Color restoration method and device in cultural relic three-dimensional digitization

Abstract

The invention discloses a color restoration method and device in cultural relic three-dimensional digitization, and belongs to the technical field of cultural relic protection, and the method comprises the following steps: S1, collecting three-dimensional point cloud data; s2, data preprocessing; s3, constructing a cultural relic three-dimensional model; s4, collecting hyperspectral imaging data of the cultural relics; s5, data preprocessing; s6, the preprocessed hyperspectral imaging data texture is mapped to the constructed cultural relic three-dimensional model, and visualized display is carried out; according to the method, the three-dimensional point cloud data of the cultural relics are acquired based on the three-dimensional laser scanning equipment, the three-dimensional model of the cultural relics is constructed based on the three-dimensional point cloud data of the cultural relics, the hyperspectral imaging data of the cultural relics are acquired based on the hyperspectral imager, and the texture of the hyperspectral imaging data of the cultural relics is mapped to the three-dimensional model of the cultural relics for visual display. And the color and texture of the cultural relic can be accurately restored.

Description

Technical Field

[0001] The invention belongs to the technical field of cultural relics protection, and in particular relates to a color restoration method and device in three-dimensional digitization of cultural relics. Background Art

[0002] Cultural relics refer to all valuable material remains formed in the process of social and historical development, created and produced by humans or leaving traces of human activities. They have historical, artistic and scientific value, and are records and witnesses of human social activities.

[0003] Cultural relics will fade and darken due to their age and the influence of factors such as natural erosion. In order to preserve and protect cultural relic information more authentically, digital color restoration of cultural relics is particularly important.

[0004] The current method for digital color restoration of cultural relics is generally to obtain point cloud data through 3D laser scanning to construct a 3D model for the preservation and protection of cultural relics. This method has difficulty capturing subtle color changes on the surface of cultural relics and is often unable to accurately restore the original color of cultural relics, especially when the color on the surface of cultural relics changes due to time and environmental factors, resulting in large differences in color between digitized cultural relics and the originals.

[0005] In view of this, a color restoration method and device for three-dimensional digitization of cultural relics are designed to solve the above problems. Summary of the invention

[0006] In order to solve the problems raised in the above background technology, the present invention provides a method and device for color restoration in three-dimensional digitization of cultural relics, which has the characteristics of accurately restoring the color and texture of the cultural relics themselves.

[0007] Another object of the present invention is to provide a color restoration device for three-dimensional digitization of cultural relics.

[0008] To achieve the above object, the present invention provides the following technical solution: a color restoration method in three-dimensional digitization of cultural relics, comprising the following steps:

[0009] S1: Collect 3D point cloud data of cultural relics through 3D laser scanning equipment;

[0010] S2: Preprocessing the collected 3D point cloud data of cultural relics;

[0011] S3: Extract feature points of the preprocessed 3D point cloud data of cultural relics based on the adaptive local surface variation factor, describe the extracted feature points through the FPFH algorithm, roughly align the 3D point cloud data of cultural relics through the random sampling consistency algorithm, and finely align the 3D point cloud data of cultural relics through the ICP algorithm. Use the multi-scale transformation algorithm to gradually fuse the 3D point cloud data of cultural relics from coarse to fine and convert it into a triangular mesh surface model, optimize the triangular mesh surface model, and obtain the 3D model of cultural relics;

[0012] S4: Collecting hyperspectral imaging data of cultural relics through hyperspectral imager;

[0013] S5: preprocessing the collected hyperspectral imaging data;

[0014] S6: Texture mapping of the pre-processed hyperspectral imaging data onto the constructed three-dimensional model of the cultural relics is performed and visualized.

[0015] Furthermore, in step S3, the specific steps of extracting feature points of the pre-processed cultural relic three-dimensional point cloud data based on the adaptive local surface change factor include:

[0016] Assume that any point P of the 3D point cloud data of the cultural relic i The covariance matrix C of the field is expressed as:

[0017]

[0018] Where: k represents the total number of points in the three-dimensional point cloud data of cultural relics, P ij Represented as the coordinate vector of the point cloud,

[0019] is represented as the centroid of the point cloud, It is represented as a decreasing function, and r is represented as point P i Field radius;

[0020] Based on the covariance matrix C, the eigenvalues ​​λ1, λ2 and λ3 are obtained;

[0021] Assuming λ1<λ2<λ3, then point P i The variation factor of the surface in the neighborhood σ(P i ) is:

[0022]

[0023] Where: λ1, λ2 and λ3 are represented as eigenvalues;

[0024] Select a reasonable threshold ε and keep σ(P i )>ε is the extracted feature point, where the expression of the threshold ε is:

[0025]

[0026] Where: k represents the total number of points in the three-dimensional point cloud data of cultural relics, σ(P i ) is represented by point P i The variation factor of the surface within the neighborhood.

[0027] Furthermore, in step S6, the specific steps of texture mapping the pre-processed hyperspectral imaging data onto the constructed three-dimensional model of cultural relics include:

[0028] By using the barycentric coordinate method, the corresponding texture coordinates are assigned to each vertex in the three-dimensional model of the cultural relic;

[0029] Through the SIFT algorithm, multiple texture images are stitched into a complete texture map based on the texture coordinates assigned to each vertex;

[0030] Perform color correction on the spliced ​​texture image through a color space conversion algorithm;

[0031] The stitched texture image is filtered and denoised through the bilateral filtering image processing algorithm to improve the smoothness of the texture image while retaining its edge features, thereby optimizing the expression of surface details of the cultural relics.

[0032] Furthermore, in step S6, the specific steps of assigning corresponding texture coordinates to each vertex in the three-dimensional model of the cultural relic by the barycentric coordinate method include:

[0033] Assume that the texture coordinates corresponding to vertices A, B, and C are (U A ,V A )、(U B ,V B ) and (U C ,V C );

[0034] For any point P in the triangle, its centroid coordinates are (α, β, γ), and the calculation expression is:

[0035]

[0036] Where: Area(PBC) represents the area of ​​triangle PBC, Area(PCA) represents the area of ​​triangle PCA, Area(PAB) represents the area of ​​triangle PAB, and Area(ABC) represents the area of ​​triangle ABC, which can be calculated by vector cross multiplication;

[0037] Then the texture coordinates of point P (U P ,V P ) can be obtained by interpolation, the expression is:

[0038] U P =αUA +βU B +γU C

[0039] V P =αV A +βV+γV C

[0040] Where: (α, β, γ) represents the center coordinates of point P, (U A ,V A )、(U B ,V B ) and (U C ,V C ) are represented as texture coordinates of vertices A, B, and C;

[0041] The texture coordinate distribution is optimized by Laplace parameterization to obtain the vertex texture coordinates.

[0042] Furthermore, in step S6, the specific steps of stitching multiple texture images into a complete texture map based on the texture coordinates assigned to each vertex by using the SIFT algorithm include:

[0043] The key feature points of each texture image are detected by SIFT algorithm, and its feature descriptor is calculated;

[0044] By comparing the feature point descriptors between different images, matching feature point pairs are found;

[0045] Calculate the homography matrix based on the matched feature point pairs through the random sampling consistency algorithm;

[0046] Through the homography matrix, all texture images are transformed into a common coordinate system, and the stitching and fusion of multiple texture images are completed based on the texture coordinates.

[0047] Furthermore, in step S6, the expression of the bilateral filtering image processing algorithm is:

[0048]

[0049] Where: W p is represented as a normalization factor, Ω is represented as a neighborhood centered on pixel x, Expressed as a spatial Gaussian function, It is represented as a range Gaussian function, and I(x) is represented as the grayscale value of pixel x.

[0050] The color restoration device in the three-dimensional digitization of cultural relics comprises:

[0051] Point cloud data acquisition module, which collects 3D point cloud data of cultural relics through 3D laser scanning equipment;

[0052] Point cloud data preprocessing module, which preprocesses the collected 3D point cloud data of cultural relics;

[0053] The cultural relic 3D model building module extracts the feature points of the pre-processed cultural relic 3D point cloud data based on the adaptive local surface change factor, describes the extracted feature points through the FPFH algorithm, performs coarse registration of the cultural relic 3D point cloud data through the random sampling consistency algorithm, performs fine registration of the cultural relic 3D point cloud data through the ICP algorithm, gradually fuses the cultural relic 3D point cloud data from coarse to fine through the multi-scale transformation algorithm and converts it into a triangular mesh surface model, optimizes the triangular mesh surface model, and obtains the cultural relic 3D model;

[0054] Hyperspectral imaging data acquisition module, which collects hyperspectral imaging data of cultural relics through a hyperspectral imager;

[0055] Hyperspectral imaging data preprocessing module, which preprocesses the collected hyperspectral imaging data;

[0056] Texture mapping module, which maps the pre-processed hyperspectral imaging data texture onto the constructed three-dimensional model of cultural relics;

[0057] The model display module outputs the three-dimensional model mapped with the texture of the cultural relics to the platform for visual display.

[0058] Compared with the prior art, the present invention has the following beneficial effects:

[0059] The present invention obtains three-dimensional point cloud data of cultural relics based on a three-dimensional laser scanning device, constructs a three-dimensional model of the cultural relics based on the three-dimensional point cloud data of the cultural relics, obtains hyperspectral imaging data of the cultural relics based on a hyperspectral imager, and maps the texture of the hyperspectral imaging data of the cultural relics onto the three-dimensional model of the cultural relics for visual display. Compared with the existing technology, the present invention can accurately restore the color and texture of the cultural relics themselves. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 is a flow chart of the method of the present invention;

[0061] Figure 2 It is a system framework diagram of the present invention;

[0062] In the figure: 1. Point cloud data acquisition module; 2. Point cloud data preprocessing module; 3. Cultural relics three-dimensional model construction module; 4. Hyperspectral imaging data acquisition module; 5. Hyperspectral imaging data preprocessing module; 6. Texture mapping module; 7. Model display module. DETAILED DESCRIPTION

[0063] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0064] Embodiment 1

[0065] See attached Figure 1 The present invention provides the following technical solution: A color restoration method in three-dimensional digitization of cultural relics, comprising the following steps:

[0066] S1: Collect 3D point cloud data of cultural relics through 3D laser scanning equipment;

[0067] S2: Preprocessing the collected 3D point cloud data of cultural relics;

[0068] Preprocessing includes noise removal;

[0069] Noise removal refers to eliminating, weakening or suppressing random changes in images. Interference is usually caused by measurement errors and some uncertain random factors, collectively referred to as noise.

[0070] S3: Extract feature points of the preprocessed 3D point cloud data of cultural relics based on the adaptive local surface variation factor, describe the extracted feature points through the FPFH algorithm, roughly align the 3D point cloud data of cultural relics through the random sampling consistency algorithm, and finely align the 3D point cloud data of cultural relics through the ICP algorithm. Use the multi-scale transformation algorithm to gradually fuse the 3D point cloud data of cultural relics from coarse to fine and convert it into a triangular mesh surface model, optimize the triangular mesh surface model, and obtain the 3D model of cultural relics;

[0071] S4: Collecting hyperspectral imaging data of cultural relics through hyperspectral imager;

[0072] S5: preprocessing the collected hyperspectral imaging data;

[0073] Preprocessing includes noise removal, smoothing, image denoising, radiation correction and geometric correction;

[0074] Noise removal refers to eliminating, weakening or suppressing random changes in images. Interference is usually caused by measurement errors and some uncertain random factors, collectively referred to as noise.

[0075] Smoothing refers to fitting the original data and generating smoother data;

[0076] Image denoising refers to the use of algorithms to remove noise from data to improve quality;

[0077] Radiation correction refers to the correction of systematic and random radiation distortion or aberration caused by external factors, data acquisition and transmission systems, aiming to eliminate or correct the image distortion caused by radiation error;

[0078] Geometric correction refers to the process of eliminating or correcting image geometric errors;

[0079] S6: Texture mapping of the pre-processed hyperspectral imaging data onto the constructed three-dimensional model of the cultural relics is performed and visualized.

[0080] Specifically, in step S3, the specific steps of extracting feature points of the pre-processed cultural relic three-dimensional point cloud data based on the adaptive local surface change factor include:

[0081] Assume that any point P of the 3D point cloud data of the cultural relic i The covariance matrix C of the field is expressed as:

[0082]

[0083] Where: k represents the total number of points in the three-dimensional point cloud data of cultural relics, P ij Represented as the coordinate vector of the point cloud,

[0084] is represented as the centroid of the point cloud, It is represented as a decreasing function, and r is represented as point P i Field radius;

[0085] Based on the covariance matrix C, the eigenvalues ​​λ1, λ2 and λ3 are obtained;

[0086] Assuming λ1<λ2<λ3, then point P i The variation factor of the surface in the neighborhood σ(P i ) is:

[0087]

[0088] Where: λ1, λ2 and λ3 are represented as eigenvalues;

[0089] Select a reasonable threshold ε and keep σ(P i )>ε is the extracted feature point, where the expression of the threshold ε is:

[0090]

[0091] Where: k represents the total number of points in the three-dimensional point cloud data of cultural relics, σ(P i ) is represented by point P i The variation factor of the surface within the neighborhood.

[0092] Specifically, in step S6, the specific steps of texture mapping the pre-processed hyperspectral imaging data onto the constructed three-dimensional model of cultural relics include:

[0093] By using the barycentric coordinate method, the corresponding texture coordinates are assigned to each vertex in the three-dimensional model of the cultural relic;

[0094] Through the SIFT algorithm, multiple texture images are stitched into a complete texture map based on the texture coordinates assigned to each vertex;

[0095] Perform color correction on the spliced ​​texture image through a color space conversion algorithm;

[0096] The stitched texture image is filtered and denoised through the bilateral filtering image processing algorithm to improve the smoothness of the texture image while retaining its edge features, thereby optimizing the expression of surface details of the cultural relics.

[0097] Specifically, in step S6, the specific steps of assigning corresponding texture coordinates to each vertex in the three-dimensional model of the cultural relic by the barycentric coordinate method include:

[0098] Assume that the texture coordinates corresponding to vertices A, B, and C are (U A ,V A )、(U B ,V B ) and (U C ,V C );

[0099] For any point P in the triangle, its centroid coordinates are (α, β, γ), and the calculation expression is:

[0100]

[0101]

[0102] Where: Area(PBC) represents the area of ​​triangle PBC, Area(PCA) represents the area of ​​triangle PCA, Area(PAB) represents the area of ​​triangle PAB, and Area(ABC) represents the area of ​​triangle ABC, which can be calculated by vector cross multiplication;

[0103] Then the texture coordinates of point P (U P ,V P ) can be obtained by interpolation, the expression is:

[0104] U P =αU A +βU B +γU C

[0105] V P =αV A+βV+γV C

[0106] Where: (α, β, γ) represents the center coordinates of point P, (U A ,V A )、(U B ,V B ) and (U C ,V C ) are represented as texture coordinates of vertices A, B, and C;

[0107] The texture coordinate distribution is optimized by Laplace parameterization to obtain the vertex texture coordinates.

[0108] Specifically, in step S6, the specific steps of stitching multiple texture images into a complete texture map based on the texture coordinates assigned to each vertex by using the SIFT algorithm include:

[0109] The key feature points of each texture image are detected by SIFT algorithm, and its feature descriptor is calculated;

[0110] By comparing the feature point descriptors between different images, matching feature point pairs are found;

[0111] Calculate the homography matrix based on the matched feature point pairs through the random sampling consistency algorithm;

[0112] Through the homography matrix, all texture images are transformed into a common coordinate system, and the stitching and fusion of multiple texture images are completed based on the texture coordinates.

[0113] Specifically, in step S6, the expression of the bilateral filtering image processing algorithm is:

[0114]

[0115] Where: W p is represented as a normalization factor, Ω is represented as a neighborhood centered on pixel x, G σs Expressed as a spatial Gaussian function, It is represented as a range Gaussian function, and I(x) is represented as the grayscale value of pixel x.

[0116] Embodiment 2

[0117] See attached Figure 2 , a color restoration device for three-dimensional digitization of cultural relics, comprising:

[0118] Point cloud data acquisition module 1, collects three-dimensional point cloud data of cultural relics through three-dimensional laser scanning equipment;

[0119] Point cloud data preprocessing module 2, preprocessing the collected three-dimensional point cloud data of cultural relics;

[0120] The cultural relic 3D model construction module 3 extracts the feature points of the pre-processed cultural relic 3D point cloud data based on the adaptive local surface change factor, describes the extracted feature points through the FPFH algorithm, performs coarse registration of the cultural relic 3D point cloud data through the random sampling consistency algorithm, performs fine registration of the cultural relic 3D point cloud data through the ICP algorithm, gradually fuses the cultural relic 3D point cloud data from coarse to fine through the multi-scale transformation algorithm and converts it into a triangular mesh surface model, optimizes the triangular mesh surface model, and obtains the cultural relic 3D model;

[0121] A hyperspectral imaging data acquisition module 4 collects hyperspectral imaging data of cultural relics through a hyperspectral imager;

[0122] A hyperspectral imaging data preprocessing module 5 is used to preprocess the collected hyperspectral imaging data;

[0123] Texture mapping module 6, mapping the pre-processed hyperspectral imaging data texture onto the constructed three-dimensional model of cultural relics;

[0124] The model display module 7 outputs the three-dimensional model with the texture of the cultural relics mapped to the platform for visual display.

[0125] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A color restoration method in three-dimensional digitization of cultural relics, characterized in that: The following steps are involved: S1: Collect 3D point cloud data of cultural relics through 3D laser scanning equipment; S2: Preprocessing the collected 3D point cloud data of cultural relics; S3: Extract feature points of the preprocessed 3D point cloud data of cultural relics based on the adaptive local surface variation factor, describe the extracted feature points through the FPFH algorithm, coarsely align the 3D point cloud data of cultural relics through the random sampling consistency algorithm, finely align the 3D point cloud data of cultural relics through the ICP algorithm, gradually fuse the 3D point cloud data of cultural relics from coarse to fine through the multi-scale transformation algorithm and convert it into a triangular mesh surface model, optimize the triangular mesh surface model, and obtain the 3D model of cultural relics; S4: Collecting hyperspectral imaging data of cultural relics through hyperspectral imager; S5: preprocessing the collected hyperspectral imaging data; S6: Texture mapping of the pre-processed hyperspectral imaging data onto the constructed three-dimensional model of the cultural relics is performed and visualized.

2. The color restoration method in three-dimensional digitization of cultural relics according to claim 1, characterized in that: In step S3, the specific steps of extracting feature points of the pre-processed cultural relic three-dimensional point cloud data based on the adaptive local surface change factor include: Assume that any point P of the 3D point cloud data of the cultural relic i The covariance matrix C of the field is expressed as: Where: k represents the total number of points in the three-dimensional point cloud data of cultural relics, P ij Represented as the coordinate vector of the point cloud, is represented as the centroid of the point cloud, It is represented as a decreasing function, and r is represented as point P i Field radius; Based on the covariance matrix C, the eigenvalues ​​λ1, λ2 and λ3 are obtained; Assuming λ1<λ2<λ3, then point P i The variation factor of the surface in the neighborhood σ(P i ) is: Where: λ1, λ2 and λ3 are represented as eigenvalues; Select a reasonable threshold ε and keep σ(P i )>ε is the extracted feature point, where the expression of the threshold ε is: Where: k represents the total number of points in the three-dimensional point cloud data of cultural relics, σ(P i ) is represented by point P i The variation factor of the surface within the neighborhood.

3. The color restoration method in three-dimensional digitization of cultural relics according to claim 1, characterized in that: In step S6, the specific steps of texture mapping the pre-processed hyperspectral imaging data onto the constructed three-dimensional model of cultural relics include: By using the barycentric coordinate method, the corresponding texture coordinates are assigned to each vertex in the three-dimensional model of the cultural relic; Through the SIFT algorithm, multiple texture images are stitched into a complete texture map based on the texture coordinates assigned to each vertex; Perform color correction on the spliced ​​texture image through a color space conversion algorithm; The stitched texture image is filtered and denoised through the bilateral filtering image processing algorithm to improve the smoothness of the texture image while retaining its edge features, thereby optimizing the expression of surface details of the cultural relics.

4. The color restoration method in three-dimensional digitization of cultural relics according to claim 3, characterized in that: In step S6, the specific steps of assigning corresponding texture coordinates to each vertex in the three-dimensional model of the cultural relic by the barycentric coordinate method include: Assume that the texture coordinates corresponding to vertices A, B, and C are (U A ,V A )、(U B ,V B ) and (U C ,V C ); For any point P in the triangle, its centroid coordinates are (α, β, γ), and the calculation expression is: Where: Area(PBC) represents the area of ​​triangle PBC, Area(PCA) represents the area of ​​triangle PCA, Area(PAB) represents the area of ​​triangle PAB, and Area(ABC) represents the area of ​​triangle ABC, which can be calculated by vector cross multiplication; Then the texture coordinates of point P (U P ,V P ) can be obtained by interpolation, the expression is: U P =αU A +βU B +γU C V P =αV A +βV+γV C Where: (α, β, γ) represents the center coordinates of point P, (U A ,V A )、(U B ,V B ) and (U C ,V C ) are represented as texture coordinates of vertices A, B, and C; The texture coordinate distribution is optimized by Laplace parameterization to obtain the vertex texture coordinates.

5. The color restoration method in three-dimensional digitization of cultural relics according to claim 3, characterized in that: In step S6, the specific steps of stitching multiple texture images into a complete texture map based on the texture coordinates assigned to each vertex by using the SIFT algorithm include: The key feature points of each texture image are detected by SIFT algorithm, and its feature descriptor is calculated; By comparing the feature point descriptors between different images, matching feature point pairs are found; Calculate the homography matrix based on the matched feature point pairs through the random sampling consistency algorithm; Through the homography matrix, all texture images are transformed into a common coordinate system, and the stitching and fusion of multiple texture images are completed based on the texture coordinates.

6. The color restoration method in three-dimensional digitization of cultural relics according to claim 3, characterized in that: In step S6, the expression of the bilateral filtering image processing algorithm is: Where: W p is represented as a normalization factor, Ω is represented as a neighborhood centered on pixel x, Expressed as a spatial Gaussian function, It is represented as a range Gaussian function, and I(x) is represented as the grayscale value of pixel x.

7. A color restoration device for three-dimensional digitization of cultural relics according to any one of claims 1 to 6, characterized in that: include: A point cloud data acquisition module (1) collects three-dimensional point cloud data of cultural relics through a three-dimensional laser scanning device; Point cloud data preprocessing module (2), preprocessing the collected three-dimensional point cloud data of cultural relics; The cultural relic three-dimensional model building module (3) extracts feature points of the pre-processed cultural relic three-dimensional point cloud data based on the adaptive local surface change factor, describes the extracted feature points through the FPFH algorithm, performs coarse alignment of the cultural relic three-dimensional point cloud data through the random sampling consistency algorithm, performs fine alignment of the cultural relic three-dimensional point cloud data through the ICP algorithm, gradually fuses the cultural relic three-dimensional point cloud data from coarse to fine through the multi-scale transformation algorithm and converts it into a triangular mesh surface model, optimizes the triangular mesh surface model, and obtains the cultural relic three-dimensional model; A hyperspectral imaging data acquisition module (4) collects hyperspectral imaging data of cultural relics through a hyperspectral imager; A hyperspectral imaging data preprocessing module (5) for preprocessing the collected hyperspectral imaging data; A texture mapping module (6) maps the pre-processed hyperspectral imaging data texture onto the constructed three-dimensional model of the cultural relic; The model display module (7) outputs the three-dimensional model with the texture of the cultural relics mapped to the platform for visual display.

Images