302 is specifically configured to: Dividing the surface of the target image into a preset number of grid units according to a preset size; Comparing the three-dimensional surface topography data of the grid unit with the corresponding preset plane model to obtain the deformation offset of the grid unit; Marking the grid cells whose deformation offset exceeds a preset threshold as geometric distortion areas; Boundary features of all the geometric distortion areas are extracted to generate the geometric distortion distribution map.

[0069] Optionally, the geometric correction module 303 is specifically configured to: Extracting edge feature points and texture key points in the target image to generate an initial geometric deformation mapping relationship; Constructing a physical deformation model based on the material properties of the target image, wherein the physical deformation model includes a material elastic modulus, an environmental humidity parameter, and a force deformation curve; Iteratively adjusting the initial geometric deformation mapping relationship according to the physical deformation model to generate the geometric correction parameters; The pixels of the geometric distortion distribution map are reconstructed according to an interpolation algorithm and the geometric correction parameters to obtain reconstructed image data.

[0070] Optionally, the spectral color correction module 304 is specifically configured to: Converting the color representation of the spectral reflectance data into a preset color space to extract the chromaticity component of the target image; Obtaining the material type and lighting conditions of the target image; Retrieving a standard color sample that matches the material type and lighting conditions from a pre-established color database, and obtaining a corresponding standard chromaticity component; Comparing the chromaticity component of the target image with the standard chromaticity component, and generating a color adjustment parameter according to the comparison result; Based on the color adjustment parameter, the chroma component of the target image is corrected so that the difference between the corrected chroma component and the standard chroma component is less than a preset threshold; The corrected color components are output to form corrected image data.

[0071] Optionally, the spectral color correction module 304 is specifically configured to: Calculating a color difference vector between the chroma component of the target image and the standard chroma component, wherein the color difference vector includes a brightness difference and a chroma difference; Based on the color difference vector, a color adjustment matrix is ​​generated by a preset color correction model, wherein the color correction model includes linear transformation coefficients and nonlinear compensation parameters; The weight of the color adjustment matrix is ​​adjusted according to the material reflectance parameter of the target image to generate a color adjustment parameter.

[0072] Optionally, the system further includes a data compression and storage module 305, specifically configured to: Decomposing the corrected image data into a data set comprising at least two levels; Encoding the first level data in the data set using a transformation-based compression algorithm and encoding the second level data using a sparse feature-based compression algorithm to form a layered compressed data stream; generating a unique identifier based on the layered compressed data stream, wherein the unique identifier includes a data feature summary, a timestamp, and processing device information; The unique identifier is bound to the layered compressed data stream and stored in a storage network.

[0073] It should be noted that the above embodiments provide devices that implement their functions using only the division of the above functional modules as examples. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the device and method embodiments provided in the above embodiments are based on the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0074] This embodiment also discloses an electronic device, referring to Figure 4 The electronic device may include: at least one processor 401 , at least one communication bus 402 , a user interface 403 , a network interface 404 , and at least one memory 405 .

[0075] The communication bus 402 is used to implement the connection and communication between these components.

[0076] The user interface 403 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 403 may also include a standard wired interface and a wireless interface.

[0077] The network interface 404 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0078] Processor 401 may include one or more processing cores. Using various interfaces and circuits, processor 401 connects to various components within the server. It executes instructions, programs, code sets, or instruction sets stored in memory 405, as well as accesses data stored in memory 405, to perform various server functions and process data. Optionally, processor 401 may be implemented using at least one of the following hardware forms: a digital signal processing (DSP), a field-programmable gate array (FPGA), or a programmable logic array (PLA). Processor 401 may integrate one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. The CPU primarily processes the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing content displayed on the display screen; and the modem handles wireless communications. It is understood that the modem may also be implemented independently of the processor 401 and implemented as a separate chip.

[0079] Among them, the memory 405 may include a random access memory (RAM) or a read-only memory (Read-Only Memory). Optionally, the memory 405 includes a non-transitory computer-readable storage medium. The memory 405 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 405 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 405 may also be optionally at least one storage device located away from the aforementioned processor 401. As Figure 4 As shown, the memory 405 as a computer storage medium may include an operating system, a network communication module, a user interface module, and an image digitization application.

[0080] exist Figure 4In the electronic device shown, the user interface 403 is mainly used to provide an input interface for the user and obtain data input by the user; and the processor 401 can be used to call an image digitization application stored in the memory 405. When executed by one or more processors 401, the electronic device executes one or more methods in the above embodiments.

[0081] It should be noted that for the aforementioned method embodiments, for simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required for this application.

[0082] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0083] In the several embodiments provided in this application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are merely schematic, such as the division of units, which is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interface, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0084] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0085] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0086] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of this application, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory 405 and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the method of this application. The aforementioned memory 405 includes various media that can store program code, such as a USB flash drive, a mobile hard drive, a magnetic disk, or an optical disk.

[0087] The above is only an exemplary embodiment of the present disclosure and cannot be used to limit the scope of the present disclosure. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the disclosure of the specification, those skilled in the art will easily think of other embodiments of the present disclosure. This application is intended to cover any variations, uses or adaptive changes of the present disclosure, which follow the general principles of the present disclosure and include common knowledge or customary technical means in the technical field that are not recorded in the present disclosure. The description and examples are to be regarded as exemplary only, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A method for digitizing an image, characterized in that: Applied to a server, the method includes: Obtain spectral data, three-dimensional topography data and material reflectance parameters of the target image; Analyzing the surface deformation characteristics of the target image based on the three-dimensional topography data to generate a geometric distortion distribution map; generating geometric correction parameters according to the surface deformation characteristics, and reconstructing pixel positions of the geometric distortion distribution map to obtain reconstructed image data; The spectral data is mapped to a preset color space, matched with a pre-established color database, and the color components of the reconstructed image data are adjusted to obtain corrected image data.

2. The method according to claim 1, characterized in that The spectral data, three-dimensional shape data and material reflectivity parameters of the target image are obtained as follows: Scanning the target image to obtain the spectral data of each band and the three-dimensional coordinate data of the surface, and constructing the three-dimensional topography data of the target image according to the three-dimensional coordinate data; The target image is illuminated by a light source, the incident angle and polarization direction of the light source are adjusted, and the spectral distribution of the reflected light is detected by a spectrometer to obtain the material reflectivity parameters of the material surface.

3. The method according to claim 1, characterized in that The step of analyzing the surface deformation characteristics of the target image based on the three-dimensional topography data to generate a geometric distortion distribution map includes: Dividing the surface of the target image into a preset number of grid units according to a preset size; Comparing the three-dimensional surface topography data of the grid unit with the corresponding preset plane model to obtain the deformation offset of the grid unit; Marking the grid cells whose deformation offset exceeds a preset threshold as geometric distortion areas; Boundary features of all the geometric distortion areas are extracted to generate the geometric distortion distribution map.

4. The method according to claim 1, wherein Generating geometric correction parameters according to the surface deformation features and reconstructing pixel positions of the geometric distortion distribution map to obtain reconstructed image data further includes: Extracting edge feature points and texture key points in the target image to generate an initial geometric deformation mapping relationship; Constructing a physical deformation model based on the material properties of the target image, wherein the physical deformation model includes a material elastic modulus, an environmental humidity parameter, and a force deformation curve; Iteratively adjusting the initial geometric deformation mapping relationship according to the physical deformation model to generate the geometric correction parameters; The pixels of the geometric distortion distribution map are reconstructed according to an interpolation algorithm and the geometric correction parameters to obtain reconstructed image data.

5. The method according to claim 1, wherein Mapping the spectral data to a preset color space, matching a pre-established color database, and adjusting the color components of the reconstructed image data to obtain corrected image data further includes: Converting the color representation of the spectral reflectance data into a preset color space to extract the chromaticity component of the target image; Obtaining the material type and lighting conditions of the target image; Retrieving a standard color sample that matches the material type and lighting conditions from a pre-established color database, and obtaining a corresponding standard chromaticity component; Comparing the chromaticity component of the target image with the standard chromaticity component, and generating a color adjustment parameter according to the comparison result; Based on the color adjustment parameter, the chroma component of the target image is corrected so that the difference between the corrected chroma component and the standard chroma component is less than a preset threshold; The corrected color components are output to form corrected image data.

6. The method according to claim 5, characterized in that The step of comparing the chromaticity component of the target image with the standard chromaticity component and generating the color adjustment parameter according to the comparison result further comprises: Calculating a color difference vector between the chroma component of the target image and the standard chroma component, wherein the color difference vector includes a brightness difference and a chroma difference; Based on the color difference vector, a color adjustment matrix is ​​generated by a preset color correction model, wherein the color correction model includes linear transformation coefficients and nonlinear compensation parameters; The weight of the color adjustment matrix is ​​adjusted according to the material reflectance parameter of the target image to generate a color adjustment parameter.

7. The method according to claim 1, characterized in that The method further comprises: Decomposing the corrected image data into a data set comprising at least two levels; Encoding the first level data in the data set using a transformation-based compression algorithm and encoding the second level data using a sparse feature-based compression algorithm to form a layered compressed data stream; generating a unique identifier based on the layered compressed data stream, wherein the unique identifier includes a data feature summary, a timestamp, and processing device information; The unique identifier is bound to the layered compressed data stream and stored in a storage network.

8. A system for digitizing an image, characterized in that: Specifically include: Data acquisition module, used to obtain spectral data, three-dimensional shape data and material reflectance parameters of the target image; a feature extraction module, configured to analyze the surface deformation features of the target image based on the three-dimensional shape data and generate a geometric distortion distribution map; a geometric correction module, configured to generate geometric correction parameters according to the surface deformation characteristics, reconstruct pixel positions of the geometric distortion distribution map, and obtain reconstructed image data; The spectral color correction module is used to map the spectral data to a preset color space, match it with a pre-established color database, and adjust the color components of the reconstructed image data to obtain corrected image data.

9. An electronic device, characterized in that: The electronic device comprises a processor, a memory, a user interface and a network interface, wherein the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device executes the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores instructions, and when the instructions are executed, the method according to any one of claims 1 to 7 is executed.

Images