Chinese art paper and traditional Chinese painting identification method and device based on similarity calculation
By using a similarity-based method, ultra-high-definition cameras are used to capture and analyze detailed images of traditional Chinese paintings, and multiple similarity indices are calculated. This solves the problem of relying on expert experience for the authentication of traditional Chinese paintings, and achieves more objective and fair authentication results.
Patent Information
- Application Number
- CN202410980313.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-22
- Publication Date
- 2024-11-22
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The lack of a complete set of standards for the authentication of traditional Chinese paintings in the current technology has led to an over-reliance on expert experience, which increases human error and uncertainty.
A similarity-based method was adopted, which used an ultra-high-definition camera to capture detailed images of traditional Chinese paintings, recorded the data, encrypted and stored it, and then extracted the details after decryption. The edge SSIM index, structural similarity index, contour similarity index, texture contrast index and color distribution contrast index were calculated, and weighted analysis was performed to obtain the identification results.
This has achieved objectivity and impartiality in the authentication of traditional Chinese paintings, reduced human error, and improved the stability and consistency of the authentication process.
Smart Images

Figure CN119006372A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of traditional Chinese painting identification, and in particular to a method and device for identifying rice paper traditional Chinese paintings based on similarity calculation. Background Art
[0002] In today's social life, traditional culture and art have begun to attract more attention and re-examination. Among them, Chinese painting, as a unique art form in China, has important value and significance for inheriting and carrying forward Chinese culture. Among them, Xuan paper is one of the four famous papers in China. The main raw material is the fiber from the bark of the Chinese cedar tree. The fiber contains less lignin and higher cellulose, which makes Xuan paper not easy to acidify, anti-aging, and has the advantages of fixed fiber structure.
[0003] However, Chinese paintings face many challenges and problems in identification and evaluation. In my country, calligraphy and painting collection has a long history, but there is still no complete set of evaluation standards. This has led to the identification of Chinese paintings relying too much on the identification level and experience of experts and identification institutions, and at the same time, too much human influence has been added, making the probability of "mistakes" exceed people's expectations. Summary of the invention
[0004] In view of the above problems, an object of the present invention is to provide a method for identifying traditional Chinese paintings on rice paper based on similarity calculation.
[0005] The second object of the present invention is to provide a rice paper Chinese painting identification device based on similarity calculation.
[0006] The first technical solution adopted by the present invention is: S1, based on the position parameters of the authentic Chinese painting, using an ultra-clear camera to randomly shoot the details of the Chinese painting to be identified, record the relevant data and encrypt and store them;
[0007] S2, decrypting the encrypted data, and extracting details according to the data to obtain a plurality of high-definition pictures of details of the Chinese painting to be identified;
[0008] S3, grouping the plurality of high-definition images of details of Chinese paintings to be identified and the plurality of authentic Chinese paintings' details into corresponding groups, and calculating the edge SSIM index, structural similarity index, contour similarity index, foot point contrast index, texture contrast index and color distribution contrast index of each group of images;
[0009] S4. Weight all the indexes to obtain the similarity index of each group, and perform analysis and calculation to obtain the identification result.
[0010] Optionally, S2 includes:
[0011] S21, decrypting the encrypted related data using a symmetric encryption algorithm;
[0012] S22, extracting details of the Chinese painting to be identified based on the position of the detail image of the authentic Chinese painting, the distance between the ultra-high-definition camera and the painting, and the camera magnification data, to obtain a high-definition image of the details of the Chinese painting to be identified.
[0013] Optionally, the S3 includes:
[0014] S301, using the Sobel operator to process the high-definition image of the details of the Chinese painting to be identified and the image of the detail of the authentic Chinese painting respectively, to obtain a processed image of the detail to be identified and a processed image of the authentic detail;
[0015] S302, applying two convolution filters to filter the processed detail image to be identified and the processed detail image of the authentic product respectively, to obtain corresponding vertical edge gradient values and horizontal edge gradient values;
[0016] S303, performing calculation based on the vertical edge gradient value and the horizontal edge gradient value to obtain a corresponding gradient strength;
[0017] S304, generating an edge image based on the gradient strength, and calculating the edge SSIM index through quantitative analysis, wherein the edge SSIM index calculation formula is as follows:
[0018]
[0019] In the formula, μ x is the average value of x; μ y is the average value of y; σ x 2 is the variance of x; σ y 2 is the variance of y; σ xy is the covariance of x and y; c1=(k1L) 2 ; c2=(k2L) 2 ; Where L is the dynamic range of pixel values; ; k1 and k2 are preset coefficients.
[0020] Optionally, S3 further includes:
[0021] S305, extracting the high-definition detailed image of the Chinese painting to be identified and the detailed image of the authentic Chinese painting respectively by grayscale, and obtaining the fiber structure and fiber coloring structure of the detailed image;
[0022] S306, using a Canny edge detector to extract edge information of the fiber structure and the fiber coloring structure, and calculating the structure similarity index.
[0023] Optionally, S3 further includes:
[0024] S307, extracting the contours of the high-definition detailed image of the Chinese painting to be identified and the detailed image of the authentic Chinese painting through the findContours function of OpenCV;
[0025] S308, calculating the Hu moment and the Euclidean distance of the Hu moment of the authentic Chinese painting and the Chinese painting to be authenticated, and obtaining the contour similarity index through calculation using a negative exponential function.
[0026] Optionally, S3 further includes:
[0027] S309, graying the high-definition detailed image of the Chinese painting to be authenticated and the detailed image of the authentic Chinese painting to obtain a grayscale detailed image;
[0028] S310, extracting texture features of the grayscale detail image using a grayscale co-occurrence matrix;
[0029] S311, converting the texture features into vectors, calculating the Euclidean distance of the vectors of each group of detail images, and obtaining the texture contrast index by calculating through a negative exponential function.
[0030] Optionally, S3 further includes:
[0031] S312, using a Gaussian filter to reduce noise on the high-definition image of the details of the Chinese painting to be identified and the image of the detail of the authentic Chinese painting to obtain a noise-reduced detail image;
[0032] S313, detecting corner points of the noise-reduced detail image respectively by using Harris detection algorithm to obtain key local features;
[0033] S314, extracting surrounding area features using a scale-invariant feature conversion method to obtain a feature description;
[0034] S315, using a KD-tree matching algorithm to compare the feature description of the Chinese painting to be identified with the feature description of the authentic Chinese painting, and finding corresponding corner point pairs;
[0035] S316: Perform a comparison score using the internal point ratio score to obtain the corner point comparison index.
[0036] Optionally, S3 further includes:
[0037] S317, performing color space conversion on the high-definition image of the details of the Chinese painting to be identified and the image of the details of the authentic Chinese painting, and calculating the color normalization histogram of each image using OpenCV;
[0038] S318: Flatten the histogram into a one-dimensional vector and calculate the cosine similarity to obtain the color distribution contrast index.
[0039] Optionally, the S4 includes:
[0040] S41, performing preset weight matching and weighting on the edge SSIM index, structure similarity index, contour similarity index, foot point contrast index, texture contrast index and color distribution contrast index to obtain the similarity index of each group;
[0041] S42, calculating the coefficient of variation for all the similarity indexes to obtain the coefficient of variation;
[0042] S43, averaging the similarity indexes of all groups to obtain an average similarity index;
[0043] S44, making a judgment based on the coefficient of variation and the average similarity index to obtain the identification result.
[0044] The second technical solution adopted by the present invention is: a rice paper Chinese painting identification device based on similarity calculation, comprising: a shooting module, which is used to randomly shoot the details of the Chinese painting to be identified by using an ultra-clear camera based on the position parameters of the authentic Chinese painting, record the relevant data and encrypt and store it;
[0045] A detail extraction module is used to decrypt the encrypted data and extract details based on the data to obtain a plurality of high-definition pictures of details of the Chinese painting to be identified;
[0046] An index calculation module is used to group the plurality of high-definition images of details of Chinese paintings to be identified and the plurality of authentic Chinese paintings into groups, and calculate the edge SSIM index, structural similarity index, contour similarity index, foot point contrast index, texture contrast index and color distribution contrast index of each group of images;
[0047] The identification module is used to weight all the indexes to obtain the similarity index of each group, and perform analysis and calculation to obtain the identification result.
[0048] Beneficial effects of the above technical solution:
[0049] The present invention randomly shoots the details of the Chinese painting to be identified by using an ultra-clear camera based on the position parameters of the authentic Chinese painting, records the relevant data for encryption and storage, decrypts the encrypted relevant data, and extracts the details according to the relevant data to obtain a plurality of high-definition details of the Chinese painting to be identified, and the plurality of high-definition details of the Chinese painting to be identified and the plurality of authentic details of the Chinese painting are grouped accordingly, and the edge SSIM index, structural similarity index, contour similarity index, foot point comparison index, texture comparison index and color distribution comparison index of each group of pictures are calculated, and weight matching and weighting are performed based on all the indexes to obtain the similarity index of each group, and the similarity index of each group is obtained, and the analysis and calculation are performed to obtain the identification result. Multi-dimensional detail identification is guaranteed, and the identification process is more objective and fair. BRIEF DESCRIPTION OF THE DRAWINGS
[0050] Figure 1 A flow chart of a method for identifying traditional Chinese paintings on rice paper based on similarity calculation provided by the present invention;
[0051] Figure 2 A schematic diagram of the structure of a rice paper traditional Chinese painting identification device based on similarity calculation provided by the present invention;
[0052] Figure 3 It is a schematic diagram of the tree structure process of the present invention. DETAILED DESCRIPTION
[0053] The following detailed description of the embodiments of the present invention is further described in detail in conjunction with the accompanying drawings and examples. The detailed description of the following embodiments and the accompanying drawings are used to exemplarily illustrate the principles of the present invention, but cannot be used to limit the scope of the present invention, that is, the present invention is not limited to the preferred embodiments described, and the scope of the present invention is defined by the claims.
[0054] In the description of the present invention, it should be noted that, unless otherwise specified, “plurality” means two or more than two; the terms “first”, “second”, etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance; for ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to the specific circumstances.
[0055] Embodiment 1
[0056] An embodiment of the present invention provides a method for identifying rice paper traditional Chinese paintings based on similarity calculation, such as Figure 1 As shown,
[0057] With the help of machine vision and image feature extraction (texture features, color features, contour features, corner features), we can upload ultra-high-definition images of specific locations on rice paper Chinese paintings collected by devices carrying ultra-high-definition cameras to the system, where the system analyzes the images at specific locations, extracts features, and finally encrypts and stores them. When performing identification, we collect ultra-high-definition detail images of the same location from the Chinese painting to be identified, analyze, extract, and compare features with the identity information of the corresponding Chinese painting in the system, thereby ensuring the consistency and stability of the identification process.
[0058] GCP: Genuine Chinese Painting;
[0059] CPAA: Chinese Painting Awaiting Authentication;
[0060] SSIM: Structural Similarity, is a metric that measures the similarity between two images.
[0061] Cosine similarity: Cosine similarity is calculated as the dot product of two non-zero vectors divided by their norm.
[0062] Coefficient of variation COVC: A normalized measure of the dispersion of a probability distribution, defined as the ratio of the standard deviation to the mean.
[0063] S1. Based on the position parameters of the authentic Chinese painting, use an ultra-clear camera to randomly shoot the details of the Chinese painting to be identified, record the relevant data and encrypt and store it.
[0064] Use an ultra-high-definition camera to randomly take 20 to 100 ultra-high-definition horizontal detail images at GCP. Randomly collect no less than 20 and no more than 100 ultra-high-definition (magnification 200 to 1000 times) detail images of GCP (images can be overlapped). Collect detail images of specific locations of the rice paper Chinese painting to be identified.
[0065] The detailed data of CPAA is collected according to the position parameters (horizontal position, vertical height) and magnification of GCP to ensure that the sampling position and content of the detailed map of CPAA and the detailed map of GCP are highly similar.
[0066] The ultra-high-definition detail images, detail images taken at the GCP location, and the high-end records of the camera's distance from the GCP are submitted to the system, and the data is encrypted and stored using AES.
[0067] It should be noted that the camera is ultra-high definition and will not cause damage to rice paper during the photo-taking process.
[0068] S2. Decrypt the encrypted data and extract details based on the data to obtain a plurality of high-definition pictures of details of the Chinese painting to be identified.
[0069] S21, decrypting the encrypted related data using a symmetric encryption algorithm;
[0070] Use AES to decrypt the GCP data corresponding to CPAA.
[0071] S22, extracting details of the Chinese painting to be identified based on the position of the detail image of the authentic Chinese painting, the distance between the ultra-high-definition camera and the painting, and the camera magnification data, to obtain a high-definition image of the details of the Chinese painting to be identified.
[0072] According to the location of each detail image of GCP, the distance and height between the camera and the rice paper painting, and the camera magnification data, the corresponding detailed high-definition image is extracted from CPAA, and the extracted high-definition image and location information are uploaded through the device. After collecting all the detailed data, it is submitted for identification.
[0073] S3. Group the multiple high-definition images of details of the Chinese paintings to be identified and the multiple authentic Chinese paintings' details into corresponding groups, and calculate the edge SSIM index, structural similarity index, contour similarity index, foot point contrast index, texture contrast index and color distribution contrast index of each group of images.
[0074] Each group of detail images to be compared (such as the detail image at location A of the authentic Chinese painting and the detail image at location A of the Chinese painting to be authenticated) is analyzed and compared in six dimensions to obtain a similarity index between 0 and 1.
[0075] S301, using the Sobel operator to process the high-definition image of the details of the Chinese painting to be identified and the image of the detail of the authentic Chinese painting respectively, to obtain a processed image of the detail to be identified and a processed image of the authentic detail;
[0076] Sobel operator comparison,Apply the Sobel operator to the two corresponding detail images respectively.
[0077] S302, applying two convolution filters to filter the processed detail image to be identified and the processed authentic detail image respectively, to obtain corresponding vertical edge gradient values and horizontal edge gradient values.
[0078] Two convolution filters are applied to the image: one for detecting horizontal edges (in the x direction) and the other for detecting vertical edges (in the y direction), and then the gradient values in the x and y directions are obtained.
[0079] S303: Calculate based on the vertical edge gradient value and the horizontal edge gradient value to obtain corresponding gradient strength.
[0080] Calculate the gradient strength and direction, that is, the square root of the sum of the gradient values in the x and y directions, that is:
[0081]
[0082] S304, generating an edge image based on the gradient strength, and calculating the edge SSIM index through quantitative analysis, wherein the edge SSIM index calculation formula is as follows:
[0083]
[0084] In the formula, μ x is the average value of x; μ y is the average value of y; σ x 2 is the variance of x; σ y 2 is the variance of y; σ xy is the covariance of x and y; c1=(k1L) 2 ; c2=(k2L) 2; Where L is the dynamic range of pixel values; ; k1 and k2 are preset coefficients. Default values: k1 = 0.01, k2 = 0.03.
[0085] The edge image is generated using the calculated gradient strength, and the brightness, contrast, structure and SSIM index are calculated by quantitative analysis of the edge map extracted from the two images. If the SSIM result is less than 0, it is recorded as 0, and the SSIM index is converted to the range of 0 to 1.
[0086] S305, extracting the high-definition detailed image of the Chinese painting to be identified and the detailed image of the authentic Chinese painting respectively by grayscale, and obtaining the fiber structure and fiber coloring structure of the detailed image.
[0087] Canny uses grayscale to read images and extract the fiber structure of detailed images (including the coloring structure on the fibers).
[0088] S306, using a Canny edge detector to extract edge information of the fiber structure and the fiber coloring structure, and calculating the structure similarity index.
[0089] The Canny edge detector is used to extract edge information from grayscale images. It is important to choose a suitable threshold because different thresholds will produce different edge detection results. If the SSIM result is less than 0, it is recorded as 0, and the SSIM index is converted to the range of 0 to 1.
[0090] S307, extracting the contours of the high-definition detailed image of the Chinese painting to be identified and the detailed image of the authentic Chinese painting through the findContours function of OpenCV;
[0091] Using the contour approximation method, it compresses the horizontal, vertical and diagonal segments and only keeps the endpoints of these segments as the points of the contour.
[0092] S308, calculating the Hu moment and the Euclidean distance of the Hu moment of the authentic Chinese painting and the Chinese painting to be authenticated, and obtaining the contour similarity index through calculation using a negative exponential function.
[0093] The Hu moments of GCP and CPAA were calculated, and the Euclidean distance of the Hu moments of GCP and CPAA was calculated and converted into a (0-1) score through a negative exponential function.
[0094] S309, graying the high-definition detailed image of the Chinese painting to be authenticated and the detailed image of the authentic Chinese painting to obtain a grayscale detailed image;
[0095] Grayscale the image.
[0096] S310: Extracting texture features of the grayscale detail image using a grayscale co-occurrence matrix.
[0097] The Gray Level Co-occurrence Matrix (GLCM) is used to extract texture features from images, and texture features such as contrast, uniformity, homogeneity, entropy, etc. can be obtained.
[0098] S311, converting the texture features into vectors, calculating the Euclidean distance of the vectors of each group of detail images, and obtaining the texture contrast index by calculating through a negative exponential function.
[0099] The texture features are represented in the form of vectors, and the Euclidean distance of each set of vectors is calculated and converted into a (0-1) score through a negative exponential function.
[0100] S312, using a Gaussian filter to reduce noise on the high-definition image of the details of the Chinese painting to be identified and the image of the detail of the authentic Chinese painting to obtain a noise-reduced detail image.
[0101] Use a Gaussian filter (kernel size is 3X3) to reduce image noise.
[0102] S313, detecting corner points of the noise-reduced detail image respectively by using Harris detection algorithm to obtain key local features.
[0103] Harris' detection algorithm is used to detect corner points (i.e., key local features of the content structure).
[0104] S314, extracting surrounding area features using a scale-invariant feature conversion method to obtain a feature description;
[0105] Feature description, SIFT (Scale Invariant Feature Transform) is used to extract surrounding area features.
[0106] S315, using a KD-tree matching algorithm to compare the feature description of the Chinese painting to be identified with the feature description of the authentic Chinese painting, and finding corresponding corner point pairs;
[0107] The KD-tree matching algorithm is used to compare the feature descriptors of the two images and find the corresponding corner point pairs.
[0108] S316: Perform a comparison score using the internal point ratio score to obtain the corner point comparison index.
[0109] Compare the scores using the inlier ratio score.
[0110] S317, performing color space conversion on the high-definition image of the details of the Chinese painting to be identified and the image of the details of the authentic Chinese painting, and calculating the color normalization histogram of each image using OpenCV;
[0111] Convert images from BGR or RGB color space to HSV or Lab color space, etc., and use OpenCV to calculate the color normalized histogram of each image.
[0112] S318: Flatten the histogram into a one-dimensional vector and calculate the cosine similarity to obtain the color distribution contrast index.
[0113] Flatten the histogram into a one-dimensional vector, calculate the cosine similarity (range between -1 and 1), and convert the calculation result into the range of 0 to 1 (that is, less than 0 is recorded as 0).
[0114] S4. Weight all the indexes to obtain the similarity index of each group, and perform analysis and calculation to obtain the identification result.
[0115] The scores of each group of detail images are weighted to calculate the similarity index, and the coefficient of variation COVC and average index of the similarity index of all groups are calculated to finally draw a conclusion.
[0116] S41, the edge SSIM index, structure similarity index, contour similarity index, foot point comparison index, texture comparison index and color distribution comparison index are weighted and matched with preset weights to obtain the similarity index of each group.
[0117] The similarity index of details is obtained by weighting the six similarities of the Sobel operator's edge SSIM index, structural similarity index, contour similarity index, corner contrast index, texture contrast index, and color distribution contrast index (15%, 15%, 20%, 15%, 20%, 15%).
[0118] S42, calculating the coefficient of variation for all the similarity indexes to obtain the coefficient of variation;
[0119] The coefficient of variation (COVC) was calculated for all detail contrast scores.
[0120] S43, averaging the similarity indexes of all groups to obtain an average similarity index;
[0121] The similarity index is averaged over all detail comparison scores.
[0122] S44, making a judgment based on the coefficient of variation and the average similarity index to obtain the identification result.
[0123] The following conclusions lead to a final evaluation of CPAA:
[0124] For those with a coefficient of variation COVC exceeding 0.15 (high coefficient of variation) and a similarity index exceeding 0.75, the CPAA details with lower scores were resampled.
[0125] Those with a coefficient of variation COVC exceeding 0.15 (high coefficient of variation) and a similarity index below 0.75 were marked as counterfeits and subjected to manual intervention.
[0126] Those with a coefficient of variation COVC of less than 0.15 (low coefficient of variation) and a similarity index lower than 0.75 were marked as counterfeits and subjected to manual intervention.
[0127] The coefficient of variation COVC does not exceed 0.15 (low coefficient of variation) and the similarity index is higher than 0.75, which is marked as authentic.
[0128] This embodiment uses a high-definition camera to randomly shoot details of Chinese paintings to be identified based on the position parameters of authentic Chinese paintings, records relevant data for encryption and storage, decrypts the encrypted relevant data, and extracts details based on the relevant data to obtain multiple high-definition details of Chinese paintings to be identified. Multiple high-definition details of Chinese paintings to be identified and multiple authentic details of Chinese paintings are grouped together, and the edge SSIM index, structural similarity index, contour similarity index, foot point comparison index, texture comparison index and color distribution comparison index of each group of images are calculated. Weights are matched and weighted based on all indexes to obtain the similarity index of each group, and analysis and calculation are performed to obtain the identification results. Multi-dimensional detail identification is guaranteed, making the identification process more objective and fair.
[0129] Embodiment 2
[0130] An embodiment of the present invention provides a rice paper Chinese painting identification device 200 based on similarity calculation, such as Figure 2 As shown, including:
[0131] The shooting module 201 is used to randomly shoot details of the Chinese painting to be identified using an ultra-clear camera based on the position parameters of the authentic Chinese painting, record relevant data and encrypt and store them;
[0132] A detail extraction module 202 is used to decrypt the encrypted data and extract details based on the data to obtain a plurality of high-definition pictures of details of the Chinese painting to be identified;
[0133] The index calculation module 203 is used to group the plurality of high-definition images of details of Chinese paintings to be identified and the plurality of authentic Chinese paintings into groups, and calculate the edge SSIM index, structural similarity index, contour similarity index, foot point contrast index, texture contrast index and color distribution contrast index of each group of images;
[0134] The identification module 204 is used to weight all the indexes to obtain the similarity index of each group, and perform analysis and calculation to obtain the identification result.
[0135] This embodiment is a rice paper Chinese painting identification device based on similarity calculation, including a shooting module, which is used to randomly shoot the details of the Chinese painting to be identified with an ultra-clear camera based on the position parameters of the authentic Chinese painting, record the relevant data for encryption and storage; a detail extraction module, which is used to decrypt the encrypted relevant data, and extract details according to the relevant data to obtain multiple high-definition details of the Chinese painting to be identified; an index calculation module, which is used to group the multiple high-definition details of the Chinese painting to be identified and the multiple authentic details of the Chinese painting, and calculate the edge SSIM index, structural similarity index, contour similarity index, foot point comparison index, texture comparison index and color distribution comparison index of each group of pictures; an identification module, which is used to perform weight matching and weighting based on all indexes, obtain the similarity index of each group, and perform analysis and calculation to obtain the identification result. Multi-dimensional detail identification is guaranteed, making the identification process more objective and fair.
[0136] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.
Claims
1. A method for identifying traditional Chinese paintings on rice paper based on similarity calculation. It is characterized in that include: S1. Based on the position parameters of the authentic Chinese painting, use an ultra-clear camera to randomly shoot the details of the Chinese painting to be identified, record the relevant data and encrypt and store it; S2, decrypting the encrypted data, and extracting details according to the data to obtain a plurality of high-definition pictures of details of the Chinese painting to be identified; S3, grouping the plurality of high-definition images of details of Chinese paintings to be identified and the plurality of authentic Chinese paintings' details into corresponding groups, and calculating the edge SSIM index, structural similarity index, contour similarity index, foot point contrast index, texture contrast index and color distribution contrast index of each group of images; S4. Weight all the indexes to obtain the similarity index of each group, and perform analysis and calculation to obtain the identification result.
2. The method for identifying traditional Chinese paintings on rice paper based on similarity calculation according to claim 1, It is characterized in that The S2 includes: S21, decrypting the encrypted related data using a symmetric encryption algorithm; S22, extracting details of the Chinese painting to be identified based on the position of the detail image of the authentic Chinese painting, the distance between the ultra-high-definition camera and the painting, and the camera magnification data, to obtain a high-definition image of the details of the Chinese painting to be identified.
3. The method for identifying traditional Chinese paintings on rice paper based on similarity calculation according to claim 1, It is characterized in that The S3 includes: S301, using the Sobel operator to process the high-definition image of the details of the Chinese painting to be identified and the image of the detail of the authentic Chinese painting respectively, to obtain a processed image of the detail to be identified and a processed image of the authentic detail; S302, applying two convolution filters to filter the processed detail image to be identified and the processed detail image of the authentic product respectively, to obtain corresponding vertical edge gradient values and horizontal edge gradient values; S303, performing calculation based on the vertical edge gradient value and the horizontal edge gradient value to obtain a corresponding gradient strength; S304, generating an edge image based on the gradient strength, and calculating the edge SSIM index through quantitative analysis, wherein the edge SSIM index calculation formula is as follows: In the formula, μ x is the average value of x; μ y is the average value of y; σ x 2 is the variance of x; σ y 2 is the variance of y; σ xy is the covariance of x and y; c 1 =(k 1 L) 2 ;c 2 =(k 2 L) 2 ; Where L is the dynamic range of pixel values; k 1 and k 2 is the preset coefficient.
4. The method for identifying traditional Chinese paintings on rice paper based on similarity calculation according to claim 1, It is characterized in that The S3 further includes: S305, extracting the high-definition detailed image of the Chinese painting to be identified and the detailed image of the authentic Chinese painting respectively by grayscale, and obtaining the fiber structure and fiber coloring structure of the detailed image; S306, using a Canny edge detector to extract edge information of the fiber structure and the fiber coloring structure, and calculating the structure similarity index.
5. The method for identifying traditional Chinese paintings on rice paper based on similarity calculation according to claim 1, It is characterized in that The S3 further includes: S307, extracting the contours of the high-definition detailed image of the Chinese painting to be identified and the detailed image of the authentic Chinese painting through the findContours function of OpenCV; S308, calculating the Hu moment and the Euclidean distance of the Hu moment of the authentic Chinese painting and the Chinese painting to be authenticated, and obtaining the contour similarity index through calculation using a negative exponential function.
6. The method for identifying traditional Chinese paintings on rice paper based on similarity calculation according to claim 1, It is characterized in that The S3 further includes: S309, graying the high-definition detailed image of the Chinese painting to be authenticated and the detailed image of the authentic Chinese painting to obtain a grayscale detailed image; S310, extracting texture features of the grayscale detail image using a grayscale co-occurrence matrix; S311, converting the texture features into vectors, calculating the Euclidean distance of the vectors of each group of detail images, and obtaining the texture contrast index by calculating through a negative exponential function.
7. The method for identifying traditional Chinese paintings on rice paper based on similarity calculation according to claim 1, It is characterized in that The S3 further includes: S312, using a Gaussian filter to reduce noise on the high-definition image of the details of the Chinese painting to be identified and the image of the detail of the authentic Chinese painting to obtain a noise-reduced detail image; S313, detecting corner points of the noise-reduced detail image respectively by using Harris detection algorithm to obtain key local features; S314, extracting surrounding area features using a scale-invariant feature conversion method to obtain a feature description; S315, using a KD-tree matching algorithm to compare the feature description of the Chinese painting to be identified with the feature description of the authentic Chinese painting, and finding corresponding corner point pairs; S316: Perform a comparison score using the internal point ratio score to obtain the corner point comparison index.
8. The method for identifying traditional Chinese paintings on rice paper based on similarity calculation according to claim 1, It is characterized in that The S3 further includes: S317, performing color space conversion on the high-definition image of the details of the Chinese painting to be identified and the image of the details of the authentic Chinese painting, and calculating the color normalization histogram of each image using OpenCV; S318: Flatten the histogram into a one-dimensional vector and calculate the cosine similarity to obtain the color distribution contrast index.
9. The method for identifying traditional Chinese paintings on rice paper based on similarity calculation according to claim 1, It is characterized in that The S4 includes: S41, performing preset weight matching and weighting on the edge SSIM index, structure similarity index, contour similarity index, foot point contrast index, texture contrast index and color distribution contrast index to obtain the similarity index of each group; S42, calculating the coefficient of variation for all the similarity indexes to obtain the coefficient of variation; S43, averaging the similarity indexes of all groups to obtain an average similarity index; S44, making a judgment based on the coefficient of variation and the average similarity index to obtain the identification result.
10. A device for identifying traditional Chinese paintings on rice paper based on similarity calculation. It is characterized in that include: A shooting module is used to randomly shoot details of the Chinese painting to be identified using an ultra-clear camera based on the position parameters of the authentic Chinese painting, record the relevant data and encrypt and store it; A detail extraction module is used to decrypt the encrypted data and extract details based on the data to obtain a plurality of high-definition pictures of details of the Chinese painting to be identified; An index calculation module is used to group the plurality of high-definition images of details of Chinese paintings to be identified and the plurality of authentic Chinese paintings into groups, and calculate the edge SSIM index, structural similarity index, contour similarity index, foot point contrast index, texture contrast index and color distribution contrast index of each group of images; The identification module is used to weight all the indexes to obtain the similarity index of each group, and perform analysis and calculation to obtain the identification result.
