Cultural relic hyperspectral image reconstruction method and device, hyperspectral area array camera and medium

Abstract

The application provides a cultural relic hyperspectral image reconstruction method and device, a hyperspectral area array camera and a medium, relates to the technical field of hyperspectral imaging and cultural relic digital protection, and collects the RGB image of the surface of the cultural relic under the illumination of uniform and continuous spectrum; a plurality of sampling points are selected in the RGB image, and the hyperspectral data at the plurality of sampling points is collected; the spatial registration of the RGB image and the hyperspectral data is carried out, and the corresponding relationship between the pixel coordinates and the spectral coordinates is established; the spatial information of the RGB image and the hyperspectral data are fused according to the corresponding relationship between the pixel coordinates and the spectral coordinates; the fusion result is combined with the prior constraint of the preset cultural relic pigment standard spectral database to reconstruct the complete hyperspectral image of the surface of the cultural relic, complete hyperspectral area array images are quickly generated under the condition of a small amount of sampling, have the advantages of lightweight equipment, efficient collection and reliable results, and can be widely applied to cultural relic protection scenes such as murals, calligraphy and painting, ceramics and textiles.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of hyperspectral imaging and digital protection of cultural relics, and particularly relates to a hyperspectral image reconstruction method and device for cultural relics based on a hyperspectral area array camera, a hyperspectral area array camera and a medium. BACKGROUND

[0002] As a non-destructive testing method combining imaging technology and spectral technology, hyperspectral imaging technology has shown great application potential in the field of cultural relic protection and scientific research. It can simultaneously obtain spatial information and continuous spectral information of the surface of cultural relics, thereby realizing fine analysis of cultural relics such as pigment type identification, aging condition evaluation, repair trace detection, and authenticity identification. The traditional hyperspectral imaging technology applied in the field of cultural relic protection is mainly based on scanning principle imaging hyperspectral camera. This kind of equipment constructs a spectral data cube through line array or point array scanning method, and although it can obtain spatial distribution information with high spectral resolution, the inherent scanning mechanism leads to low data acquisition efficiency, and it takes a long time to complete a complete image. In addition, such systems are usually large in size, complex in structure, high in cost, and extremely demanding on environmental vibration and light source stability. These shortcomings make them difficult to be widely used in cultural relic protection scenes such as museums and archaeological sites that require rapid and flexible operation. In related technologies, a point spectrometer is used for hyperspectral imaging of cultural relics. This kind of equipment collects reflectance spectral data at a specific point through an optical fiber probe, and has the advantages of portability, flexibility and high spectral resolution. However, its essential defect is that it can only obtain spectral information of discrete points, and lacks complete spatial information. SUMMARY

[0003] The present application provides a cultural relic hyperspectral image reconstruction method, device, hyperspectral area array camera and medium to solve the defects of traditional hyperspectral imaging equipment that takes a long time to complete imaging, or can only obtain spectral information of discrete points and lacks complete spatial information.

[0004] The present application provides a cultural relic hyperspectral image reconstruction method, comprising:

[0005] Under uniform continuous spectral illumination, an RGB image of the surface of a cultural relic is collected;

[0006] A plurality of sampling points are selected in the RGB image, and hyperspectral data at the plurality of sampling points is collected;

[0007] The RGB image and the hyperspectral data are spatially registered to establish a corresponding relationship between pixel coordinates and spectral coordinates;

[0008] Fusing the spatial information of the RGB image and the hyperspectral data according to the correspondence between the pixel coordinates and the spectral coordinates, and reconstructing a complete hyperspectral image of the cultural relic surface by combining a priori constraint of a preset cultural relic pigment standard spectrum database.

[0009] The method for reconstructing a hyperspectral image of a cultural relic according to the application further comprises the following steps after the complete hyperspectral image of the cultural relic surface is reconstructed:

[0010] Matching the spectral curve of each pixel in the complete hyperspectral image with the preset cultural relic pigment standard spectrum database, and outputting a pigment category and distribution map of the cultural relic surface.

[0011] The method for reconstructing a hyperspectral image of a cultural relic according to the application, the spatial registration of the RGB image and the hyperspectral data and the establishment of the correspondence between the pixel coordinates and the spectral coordinates comprise the following steps:

[0012] Obtaining initial pixel coordinates of the sampling point on the RGB image;

[0013] Finding an extreme point in a continuous space by fitting discrete pixel similarity values within a preset coordinate range of the initial pixel coordinates, and taking the coordinates of the extreme point as accurate pixel coordinates of the sampling point;

[0014] Establishing the correspondence between the pixel coordinates and the spectral coordinates based on the accurate pixel coordinates and the hyperspectral data.

[0015] The method for reconstructing a hyperspectral image of a cultural relic according to the application, the fusing of the spatial information of the RGB image and the hyperspectral data according to the correspondence between the pixel coordinates and the spectral coordinates comprises the following steps:

[0016] Extracting multi-scale spatial features of the RGB image to obtain a spatial feature map;

[0017] Embedding the hyperspectral data as a high-precision spectral supervision signal into a corresponding coordinate position in the spatial feature map according to the correspondence between the pixel coordinates and the spectral coordinates, to generate a spatial feature map with a spectral supervision signal;

[0018] Constructing a fusion network, inputting the spatial feature map with the spectral supervision signal into the fusion network, and outputting a fusion result.

[0019] The method for reconstructing a hyperspectral image of a cultural relic according to the application, the fusion result comprising initial estimated spectra of each pixel in the RGB image, and the reconstruction of the complete hyperspectral image of the cultural relic surface by combining a priori constraint of a preset cultural relic pigment standard spectrum database comprises the following steps:

[0020] linear combination constraint fitting is performed on the initial estimated spectrum of each pixel in the RGB image under the prior constraint of the preset cultural relic pigment standard spectrum database;

[0021] optimizing the result of the linear combination constraint fitting based on a joint optimization function to obtain a complete hyperspectral image of the cultural relic surface.

[0022] According to the cultural relic hyperspectral image reconstruction method provided by the application, the joint optimization function comprises a first loss term based on hyperspectral data supervision, a second loss term based on spectral library prior constraint and a third loss term based on spatial smoothing regularization.

[0023] According to the cultural relic hyperspectral image reconstruction method provided by the application, the fusion network is a deep learning model, and the deep learning model is trained based on an RGB image and a point spectrum supervision signal.

[0024] The application provides a cultural relic hyperspectral image reconstruction device, comprising:

[0025] A first acquisition module is configured to acquire an RGB image of a cultural relic surface under uniform continuous spectrum illumination.

[0026] A second acquisition module is configured to select a plurality of sampling points in the RGB image and acquire hyperspectral data at the sampling points.

[0027] A building module is configured to perform spatial registration on the RGB image and the hyperspectral data to establish a corresponding relationship between pixel coordinates and spectral coordinates.

[0028] A reconstruction module is configured to fuse spatial information of the RGB image and the hyperspectral data according to the corresponding relationship between the pixel coordinates and the spectral coordinates, and reconstruct a complete hyperspectral image of the cultural relic surface by combining a fusion result with prior constraints of a preset cultural relic pigment standard spectrum database.

[0029] The application further provides a hyperspectral area array camera, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, and the processor implements the cultural relic hyperspectral image reconstruction method according to any one of the above.

[0030] The application further provides a non-transitory computer readable storage medium having a computer program stored thereon, and the computer program is executed by a processor to implement the cultural relic hyperspectral image reconstruction method according to any one of the above.

[0031] The application provides a cultural relic hyperspectral image reconstruction method and device, a hyperspectral area array camera and a medium. BRIEF DESCRIPTION OF DRAWINGS

[0032] In order to more clearly illustrate the technical solutions in the present application or prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort on the basis of these drawings.

[0033] Figure 1 FIG. 1 is a flowchart of a cultural relic hyperspectral image reconstruction method provided by an embodiment of the present application;

[0034] Figure 2 FIG. 2 is a functional structure diagram of a cultural relic hyperspectral image reconstruction device provided by an embodiment of the present application;

[0035] Figure 3 FIG. 3 is a functional structure diagram of a hyperspectral area array camera provided by an embodiment of the present application. DETAILED DESCRIPTION

[0036] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be described clearly and completely below with reference to the drawings in the present application. Obviously, the described embodiments are some embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without any creative effort belong to the protection scope of the present application.

[0037] Figure 1 FIG. 4 is a functional structure diagram of a cultural relic hyperspectral image reconstruction method provided by an embodiment of the present application, as shown in FIG. 4, the cultural relic hyperspectral image reconstruction method provided by the embodiment of the present application comprises: Figure 1

[0038] ​Step 101, under uniform continuous spectrum illumination, collect the RGB image of the surface of the cultural relic;

[0039] In the embodiment of the present application, the halogen light source module is started to provide uniform continuous spectrum illumination, and the RGB image of the surface of the cultural relic is obtained.

[0040] Step 102, select a plurality of sampling points in the RGB image, and collect hyperspectral data at the plurality of sampling points;

[0041] Step 103, spatially register the RGB image and the hyperspectral data to establish the correspondence between the pixel coordinates and the spectral coordinates;

[0042] In the embodiment of the present application, the point spectrum sampling position and the RGB image are geometrically registered to establish spatial coordinate consistency, so as to ensure one-to-one correspondence between the spectral value and the image pixel.

[0043] Step 104, according to the correspondence between the pixel coordinates and the spectral coordinates, fuse the spatial information of the RGB image and the hyperspectral data, and combine the fusion result with the prior constraint of the preset cultural relic pigment standard spectrum database to reconstruct the complete hyperspectral image of the surface of the cultural relic.

[0044] The traditional cultural relic hyperspectral image reconstruction method is based on the imaging hyperspectral equipment based on the scanning principle. This kind of equipment constructs a spectral data cube through line array or point array scanning mode, and although it can obtain spatial distribution information with high spectral resolution, the inherent scanning mechanism leads to low data acquisition efficiency, and it takes a long time to complete a complete image. In addition, such systems are usually large in size, complex in structure, high in cost, and extremely demanding on environmental vibration and light source stability. These shortcomings make them difficult to be widely used in cultural relic protection scenes such as museums and archaeological sites which require rapid and flexible operation. In the related art, a point spectrometer is used for cultural relic hyperspectral imaging. This kind of equipment collects reflected spectrum data at a specific point through an optical fiber probe, and has the advantages of portability, flexibility and high spectral resolution. However, its essential defect is that it can only obtain spectral information of discrete points, and lacks complete spatial information.

[0045] The cultural relic hyperspectral image reconstruction method provided in the embodiments of the present application comprises the following steps: collecting an RGB image of a cultural relic surface under uniform continuous spectrum illumination; selecting a plurality of sampling points in the RGB image and collecting hyperspectral data at the plurality of sampling points; performing spatial registration on the RGB image and the hyperspectral data to establish a corresponding relationship between pixel coordinates and spectral coordinates; fusing spatial information of the RGB image and the hyperspectral data according to the corresponding relationship between the pixel coordinates and the spectral coordinates; and reconstructing a complete hyperspectral image of the cultural relic surface by combining a fusion result with a prior constraint of a preset cultural relic pigment standard spectrum database. The complete hyperspectral image is quickly generated under the condition of a small amount of nondestructive sampling, and has the advantages of lightweight equipment, efficient collection and reliable results, and can be widely applied to cultural relic protection scenes such as murals, paintings, ceramics and textiles.

[0046] Based on any of the above embodiments, after the complete hyperspectral image of the cultural relic surface is reconstructed, the method further comprises:

[0047] Matching the spectral curve of each pixel in the complete hyperspectral image with the preset cultural relic pigment standard spectrum database to output a pigment category and a distribution map of the cultural relic surface.

[0048] In the embodiments of the present application, the pixel-level reconstructed spectrum is matched with the standard spectrum library to output the pigment category and the distribution map, and the standard spectrum library is used to complete automatic pigment identification, aging analysis and repair evaluation.

[0049] Based on any of the above embodiments, the spatial registration of the RGB image and the hyperspectral data to establish the corresponding relationship between the pixel coordinates and the spectral coordinates comprises:

[0050] Step 201: obtaining initial pixel coordinates of the sampling points on the RGB image;

[0051] Step 202: within a preset coordinate range of the initial pixel coordinates, an extreme point in a continuous space is found by fitting discrete pixel similarity values, and the coordinate of the extreme point is taken as the accurate pixel coordinates of the sampling points.

[0052] Step 203: establishing the corresponding relationship between the pixel coordinates and the spectral coordinates based on the accurate pixel coordinates and the hyperspectral data.

[0053] The embodiment of the application greatly compresses the calculation range by acquiring initial pixel coordinates and limiting search within a preset coordinate range, avoids huge calculation overhead brought by global search, and thus significantly improves registration efficiency, and is especially suitable for efficient processing of high-resolution cultural relic images. The sub-pixel level registration method based on fitting discrete pixel similarity values to find continuous spatial extreme points breaks through the precision limit of traditional integer pixel coordinates, and can acquire accurate coordinate positions up to sub-pixel level. This improvement enables the spatial correspondence between hyperspectral data and RGB image pixels to reach unprecedented accuracy, and fundamentally eliminates the spectral spatial misplacement problem caused by registration error. The spectral coordinate correspondence established based on the accurate pixel coordinates lays a reliable spatial geometric foundation for subsequent multi-source data fusion and high-precision spectral reconstruction, and ensures that each spectral data in the fusion process can be accurately corresponded to the correct physical position on the cultural relic surface, thereby improving the geometric fidelity of hyperspectral image reconstruction as a whole.

[0054] According to any one of the above embodiments, the fusion of the spatial information of the RGB image and the hyperspectral data according to the correspondence between the pixel coordinates and the spectral coordinates comprises:

[0055] Step 301, extracting multi-scale spatial features of the RGB image to obtain a spatial feature map;

[0056] Step 302, embedding the hyperspectral data as high-precision spectral supervision signals into the corresponding coordinate positions in the spatial feature map according to the correspondence between the pixel coordinates and the spectral coordinates, to generate a spatial feature map with spectral supervision signals;

[0057] Step 303, constructing a fusion network, inputting the spatial feature map with spectral supervision signals into the fusion network, and outputting a fusion result.

[0058] In the embodiment of the application, the fusion network is a deep learning model, which is obtained by training based on an RGB image and point spectral supervision signals.

[0059] In the embodiment of the application, a deep network Predicting a preliminary reconstruction result of the hyperspectral image:

[0060]

[0061] wherein, is the preliminary reconstruction result of the hyperspectral image, RGB image, H is the height of the hyperspectral image, W is the width of the hyperspectral image, and 3 is the channel number of the hyperspectral image; point hyperspectral data, B is the number of spectral bands, and there are N sampling points, is a pixel point, is a wavelength.

[0062] Based on any of the above embodiments, the fusion result includes an initial estimated spectrum of each pixel in the RGB image, and the combination of the fusion result and the prior constraint of the preset cultural relic pigment standard spectrum database reconstructs a complete hyperspectral image of the cultural relic surface, including:

[0063] Step 401, linear combination constraint fitting is performed on the initial estimated spectrum of each pixel in the RGB image by using the prior constraint of the preset cultural relic pigment standard spectrum database;

[0064] Step 402, the result after the linear combination constraint fitting is optimized based on a joint optimization function, to obtain a complete hyperspectral image of the cultural relic surface.

[0065] In the embodiment of the present application, the initial estimated spectrum of each pixel in the RGB image is linearly combined with the prior constraint of the preset cultural relic pigment standard spectrum database. Further fitting spectrum library linear combination:

[0066] Wherein

[0067] Wherein, is a combination coefficient, is a prior standard spectrum.

[0068] In the embodiment of the present application, the joint optimization function includes a first loss term based on hyperspectral data supervision, a second loss term based on spectral library prior constraint, and a third loss term based on spatial smoothing regularization.

[0069] Let the joint optimization function be:

[0070]

[0071] Wherein: is the first loss term based on hyperspectral data supervision: point spectrum supervision, at the point spectrum, the predicted value is required to be consistent with the true point spectrum;

[0072] is the second loss term based on spectral library prior constraint: spectral library constraint, the whole image spectrum should be close to the combination fitting of the spectral library;

[0073] is the third loss term based on spatial smoothing regularization: spatial smoothing / edge preserving regularization term, guiding the image reconstruction to have spatial continuity or edge clarity;

[0074] is the true point spectrum, and is an optimization coefficient, Spatial smoothing / edge-preserving regularization term.

[0075] The artifact hyperspectral image reconstruction method provided by the embodiment of the present application uses the artifact pigment standard spectrum database to perform linear combination constraint fitting on the initial estimated spectrum, deeply embeds physical priori knowledge into the calculation framework, and forces the reconstructed spectrum to conform to the spectral morphological characteristics of the real pigment, fundamentally avoiding physically unreasonable spectral reconstruction results, and significantly improving the interpretability and reliability of the reconstruction results; the joint optimization function fusing multiple constraint factors is adopted, the first loss term is used to ensure that the reconstruction results at the point spectrum sampling position are strictly consistent with the measured high-precision data, the second loss term is used to ensure that the spectral morphology of the whole image conforms to the physical priori of the pigment library, and the third loss term is used to introduce spatial context constraints to suppress noise and maintain the material boundary definition, realizing the collaborative optimization of data driving and knowledge driving, effectively solving the overfitting or non-unique solution problem caused by a single supervision signal, providing a solid data foundation for subsequent pigment identification, aging evaluation and other quantitative analysis through the unique RGB+point spectrum fusion reconstruction and standard spectrum priori constraint, and significantly improving the scientificity and work efficiency of artifact pigment identification and aging evaluation.

[0076] The artifact hyperspectral image reconstruction device provided by the present application is described below, and the artifact hyperspectral image reconstruction device described below can be correspondingly referred to the artifact hyperspectral image reconstruction method described above.

[0077] Figure 2 The functional structure diagram of the artifact hyperspectral image reconstruction device provided by the embodiment of the present application is shown in Figure 2 The artifact hyperspectral image reconstruction device provided by the embodiment of the present application comprises:

[0078] The first acquisition module 201 is used to acquire the RGB image of the artifact surface under uniform continuous spectrum illumination.

[0079] The second acquisition module 202 is used to select a plurality of sampling points in the RGB image and acquire hyperspectral data at the sampling points.

[0080] The establishment module 203 is used to perform spatial registration on the RGB image and the hyperspectral data, and establish the corresponding relationship between the pixel coordinates and the spectral coordinates.

[0081] The reconstruction module 204 is used to fuse the spatial information of the RGB image and the hyperspectral data according to the corresponding relationship between the pixel coordinates and the spectral coordinates, and reconstruct the complete hyperspectral image of the artifact surface by combining the fusion result with the priori constraint of the preset artifact pigment standard spectrum database.

[0082] The cultural relic hyperspectral image reconstruction device provided by the embodiment of the present application can quickly generate a complete hyperspectral area array image under the condition of a small amount of nondestructive sampling, has the advantages of lightweight equipment, efficient acquisition and reliable results, and can be widely applied to cultural relic protection scenes such as murals, paintings, ceramics and textiles.

[0083] Figure 3 An example of a schematic diagram of the physical structure of a hyperspectral area array camera is shown in FIG. 1, which can include a processor 310, a communication interface 320, a memory 330 and a communication bus 340, wherein the processor 310, the communication interface 320 and the memory 330 can communicate with each other through the communication bus 340. The memory 330 includes a computer program, an operating system and acquired data, and the processor 310 can call the logical instructions in the memory 330 to execute the cultural relic hyperspectral image reconstruction method, which includes the following steps: acquiring an RGB image of a cultural relic surface under uniform continuous spectrum illumination; selecting a plurality of sampling points in the RGB image and acquiring hyperspectral data at the plurality of sampling points; performing spatial registration on the RGB image and the hyperspectral data to establish a correspondence between pixel coordinates and spectral coordinates; and fusing the spatial information of the RGB image and the hyperspectral data according to the correspondence between the pixel coordinates and the spectral coordinates, and reconstructing a complete hyperspectral image of the cultural relic surface by combining the fusion result with a prior constraint of a preset cultural relic pigment standard spectrum database. Figure 3

[0084] In addition, the logical instructions in the memory 330 described above can be implemented in the form of a software functional unit and sold or used as an independent product, and can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the part that contributes to the related art can be embodied in the form of a software product, and the computer software product is stored in a storage medium.

[0085] ​In another aspect, the present application also provides a non-transitory computer readable storage medium having stored thereon a computer program, which, when executed by a processor, implements a cultural relic hyperspectral image reconstruction method provided by each of the above methods, and the method comprises: collecting an RGB image of a surface of a cultural relic under uniform continuous spectral illumination; selecting a plurality of sampling points in the RGB image, and collecting hyperspectral data at the plurality of sampling points; performing spatial registration on the RGB image and the hyperspectral data to establish a corresponding relationship between pixel coordinates and spectral coordinates; fusing spatial information of the RGB image and the hyperspectral data according to the corresponding relationship between the pixel coordinates and the spectral coordinates; and reconstructing a complete hyperspectral image of the surface of the cultural relic by combining a fusion result with a priori constraint of a preset cultural relic pigment standard spectral database.

[0086] The device embodiments described above are merely illustrative, wherein the units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiment scheme according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0087] From the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be realized by means of software and necessary general hardware platforms, and of course can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes a plurality of instructions to make a computer device (which can be a personal computer, server, or network device, etc.) execute the methods described in each embodiment or some parts of the embodiment.

[0088] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for reconstructing a cultural relic hyperspectral image, characterized in that, The method comprises the following steps: Under uniform continuous spectrum illumination, an RGB image of an artifact surface is collected; A plurality of sampling points are selected in the RGB image, and hyperspectral data at the plurality of sampling points is collected; The RGB image and the hyperspectral data are spatially registered to establish a correspondence between pixel coordinates and spectral coordinates; The spatial information of the RGB image and the hyperspectral data are fused according to the correspondence between the pixel coordinates and the spectral coordinates, and a complete hyperspectral image of the artifact surface is reconstructed by combining a priori constraints of a preset artifact pigment standard spectrum database; The spatial registration of the RGB image and the hyperspectral data to establish the correspondence between the pixel coordinates and the spectral coordinates comprises the following steps: An initial pixel coordinate of the sampling point on the RGB image is obtained; Within a preset coordinate range of the initial pixel coordinate, an extreme point in a continuous space is found by fitting discrete pixel similarity values, and the coordinate of the extreme point is taken as an accurate pixel coordinate of the sampling point; The correspondence between the pixel coordinates and the spectral coordinates is established based on the accurate pixel coordinate and the hyperspectral data; The fusion of the spatial information of the RGB image and the hyperspectral data according to the correspondence between the pixel coordinates and the spectral coordinates comprises the following steps: Multi-scale spatial features of the RGB image are extracted to obtain a spatial feature map; According to the correspondence between the pixel coordinates and the spectral coordinates, the hyperspectral data is embedded into a corresponding coordinate position in the spatial feature map as a high-precision spectral supervision signal to generate a spatial feature map with a spectral supervision signal; A fusion network is constructed, the spatial feature map with the spectral supervision signal is input into the fusion network, and a fusion result is output.

2. The cultural relic hyperspectral image reconstruction method according to claim 1, characterized in that, After the complete hyperspectral image of the artifact surface is reconstructed, the following steps are further included: The spectral curve of each pixel in the complete hyperspectral image is matched with the preset artifact pigment standard spectrum database, and a pigment category and a distribution map of the artifact surface are output.

3. The cultural relic hyperspectral image reconstruction method according to claim 1, characterized in that, The fusion result includes an initial estimated spectrum of each pixel in the RGB image, and the reconstruction of the complete hyperspectral image of the artifact surface from the fusion result combined with the a priori constraints of the preset artifact pigment standard spectrum database comprises the following steps: The initial estimated spectrum of each pixel in the RGB image is linearly combined and constrained fitted by using the a priori constraints of the preset artifact pigment standard spectrum database; The result after the linear combination and constrained fitting is optimized based on a joint optimization function to obtain the complete hyperspectral image of the artifact surface.

4. The cultural relic hyperspectral image reconstruction method according to claim 3, characterized in that, The joint optimization function comprises a first loss term based on hyperspectral data supervision, a second loss term based on spectral library a priori constraints, and a third loss term based on spatial smoothing regularization.

5. The cultural relic hyperspectral image reconstruction method according to claim 1, characterized in that, The fusion network is a deep learning model, which is trained based on an RGB image and a point spectrum supervision signal.

6. An apparatus for reconstructing a cultural relic hyperspectral image, adapted to the method for reconstructing a cultural relic hyperspectral image according to any one of claims 1-5, characterized in that, The method comprises the following steps: A first acquisition module is configured to collect an RGB image of an artifact surface under uniform continuous spectrum illumination; A second acquisition module is configured to select a plurality of sampling points in the RGB image and collect hyperspectral data at the sampling points; The establishing module is configured to perform spatial registration on the RGB image and the hyperspectral data, and establish a correspondence between pixel coordinates and spectral coordinates; The reconstructing module is configured to fuse spatial information of the RGB image and the hyperspectral data according to the correspondence between the pixel coordinates and the spectral coordinates, and reconstruct a complete hyperspectral image of an artifact surface by combining a fusion result with a priori constraint of a preset artifact pigment standard spectrum database.

7. A hyperspectral area array camera comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that, The processor implements the artifact hyperspectral image reconstruction method according to any one of claims 1 to 5 when executing the program.

8. A non-transitory readable storage medium having stored thereon a computer program, characterized in that, The computer program implements the artifact hyperspectral image reconstruction method according to any one of claims 1 to 5 when executed by the processor.

Images