Super-depth-of-field microscopic scanning and AI image analysis method and system for ancient painting and calligraphy repair

Abstract

The invention provides an ultra-depth-of-field microscopic scanning and AI image analysis method and system for ancient painting and calligraphy repair, and relates to the field of image processing.The method comprises the steps that a scanning track and scanning parameters are determined according to scanning related feature information of a to-be-repaired ancient painting and calligraphy, the scanning track comprises a plurality of scanning track points, and the scanning track points are used for scanning the to-be-repaired ancient painting and calligraphy; the scanning parameters comprise an initial scanning height corresponding to each scanning track point; according to the scanning track and the initial scanning height corresponding to each scanning track point, adjusting the spatial position of the super-depth-of-field three-dimensional microscope, and obtaining a multi-layer image at each scanning track point through the super-depth-of-field three-dimensional microscope; stacking the multi-layer image of each scanning track point to generate a region image corresponding to each scanning track point, and splicing the region images corresponding to each scanning track point to generate a scanning image of the ancient painting and calligraphy to be repaired; and determining the damage characteristics of the to-be-repaired ancient painting and calligraphy through the damage analysis model. The method has the advantage of improving the efficiency and quality of repairing the ancient painting and calligraphy.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and in particular to a method and system for ultra-depth-of-field microscopy scanning and AI image analysis for the restoration of ancient calligraphy and painting. Background Art

[0002] In the existing restoration process of ancient calligraphy and paintings, microscopes are indispensable auxiliary tools. Traditional optical microscopes and super-depth-of-field microscopes are often used in combination to meet different observation needs. Optical microscopes are usually equipped with multiple objective lenses, with magnifications ranging from 4x to 100x. When using, you need to first locate the target area and adjust the clarity through a low-power microscope, and then gradually switch to a high-power microscope for detailed observation. However, the field of view of a high-power microscope is small, and it is difficult to directly locate the target area. In addition, the working distance is short, and improper operation can easily damage the sample or lens. In recent years, the application of super-depth-of-field microscopes in the restoration of ancient calligraphy and paintings has gradually increased. Compared with traditional microscopes, super-depth-of-field microscopes can achieve free zoom through a single objective lens, which greatly simplifies the operation process. At the same time, the super-depth-of-field microscope combines CCD elements for digital image presentation, which not only has higher resolution, but also can show microscopic details more clearly, which is of great significance for analyzing microscopic damage to ancient calligraphy and paintings.

[0003] In the existing technology, ultra-depth of field microscopy has extremely high requirements for the flatness of the scanning surface, and the microscopic unevenness on the surface of ancient calligraphy and paintings makes focusing and focusing during the scanning process extremely difficult, thus affecting the scanning quality and efficiency. In addition, the fixed scanning platform size specifications of existing ultra-depth of field equipment are fixed, which makes it difficult to adapt to the format requirements of large calligraphy and paintings. For example, the scanning platform of existing ultra-depth of field microscopes usually only supports samples within A3 size (297mm×420mm), while ancient calligraphy and paintings can reach several meters or even ten meters. As a result, a single scan can only cover a local area, and complete digitization must be achieved through multiple stitching. However, the stitching process is prone to introduce problems such as geometric distortion and color distortion. Although handheld depth of field meters are portable, their working mode is essentially "point sampling", which can only magnify and observe local areas (such as a single scanning range of about 10mm×10mm). It is impossible to achieve full-format depth of field information collection, resulting in fragmented digitization results, which makes it difficult to meet the requirements of global depth of field consistency and color continuity for high-definition digitization of ancient calligraphy and paintings. Furthermore, a large-scale AI image recognition system suitable for identifying damage characteristics in ancient calligraphy and paintings has yet to be developed. Existing AI image recognition systems are mostly trained and identified for general objects or scenes, lacking specialized training and identification capabilities for damage to ancient calligraphy and paintings. Therefore, they are unable to meet the actual needs of the ancient calligraphy and painting restoration field.

[0004] Therefore, it is necessary to provide an ultra-depth-of-field microscopic scanning and AI image analysis method and system for the restoration of ancient calligraphy and paintings, so as to improve the efficiency and quality of the restoration of ancient calligraphy and paintings. Summary of the Invention

[0005] The present invention provides an ultra-depth-of-field microscopic scanning and AI image analysis system for the restoration of ancient calligraphy and painting, comprising: an information acquisition module for acquiring scanning-related feature information of the ancient calligraphy and painting to be restored; an automatic scanning module comprising a scanning device, an ultra-depth-of-field three-dimensional microscope, and a scanning control unit, wherein the ultra-depth-of-field three-dimensional microscope is mounted on the scanning device, and the scanning control unit is configured to determine a scanning trajectory and scanning parameters based on the scanning-related feature information of the ancient calligraphy and painting to be restored, wherein the scanning trajectory includes a plurality of scanning trajectory points, and the scanning parameters include a starting scanning height corresponding to each scanning trajectory point, and the scanning device is configured to adjust the spatial position of the ultra-depth-of-field three-dimensional microscope based on the scanning trajectory and the starting scanning height corresponding to each scanning trajectory point, and the ultra-depth-of-field three-dimensional microscope is configured to acquire a multi-layer image at each scanning trajectory point; an image processing module is configured to stack the multi-layer images of each scanning trajectory point to generate a regional image corresponding to each scanning trajectory point, and to splice the regional images corresponding to each scanning trajectory point to generate a scanned image of the ancient calligraphy and painting to be restored; and a damage analysis module is configured to determine damage features of the ancient calligraphy and painting to be restored based on the scanned image of the ancient calligraphy and painting to be restored using a damage analysis model.

[0006] Furthermore, the scanning device includes a frame, a vacuum suction cup honeycomb panel, a position adjustment component and a laser ranging component, wherein the vacuum suction cup honeycomb panel and the position adjustment component are both arranged on the frame, the position adjustment component is arranged above the vacuum suction cup honeycomb panel, the ancient calligraphy and painting to be restored is placed on the vacuum suction cup honeycomb panel, the position adjustment component includes an X-axis coordinate adjustment device, a Y-axis coordinate adjustment device and a Z-axis coordinate adjustment device, the Y-axis coordinate adjustment device is arranged on the X-axis coordinate adjustment device, the Z-axis coordinate adjustment device is arranged on the Y-axis coordinate adjustment device, and the super-depth of field three-dimensional microscope is installed on the Z-axis coordinate adjustment device; the laser ranging component is arranged on the Z-axis coordinate adjustment device, and the laser ranging component includes a laser ranging dot matrix composed of multiple laser ranging devices.

[0007] Furthermore, the scanning device is also used to adjust the position of the laser ranging component in the XY plane according to a preset ranging trajectory, wherein the XY plane is parallel to the vacuum suction cup honeycomb panel, the preset ranging trajectory includes multiple ranging trajectory points, and the laser ranging component is used to obtain regional ranging information at each ranging trajectory point, and the scanning-related feature information of the ancient calligraphy and painting to be restored includes at least the regional ranging information corresponding to each ranging trajectory point; the scanning control unit determines the scanning trajectory according to the scanning-related feature information of the ancient calligraphy and painting to be restored, including: determining the scanning trajectory according to the regional ranging information corresponding to each ranging trajectory point.

[0008] Furthermore, the scanning control unit determines the scanning trajectory according to the regional ranging information corresponding to each ranging trajectory point, including: generating the concave and convex heights of multiple positions of the ancient calligraphy and painting to be restored according to the regional ranging information corresponding to each ranging trajectory point; establishing a trajectory generation constraint condition set, wherein the trajectory generation constraint condition set includes at least the shortest distance and the longest distance between two adjacent scanning trajectory points; establishing a multi-objective reward function, wherein the multi-objective reward function is related to the concave and convex height difference between two adjacent scanning trajectory points, the concave and convex height of each scanning trajectory point and the distance between two adjacent scanning trajectory points; and generating a scanning trajectory through a trajectory generation model according to the concave and convex heights of multiple positions of the ancient calligraphy and painting to be restored, the trajectory generation constraint condition set and the multi-objective reward function.

[0009] Furthermore, the scanning control unit generates a scanning trajectory through a trajectory generation model according to the concave-convex heights of multiple positions of the ancient calligraphy and painting to be restored, a trajectory generation constraint set and a multi-objective reward function, including: for each position, determining the concave-convex height difference value of the position according to the concave-convex heights of adjacent positions; normalizing the concave-convex height difference value of each position to generate a normalized concave-convex height difference value for each position; determining multiple trajectory generation starting positions from multiple positions according to the normalized concave-convex height difference value of each position; for each trajectory generation starting position, generating a scanning trajectory corresponding to the trajectory generation starting position through the trajectory generation model according to the concave-convex heights of multiple positions of the ancient calligraphy and painting to be restored, a trajectory generation constraint set and a multi-objective reward function; establishing a fitness function, wherein the fitness function is related to the concave-convex height difference value between two adjacent scanning trajectory points and the scanning overlap area between any two scanning trajectory points; determining the fitness value of the scanning trajectory corresponding to each trajectory generation starting position; and generating a scanning trajectory through a genetic algorithm based on the fitness value of the scanning trajectory corresponding to each trajectory generation starting position.

[0010] Furthermore, the scanning control unit determines the starting scanning height corresponding to the scanning trajectory point based on the scanning-related feature information of the ancient calligraphy and painting to be restored, including: interpolating the concave and convex heights of multiple positions of the ancient calligraphy and painting to be restored to generate the concave and convex heights corresponding to the scanning trajectory point; calculating the height difference between the concave and convex height corresponding to the scanning trajectory point and the reference height; and determining the starting scanning height corresponding to the scanning trajectory point based on the height difference between the concave and convex height corresponding to the scanning trajectory point and the reference height.

[0011] Furthermore, the scanning-related characteristic information of the ancient calligraphy and painting to be restored also includes material information, pigment type information and color distribution characteristics; the scanning device also includes a fill light device; the scanning control unit determines the scanning parameters based on the scanning-related characteristic information of the ancient calligraphy and painting to be restored, including: determining the fill light parameters based on the material information, pigment type information and color distribution characteristics, wherein the scanning parameters include fill light parameters, and the fill light parameters include color temperature and light intensity.

[0012] Furthermore, the scanning control unit determines the fill light parameters according to the material information, pigment type information and color distribution characteristics, including: obtaining the material information, pigment type information, color distribution characteristics and optimal fill light parameters of multiple sample ancient calligraphy and paintings; dividing the multiple sample ancient calligraphy and paintings into multiple sample groups according to the material information, pigment type information and color distribution characteristics of the multiple sample ancient calligraphy and paintings through split hierarchical clustering; determining the sample group to which the ancient calligraphy and painting to be restored belongs according to the material information, pigment type information and color distribution characteristics of the ancient calligraphy and painting to be restored; determining similar sample ancient calligraphy and paintings to be restored according to the material information, pigment type information and color distribution characteristics of the ancient calligraphy and painting to be restored and the material information, pigment type information and color distribution characteristics of the sample ancient calligraphy and paintings included in the sample group to which the ancient calligraphy and painting to be restored belongs; and determining the fill light parameters based on the optimal fill light parameters of the similar sample ancient calligraphy and paintings to be restored.

[0013] Furthermore, the damage analysis model includes a defect analysis unit, a fracture analysis unit, a discoloration analysis unit and an insect pest analysis unit, wherein the defect analysis unit is used to generate a defect distribution map based on the scanned image of the ancient calligraphy and painting to be repaired, the fracture analysis unit is used to generate a fracture distribution map based on the scanned image of the ancient calligraphy and painting to be repaired, the color change analysis unit is used to generate a color change distribution map based on the scanned image of the ancient calligraphy and painting to be repaired, and the insect pest analysis unit is used to generate an insect pest distribution map based on the scanned image of the ancient calligraphy and painting to be repaired.

[0014] The present invention provides a super-depth-of-field microscopic scanning and AI image analysis method for the restoration of ancient calligraphy and painting, which is applied to the above-mentioned super-depth-of-field microscopic scanning and AI image analysis system for the restoration of ancient calligraphy and painting, and is characterized in that it includes: obtaining scanning-related feature information of the ancient calligraphy and painting to be restored; determining a scanning trajectory and scanning parameters based on the scanning-related feature information of the ancient calligraphy and painting to be restored, wherein the scanning trajectory includes multiple scanning trajectory points, and the scanning parameters include a starting scanning height corresponding to each scanning trajectory point; adjusting the spatial position of a super-depth-of-field three-dimensional microscope based on the scanning trajectory and the starting scanning height corresponding to each scanning trajectory point, and obtaining a multi-layer image at each scanning trajectory point through the super-depth-of-field three-dimensional microscope; stacking the multi-layer images of each scanning trajectory point to generate a regional image corresponding to each scanning trajectory point, and splicing the regional images corresponding to each scanning trajectory point to generate a scanned image of the ancient calligraphy and painting to be restored; determining the damage characteristics of the ancient calligraphy and painting to be restored based on the scanned image of the ancient calligraphy and painting to be restored through a damage analysis model.

[0015] Compared with the existing technology, the ultra-depth-of-field microscopy scanning and AI image analysis method and system for ancient calligraphy and painting restoration provided by the present invention have at least the following beneficial effects:

[0016] Obtain relevant characteristic information of the scans of the ancient calligraphy and paintings to be restored. This information covers many key factors such as the materials, pigments, color distribution, etc. of the ancient calligraphy and paintings, providing comprehensive and accurate basic data for subsequent scanning, image processing and damage analysis. It helps to grasp the characteristics of ancient calligraphy and paintings as a whole and formulate restoration strategies that are more in line with the actual situation.

[0017] Based on the acquired feature information, the scanning trajectory and scanning parameters are intelligently determined. Based on the characteristics of different ancient calligraphy and paintings, a trajectory consisting of multiple scanning points is planned, and an appropriate starting scanning height is set for each track point. This makes the scanning process more targeted and accurate, avoiding the blindness and unevenness that may exist in traditional scanning methods, and greatly improving scanning quality.

[0018] According to the determined scanning trajectory and starting scanning height, the spatial position of the ultra-depth-of-field three-dimensional microscope is precisely adjusted, realizing automated and intelligent scanning operations, reducing manual intervention, improving scanning efficiency, and ensuring the stability and repeatability of the scanning process.

[0019] The ultra-depth-of-field 3D microscope acquires multiple layers of images at each scanning trajectory point. This multi-layered image acquisition method captures detailed information at varying depths on the surface of ancient calligraphy and paintings, providing a rich data source for subsequent image stacking, helping to more fully reveal the original appearance of the ancient calligraphy and paintings. The multi-layer images at each scanning trajectory point are stacked, and a specific algorithm is used to fuse the multi-layer image information into a single high-quality regional image, effectively reducing image noise, highlighting detailed information, and making the image clearer and more accurate. The regional images at each trajectory point are then spliced ​​together to generate a complete scanned image of the ancient calligraphy and painting to be restored.

[0020] Analyzing scanned images using a damage analysis model can accurately determine the damage characteristics of the ancient calligraphy and painting to be restored, such as the type of damage (e.g., breakage, fading, insect damage), location, and extent of damage. This scientific model-based analysis method avoids the subjectivity and errors inherent in manual judgment and provides restorers with objective and accurate damage information. Accurate damage characteristic information provides a strong basis for restorers to formulate restoration plans, enabling them to select appropriate restoration materials and techniques based on the damage situation, improving the accuracy and success rate of restoration and maximizing the restoration of the original appearance and artistic value of the ancient calligraphy and painting. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] This specification will be further described in the form of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting, and in these embodiments, like numbers represent like structures, wherein:

[0022] Figure 1 This is a module diagram of a super-depth-of-field microscopy scanning and AI image analysis system for ancient calligraphy and painting restoration according to some embodiments of this specification;

[0023] Figure 2 is a schematic structural diagram of a scanning device according to some embodiments of this specification;

[0024] Figure 3 is a schematic diagram of a process for generating a scanning trajectory according to some embodiments of this specification;

[0025] Figure 4 is a schematic structural diagram of a damage analysis model according to some embodiments of this specification;

[0026] Figure 5 is a schematic diagram of a user interface of a scanning control unit according to some embodiments of this specification;

[0027] Figure 6 is a schematic diagram of another user interface of a scanning control unit according to some embodiments of this specification;

[0028] Figure 7 is a schematic diagram of damage features of ancient calligraphy and painting to be restored according to some embodiments of this specification;

[0029] Figure 8-10 is a schematic diagram of a regional image according to some embodiments of this specification;

[0030] Figure 11 It is a flowchart of the ultra-depth-of-field microscopic scanning and AI image analysis method for the restoration of ancient calligraphy and painting according to some embodiments of this specification.

[0031] In the figure, 1. Rack; 2. Vacuum suction cup honeycomb panel; 3. X-axis coordinate adjustment device; 4. Y-axis coordinate adjustment device; 5. Z-axis coordinate adjustment device; 6. Ultra-depth of field three-dimensional microscope. DETAILED DESCRIPTION

[0032] To more clearly illustrate the technical solutions of the embodiments of this specification, the following briefly describes the drawings required for describing the embodiments. Obviously, the drawings described below are merely examples or embodiments of this specification. Those skilled in the art can apply this specification to other similar scenarios based on these drawings without inventive effort. Unless otherwise apparent from the context or otherwise noted, the same reference numerals in the figures represent the same structure or operation.

[0033] Figure 1 Schematic diagram of a module of a super-depth-of-field microscopy scanning and AI image analysis system for ancient calligraphy and painting restoration according to some embodiments of this specification. Figure 1 As shown, the ultra-depth-of-field microscopic scanning and AI image analysis system for the restoration of ancient calligraphy and paintings may include an information acquisition module, an automatic scanning module, an image processing module, and a damage analysis module.

[0034] The information acquisition module is used to obtain the scan-related feature information of the ancient calligraphy and painting to be restored.

[0035] Specifically, the scanning-related feature information of the ancient calligraphy and painting to be restored may include regional ranging information corresponding to each ranging trajectory point.

[0036] The scan-related feature information of the ancient calligraphy and painting to be restored may also include material information, pigment type information and color distribution characteristics.

[0037] Material information refers to the type of material used to support ancient calligraphy and paintings, such as paper (rice paper, leather paper, hemp paper, etc.) or silk (raw silk, cooked silk, etc.). Different materials have varying physical and chemical properties. Techniques such as infrared spectroscopy and Raman spectroscopy can be used to analyze the chemical composition of the material and determine the material information of the ancient calligraphy and paintings to be restored.

[0038] Pigment type information refers to the mineral pigments (such as cinnabar, azurite, and malachite), plant pigments (such as cyanine and gamboge), or chemically synthesized pigments used in ancient calligraphy and painting. This information is determined by analyzing the pigment's chemical composition using techniques such as X-ray fluorescence spectroscopy and laser-induced breakdown spectroscopy.

[0039] Color distribution characteristics refer to the spatial distribution patterns of colors in ancient calligraphy and paintings, including color type, concentration, and distribution area. The color distribution characteristics are quantified by counting the pixel distribution of each color channel in the image.

[0040] The automatic scanning module includes a scanning device, a super-depth-of-field three-dimensional microscope 6 and a scanning control unit. The super-depth-of-field three-dimensional microscope 6 is mounted on the scanning device.

[0041] Figure 2 is a schematic diagram of the structure of a scanning device according to some embodiments of this specification, such as Figure 2 As shown, the scanning device includes a frame 1, a vacuum suction cup honeycomb panel 2, a position adjustment component, and a laser ranging component. The vacuum suction cup honeycomb panel 2 and the position adjustment component are both mounted on the frame 1. The position adjustment component is mounted above the vacuum suction cup honeycomb panel 2. The ancient calligraphy or painting to be restored is placed on the vacuum suction cup honeycomb panel 2. The position adjustment component includes an X-axis coordinate adjustment device 3, a Y-axis coordinate adjustment device 4, and a Z-axis coordinate adjustment device 5. The Y-axis coordinate adjustment device 4 is mounted on the X-axis coordinate adjustment device 3, and the Z-axis coordinate adjustment device 5 is mounted on the Y-axis coordinate adjustment device 4. The ultra-depth-of-field three-dimensional microscope 6 is mounted on the Z-axis coordinate adjustment device 5. The laser ranging component is mounted on the Z-axis coordinate adjustment device 5 and includes a laser ranging dot matrix composed of multiple laser ranging devices.

[0042] Specifically, the vacuum suction cup honeycomb panel 2 is made of two thin panels firmly bonded to both sides of a thicker honeycomb core material, forming a honeycomb sandwich structure. The honeycomb sandwich structure serves as the main structure of the vacuum suction cup. The honeycomb sandwich structure is designed with airways or cavities inside, which are connected to equipment such as a vacuum pump to form a vacuum adsorption function. Operators can place the ancient calligraphy and painting to be restored on the vacuum suction cup honeycomb panel 2, which adsorbs and fixes the ancient calligraphy and painting to be restored, preventing it from moving during subsequent scanning.

[0043] The main function of the X-axis coordinate adjustment device 3 is to control the movement of the ultra-depth-of-field three-dimensional microscope 6 in the horizontal direction (i.e., the X-axis direction). By adjusting the position of the microscope on the X-axis, scanning of different lateral areas of ancient calligraphy and painting can be achieved. The X-axis coordinate adjustment device 3 can be composed of components such as a motor, a guide rail, and a lead screw. The motor provides power to drive the lead screw to rotate, thereby driving the moving parts (such as a slider or platform) mounted on the guide rail to move along the X-axis direction. The guide rail plays a supporting and guiding role to ensure the smooth operation of the moving parts.

[0044] The Y-axis coordinate adjustment device 4 is responsible for controlling the movement of the ultra-depth-of-field three-dimensional microscope 6 in another horizontal direction (i.e., the Y-axis direction). It works in conjunction with the X-axis coordinate adjustment device 3 to enable the microscope to move freely on the two-dimensional plane of ancient calligraphy and painting, thereby achieving coverage scanning of the entire picture. The structure of the Y-axis coordinate adjustment device 4 is similar to that of the X-axis coordinate adjustment device 3, and is also composed of components such as a motor, a guide rail, and a lead screw. The difference is that it is installed on the moving component of the X-axis coordinate adjustment device 3, so that when the X-axis coordinate adjustment device 3 moves, the Y-axis coordinate adjustment device 4 and the microscope thereon will also move in the X-axis direction; and when the motor of the Y-axis coordinate adjustment device 4 itself works, it can drive the microscope to move independently in the Y-axis direction.

[0045] The main function of the Z-axis coordinate adjustment device 5 is to control the movement of the super-depth-of-field three-dimensional microscope 6 in the vertical direction (i.e., the Z-axis direction). By adjusting the position of the super-depth-of-field three-dimensional microscope 6 on the Z-axis, the distance between the super-depth-of-field three-dimensional microscope 6 and the surface of the ancient calligraphy and painting can be adjusted to obtain the best scanning effect. The Z-axis coordinate adjustment device 5 may include an electric push rod and is also equipped with a high-precision position sensor (such as a grating ruler, an encoder, etc.) to achieve precise control of the position of the super-depth-of-field three-dimensional microscope 6 in the Z-axis direction. The electric push rod drives the super-depth-of-field three-dimensional microscope 6 to move along the Z-axis direction, and the position sensor feeds back the actual position information of the super-depth-of-field three-dimensional microscope 6 in real time. The scanning control unit adjusts the extension length of the electric push rod according to the feedback information to ensure that the super-depth-of-field three-dimensional microscope 6 can accurately reach the target position.

[0046] In some embodiments, the scanning device can also be used to adjust the position of the laser ranging component in the XY plane according to a preset ranging trajectory, wherein the XY plane is parallel to the vacuum suction cup honeycomb panel 2, the preset ranging trajectory includes multiple ranging trajectory points, and the laser ranging component is used to obtain regional ranging information at each ranging trajectory point. The scanning-related feature information of the ancient calligraphy and painting to be restored includes at least the regional ranging information corresponding to each ranging trajectory point, wherein the regional ranging information may include the distance to the surface of the ancient calligraphy and painting to be restored measured by each laser ranging device included in the laser ranging dot array when the laser ranging component is located at the ranging trajectory point.

[0047] Specifically, the preset ranging trajectory is a series of pre-planned path points in the XY plane. These points are called ranging trajectory points, which are used to guide the laser ranging component to move in an orderly manner within a specific area to ensure that the ranging information of different positions of the ancient calligraphy and painting to be restored can be obtained comprehensively and systematically. According to the length, width and other dimensional parameters of the ancient calligraphy and painting to be restored, as well as its approximate shape (such as rectangle, circle, etc.), a ranging trajectory that can cover the entire surface of the ancient calligraphy and painting can be planned. For example, for a rectangular ancient calligraphy and painting, a serpentine or grid-shaped trajectory from left to right and from top to bottom can be planned. Assuming there is a rectangular ancient calligraphy and painting that is 100 cm long and 80 cm wide, the preset ranging trajectory can be set to set a ranging trajectory point every 10 cm to form a 10×8 grid, with a total of 80 ranging trajectory points.

[0048] The scanning control unit is used to determine the scanning trajectory and scanning parameters based on the scanning-related feature information of the ancient calligraphy and painting to be restored, wherein the scanning trajectory includes multiple scanning trajectory points, and the scanning parameters include the starting scanning height corresponding to each scanning trajectory point. The scanning device is used to adjust the spatial position of the super-depth-of-field three-dimensional microscope according to the scanning trajectory and the starting scanning height corresponding to each scanning trajectory point. The super-depth-of-field three-dimensional microscope is used to obtain multi-layer images at each scanning trajectory point.

[0049] In some embodiments, the scanning control unit may determine the scanning trajectory according to the regional ranging information corresponding to each ranging trajectory point, specifically including:

[0050] Generate the concave and convex heights of multiple locations of the ancient calligraphy and painting to be restored based on the regional distance measurement information corresponding to each distance measurement trajectory point;

[0051] Establishing a trajectory generation constraint condition set, wherein the trajectory generation constraint condition set at least includes a shortest distance and a longest distance between two adjacent scanning trajectory points;

[0052] Establishing a multi-objective reward function, wherein the multi-objective reward function is related to the difference in concave-convex height between two adjacent scanning trajectory points, the concave-convex height of each scanning trajectory point, and the distance between two adjacent scanning trajectory points;

[0053] The trajectory generation model is used to generate a scanning trajectory based on the concave and convex heights of multiple locations of the ancient calligraphy and painting to be restored, the trajectory generation constraint set and the multi-objective reward function.

[0054] Specifically, check whether the regional ranging information corresponding to each ranging trajectory point is complete, that is, whether the data measured by each laser ranging device in the laser ranging dot array exists. If there is missing data, it can be interpolated based on the data of adjacent ranging points. For example, linear interpolation is used, assuming that the distance and height changes between adjacent ranging points are linear, and the missing values ​​are estimated by calculating the slope of adjacent valid data points. The laser ranging process may be affected by factors such as environmental interference and equipment errors, resulting in noise in the measurement data. A filtering algorithm, such as median filtering, can be used to process the regional ranging information of each ranging trajectory point. Median filtering replaces the measurement value of each ranging point with the median of all measurement values ​​in its neighborhood. It can effectively remove impulse noise and make the data smoother.

[0055] The regional ranging information corresponding to each ranging trajectory point is spliced ​​to generate the overall ranging information.

[0056] To generate surface height data for ancient calligraphy and paintings, a reference plane is required. The least squares method can be used to fit the three-dimensional coordinates of all measurement points to a single best-fit plane, which serves as the reference plane. The goal of the least squares method is to minimize the sum of the squared distances from all measurement points to the plane, thereby obtaining a plane that best represents the overall height trend of the ancient calligraphy and painting.

[0057] Sample multiple positions, and for each position, calculate its vertical distance to the reference plane. This distance is the concave-convex height of the measuring point relative to the reference plane. If the position is above the reference plane, the concave-convex height is a positive value, indicating that the position is convex; if the position is below the reference plane, the concave-convex height is a negative value, indicating that the position is concave. Among them, the surface of the ancient calligraphy and painting to be restored can be divided into several regular areas, such as a grid, and then a position is sampled in the center of each area. Make sure that the sampling points are evenly distributed over the entire surface to avoid excessive or insufficient sampling in local areas. For example, for an ancient calligraphy and painting that is 100 cm long and 80 cm wide, it can be divided into a grid of 10 rows and 8 columns, and the size of each grid is 10 cm × 10 cm. Then, sampling points are set at the intersection of each grid, resulting in a total of 80 sampling positions.

[0058] The trajectory generation constraint set is a collection of various limiting factors in the scanning trajectory generation process. It ensures that the generated scanning trajectory not only meets the actual operating capabilities of the scanning device, but also meets the needs of effective scanning of ancient calligraphy and paintings to be restored.

[0059] The minimum distance between two adjacent scanning track points refers to the minimum allowable distance between the projections of any two adjacent scanning track points onto the surface of an ancient calligraphy or painting during the scanning process. If adjacent scanning track points are too close together, the scanner may frequently start and stop or adjust its position during movement, affecting scanning stability and accuracy. Setting a minimum distance ensures sufficient space for the scanner to operate smoothly during movement, reducing scanning deviations caused by mechanical motion errors. Furthermore, overly dense scanning tracks increase scanning time and data processing volume, reducing scanning efficiency. By specifying a minimum distance, unnecessary repeated scans can be avoided, improving scanning efficiency.

[0060] The maximum distance between two adjacent scanning track points is the maximum allowable distance between the projections of any two adjacent scanning track points onto the surface of an ancient painting or calligraphy. If the distance between adjacent scanning track points is too far, some areas may be missed, preventing the full acquisition of surface information. Setting the maximum distance ensures that the scanning track covers the entire surface of the painting or calligraphy, avoiding blind spots.

[0061] In addition to the shortest and longest distances between two adjacent scanning trajectory points, the trajectory generation constraint set can also include some other constraints, such as:

[0062] The distance between the scanning track and the edge of ancient calligraphy and painting: In order to avoid the scanning device colliding with the edge of ancient calligraphy and painting during movement, it is necessary to specify the minimum safe distance between the scanning track and the edge of ancient calligraphy and painting.

[0063] Curvature limitation of the scanning track: In order to ensure that the scanning device can move smoothly along the scanning track, the curvature of the scanning track needs to be limited to avoid excessive bending.

[0064] If the height difference between adjacent track points is too large, the scanner will need to frequently adjust its height during movement, which will not only increase scanning difficulty and errors, but may also damage the scanner. A smaller height difference between adjacent track points will make the scanning process smoother, improving scanning accuracy and stability.

[0065] Ultra-depth-of-field microscopy requires extremely high flatness of the scanning surface, and the microscopic unevenness on the surface of ancient calligraphy and paintings makes focusing and focusing during the scanning process extremely difficult, thus affecting the scanning quality and efficiency. In areas where the height variation of the unevenness is small, the scanning device can relatively easily maintain the focus and focusing state, thereby obtaining a clear image. In areas where the height variation of the unevenness is large, if the scanning trajectory cannot be reasonably planned according to the unevenness height, the image may appear locally blurred or distorted. By incorporating the unevenness height of the scanning trajectory points into the multi-objective reward function, the trajectory generation model can be guided to generate a more reasonable scanning trajectory, so that the scanning device can better adapt to the surface undulations during the scanning process, thereby improving the overall clarity of the image.

[0066] The distance between two adjacent scanning track points affects the efficiency and quality of the scan. A short distance between two adjacent scanning track points will increase the scanning time and data processing volume, while a reasonable distance between two adjacent scanning track points can ensure full coverage and high efficiency of the scanning area.

[0067] As an example only, the multi-objective reward function can be:

[0068]

[0069] Among them, R(S t ,a t ,S t+1 ) is the multi-target reward, the current state S t Contains the concave and convex height information H of the current scanning trajectory point t , action a t It is the action of selecting the next scanning trajectory point, which causes the state to transfer to S t+1 , the concave-convex height is H t+1 , H max is the maximum concave-convex height of the ancient calligraphy and painting to be restored, △L (t,t+1) is the distance between the current scanning track point and the next scanning track point, △L max is the preset maximum distance between two adjacent scanning trajectory points, a1, a2 and a3 are weights, a1, a2 and a3 are greater than 0, and a1+a2+a3=1.

[0070] Understandably, the multi-objective reward function above gives lower rewards for situations with large height differences. Large height differences between adjacent scanning trajectory points require frequent height adjustments during movement, increasing scanning difficulty, errors, and even damage to the device. Therefore, it is desirable to minimize the height differences between adjacent scanning trajectory points to ensure a smoother scanning process and improve scanning accuracy and stability. The guided trajectory generation model tends to select a point with a similar height difference to the current trajectory point when selecting the next scanning trajectory point, thereby reducing the frequency of height adjustments during the scanning process.

[0071] The greater the concave-convex height of the next scanning trajectory point, the greater the reward. The guided trajectory generation model tends to select a trajectory point with a larger concave-convex height when selecting the next scanning trajectory point, so that the scanning device can better adapt to the surface undulations during the scanning process, avoiding local blur or distortion of the scanned image.

[0072] A short distance between two adjacent scanning track points will increase the scanning time and data processing volume, while a reasonable distance between two adjacent scanning track points can ensure full coverage and high efficiency of the scanning area. Therefore, it is hoped that the distance between adjacent track points is neither too large nor too small, but close to the preset maximum distance to ensure the efficiency and quality of the scanning. By setting a3×(△L (t,t+1) -△L max ) 2 ,The guidance trajectory generation model selects appropriate distance between ,adjacent trajectory points, so that the scanning trajectory covers the entire ,scanning area while avoiding the increase of scanning time and ,data processing volume due to a short distance, thereby improving ,the scanning efficiency.

[0073] Figure 3 is a schematic diagram of a process for generating a scanning trajectory according to some embodiments of this specification, such as Figure 3 As shown, in some embodiments, the scanning control unit generates a scanning trajectory through a trajectory generation model according to the concave and convex heights of multiple positions of the ancient calligraphy and painting to be restored, a trajectory generation constraint set, and a multi-objective reward function, including:

[0074] For each position, determine the concave-convex height difference value of the position based on the concave-convex heights of adjacent positions;

[0075] Normalizing the concave-convex height difference value at each position to generate a normalized concave-convex height difference value at each position;

[0076] Determine a plurality of trajectory generation starting positions from the plurality of positions according to the normalized concave-convex height difference value of each position;

[0077] For each trajectory generation starting position, the trajectory generation model is used to generate a scanning trajectory corresponding to the trajectory generation starting position based on the concave and convex heights of multiple positions of the ancient calligraphy and painting to be restored, the trajectory generation constraint set, and the multi-objective reward function;

[0078] Establishing a fitness function, wherein the fitness function is related to the difference in concave-convex height between two adjacent scanning trajectory points and the scanning overlap area between any two scanning trajectory points;

[0079] Determine the fitness value of the scanning trajectory corresponding to the starting position of each trajectory generation;

[0080] The scanning trajectory is generated by using a genetic algorithm based on the fitness value of the scanning trajectory corresponding to the starting position of each trajectory.

[0081] Specifically, the adjacent position may be a position whose distance from the position is less than a preset distance threshold. For each position, the variance of the concave-convex heights of the adjacent positions may be calculated as the concave-convex height difference value of the position. The concave-convex height difference values ​​of different positions may have different orders of magnitude. Directly using these difference values ​​for comparison and analysis may lead to inaccurate results. The purpose of normalization is to map the concave-convex height difference values ​​of different positions to a unified range (usually [0,1]), eliminate the influence of the order of magnitude, and make subsequent processing (such as determining the starting position of trajectory generation) more reasonable and effective. Min-Max Normalization can be used to normalize the concave-convex height difference value of each position.

[0082] The position where the normalized concave-convex height difference value is greater than a preset concave-convex height difference value threshold can be used as the starting position for trajectory generation.

[0083] As can be understood, the bump height difference value reflects the complexity of the surface of the ancient calligraphy or painting being restored, both at a specific location and in its adjacent areas. A larger difference value indicates a more dramatic surface bump variation, leading to increased scanning difficulty and complexity. A preset bump height difference threshold serves as a critical value for identifying complex surface areas. When the normalized bump height difference value at a location exceeds this threshold, that location and its adjacent areas are considered to have high surface complexity. Selecting an area with high surface complexity as the starting location for trajectory generation ensures that the scanning trajectory begins in the most challenging areas, better addressing microscopic surface irregularities. Selecting a location with a smaller bump height difference value as the starting location may cause the scanning trajectory to repeatedly scan relatively flat areas, increasing scanning time and data processing. Selecting a location with a larger bump height difference value as the starting location ensures that the scanning trajectory covers more complex areas, reduces redundant scanning, and improves scanning efficiency. By determining multiple starting locations for trajectory generation, as many distinct scanning trajectories as possible are generated, providing more effective information for subsequent scanning trajectory generation.

[0084] As an example only, the fitness function may be:

[0085]

[0086] Among them, F(S) is the fitness value, H n is the concave-convex height corresponding to the nth scanning trajectory point, H n+1 is the concave-convex height corresponding to the n+1th scanning track point, N is the total number of scanning track points, △S ij is the overlapping area between the scanning area of ​​the i-th scanning trajectory point and the scanning area of ​​the j-th scanning trajectory point, b1, b2 and b3 are weights, b1, b2 and b3 are greater than 0, and b1+b2+b3=1.

[0087] Understandably, The average sum of the squared differences in the heights of adjacent trajectory points is calculated. This difference reflects the degree of height variation in the scanned trajectory. Larger differences indicate more dramatic height variations, potentially increasing scanning difficulty and error. By averaging the sum of squares, we determine that scanned trajectories with greater height variations have lower fitness values. The average of the sum of the overlapping areas of the scanned regions between all trajectory points was calculated. Scanned region overlap reflects the degree of redundancy within the scanned trajectory. A larger overlap indicates more repeated scans, increasing scanning time and data processing. By averaging the overlapped areas, the fitness value of scanned trajectories with larger overlapped areas is reduced. The number of track points is taken into account. Given a fixed scanning area, a greater number of track points may mean a denser scanning trajectory, but this may also increase scanning time and data processing. By taking the inverse, the fitness value of a scanning trajectory with a greater number of track points is reduced.

[0088] A genetic algorithm is an optimization algorithm that simulates natural selection and heredity mechanisms and is suitable for solving complex optimization problems. The genetic algorithm generates a fitness value for the scanning trajectory corresponding to the starting position based on each trajectory. The specific process of generating the scanning trajectory is as follows:

[0089] Initialize population: Randomly generate a set of initial scanning trajectories as the initial population of the genetic algorithm.

[0090] Evaluate fitness: Use the fitness function to evaluate the fitness of each individual in the population (i.e., each scanning trajectory).

[0091] Selection operation: Select excellent individuals to enter the next generation population based on their fitness values. Individuals with higher fitness values ​​have a greater probability of being selected.

[0092] Crossover operation: Perform a crossover operation on the selected individuals to generate new individuals (i.e., new scanning trajectories). The crossover operation simulates the gene recombination process in biological genetics and helps to produce better individuals.

[0093] Mutation: Mutating newly generated individuals introduces a degree of randomness to increase population diversity. Mutation simulates the genetic mutation process in biological inheritance and helps prevent the algorithm from falling into a local optimal solution.

[0094] Iterative optimization: Repeat the selection, crossover, mutation and other operations until the termination condition is met (such as reaching the maximum number of iterations or the fitness value reaches the preset threshold). The optimal individual in the population is the generated scanning trajectory.

[0095] In some embodiments, the scanning control unit determines the starting scanning height corresponding to the scanning trajectory point based on the scanning-related feature information of the ancient calligraphy and painting to be restored, including:

[0096] Interpolate the concave and convex heights of multiple locations of the ancient calligraphy and painting to be restored to generate the concave and convex heights corresponding to the scanning trajectory points;

[0097] Calculate the height difference between the concave and convex height corresponding to the scanning trajectory point and the reference height;

[0098] The starting scanning height corresponding to the scanning track point is determined according to the height difference between the concave-convex height corresponding to the scanning track point and the reference height.

[0099] Specifically, interpolation algorithms such as linear interpolation, polynomial interpolation, and spline interpolation can be used to interpolate the concave and convex heights at multiple locations on the ancient calligraphy and painting to be restored, generating the concave and convex heights corresponding to the scan trajectory points. The base height is a preset reference height. By calculating the difference between the concave and convex heights corresponding to the scan trajectory points and the base height, the degree of deviation of each scan trajectory point from the base height can be determined. The height difference can be obtained by simply subtracting the concave and convex height value corresponding to the scan trajectory point from the base height value.

[0100] When the height difference between the concave-convex height corresponding to the scanning trajectory point and the reference height is greater than the preset height interpolation threshold, the scanning trajectory point is uneven. The height difference between the concave-convex height corresponding to the scanning trajectory point and the reference height is summed with the concave-convex height to obtain the starting scanning height corresponding to the scanning trajectory point.

[0101] It is understandable that due to the damage on the surface of ancient calligraphy and painting and the stress of the paper, the paper appears uneven. This uneven surface will make it difficult for the super-depth-of-field microscope to maintain a stable focusing distance during the scanning process. If a unified scanning starting height is used, some areas will inevitably be in focus while some areas will be blurred. The starting scanning height of each scanning trajectory point is determined by the above method, so that the microscope head can automatically adjust the scanning height according to the different concave and convex conditions of the surface of ancient calligraphy and painting during the scanning process, ensuring that each point can start scanning at a constant height. This can effectively avoid the image blurring caused by differences in focus position, ensure that the scanned image is clear and accurate, and provide high-quality image data for subsequent ancient calligraphy and painting restoration work.

[0102] For each scanning track point, the super-depth of field 3D microscope is adjusted to the starting scanning height corresponding to the scanning track point through the X-axis coordinate adjustment device and the Y-axis coordinate adjustment device. The microscope is then adjusted to the starting scanning height corresponding to the scanning track point through the Z-axis coordinate adjustment device, and moves downward with a fixed step length (for example, 20 μm). Each time it moves to a new photographic position, the camera is triggered to take a photo. For example, if 15 photos need to be taken at each point, the super-depth of field 3D microscope will move downward along the z-axis 14 times, each time moving 20 μm. This allows for the acquisition of multiple layers of images at each scanning track point.

[0103] In some embodiments, the scanning device further includes a fill light device.

[0104] In some embodiments, the scanning control unit determines scanning parameters based on the scanning-related feature information of the ancient calligraphy and painting to be restored, including:

[0105] The fill light parameters are determined according to the material information, pigment type information and color distribution characteristics, wherein the scanning parameters include the fill light parameters, and the fill light parameters include color temperature and light intensity.

[0106] Understandably, the materials, pigments, and color distributions of ancient calligraphy and paintings vary. Fill-in lighting equipment can adjust color temperature and intensity to highlight detailed features of ancient calligraphy and paintings, such as brushstrokes and textures, facilitating subsequent image analysis and restoration. Specifically, different paper materials have different light reflection and absorption properties. For example, rice paper is thin and light-transmitting, so it may require lower light intensities to avoid overexposure. Thicker paper, on the other hand, may require higher light intensities to maintain image brightness. If ancient calligraphy or paintings are painted on fabric, the texture and fiber structure of the fabric will affect light scattering and reflection. In this case, the fill-in lighting equipment's color temperature and intensity must be adjusted based on the fabric's characteristics to highlight the texture and details of the painting or calligraphy. Mineral pigments typically have high reflectivity and vibrant colors. To accurately reproduce the colors of these pigments, it's necessary to select an appropriate color temperature so that the fill-in lighting matches the pigment's spectral characteristics. For example, some red mineral pigments appear most realistic at specific color temperatures. Plant pigments have relatively subdued colors and are more sensitive to light. Excessive light intensity can cause plant pigments to fade or distort. Therefore, it's necessary to adjust the light intensity based on the pigment's characteristics and select an appropriate color temperature to highlight its color characteristics. For ancient calligraphy and paintings with rich colors and high contrast, higher light intensity is required to highlight the layering and details of the colors. Furthermore, selecting an appropriate color temperature based on the distribution of warm and cool colors can make the colors more vivid and realistic. For ancient calligraphy and paintings with monotonous colors and low contrast, the light intensity can be appropriately reduced to avoid overexposure caused by excessive light. Furthermore, adjusting the color temperature can enhance color contrast and clarity.

[0107] In some embodiments, the scanning control unit determines fill light parameters based on material information, pigment type information, and color distribution characteristics, including:

[0108] Obtain material information, pigment type information, color distribution characteristics and optimal fill light parameters for multiple samples of ancient calligraphy and painting;

[0109] Through split hierarchical clustering, multiple samples of ancient calligraphy and painting are divided into multiple sample groups according to their material information, pigment type information, and color distribution characteristics;

[0110] Determine the sample group to which the ancient calligraphy and painting to be restored belongs based on the material information, pigment type information, and color distribution characteristics of the ancient calligraphy and painting to be restored;

[0111] Determine similar sample ancient calligraphy and paintings to be restored based on the material information, pigment type information, and color distribution characteristics of the ancient calligraphy and painting to be restored and the material information, pigment type information, and color distribution characteristics of the sample ancient calligraphy and paintings included in the sample group to which the ancient calligraphy and painting to be restored belongs;

[0112] The fill light parameters are determined based on the optimal fill light parameters of similar sample ancient calligraphy and painting to be restored.

[0113] Specifically, the material and pigment type information of the sample ancient calligraphy and paintings is encoded, and the color distribution characteristics are normalized. Then, appropriate distance metrics (such as Euclidean distance and cosine similarity) are used to calculate the distance between the sample ancient calligraphy and paintings, and then cluster them. Through divisive hierarchical clustering, multiple sample ancient calligraphy and paintings are divided into multiple sample groups based on the distance between them.

[0114] The characteristics of the ancient calligraphy and painting to be restored are compared with those of each sample group, and the similarity between the sample group and each sample group is calculated. The same distance metric used in the clustering process can be used. The sample group with the highest similarity is selected as the sample group to which the ancient calligraphy and painting to be restored belongs. This means that the ancient calligraphy and painting to be restored can find samples within this sample group that have the most similar characteristics to it. Within the sample group to which the ancient calligraphy and painting to be restored belongs, the similarity between the ancient calligraphy and painting to be restored and each sample is further calculated. Based on a preset similarity threshold, samples with a high similarity to the ancient calligraphy and painting to be restored are selected as similar samples. Similar samples closely resemble the ancient calligraphy and painting to be restored in terms of material, pigment type, and color distribution. Therefore, their lighting parameters are of high reference value for determining the lighting parameters for the ancient calligraphy and painting to be restored. Statistical analysis of the optimal lighting parameters for similar samples can be performed using methods such as average and weighted average. For example, the average of the optimal lighting parameters of all similar samples can be calculated as the reference value for the lighting parameters for the ancient calligraphy and painting to be restored.

[0115] The image processing module is used to stack the multi-layer images of each scanning trajectory point to generate the regional image corresponding to each scanning trajectory point, and to splice the regional images corresponding to each scanning trajectory point to generate the scanned image of the ancient calligraphy and painting to be restored.

[0116] Specifically, the stacking operation is to integrate the multi-layer images of the scanning trajectory points and fuse them into one image through a specific algorithm. Its basic principle is to analyze the information of each pixel in each layer of the image and synthesize the final pixel value according to certain rules (such as taking the average, maximum, minimum, or weighted average based on the quality of the pixel points). After the stacking operation, each trajectory point obtains an image that integrates the information of multiple layers of images, such as Figure 8-10 As shown in the figure, these images are the regional images corresponding to the corresponding track points. Regional images contain comprehensive information about the area where the track point is located under various scanning conditions, and can more comprehensively and accurately reflect the characteristics of the ancient calligraphy and painting in that area, such as color, texture, and details. These regional images serve as the basis for subsequent image stitching.

[0117] The order in which the regional images are stitched is determined based on the layout of the scan trajectory points. For example, if scanning is performed from left to right and top to bottom, the stitching can be performed in this order. Starting with adjacent regional images, feature matching, transformation model estimation, and image fusion are performed to stitch the two images into a larger image. This stitched image is then stitched with the next adjacent regional image, and so on, until all regional images are stitched.

[0118] Figure 5 、 Figure 6 is a schematic diagram of a user interface of a scanning control unit according to some embodiments of this specification, such as Figure 5 、 Figure 6 As shown, the operator can control the scanning status of the scanning device (for example, start, pause, stop, etc.) through the user interface, and can also obtain the working information of the scanning device through the user interface, such as the X-axis position, Y-axis position, Z-axis position, etc.

[0119] The damage analysis module is used to determine the damage characteristics of the ancient calligraphy and painting to be restored based on the scanned images of the ancient calligraphy and painting to be restored through a damage analysis model.

[0120] Figure 4 is a schematic diagram of the structure of the damage analysis model shown in some embodiments of this specification, such as Figure 4 As shown, in some embodiments, the damage analysis model includes a defect analysis unit, a fracture analysis unit, a discoloration analysis unit and an insect pest analysis unit, wherein the defect analysis unit is used to generate a defect distribution map based on the scanned image of the ancient calligraphy and painting to be repaired, the fracture analysis unit is used to generate a fracture distribution map based on the scanned image of the ancient calligraphy and painting to be repaired, the color change analysis unit is used to generate a color change distribution map based on the scanned image of the ancient calligraphy and painting to be repaired, and the insect pest analysis unit is used to generate an insect pest distribution map based on the scanned image of the ancient calligraphy and painting to be repaired.

[0121] Specifically, the defect analysis unit is used to identify and locate damaged areas on the surface of ancient calligraphy and paintings. Damage may appear in the form of wear, missing, scratches, etc. These damages will destroy the integrity and artistic value of ancient calligraphy and paintings. The defect analysis unit performs image processing and analysis on the scanned image, uses edge detection algorithms to identify the boundaries of the damaged area, and distinguishes the damaged parts by comparing the pixel features (such as color, texture, etc.) of the normal area and the damaged area. For example, the color of the worn area is lighter than the surrounding normal area, and the texture changes. The defect analysis unit can capture these differences and generate a defect distribution map. The defect distribution map intuitively shows the location, shape and range of the damaged area on the ancient calligraphy and painting. According to the distribution map, the restorers can understand the severity and distribution of the damage, which provides an important reference for subsequent restoration work, such as determining the area that needs to be repaired and selecting appropriate repair materials.

[0122] The Fracture Analysis Unit is used to identify fractures in ancient calligraphy and paintings. Fractures are one of the most serious forms of damage to ancient calligraphy and paintings, potentially resulting in the loss of content and structural damage. The Fracture Analysis Unit performs edge detection and morphological analysis on scanned images to identify the boundaries and features of fractured areas. Fractured areas typically appear as visible cracks or fracture lines in the image, with pixel characteristics significantly different from those of normal areas. By analyzing information such as the length, width, and direction of the fracture lines, the extent and location of the fracture can be determined. A fracture distribution map details the location and morphology of fractures on ancient calligraphy and paintings. Restorers can use this distribution map to assess the severity of the fracture and select appropriate restoration methods, such as splicing or filling, to restore the integrity of the ancient calligraphy and painting.

[0123] The color change analysis unit is used to detect color changes in ancient calligraphy and paintings. Due to the passage of time and environmental factors (such as light, humidity, and temperature), the colors of ancient calligraphy and paintings may fade and discolor, affecting their artistic quality and historical value. The color change analysis unit performs color space conversion and color feature extraction on the scanned image, converting the image from the RGB color space to another color space more suitable for color analysis (such as HSV, Lab, etc.). It then determines the area and extent of color change by comparing color feature values ​​of different regions, such as hue, saturation, and brightness. For example, a color histogram can intuitively display the distribution of colors in an image. By comparing the color histograms of normal and discolored areas, differences in color change can be detected. The color change distribution map clearly shows the area and extent of color change in ancient calligraphy and paintings. Based on the distribution map, restorers can understand the type and extent of color change and select appropriate restoration materials and methods, such as color inpainting and fading treatment, to restore the original colors of the ancient calligraphy and paintings.

[0124] The Pest Analysis Unit is used to detect insect infestation in ancient calligraphy and paintings, as well as the extent of the infestation. Pests are a common threat to the preservation of ancient calligraphy and paintings. Pests can eat paper and pigments, causing damage such as holes and defects. The Pest Analysis Unit performs texture analysis and shape recognition on scanned images to identify traces of insect damage. Insect damage typically has specific shape and texture characteristics, such as holes and insect tracks. Image processing algorithms, such as template matching and machine learning, can identify these features and determine the location and extent of the infestation. For example, images of common insect damage can be collected as templates, and template matching algorithms can be used to locate similar areas in the scanned image. Pest distribution maps visually display the distribution of insect damage on ancient calligraphy and paintings. Based on these distribution maps, conservators can assess the impact of insect damage on ancient calligraphy and paintings and take appropriate preventive measures, such as insecticide treatment and repair of damage caused by insects, to protect the calligraphy and paintings.

[0125] The damage analysis module can fuse the defect distribution map, fracture distribution map, color change distribution map and pest distribution map to generate Figure 6 Schematic diagram of the damage characteristics of the ancient calligraphy and painting to be restored.

[0126] Figure 11 This is a flow chart of the ultra-depth-of-field microscopy scanning and AI image analysis method for ancient calligraphy and painting restoration according to some embodiments of this specification, such as Figure 11 As shown, the ultra-depth-of-field microscopy scanning and AI image analysis method for ancient calligraphy and painting restoration may include the following steps:

[0127] Obtaining scan-related feature information of the ancient calligraphy and painting to be restored;

[0128] Determine a scanning trajectory and scanning parameters based on the scanning-related feature information of the ancient calligraphy and painting to be restored, wherein the scanning trajectory includes a plurality of scanning trajectory points, and the scanning parameters include a starting scanning height corresponding to each scanning trajectory point;

[0129] According to the scanning trajectory and the starting scanning height corresponding to each scanning trajectory point, the spatial position of the ultra-depth-of-field three-dimensional microscope is adjusted, and a multi-layer image is acquired at each scanning trajectory point by the ultra-depth-of-field three-dimensional microscope;

[0130] The multi-layer images of each scanning trajectory point are stacked to generate the regional image corresponding to each scanning trajectory point, and the regional images corresponding to each scanning trajectory point are spliced ​​to generate the scanned image of the ancient calligraphy and painting to be restored;

[0131] The damage analysis model is used to determine the damage characteristics of the ancient calligraphy and painting to be restored based on the scanned images of the ancient calligraphy and painting to be restored.

[0132] The super-depth-of-field microscopic scanning and AI image analysis method used for the restoration of ancient calligraphy and paintings can be applied to the super-depth-of-field microscopic scanning and AI image analysis system used for the restoration of ancient calligraphy and paintings, and will not be repeated here.

[0133] Finally, it should be understood that the embodiments described in this specification are intended only to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, by way of example and not limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly described and illustrated in this specification.

Claims

1. Ultra-depth-of-field microscopy and AI image analysis system for the restoration of ancient calligraphy and painting, characterized by: include: An information acquisition module is used to obtain scan-related feature information of the ancient calligraphy and painting to be restored; An automatic scanning module includes a scanning device, a super-depth-of-field 3D microscope, and a scanning control unit. The super-depth-of-field 3D microscope is mounted on the scanning device. The scanning control unit is used to determine a scanning trajectory and scanning parameters based on scanning-related feature information of the ancient calligraphy and painting to be restored. The scanning trajectory includes multiple scanning trajectory points, and the scanning parameters include a starting scanning height corresponding to each scanning trajectory point. The scanning device is used to adjust the spatial position of the super-depth-of-field 3D microscope based on the scanning trajectory and the starting scanning height corresponding to each scanning trajectory point. The super-depth-of-field 3D microscope is used to acquire a multi-layer image at each scanning trajectory point. An image processing module is used to stack the multi-layer images of each scanning trajectory point to generate a regional image corresponding to each scanning trajectory point, and to splice the regional images corresponding to each scanning trajectory point to generate a scanned image of the ancient calligraphy and painting to be restored; The damage analysis module is used to determine the damage characteristics of the ancient calligraphy and painting to be restored based on the scanned images of the ancient calligraphy and painting to be restored through a damage analysis model.

2. The ultra-depth-of-field microscopy and AI image analysis system for ancient calligraphy and painting restoration according to claim 1 is characterized in that: The scanning device includes a frame, a vacuum suction cup honeycomb panel, a position adjustment component and a laser ranging component, wherein the vacuum suction cup honeycomb panel and the position adjustment component are both arranged on the frame, the position adjustment component is arranged above the vacuum suction cup honeycomb panel, the ancient calligraphy and painting to be restored is placed on the vacuum suction cup honeycomb panel, the position adjustment component includes an X-axis coordinate adjustment device, a Y-axis coordinate adjustment device and a Z-axis coordinate adjustment device, the Y-axis coordinate adjustment device is arranged on the X-axis coordinate adjustment device, the Z-axis coordinate adjustment device is arranged on the Y-axis coordinate adjustment device, and the super-depth-of-field three-dimensional microscope is installed on the Z-axis coordinate adjustment device; The laser distance measuring component is arranged on the Z-axis coordinate adjustment device, and the laser distance measuring component includes a laser distance measuring dot matrix composed of a plurality of laser distance measuring devices.

3. The ultra-depth-of-field microscopy and AI image analysis system for ancient calligraphy and painting restoration according to claim 2 is characterized in that: The scanning device is further configured to adjust the position of the laser ranging assembly in an XY plane according to a preset ranging trajectory, wherein the XY plane is parallel to the vacuum suction cup honeycomb plate, the preset ranging trajectory includes a plurality of ranging trajectory points, the laser ranging assembly is configured to obtain regional ranging information at each ranging trajectory point, and the scanning-related feature information of the ancient calligraphy and painting to be restored includes at least the regional ranging information corresponding to each ranging trajectory point; The scanning control unit determines a scanning trajectory based on the scanning-related feature information of the ancient calligraphy and painting to be restored, including: The scanning trajectory is determined based on the regional ranging information corresponding to each ranging trajectory point.

4. The ultra-depth-of-field microscopy and AI image analysis system for ancient calligraphy and painting restoration according to claim 3 is characterized in that: The scanning control unit determines a scanning trajectory according to regional ranging information corresponding to each ranging trajectory point, including: Generate the concave and convex heights of multiple locations of the ancient calligraphy and painting to be restored based on the regional distance measurement information corresponding to each distance measurement trajectory point; Establishing a trajectory generation constraint condition set, wherein the trajectory generation constraint condition set at least includes the shortest distance and the longest distance between two adjacent scanning trajectory points; Establishing a multi-objective reward function, wherein the multi-objective reward function is related to the difference in concave-convex height between two adjacent scanning trajectory points, the concave-convex height of each scanning trajectory point, and the distance between two adjacent scanning trajectory points; The trajectory generation model is used to generate a scanning trajectory based on the concave and convex heights of multiple locations of the ancient calligraphy and painting to be restored, the trajectory generation constraint set and the multi-objective reward function.

5. The ultra-depth-of-field microscopy and AI image analysis system for ancient calligraphy and painting restoration according to claim 4 is characterized in that: The scanning control unit generates a scanning trajectory through a trajectory generation model according to the concave and convex heights of multiple positions of the ancient calligraphy and painting to be restored, a trajectory generation constraint set, and a multi-objective reward function, including: For each position, determine the concave-convex height difference value of the position based on the concave-convex heights of adjacent positions; Normalizing the concave-convex height difference value at each position to generate a normalized concave-convex height difference value at each position; Determine a plurality of trajectory generation starting positions from the plurality of positions according to the normalized concave-convex height difference value of each position; For each trajectory generation starting position, the trajectory generation model is used to generate a scanning trajectory corresponding to the trajectory generation starting position based on the concave and convex heights of multiple positions of the ancient calligraphy and painting to be restored, the trajectory generation constraint set, and the multi-objective reward function; Establishing a fitness function, wherein the fitness function is related to the difference in concave-convex height between two adjacent scanning trajectory points and the scanning overlap area between any two scanning trajectory points; Determine the fitness value of the scanning trajectory corresponding to the starting position of each trajectory generation; The scanning trajectory is generated by using a genetic algorithm based on the fitness value of the scanning trajectory corresponding to the starting position of each trajectory.

6. The ultra-depth-of-field microscopy and AI image analysis system for ancient calligraphy and painting restoration according to claim 4 is characterized in that: The scanning control unit determines the starting scanning height corresponding to the scanning trajectory point according to the scanning-related feature information of the ancient calligraphy and painting to be restored, including: Interpolate the concave and convex heights of multiple locations of the ancient calligraphy and painting to be restored to generate the concave and convex heights corresponding to the scanning trajectory points; Calculate the height difference between the concave and convex height corresponding to the scanning trajectory point and the reference height; The starting scanning height corresponding to the scanning track point is determined according to the height difference between the concave-convex height corresponding to the scanning track point and the reference height.

7. The ultra-depth-of-field microscopy scanning and AI image analysis system for ancient calligraphy and painting restoration according to any one of claims 1 to 6, characterized in that: The scan-related feature information of the ancient calligraphy and painting to be restored also includes material information, pigment type information and color distribution characteristics; The scanning device also includes a fill light device; The scanning control unit determines scanning parameters according to the scanning-related feature information of the ancient calligraphy and painting to be restored, including: The fill light parameters are determined according to the material information, the pigment type information and the color distribution characteristics, wherein the scanning parameters include the fill light parameters, and the fill light parameters include color temperature and light intensity.

8. The ultra-depth-of-field microscopy and AI image analysis system for ancient calligraphy and painting restoration according to claim 7 is characterized in that: The scanning control unit determines fill light parameters according to material information, pigment type information and color distribution characteristics, including: Obtain material information, pigment type information, color distribution characteristics and optimal fill light parameters for multiple samples of ancient calligraphy and painting; Through split hierarchical clustering, multiple samples of ancient calligraphy and painting are divided into multiple sample groups according to their material information, pigment type information, and color distribution characteristics; Determine the sample group to which the ancient calligraphy and painting to be restored belongs based on the material information, pigment type information, and color distribution characteristics of the ancient calligraphy and painting to be restored; Determine similar sample ancient calligraphy and paintings to be restored based on the material information, pigment type information, and color distribution characteristics of the ancient calligraphy and painting to be restored and the material information, pigment type information, and color distribution characteristics of the sample ancient calligraphy and paintings included in the sample group to which the ancient calligraphy and painting to be restored belongs; The fill light parameters are determined based on the optimal fill light parameters of similar sample ancient calligraphy and painting to be restored.

9. The ultra-depth-of-field microscopy scanning and AI image analysis system for ancient calligraphy and painting restoration according to any one of claims 1 to 6, characterized in that: The damage analysis model includes a defect analysis unit, a fracture analysis unit, a discoloration analysis unit and an insect pest analysis unit, wherein the defect analysis unit is used to generate a defect distribution map based on the scanned image of the ancient calligraphy and painting to be repaired, the fracture analysis unit is used to generate a fracture distribution map based on the scanned image of the ancient calligraphy and painting to be repaired, the color change analysis unit is used to generate a color change distribution map based on the scanned image of the ancient calligraphy and painting to be repaired, and the insect pest analysis unit is used to generate an insect pest distribution map based on the scanned image of the ancient calligraphy and painting to be repaired.

10. A method for ultra-depth microscopic scanning and AI image analysis for the restoration of ancient calligraphy and painting, applied to the ultra-depth microscopic scanning and AI image analysis system for the restoration of ancient calligraphy and painting as claimed in any one of claims 1 to 9, characterized in that: include: Obtaining scan-related feature information of the ancient calligraphy and painting to be restored; Determine a scanning trajectory and scanning parameters based on the scanning-related feature information of the ancient calligraphy and painting to be restored, wherein the scanning trajectory includes a plurality of scanning trajectory points, and the scanning parameters include a starting scanning height corresponding to each scanning trajectory point; According to the scanning trajectory and the starting scanning height corresponding to each scanning trajectory point, the spatial position of the ultra-depth-of-field three-dimensional microscope is adjusted, and a multi-layer image is acquired at each scanning trajectory point by the ultra-depth-of-field three-dimensional microscope; The multi-layer images of each scanning trajectory point are stacked to generate the regional image corresponding to each scanning trajectory point, and the regional images corresponding to each scanning trajectory point are spliced ​​to generate the scanned image of the ancient calligraphy and painting to be restored; The damage analysis model is used to determine the damage characteristics of the ancient calligraphy and painting to be restored based on the scanned images of the ancient calligraphy and painting to be restored.

Images