A method and related device for extracting three-dimensional features of intangible cultural heritage physical products

Abstract

The present invention discloses a method and related device for extracting three-dimensional features of a non-legacy physical product, wherein the method comprises: performing image acquisition processing on the non-legacy physical product according to different preset angles to obtain a plurality of image data; inputting the plurality of image data into a convergent edge extraction model to perform edge extraction processing on the non-legacy physical product to obtain a plurality of edge extraction results; extracting target area images from the corresponding plurality of image data based on the plurality of edge extraction results to obtain a plurality of target area images; performing three-dimensional spatial feature extraction processing on the plurality of target area images to obtain three-dimensional spatial feature data of the non-legacy physical product at different angles; fusing the three-dimensional spatial feature data at different angles to form a three-dimensional feature model of the non-legacy physical product. In an embodiment of the present invention, the three-dimensional feature extraction of the non-legacy physical product can be realized, and a complete three-dimensional feature model can be formed.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to a method and related device for extracting three-dimensional features of an intangible cultural heritage physical product. Background Art

[0002] In some intangible cultural heritage physical products, in order to present the intangible cultural heritage physical products through various virtual interactive images, it is necessary to extract various three-dimensional feature models of the intangible cultural heritage physical products, so as to facilitate the subsequent virtual interactive fusion processing. The extraction of three-dimensional feature models in the prior art is generally achieved through 3D digital scanning technology. Although this method can quickly scan and obtain the three-dimensional feature model of the intangible cultural heritage physical product, it requires professional personnel to operate the 3D digital scanning equipment to perform the scanning operation, which cannot meet the needs of a wider range of users. Now there is an urgent need for a method that can realize the three-dimensional feature extraction of intangible cultural heritage physical products through simple operations and meet the requirements of accurately expressing the form and beauty of the intangible cultural heritage physical products. Summary of the invention

[0003] The purpose of the present invention is to overcome the shortcomings of the prior art. The present invention provides a method and related devices for extracting three-dimensional features of intangible cultural heritage physical products, which can realize the three-dimensional feature extraction of intangible cultural heritage physical products and form a complete three-dimensional feature model to accurately express the shape and beauty of the intangible cultural heritage physical products.

[0004] In order to solve the above technical problems, an embodiment of the present invention provides a method for extracting three-dimensional features of an intangible cultural heritage physical product, the method comprising:

[0005] Performing image acquisition and processing on the intangible cultural heritage physical product at different preset angles to obtain a plurality of image data of the intangible cultural heritage physical product at different angles;

[0006] Inputting a plurality of image data of the intangible cultural heritage physical product at different angles into a convergent edge extraction model to perform edge extraction processing of the intangible cultural heritage physical product, and obtaining a plurality of edge extraction results of the intangible cultural heritage physical product at different angles;

[0007] Extract target area images from corresponding image data based on several edge extraction results of the intangible cultural heritage physical product at different angles, and obtain several target area images of the intangible cultural heritage physical product at different angles;

[0008] Based on the image three-dimensional spatial feature extraction model, three-dimensional spatial feature extraction processing is performed on a plurality of target area images of the intangible cultural heritage physical product at different angles to obtain three-dimensional spatial feature data of the intangible cultural heritage physical product at different angles;

[0009] The three-dimensional spatial feature data at different angles are fused to form a three-dimensional feature model of the intangible cultural heritage physical product.

[0010] Optionally, the image acquisition and processing of the intangible cultural heritage physical product according to different preset angles to obtain a plurality of image data of the intangible cultural heritage physical product at different angles includes:

[0011] Placing the intangible cultural heritage physical product on an image acquisition platform, and controlling an image acquisition device disposed above the image acquisition platform to perform image acquisition processing on the intangible cultural heritage physical product at a current angle;

[0012] After completing the image acquisition at the current angle, controlling the intangible cultural heritage physical product to rotate at a preset angle on the image acquisition platform;

[0013] After the intangible cultural heritage physical product is rotated by a preset angle, the image acquisition device is controlled to perform image acquisition processing on the intangible cultural heritage physical product after the rotation angle, until the image acquisition processing at different angles is completed, and a plurality of image data of the intangible cultural heritage physical product at different angles are obtained; wherein the preset rotation angles are 90 degrees, 180 degrees and 270 degrees corresponding to the current angle placed under the image acquisition platform, respectively.

[0014] Optionally, the step of inputting a plurality of image data of the intangible cultural heritage physical product at different angles into a convergent edge extraction model to perform edge extraction processing on the intangible cultural heritage physical product, and obtaining a plurality of edge extraction results of the intangible cultural heritage physical product at different angles, comprises:

[0015] Preprocessing a plurality of image data of the intangible cultural heritage physical product at different angles to form a plurality of preprocessed image data;

[0016] The preprocessed image data are sequentially input into a convergent edge extraction model to perform edge extraction processing on the intangible cultural heritage physical product, and a plurality of edge extraction results of the intangible cultural heritage physical product at different angles are obtained; wherein the edge extraction model is a model which sequentially includes a plurality of Conv modules, a plurality of BottleneckCSP / C3 network modules, an SPPF module, a plurality of upsampling modules and a plurality of Concat modules.

[0017] Optionally, the preprocessing of a plurality of image data of the intangible cultural heritage physical product at different angles to form a plurality of preprocessed image data includes:

[0018] Based on the Gaussian filter, a plurality of image data of the intangible cultural heritage physical product at different angles are smoothed, and the plurality of image data after smoothing are uniformly converted into the RGB space using the color space to form a plurality of RGB color image data;

[0019] Perform image pyramid processing on a plurality of RGB color image data to form a plurality of processed image data;

[0020] grayscale the plurality of processed image data to form a plurality of grayscale image data;

[0021] Perform BGR conversion on a plurality of processed image data to form a plurality of BGR image data;

[0022] Perform HSV conversion on a plurality of processed image data to form a plurality of HSV image data;

[0023] Scaling preprocessing is performed based on the plurality of grayscale image data, the plurality of BGR image data and the plurality of HSV image data to form a plurality of preprocessed image data.

[0024] Optionally, the step of extracting target area images from corresponding pieces of image data based on a plurality of edge extraction results of the intangible cultural heritage physical product at different angles to obtain a plurality of target area images of the intangible cultural heritage physical product at different angles includes:

[0025] Drawing a plurality of edge extraction results of the intangible cultural heritage physical product at different angles into a plurality of corresponding image data to form a plurality of drawing image data;

[0026] The target area image extraction process is performed by taking the drawing edge of several drawing image data as the center and expanding several pixel distances to obtain several target area images of the intangible cultural heritage physical product at different angles.

[0027] Optionally, the three-dimensional spatial feature extraction model based on the image is used to perform three-dimensional spatial feature extraction processing on a plurality of target area images of the intangible cultural heritage physical product at different angles to obtain three-dimensional spatial feature data of the intangible cultural heritage physical product at different angles, including:

[0028] Cutting each of the plurality of target area images into a plurality of non-overlapping image areas in sequence, and inputting the plurality of non-overlapping image areas corresponding to each target area image into an image three-dimensional space feature extraction model in sequence;

[0029] Perform linear projection processing in the image three-dimensional space feature extraction model to obtain feature vectors corresponding to several non-overlapping image regions in each target area image;

[0030] Performing image feature extraction processing on feature vectors corresponding to a plurality of non-overlapping image regions in each target region image, and establishing an association relationship between each non-overlapping image region based on the extracted image region feature information;

[0031] Combining and processing the extracted image region feature information based on the association relationship to obtain multiple resolution image features corresponding to each target region image;

[0032] The multiple resolution image features corresponding to each target area image are fused to form image feature data corresponding to each target area image;

[0033] Decoding the image feature data corresponding to each target area image to obtain a depth image of the same size as each target area image, wherein the values ​​of different pixel points in the depth image correspond to the distance of the position from the image acquisition device;

[0034] The depth image is converted into a series of spatial coordinate point sets in the coordinate matrix of the image acquisition device to obtain three-dimensional spatial feature data of the intangible cultural heritage physical product at different angles.

[0035] Optionally, the three-dimensional spatial feature data at different angles are fused to form a three-dimensional feature model of the intangible cultural heritage physical product, including:

[0036] Obtaining two three-dimensional spatial feature data at adjacent angles from the three-dimensional spatial feature data at different angles, and matching the two three-dimensional spatial feature data at adjacent angles to obtain common three-dimensional spatial feature points in the two three-dimensional spatial feature data at adjacent angles;

[0037] According to the common three-dimensional spatial feature points in two three-dimensional spatial feature data at adjacent angles, interpolation and fusion processing is performed using the nearest neighbor interpolation algorithm according to the angle difference between the two three-dimensional spatial feature data, and after completing the interpolation and fusion of the three-dimensional spatial feature data at different angles, a three-dimensional feature model of the intangible cultural heritage physical product is formed, wherein the angle difference is 90 degrees.

[0038] In addition, an embodiment of the present invention further provides a device for extracting three-dimensional features of an intangible cultural heritage physical product, the device comprising:

[0039] Image acquisition module: used to perform image acquisition processing on the intangible cultural heritage physical product according to different preset angles, and obtain a number of image data of the intangible cultural heritage physical product at different angles;

[0040] Edge extraction module: used for inputting a plurality of image data of the intangible cultural heritage physical product at different angles into a convergent edge extraction model to perform edge extraction processing of the intangible cultural heritage physical product, and obtaining a plurality of edge extraction results of the intangible cultural heritage physical product at different angles;

[0041] Target area extraction module: used to extract target area images from corresponding several image data based on several edge extraction results of the intangible cultural heritage physical product at different angles, and obtain several target area images of the intangible cultural heritage physical product at different angles;

[0042] Three-dimensional space extraction module: used for performing three-dimensional space feature extraction processing on a plurality of target area images of the intangible cultural heritage physical product at different angles based on the image three-dimensional space feature extraction model, so as to obtain the three-dimensional space feature data of the intangible cultural heritage physical product at different angles;

[0043] Feature fusion module: used to fuse the three-dimensional spatial feature data at different angles to form a three-dimensional feature model of the intangible cultural heritage physical product.

[0044] In addition, an embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, any of the methods described above is implemented.

[0045] In addition, an embodiment of the present invention further provides an electronic device, the electronic device comprising:

[0046] one or more processors;

[0047] Memory;

[0048] One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, and the one or more applications are configured to: execute the method according to any one of the above.

[0049] In an embodiment of the present invention, by performing image acquisition processing on an intangible cultural heritage physical product according to different preset angles; extracting the edges of several pieces of image data to form several edge extraction results, and finally extracting target area images in the corresponding several pieces of image data according to the several edge extraction results; then extracting the three-dimensional spatial feature data of the intangible cultural heritage physical product at different angles according to the several target area images, and finally fusing them to form a three-dimensional feature model of the intangible cultural heritage physical product; the three-dimensional feature extraction of the intangible cultural heritage physical product can be realized, and a complete three-dimensional feature model can be formed to accurately express the form and beauty of the intangible cultural heritage physical product. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0051] Figure 1 is a flow chart of a method for extracting three-dimensional features of an intangible cultural heritage physical product in an embodiment of the present invention;

[0052] Figure 2 Schematic diagram of the structure of a device for extracting three-dimensional features of an intangible cultural heritage physical product in an embodiment of the present invention;

[0053] Figure 3 It is a schematic diagram of the structure of an electronic device in an embodiment of the present invention. DETAILED DESCRIPTION

[0054] 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.

[0055] For example, see Figure 1 , Figure 1 It is a flowchart of a method for extracting three-dimensional features of an intangible cultural heritage physical product in an embodiment of the present invention.

[0056] like Figure 1 As shown, a method for extracting three-dimensional features of intangible cultural heritage physical products, the method comprising:

[0057] S11: performing image acquisition processing on the intangible cultural heritage physical product at different preset angles to obtain a plurality of image data of the intangible cultural heritage physical product at different angles;

[0058] In the specific implementation process of the present invention, the image acquisition processing of the intangible cultural heritage physical product according to different preset angles is performed to obtain a plurality of image data of the intangible cultural heritage physical product at different angles, including: placing the intangible cultural heritage physical product on an image acquisition platform, and controlling an image acquisition device arranged above the image acquisition platform to perform image acquisition processing on the intangible cultural heritage physical product at the current angle; after completing the image acquisition at the current angle, controlling the intangible cultural heritage physical product to rotate at a preset angle on the image acquisition platform; after the intangible cultural heritage physical product is rotated at the preset angle, controlling the image acquisition device to perform image acquisition processing on the intangible cultural heritage physical product after the rotation angle, until the image acquisition processing at different angles is completed, and a plurality of image data of the intangible cultural heritage physical product at different angles are obtained; wherein the preset rotation angles are 90 degrees, 180 degrees and 270 degrees corresponding to the current angle placed under the image acquisition platform, respectively.

[0059] Specifically, it is necessary to set up an image acquisition platform, which needs to be a platform without complex colors as the image acquisition platform, which will facilitate the subsequent image edge extraction and three-dimensional feature extraction; after determining the image acquisition platform, set the image acquisition device on the image acquisition platform, and then place the intangible cultural heritage physical product on the image acquisition platform, and then control the image acquisition device set above the image acquisition platform to perform image acquisition processing on the intangible cultural heritage physical product at the current angle, so as to obtain the image data at the current angle; and after completing the image acquisition of the current angle, control the intangible cultural heritage physical product to rotate a preset angle on the image acquisition platform; and after the intangible cultural heritage physical product rotates the preset angle, control The image acquisition device performs image acquisition processing on the intangible cultural heritage physical product after the rotation angle until the image acquisition processing at different angles is completed, so as to obtain several image data of the intangible cultural heritage physical product at different angles; wherein the preset rotation angles are 90 degrees, 180 degrees and 270 degrees corresponding to the current angle placed on the image acquisition platform, that is, the rotation angle is 90 degrees each time, and it can be selected three times. For example, the current placement angle is 0 degrees, and the next photo is rotated 90 degrees, that is, the relative angle is 90 degrees. The next time the photo is taken, it is rotated 90 degrees again, and the relative angle is 180 degrees at this time; the next time the photo is taken, it is rotated 90 degrees again, and the relative angle is 270 degrees at this time.

[0060] S12: inputting a plurality of image data of the intangible cultural heritage physical product at different angles into a convergent edge extraction model to perform edge extraction processing on the intangible cultural heritage physical product, and obtaining a plurality of edge extraction results of the intangible cultural heritage physical product at different angles;

[0061] In the specific implementation process of the present invention, the inputting of several image data of the intangible cultural heritage physical product at different angles into a convergent edge extraction model to perform edge extraction processing of the intangible cultural heritage physical product, and obtaining several edge extraction results of the intangible cultural heritage physical product at different angles, includes: preprocessing several image data of the intangible cultural heritage physical product at different angles to form several preprocessed image data; inputting the several preprocessed image data in sequence into the convergent edge extraction model to perform edge extraction processing of the intangible cultural heritage physical product, and obtaining several edge extraction results of the intangible cultural heritage physical product at different angles; wherein the edge extraction model is a model which sequentially includes several Conv modules, several Bottleneck CSP / C3 network modules, SPPF modules, several upsampling modules and several Concat modules.

[0062] Furthermore, the preprocessing of the plurality of image data of the intangible cultural heritage physical product at different angles to form a plurality of preprocessed image data includes: smoothing the plurality of image data of the intangible cultural heritage physical product at different angles based on a Gaussian filter, and uniformly converting the plurality of smoothed image data into an RGB space using a color space to form a plurality of RGB color image data; performing image pyramid processing on the plurality of RGB color image data to form a plurality of processed image data; performing grayscale processing on the plurality of processed image data to form a plurality of grayscale image data; performing BGR conversion processing on the plurality of processed image data to form a plurality of BGR image data; performing HSV conversion processing on the plurality of processed image data to form a plurality of HSV image data; performing scaling preprocessing based on the plurality of grayscale image data, the plurality of BGR image data and the plurality of HSV image data to form a plurality of preprocessed image data.

[0063] Specifically, when extracting the edges of intangible cultural heritage physical products in image data, it is first necessary to preprocess several image data of the intangible cultural heritage physical products at different angles to form several preprocessed image data; then the several preprocessed image data are sequentially input into the convergent edge extraction model to perform edge extraction processing of the intangible cultural heritage physical products, and several edge extraction results of the intangible cultural heritage physical products at different angles can be obtained.

[0064] During preprocessing, a Gaussian filter is first required to perform filtering and smoothing on a number of image data of intangible cultural heritage physical products at different angles, and after filtering and smoothing, the smoothed image data are uniformly converted to RGB space using color space to form a number of RGB color image data; this will facilitate subsequent unified processing and improve processing speed and accuracy; after conversion into a number of RGB color image data, the RGB color image data need to be subjected to image pyramid processing, wherein the image pyramid processing is a set of images of different resolutions from fine to coarse generated by the original image according to certain rules; the bottom of the pyramid is a high-resolution representation of the image, that is, the original image, and the top is a low-resolution approximation; the bottom layer has the highest resolution and the largest amount of data, and as the number of layers increases, Its resolution gradually decreases, and the amount of data also decreases proportionally; there are two methods for constructing an image pyramid: one is to automatically construct a pyramid with a multi-resolution data source; the other is that except for the bottom layer of the pyramid which is the original image data, the image data of other layers are constructed by sampling and extracting from the bottom layer data; after obtaining a number of processed image data, the several processed image data are grayed to form a number of grayscale image data; at the same time, a number of processed image data are also required to be BGR converted to form a number of BGR image data; and a number of processed image data are HSV converted to form a number of HSV image data; finally, scaling preprocessing is performed based on a number of grayscale image data, a number of BGR image data and a number of HSV image data to form a number of preprocessed image data.

[0065] Among them, the edge extraction model is a model that includes several Conv modules, several Bottleneck CSP / C3 network modules, SPPF modules, several upsampling modules and several Concat modules in sequence; the training of the edge extraction model can be performed using labeled edge images; after the preprocessed image data are sequentially input into the converged edge extraction model, the upsampling operation is performed through the upsampling module in the converged edge extraction model, and the corresponding feature image is sampled and extracted; then the extracted corresponding feature image is input into the Bottleneck CSP / C3 network module for feature extraction, and then the extracted features are input into the SPPF module, and the input extracted features are copied in the SPPF module, and the copied extracted features are connected using the Concat module to output the connected feature data; then the connected feature data is generated through the convolution kernel to generate the convolution output feature, and then after BN and Leaky Relu inputs the result into the backbone network model; in the backbone network model, multiple convolution layers are used for convolution processing, and the convolution results are extracted to the corresponding feature images through residual structure and sampling for addition processing to form added feature data; the added feature data is then added to the adjacent layers using an adaptive feature pool, and the Concat operation is performed in the Concat module, and the Concat operation results are successively calculated using GOU_Loss for regression loss calculation and non-maximum suppression algorithm calculation processing, and the calculation results are output as the edge extraction results of several pre-processed image data; in this way, several edge extraction results of intangible cultural heritage physical products at different angles can be obtained.

[0066] S13: extracting target area images from corresponding pieces of image data based on a plurality of edge extraction results of the intangible cultural heritage physical product at different angles, and obtaining a plurality of target area images of the intangible cultural heritage physical product at different angles;

[0067] In the specific implementation process of the present invention, the target area images in the corresponding several image data are extracted based on the several edge extraction results of the intangible cultural heritage physical product at different angles to obtain several target area images of the intangible cultural heritage physical product at different angles, including: drawing the several edge extraction results of the intangible cultural heritage physical product at different angles in the corresponding several image data to form several drawn image data; using the drawn edges of the several drawn image data as the center to expand several pixel distances to perform target area image extraction processing to obtain several target area images of the intangible cultural heritage physical product at different angles.

[0068] Specifically, in order to facilitate the subsequent target area extraction, it is first necessary to draw the edges in the image data according to several edge extraction results of the intangible cultural heritage physical products at different angles to obtain several drawn image data; then, the drawn edges of the several drawn image data are used as the center to expand several pixel distances to perform target area image extraction processing, and obtain several target area images of the intangible cultural heritage physical products at different angles; in this embodiment, the distance can be expanded by 50 pixels, which can ensure the presence of a complete target in the target area image, and at the same time can correspondingly reduce the image area during subsequent image processing, thereby increasing the speed of image processing.

[0069] S14: performing three-dimensional spatial feature extraction processing on a plurality of target area images of the intangible cultural heritage physical product at different angles based on the image three-dimensional spatial feature extraction model to obtain three-dimensional spatial feature data of the intangible cultural heritage physical product at different angles;

[0070] In the specific implementation process of the present invention, the three-dimensional spatial feature extraction model is used to perform three-dimensional spatial feature extraction processing on several target area images of the intangible cultural heritage physical product at different angles to obtain three-dimensional spatial feature data of the intangible cultural heritage physical product at different angles, including: cutting each of the several target area images into several non-overlapping image areas in turn, and inputting the several non-overlapping image areas corresponding to each target area image into the image three-dimensional spatial feature extraction model in turn; performing linear projection processing in the image three-dimensional spatial feature extraction model to obtain feature vectors corresponding to several non-overlapping image areas in each target area image; performing image feature extraction processing on the feature vectors corresponding to several non-overlapping image areas in each target area image, and according to the proposed The obtained image area feature information establishes an association relationship between each non-overlapping image area; based on the association relationship, the extracted image area feature information is combined and processed to obtain multiple resolution image features corresponding to each target area image; the multiple resolution image features corresponding to each target area image are fused to form image feature data corresponding to each target area image; the image feature data corresponding to each target area image is decoded to obtain a depth image with the same size as each target area image, and the values ​​of different pixel points in the depth image correspond to the distance of the position from the image acquisition device; the depth image is converted into a series of spatial coordinate point sets in the coordinate matrix of the image acquisition device to obtain three-dimensional spatial feature data of the intangible cultural heritage physical product at different angles.

[0071] Specifically, the image three-dimensional spatial feature extraction model can be a Vision Transformer model; that is, when the three-dimensional spatial feature extraction model is used to extract the three-dimensional spatial features of several target area images of the intangible cultural heritage physical product at different angles, it is first necessary to cut each of the several target area images of the intangible cultural heritage physical product at different angles into several non-overlapping image areas in turn, and then input the several non-overlapping image areas in each target area image into the image three-dimensional spatial feature extraction model; then the three-dimensional spatial feature extraction processing can be performed in the image three-dimensional spatial feature extraction model, specifically, encoding and decoding processing are performed in turn to form a corresponding depth image, and then the three-dimensional spatial feature data of the intangible cultural heritage physical product at different angles is obtained by converting the depth image into a series of spatial coordinate point sets in the coordinate matrix of the image acquisition device.

[0072] Among them, during the encoding process, the encoder performs linear projection processing on several non-overlapping image areas in each input target area image, and obtains the feature vectors corresponding to the several non-overlapping image areas in each target area image after linear projection; then a multi-layer series-connected multi-head self-attention module is used to process the feature vectors to obtain the image area feature information corresponding to the several non-overlapping image areas; then, based on the image area feature information corresponding to the several non-overlapping image areas, an association relationship between the non-overlapping image areas corresponding to the several non-overlapping image areas is established; then, based on the association relationship, the image feature information output by the multi-layer series-connected multi-head self-attention module is combined and processed to form multiple resolution image features corresponding to each target area image; then, the multiple resolution image features corresponding to each target area image are fused to form image feature data corresponding to each target area image.

[0073] During the decoding process, a decoder is used to decode the image feature data corresponding to each target area image to obtain a depth image of the same size as each target area image, wherein the values ​​of different pixel points in the depth image correspond to the distance of the position from the image acquisition device.

[0074] S15: The three-dimensional spatial feature data at different angles are fused to form a three-dimensional feature model of the intangible cultural heritage physical product.

[0075] In the specific implementation process of the present invention, the three-dimensional spatial feature data at different angles are fused to form a three-dimensional feature model of the intangible cultural heritage physical product, including: obtaining two three-dimensional spatial feature data at adjacent angles in the three-dimensional spatial feature data at different angles, and matching the two three-dimensional spatial feature data at adjacent angles to obtain the common three-dimensional spatial feature points in the two three-dimensional spatial feature data at adjacent angles; performing interpolation and fusion processing according to the angle difference between the two three-dimensional spatial feature data using the nearest neighbor interpolation algorithm based on the common three-dimensional spatial feature points in the two three-dimensional spatial feature data at adjacent angles, and after completing the interpolation and fusion of the three-dimensional spatial feature data at different angles, forming the three-dimensional feature model of the intangible cultural heritage physical product, wherein the angle difference is 90 degrees.

[0076] Specifically, after obtaining the three-dimensional spatial feature data at different angles, it is necessary to perform splicing and fusion. At this time, it is necessary to obtain two three-dimensional spatial feature data at adjacent angles in the three-dimensional spatial feature data at different angles, and then use the two three-dimensional spatial feature data at adjacent angles for matching to obtain the three-dimensional spatial feature points that exist in common in the two three-dimensional spatial feature data at adjacent angles; then the two three-dimensional spatial feature data at adjacent angles are spliced ​​and fused according to the three-dimensional spatial feature points that exist in common in the two three-dimensional spatial feature data at adjacent angles; that is, according to the three-dimensional spatial feature points that exist in common in the two three-dimensional spatial feature data at adjacent angles, the interpolation, splicing and fusion processing is performed using the nearest neighbor interpolation algorithm according to the angle difference between the two three-dimensional spatial feature data, so as to complete the splicing and fusion of the two adjacent three-dimensional spatial feature data, and the splicing and fusion of all the three-dimensional spatial feature data can be completed by using the splicing and fusion of the two adjacent three-dimensional spatial feature data in turn, so as to form the three-dimensional feature model of the intangible cultural heritage physical product, wherein the angle difference is 90 degrees.

[0077] In an embodiment of the present invention, by performing image acquisition processing on an intangible cultural heritage physical product according to different preset angles; extracting the edges of several pieces of image data to form several edge extraction results, and finally extracting target area images in the corresponding several pieces of image data according to the several edge extraction results; then extracting the three-dimensional spatial feature data of the intangible cultural heritage physical product at different angles according to the several target area images, and finally fusing them to form a three-dimensional feature model of the intangible cultural heritage physical product; the three-dimensional feature extraction of the intangible cultural heritage physical product can be realized, and a complete three-dimensional feature model can be formed to accurately express the form and beauty of the intangible cultural heritage physical product.

[0078] For example 2, please refer to Figure 2 , Figure 2 It is a schematic diagram of the structural composition of the device for extracting three-dimensional features of intangible cultural heritage physical products in an embodiment of the present invention.

[0079] like Figure 2 As shown, a device for extracting three-dimensional features of intangible cultural heritage physical products, the device comprising:

[0080] Image acquisition module 21: used to perform image acquisition processing on the intangible cultural heritage physical product according to different preset angles, and obtain a plurality of image data of the intangible cultural heritage physical product at different angles;

[0081] In the specific implementation process of the present invention, the image acquisition processing of the intangible cultural heritage physical product according to different preset angles is performed to obtain a plurality of image data of the intangible cultural heritage physical product at different angles, including: placing the intangible cultural heritage physical product on an image acquisition platform, and controlling an image acquisition device arranged above the image acquisition platform to perform image acquisition processing on the intangible cultural heritage physical product at the current angle; after completing the image acquisition at the current angle, controlling the intangible cultural heritage physical product to rotate at a preset angle on the image acquisition platform; after the intangible cultural heritage physical product is rotated at the preset angle, controlling the image acquisition device to perform image acquisition processing on the intangible cultural heritage physical product after the rotation angle, until the image acquisition processing at different angles is completed, and a plurality of image data of the intangible cultural heritage physical product at different angles are obtained; wherein the preset rotation angles are 90 degrees, 180 degrees and 270 degrees corresponding to the current angle placed under the image acquisition platform, respectively.

[0082] Specifically, it is necessary to set up an image acquisition platform, which needs to be a platform without complex colors as the image acquisition platform, which will facilitate the subsequent image edge extraction and three-dimensional feature extraction; after determining the image acquisition platform, set the image acquisition device on the image acquisition platform, and then place the intangible cultural heritage physical product on the image acquisition platform, and then control the image acquisition device set above the image acquisition platform to perform image acquisition processing on the intangible cultural heritage physical product at the current angle, so as to obtain the image data at the current angle; and after completing the image acquisition of the current angle, control the intangible cultural heritage physical product to rotate a preset angle on the image acquisition platform; and after the intangible cultural heritage physical product rotates the preset angle, control The image acquisition device performs image acquisition processing on the intangible cultural heritage physical product after the rotation angle until the image acquisition processing at different angles is completed, so as to obtain several image data of the intangible cultural heritage physical product at different angles; wherein the preset rotation angles are 90 degrees, 180 degrees and 270 degrees corresponding to the current angle placed on the image acquisition platform, that is, the rotation angle is 90 degrees each time, and it can be selected three times. For example, the current placement angle is 0 degrees, and the next photo is rotated 90 degrees, that is, the relative angle is 90 degrees. The next time the photo is taken, it is rotated 90 degrees again, and the relative angle is 180 degrees at this time; the next time the photo is taken, it is rotated 90 degrees again, and the relative angle is 270 degrees at this time.

[0083] Edge extraction module 22: used for inputting a plurality of image data of the intangible cultural heritage physical product at different angles into a convergent edge extraction model to perform edge extraction processing of the intangible cultural heritage physical product, and obtaining a plurality of edge extraction results of the intangible cultural heritage physical product at different angles;

[0084] In the specific implementation process of the present invention, the inputting of several image data of the intangible cultural heritage physical product at different angles into a convergent edge extraction model to perform edge extraction processing of the intangible cultural heritage physical product, and obtaining several edge extraction results of the intangible cultural heritage physical product at different angles, includes: preprocessing several image data of the intangible cultural heritage physical product at different angles to form several preprocessed image data; inputting the several preprocessed image data in sequence into the convergent edge extraction model to perform edge extraction processing of the intangible cultural heritage physical product, and obtaining several edge extraction results of the intangible cultural heritage physical product at different angles; wherein the edge extraction model is a model which sequentially includes several Conv modules, several Bottleneck CSP / C3 network modules, SPPF modules, several upsampling modules and several Concat modules.

[0085] Furthermore, the preprocessing of the plurality of image data of the intangible cultural heritage physical product at different angles to form a plurality of preprocessed image data includes: smoothing the plurality of image data of the intangible cultural heritage physical product at different angles based on a Gaussian filter, and uniformly converting the plurality of smoothed image data into an RGB space using a color space to form a plurality of RGB color image data; performing image pyramid processing on the plurality of RGB color image data to form a plurality of processed image data; performing grayscale processing on the plurality of processed image data to form a plurality of grayscale image data; performing BGR conversion processing on the plurality of processed image data to form a plurality of BGR image data; performing HSV conversion processing on the plurality of processed image data to form a plurality of HSV image data; performing scaling preprocessing based on the plurality of grayscale image data, the plurality of BGR image data and the plurality of HSV image data to form a plurality of preprocessed image data.

[0086] Specifically, when extracting the edges of intangible cultural heritage physical products in image data, it is first necessary to preprocess several image data of the intangible cultural heritage physical products at different angles to form several preprocessed image data; then the several preprocessed image data are sequentially input into the convergent edge extraction model to perform edge extraction processing of the intangible cultural heritage physical products, and several edge extraction results of the intangible cultural heritage physical products at different angles can be obtained.

[0087] During preprocessing, a Gaussian filter is first required to perform filtering and smoothing on a number of image data of intangible cultural heritage physical products at different angles, and after filtering and smoothing, the smoothed image data are uniformly converted to RGB space using color space to form a number of RGB color image data; this will facilitate subsequent unified processing and improve processing speed and accuracy; after conversion into a number of RGB color image data, the RGB color image data need to be subjected to image pyramid processing, wherein the image pyramid processing is a set of images of different resolutions from fine to coarse generated by the original image according to certain rules; the bottom of the pyramid is a high-resolution representation of the image, that is, the original image, and the top is a low-resolution approximation; the bottom layer has the highest resolution and the largest amount of data, and as the number of layers increases, Its resolution gradually decreases, and the amount of data also decreases proportionally; there are two methods for constructing an image pyramid: one is to automatically construct a pyramid with a multi-resolution data source; the other is that except for the bottom layer of the pyramid which is the original image data, the image data of other layers are constructed by sampling and extracting from the bottom layer data; after obtaining a number of processed image data, the several processed image data are grayed to form a number of grayscale image data; at the same time, a number of processed image data are also required to be BGR converted to form a number of BGR image data; and a number of processed image data are HSV converted to form a number of HSV image data; finally, scaling preprocessing is performed based on a number of grayscale image data, a number of BGR image data and a number of HSV image data to form a number of preprocessed image data.

[0088] Among them, the edge extraction model is a model that includes several Conv modules, several Bottleneck CSP / C3 network modules, SPPF modules, several upsampling modules and several Concat modules in sequence; the training of the edge extraction model can be performed using labeled edge images; after the preprocessed image data are sequentially input into the converged edge extraction model, the upsampling operation is performed through the upsampling module in the converged edge extraction model, and the corresponding feature image is sampled and extracted; then the extracted corresponding feature image is input into the Bottleneck CSP / C3 network module for feature extraction, and then the extracted features are input into the SPPF module, and the input extracted features are copied in the SPPF module, and the copied extracted features are connected using the Concat module to output the connected feature data; then the connected feature data is generated through the convolution kernel to generate the convolution output feature, and then after BN and Leaky Relu inputs the result into the backbone network model; in the backbone network model, multiple convolution layers are used for convolution processing, and the convolution results are extracted to the corresponding feature images through residual structure and sampling for addition processing to form added feature data; the added feature data is then added to the adjacent layers using an adaptive feature pool, and the Concat operation is performed in the Concat module, and the Concat operation results are successively calculated using GOU_Loss for regression loss calculation and non-maximum suppression algorithm calculation processing, and the calculation results are output as the edge extraction results of several pre-processed image data; in this way, several edge extraction results of intangible cultural heritage physical products at different angles can be obtained.

[0089] The target area extraction module 23 is used to extract the target area images in the corresponding several image data based on the several edge extraction results of the intangible cultural heritage physical product at different angles, and obtain several target area images of the intangible cultural heritage physical product at different angles;

[0090] In the specific implementation process of the present invention, the target area images in the corresponding several image data are extracted based on the several edge extraction results of the intangible cultural heritage physical product at different angles to obtain several target area images of the intangible cultural heritage physical product at different angles, including: drawing the several edge extraction results of the intangible cultural heritage physical product at different angles in the corresponding several image data to form several drawn image data; using the drawn edges of the several drawn image data as the center to expand several pixel distances to perform target area image extraction processing to obtain several target area images of the intangible cultural heritage physical product at different angles.

[0091] Specifically, in order to facilitate the subsequent target area extraction, it is first necessary to draw the edges in the image data according to several edge extraction results of the intangible cultural heritage physical products at different angles to obtain several drawn image data; then, the drawn edges of the several drawn image data are used as the center to expand several pixel distances to perform target area image extraction processing, and obtain several target area images of the intangible cultural heritage physical products at different angles; in this embodiment, the distance can be expanded by 50 pixels, which can ensure the presence of a complete target in the target area image, and at the same time can correspondingly reduce the image area during subsequent image processing, thereby increasing the speed of image processing.

[0092] The three-dimensional space extraction module 24 is used to perform three-dimensional space feature extraction processing on a plurality of target area images of the intangible cultural heritage physical product at different angles based on the image three-dimensional space feature extraction model, so as to obtain three-dimensional space feature data of the intangible cultural heritage physical product at different angles;

[0093] In the specific implementation process of the present invention, the three-dimensional spatial feature extraction model is used to perform three-dimensional spatial feature extraction processing on several target area images of the intangible cultural heritage physical product at different angles to obtain three-dimensional spatial feature data of the intangible cultural heritage physical product at different angles, including: cutting each of the several target area images into several non-overlapping image areas in turn, and inputting the several non-overlapping image areas corresponding to each target area image into the image three-dimensional spatial feature extraction model in turn; performing linear projection processing in the image three-dimensional spatial feature extraction model to obtain feature vectors corresponding to several non-overlapping image areas in each target area image; performing image feature extraction processing on the feature vectors corresponding to several non-overlapping image areas in each target area image, and according to the proposed The obtained image area feature information establishes an association relationship between each non-overlapping image area; based on the association relationship, the extracted image area feature information is combined and processed to obtain multiple resolution image features corresponding to each target area image; the multiple resolution image features corresponding to each target area image are fused to form image feature data corresponding to each target area image; the image feature data corresponding to each target area image is decoded to obtain a depth image with the same size as each target area image, and the values ​​of different pixel points in the depth image correspond to the distance of the position from the image acquisition device; the depth image is converted into a series of spatial coordinate point sets in the coordinate matrix of the image acquisition device to obtain three-dimensional spatial feature data of the intangible cultural heritage physical product at different angles.

[0094] Specifically, the image three-dimensional spatial feature extraction model can be a Vision Transformer model; that is, when the three-dimensional spatial feature extraction model is used to extract the three-dimensional spatial features of several target area images of the intangible cultural heritage physical product at different angles, it is first necessary to cut each of the several target area images of the intangible cultural heritage physical product at different angles into several non-overlapping image areas in turn, and then input the several non-overlapping image areas in each target area image into the image three-dimensional spatial feature extraction model; then the three-dimensional spatial feature extraction processing can be performed in the image three-dimensional spatial feature extraction model, specifically, encoding and decoding processing are performed in turn to form a corresponding depth image, and then the three-dimensional spatial feature data of the intangible cultural heritage physical product at different angles is obtained by converting the depth image into a series of spatial coordinate point sets in the coordinate matrix of the image acquisition device.

[0095] Among them, during the encoding process, the encoder performs linear projection processing on several non-overlapping image areas in each input target area image, and obtains the feature vectors corresponding to the several non-overlapping image areas in each target area image after linear projection; then a multi-layer series-connected multi-head self-attention module is used to process the feature vectors to obtain the image area feature information corresponding to the several non-overlapping image areas; then, based on the image area feature information corresponding to the several non-overlapping image areas, an association relationship between the non-overlapping image areas corresponding to the several non-overlapping image areas is established; then, based on the association relationship, the image feature information output by the multi-layer series-connected multi-head self-attention module is combined and processed to form multiple resolution image features corresponding to each target area image; then, the multiple resolution image features corresponding to each target area image are fused to form image feature data corresponding to each target area image.

[0096] During the decoding process, a decoder is used to decode the image feature data corresponding to each target area image to obtain a depth image of the same size as each target area image, wherein the values ​​of different pixel points in the depth image correspond to the distance of the position from the image acquisition device.

[0097] Feature fusion module 25: used to fuse the three-dimensional spatial feature data at different angles to form a three-dimensional feature model of the intangible cultural heritage physical product.

[0098] In the specific implementation process of the present invention, the three-dimensional spatial feature data at different angles are fused to form a three-dimensional feature model of the intangible cultural heritage physical product, including: obtaining two three-dimensional spatial feature data at adjacent angles in the three-dimensional spatial feature data at different angles, and matching the two three-dimensional spatial feature data at adjacent angles to obtain the common three-dimensional spatial feature points in the two three-dimensional spatial feature data at adjacent angles; performing interpolation and fusion processing according to the angle difference between the two three-dimensional spatial feature data using the nearest neighbor interpolation algorithm based on the common three-dimensional spatial feature points in the two three-dimensional spatial feature data at adjacent angles, and after completing the interpolation and fusion of the three-dimensional spatial feature data at different angles, forming the three-dimensional feature model of the intangible cultural heritage physical product, wherein the angle difference is 90 degrees.

[0099] Specifically, after obtaining the three-dimensional spatial feature data at different angles, it is necessary to perform splicing and fusion. At this time, it is necessary to obtain two three-dimensional spatial feature data at adjacent angles in the three-dimensional spatial feature data at different angles, and then use the two three-dimensional spatial feature data at adjacent angles for matching to obtain the three-dimensional spatial feature points that exist in common in the two three-dimensional spatial feature data at adjacent angles; then the two three-dimensional spatial feature data at adjacent angles are spliced ​​and fused according to the three-dimensional spatial feature points that exist in common in the two three-dimensional spatial feature data at adjacent angles; that is, according to the three-dimensional spatial feature points that exist in common in the two three-dimensional spatial feature data at adjacent angles, the interpolation, splicing and fusion processing is performed using the nearest neighbor interpolation algorithm according to the angle difference between the two three-dimensional spatial feature data, so as to complete the splicing and fusion of the two adjacent three-dimensional spatial feature data, and the splicing and fusion of all the three-dimensional spatial feature data can be completed by using the splicing and fusion of the two adjacent three-dimensional spatial feature data in turn, so as to form the three-dimensional feature model of the intangible cultural heritage physical product, wherein the angle difference is 90 degrees.

[0100] In an embodiment of the present invention, by performing image acquisition processing on an intangible cultural heritage physical product according to different preset angles; extracting the edges of several pieces of image data to form several edge extraction results, and finally extracting target area images in the corresponding several pieces of image data according to the several edge extraction results; then extracting the three-dimensional spatial feature data of the intangible cultural heritage physical product at different angles according to the several target area images, and finally fusing them to form a three-dimensional feature model of the intangible cultural heritage physical product; the three-dimensional feature extraction of the intangible cultural heritage physical product can be realized, and a complete three-dimensional feature model can be formed to accurately express the form and beauty of the intangible cultural heritage physical product.

[0101] A computer-readable storage medium provided in an embodiment of the present invention stores a computer program on the computer-readable storage medium, and when the program is executed by a processor, the method of any one of the embodiments described above is implemented. The computer-readable storage medium includes, but is not limited to, any type of disk (including floppy disk, hard disk, optical disk, CD-ROM, and magneto-optical disk), ROM (Read-Only Memory), RAM (Random Access Memory), EPROM (Erasable Programmable Read-Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), flash memory, magnetic card or optical card. That is, the storage device includes any medium that can store or transmit information in a readable form by a device (for example, a computer, a mobile phone), which can be a read-only memory, a disk or an optical disk, etc.

[0102] An embodiment of the present invention further provides a computer application program, which runs on a computer and is used to execute the method of any one of the above embodiments.

[0103] also, Figure 3 It is a schematic diagram of the structure of an electronic device in an embodiment of the present invention.

[0104] The embodiment of the present invention further provides an electronic device, such as Figure 3 The electronic device includes a processor 302, a memory 303, an input unit 304, a display unit 305 and other devices. Those skilled in the art will understand that Figure 3 The electronic device structure device shown does not constitute a limitation on all devices, and may include more or fewer components than shown, or combine certain components. The memory 303 can be used to store the application 301 and various functional modules, and the processor 302 runs the application 301 stored in the memory 303, thereby executing various functional applications and data processing of the device. The memory can be an internal memory or an external memory, or include both internal and external memories. The internal memory may include a read-only memory (ROM), a programmable ROM (PROM), an electrically programmable ROM (EPROM), an electrically erasable programmable ROM (EEPROM), a flash memory, or a random access memory. The external memory may include a hard disk, a floppy disk, a ZIP disk, a U disk, a magnetic tape, etc. The memory disclosed in the present invention includes but is not limited to these types of memories. The memory disclosed in the present invention is only used as an example and not as a limitation.

[0105] The input unit 304 is used to receive the input of the signal and the keyword input by the user. The input unit 304 may include a touch panel and other input devices. The touch panel can collect the touch operation of the user on or near it, and drive the corresponding connection device according to a pre-set program; other input devices may include but are not limited to one or more of a physical keyboard, a function key (such as a play control key, a switch key, etc.), a trackball, a mouse, a joystick, etc. The display unit 305 can be used to display the information input by the user or the information provided to the user and various menus of the terminal device. The display unit 305 can be in the form of a liquid crystal display, an organic light emitting diode, etc. The processor 302 is the control center of the terminal device, which uses various interfaces and lines to connect various parts of the entire device, and executes various functions and processes data by running or executing software programs and / or modules stored in the memory 303, and calling data stored in the memory.

[0106] As an embodiment, the electronic device includes: one or more processors 302, a memory 303, and one or more applications 301, wherein the one or more applications 301 are stored in the memory 303 and are configured to be executed by the one or more processors 302, and the one or more applications 301 are configured to execute the method in any one of the above-mentioned embodiments.

[0107] In an embodiment of the present invention, by performing image acquisition processing on an intangible cultural heritage physical product according to different preset angles; extracting the edges of several pieces of image data to form several edge extraction results, and finally extracting target area images in the corresponding several pieces of image data according to the several edge extraction results; then extracting the three-dimensional spatial feature data of the intangible cultural heritage physical product at different angles according to the several target area images, and finally fusing them to form a three-dimensional feature model of the intangible cultural heritage physical product; the three-dimensional feature extraction of the intangible cultural heritage physical product can be realized, and a complete three-dimensional feature model can be formed to accurately express the form and beauty of the intangible cultural heritage physical product.

[0108] In addition, the above is a detailed introduction to the method and related devices for extracting three-dimensional features of a physical product of intangible cultural heritage provided by an embodiment of the present invention. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A method for extracting three-dimensional features of intangible cultural heritage physical products, characterized in that: The method comprises: Performing image acquisition and processing on the intangible cultural heritage physical product at different preset angles to obtain a plurality of image data of the intangible cultural heritage physical product at different angles; Inputting a plurality of image data of the intangible cultural heritage physical product at different angles into a convergent edge extraction model to perform edge extraction processing of the intangible cultural heritage physical product, and obtaining a plurality of edge extraction results of the intangible cultural heritage physical product at different angles; Extract target area images from corresponding image data based on several edge extraction results of the intangible cultural heritage physical product at different angles, and obtain several target area images of the intangible cultural heritage physical product at different angles; Based on the image three-dimensional spatial feature extraction model, three-dimensional spatial feature extraction processing is performed on a plurality of target area images of the intangible cultural heritage physical product at different angles to obtain three-dimensional spatial feature data of the intangible cultural heritage physical product at different angles; The three-dimensional spatial feature data at different angles are fused to form a three-dimensional feature model of the intangible cultural heritage physical product; The three-dimensional spatial feature extraction model based on the image is used to perform three-dimensional spatial feature extraction processing on a plurality of target area images of the intangible cultural heritage physical product at different angles to obtain three-dimensional spatial feature data of the intangible cultural heritage physical product at different angles, including: Cutting each of the plurality of target area images into a plurality of non-overlapping image areas in sequence, and inputting the plurality of non-overlapping image areas corresponding to each target area image into an image three-dimensional space feature extraction model in sequence; Perform linear projection processing in the image three-dimensional space feature extraction model to obtain feature vectors corresponding to several non-overlapping image regions in each target area image; Performing image feature extraction processing on feature vectors corresponding to a plurality of non-overlapping image regions in each target region image, and establishing an association relationship between each non-overlapping image region based on the extracted image region feature information; Combining and processing the extracted image region feature information based on the association relationship to obtain multiple resolution image features corresponding to each target region image; The multiple resolution image features corresponding to each target area image are fused to form image feature data corresponding to each target area image; Decoding the image feature data corresponding to each target area image to obtain a depth image of the same size as each target area image, wherein the values ​​of different pixel points in the depth image correspond to the distance of the position from the image acquisition device; The depth image is converted into a series of spatial coordinate point sets in the coordinate matrix of the image acquisition device to obtain three-dimensional spatial feature data of the intangible cultural heritage physical product at different angles; The image three-dimensional spatial feature extraction model is the Vision Transformer model.

2. The method for extracting three-dimensional features according to claim 1, characterized in that: The image acquisition and processing of the intangible cultural heritage physical product according to different preset angles to obtain a plurality of image data of the intangible cultural heritage physical product at different angles includes: Placing the intangible cultural heritage physical product on an image acquisition platform, and controlling an image acquisition device disposed above the image acquisition platform to perform image acquisition processing on the intangible cultural heritage physical product at a current angle; After completing the image acquisition at the current angle, controlling the intangible cultural heritage physical product to rotate at a preset angle on the image acquisition platform; After the intangible cultural heritage physical product is rotated by a preset angle, the image acquisition device is controlled to perform image acquisition processing on the intangible cultural heritage physical product after the rotation angle, until the image acquisition processing at different angles is completed, and a plurality of image data of the intangible cultural heritage physical product at different angles are obtained; wherein the preset rotation angles are 90 degrees, 180 degrees and 270 degrees corresponding to the current angle placed under the image acquisition platform, respectively.

3. The method for extracting three-dimensional features according to claim 1, characterized in that: The step of inputting a plurality of image data of the intangible cultural heritage physical product at different angles into a convergent edge extraction model to perform edge extraction processing on the intangible cultural heritage physical product, and obtaining a plurality of edge extraction results of the intangible cultural heritage physical product at different angles, comprises: Preprocessing a plurality of image data of the intangible cultural heritage physical product at different angles to form a plurality of preprocessed image data; The preprocessed image data are sequentially input into a convergent edge extraction model to perform edge extraction processing on the intangible cultural heritage physical product, and a plurality of edge extraction results of the intangible cultural heritage physical product at different angles are obtained; wherein the edge extraction model is a model which sequentially includes a plurality of Conv modules, a plurality of BottleneckCSP / C3 network modules, an SPPF module, a plurality of upsampling modules and a plurality of Concat modules.

4. The method for extracting three-dimensional features according to claim 3, characterized in that: The preprocessing of a plurality of image data of the intangible cultural heritage physical product at different angles to form a plurality of preprocessed image data comprises: Based on the Gaussian filter, a plurality of image data of the intangible cultural heritage physical product at different angles are smoothed, and the plurality of image data after smoothing are uniformly converted into the RGB space using the color space to form a plurality of RGB color image data; Perform image pyramid processing on a plurality of RGB color image data to form a plurality of processed image data; grayscale the plurality of processed image data to form a plurality of grayscale image data; Perform BGR conversion on a plurality of processed image data to form a plurality of BGR image data; Perform HSV conversion on a plurality of processed image data to form a plurality of HSV image data; Scaling preprocessing is performed based on the plurality of grayscale image data, the plurality of BGR image data and the plurality of HSV image data to form a plurality of preprocessed image data.

5. The method for extracting three-dimensional features according to claim 1, characterized in that: The step of extracting target area images from corresponding pieces of image data based on a plurality of edge extraction results of the intangible cultural heritage physical product at different angles to obtain a plurality of target area images of the intangible cultural heritage physical product at different angles includes: Drawing a plurality of edge extraction results of the intangible cultural heritage physical product at different angles into a plurality of corresponding image data to form a plurality of drawing image data; The target area image extraction process is performed by taking the drawing edge of several drawing image data as the center and expanding several pixel distances to obtain several target area images of the intangible cultural heritage physical product at different angles.

6. The method for extracting three-dimensional features according to claim 1, characterized in that: The fusion processing of the three-dimensional spatial feature data at different angles to form the three-dimensional feature model of the intangible cultural heritage physical product includes: Obtaining two three-dimensional spatial feature data at adjacent angles from the three-dimensional spatial feature data at different angles, and matching the two three-dimensional spatial feature data at adjacent angles to obtain three-dimensional spatial feature points that exist in common in the two three-dimensional spatial feature data at adjacent angles; According to the three-dimensional spatial feature points that exist in common in two three-dimensional spatial feature data at adjacent angles, interpolation and fusion processing is performed using the nearest neighbor interpolation algorithm according to the angle difference between the two three-dimensional spatial feature data, and after completing the interpolation and fusion of the three-dimensional spatial feature data at different angles, a three-dimensional feature model of the intangible cultural heritage physical product is formed, wherein the angle difference is 90 degrees.

7. A device for extracting three-dimensional features of intangible cultural heritage physical products, characterized in that: The device comprises: Image acquisition module: used to perform image acquisition processing on the intangible cultural heritage physical product according to different preset angles, and obtain a number of image data of the intangible cultural heritage physical product at different angles; Edge extraction module: used for inputting a plurality of image data of the intangible cultural heritage physical product at different angles into a convergent edge extraction model to perform edge extraction processing of the intangible cultural heritage physical product, and obtaining a plurality of edge extraction results of the intangible cultural heritage physical product at different angles; Target area extraction module: used to extract target area images from corresponding several image data based on several edge extraction results of the intangible cultural heritage physical product at different angles, and obtain several target area images of the intangible cultural heritage physical product at different angles; Three-dimensional space extraction module: used for performing three-dimensional space feature extraction processing on a plurality of target area images of the intangible cultural heritage physical product at different angles based on the image three-dimensional space feature extraction model, so as to obtain the three-dimensional space feature data of the intangible cultural heritage physical product at different angles; Feature fusion module: used to fuse the three-dimensional spatial feature data at different angles to form a three-dimensional feature model of the intangible cultural heritage physical product; The three-dimensional spatial feature extraction model based on the image is used to perform three-dimensional spatial feature extraction processing on a plurality of target area images of the intangible cultural heritage physical product at different angles to obtain three-dimensional spatial feature data of the intangible cultural heritage physical product at different angles, including: Cutting each of the plurality of target area images into a plurality of non-overlapping image areas in sequence, and inputting the plurality of non-overlapping image areas corresponding to each target area image into an image three-dimensional space feature extraction model in sequence; Perform linear projection processing in the image three-dimensional space feature extraction model to obtain feature vectors corresponding to several non-overlapping image regions in each target area image; Performing image feature extraction processing on feature vectors corresponding to a plurality of non-overlapping image regions in each target region image, and establishing an association relationship between each non-overlapping image region based on the extracted image region feature information; Combining and processing the extracted image region feature information based on the association relationship to obtain multiple resolution image features corresponding to each target region image; The multiple resolution image features corresponding to each target area image are fused to form image feature data corresponding to each target area image; Decoding the image feature data corresponding to each target area image to obtain a depth image of the same size as each target area image, wherein the values ​​of different pixel points in the depth image correspond to the distance of the position from the image acquisition device; The depth image is converted into a series of spatial coordinate point sets in the coordinate matrix of the image acquisition device to obtain three-dimensional spatial feature data of the intangible cultural heritage physical product at different angles; The image three-dimensional spatial feature extraction model is the Vision Transformer model.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

9. An electronic device, characterized in that: The electronic device comprises: one or more processors; Memory; One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, and the one or more applications are configured to: execute the method according to any one of claims 1 to 6.

Images