Intelligent evaluation method and system based on ancient coin image

Abstract

The invention discloses an ancient coin image-based intelligent evaluation method and system, and the method comprises the steps: carrying out the type recognition of an ancient coin image through a deep learning model, and determining the type of a to-be-evaluated ancient coin; extracting casting identification information in the ancient coin image, and determining a casting bureau of the ancient coin by comparing with a casting bureau database; extracting texture features of the surface of the ancient coin, and determining a parent model used by the ancient coin by comparing the parent model texture database; extracting difference characteristics of different sub-money batches, and classifying the ancient coins into specific sub-money batches; and detecting a plurality of condition attributes of the ancient coin image, performing quantitative scoring on the condition attributes, and calculating the condition score of the ancient coin according to each quantitative score so as to realize intelligent evaluation of the ancient coin. According to the embodiment of the invention, type identification, casting bureau determination, parent model comparison, batch classification and condition evaluation of the ancient coins can be comprehensively realized, the accuracy and efficiency of ancient coin evaluation are improved, and the method has a wide application prospect.

Description

Technical Field

[0001] The present invention belongs to the technical field of image recognition, and in particular to an intelligent evaluation method and system based on ancient coin images. Background Art

[0002] As an important carrier of historical culture, ancient coins not only have high cultural relic value, but also have important significance in academic research, collection identification, etc. With the continuous development of the ancient coin market, the demand for identification and evaluation of ancient coins is also increasing. However, the traditional ancient coin identification method mainly relies on the experience and intuitive judgment of experts, which has certain subjectivity and limitations. Therefore, how to use modern technical means to realize the intelligent evaluation of ancient coins has become an urgent problem to be solved. Summary of the invention

[0003] The purpose of the present invention is to provide an intelligent evaluation method and system based on ancient coin images to solve the shortcomings of the prior art. It can fully realize the type identification, casting bureau determination, master mold comparison, batch classification and phase evaluation of ancient coins, improve the accuracy and efficiency of ancient coin evaluation, and has broad application prospects.

[0004] An embodiment of the present application provides an intelligent evaluation method based on ancient coin images, the method comprising:

[0005] Use deep learning models to identify the types of ancient coins images and determine the types of ancient coins to be evaluated;

[0006] Extract the casting identification information from the ancient coin image, and determine the casting bureau of the ancient coin to be evaluated by comparing it with the casting bureau database;

[0007] Extract the texture features of the surface of ancient coins, and determine the master mold used for the ancient coins to be evaluated by comparing with the master mold texture database;

[0008] Extract the difference features of different batches of sub-coins and classify the ancient coins to be evaluated into specific batches of sub-coins;

[0009] Detect multiple quality attributes of ancient coin images, quantify and score each of the multiple quality attributes, and calculate the quality score of the ancient coins based on each quantitative score to achieve intelligent evaluation of ancient coins.

[0010] Optionally, extracting the casting identification information from the ancient coin image and determining the casting bureau of the ancient coin to be evaluated by comparing it with the casting bureau database includes:

[0011] Perform image enhancement, grayscale conversion and binarization on the ancient coin image to obtain a binary image;

[0012] Detect the edge of the casting mark in the binary image, and extract the contour in the image based on the edge detection result;

[0013] According to the detected contour, the casting mark area is cut out from the binary image to obtain a separate casting mark image;

[0014] Detecting specific key points in the cropped casting mark image, wherein the specific key points describe local features of the image;

[0015] Performing feature description on the detected key points to generate a casting feature vector, wherein the casting feature vector includes texture information of the image area around the key points;

[0016] The feature vector is matched with the identification image features of each foundry in the foundry database to obtain the foundry corresponding to the best match.

[0017] Optionally, the extracting of texture features on the surface of the ancient coin and determining the master mold used for the ancient coin to be evaluated by comparing with a master mold texture database includes:

[0018] The local binary pattern (LBP) algorithm is used to extract the micro-texture features of ancient coin images and generate the master feature vector.

[0019] The principal component analysis (PCA) algorithm is used to reduce the dimension of the mother-model feature vector, and the cosine similarity between the mother-model feature vector and the feature vector in the mother-model texture database is calculated, wherein the feature vectors in the mother-model texture database are extracted by the LBP algorithm and reduced in dimension by the PCA algorithm;

[0020] The values ​​of each cosine similarity are sorted, and the mother pattern in the mother pattern texture database corresponding to the feature vector with the highest similarity is determined as the mother pattern used for the ancient coin to be evaluated.

[0021] Optionally, the appearance attributes include: surface features, patina features, notch features, and detail features; the calculation formula for the appearance score is:

[0022]

[0023] Among them, S is the quality score, C_w is the weighted appearance score, H_b is the historical background score, R is the rarity score, A is the age score, T is the known stock score, C_d is the market demand score, and K is the comprehensive adjustment factor.

[0024] Optionally, the calculation formula of the appearance weighted score is:

[0025]

[0026] Among them, S_f is the surface feature score, P is the patina score, C_p is the notch score, and D is the overall detail score.

[0027] Another embodiment of the present application provides an intelligent evaluation system based on ancient coin images, the system comprising:

[0028] The recognition module is used to use the deep learning model to identify the types of ancient coins images and determine the types of ancient coins to be evaluated;

[0029] The first extraction module is used to extract the casting identification information in the ancient coin image, and determine the casting bureau of the ancient coin to be evaluated by comparing it with the casting bureau database;

[0030] The second extraction module is used to extract the texture features of the surface of the ancient coin and determine the master mold used for the ancient coin to be evaluated by comparing it with the master mold texture database;

[0031] The third extraction module is used to extract the difference features of different sub-coin batches and classify the ancient coins to be evaluated into specific sub-coin batches;

[0032] The detection module is used to detect multiple quality attributes of ancient coin images, quantify and score the multiple quality attributes respectively, and calculate the quality score of the ancient coins based on each quantitative score to achieve intelligent evaluation of ancient coins.

[0033] Yet another embodiment of the present application provides a storage medium, wherein the storage medium stores a computer program, wherein the computer program is configured to execute any of the above methods when running.

[0034] Yet another embodiment of the present application provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute any of the methods described above.

[0035] Compared with the prior art, the present invention provides an intelligent evaluation method based on ancient coin images, which uses a deep learning model to identify the types of ancient coin images to determine the types of ancient coins to be evaluated; extracts casting identification information in the ancient coin images, and determines the casting bureau of the ancient coins to be evaluated by comparing with the casting bureau database; extracts texture features on the surface of the ancient coins, and determines the mother mold used by the ancient coins to be evaluated by comparing with the mother mold texture database; extracts difference features of different sub-coin batches, and classifies the ancient coins to be evaluated into specific sub-coin batches; detects multiple phase attributes of the ancient coin images, quantifies and scores the multiple phase attributes respectively, and calculates the phase score of the ancient coins according to each quantified score to realize intelligent evaluation of the ancient coins, thereby being able to fully realize type identification, casting bureau determination, mother mold comparison, batch classification and phase evaluation of ancient coins, improve the accuracy and efficiency of ancient coin evaluation, and has broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 A hardware structure block diagram of a computer terminal for an intelligent evaluation method based on ancient coin images provided by an embodiment of the present invention;

[0037] Figure 2 A schematic diagram of a flow chart of an intelligent evaluation method based on ancient coin images provided by an embodiment of the present invention;

[0038] Figure 3 A schematic diagram of the structure of an intelligent evaluation system based on ancient coin images provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0039] The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, but should not be construed as limiting the present invention.

[0040] The embodiment of the present invention firstly provides an intelligent evaluation method based on ancient coin images, which can be applied to electronic devices such as computer terminals, specifically ordinary computers, etc.

[0041] The following describes it in detail by taking running on a computer terminal as an example. Figure 1 The hardware structure block diagram of a computer terminal for an intelligent evaluation method based on ancient coin images provided by an embodiment of the present invention. Figure 1 As shown, the computer terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data. Optionally, the computer terminal may also include a transmission device 106 for communication functions and an input and output device 108. It can be understood by those skilled in the art that Figure 1The structure shown is only for illustration and does not limit the structure of the above-mentioned computer terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations shown.

[0042] The memory 104 can be used to store software programs and modules of application software, such as program instructions / modules corresponding to the intelligent evaluation method based on ancient coin images in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the computer terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0043] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of a computer terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0044] See also Figure 2 The embodiment of the present invention provides an intelligent evaluation method based on ancient coin images, which may include the following steps:

[0045] S201, using a deep learning model to identify the type of ancient coin images, and determine the type of the ancient coin to be evaluated; specifically, one implementation method may include:

[0046] Step 1: Data collection and annotation

[0047] -Data collection: Collect a large number of high-quality image datasets of ancient coins. Ensure that the images cover as many types of ancient coins as possible, including samples from different dynasties, different denominations, and different mints.

[0048] -Data labeling: Label the collected images and label each image with a corresponding type label. These labels should accurately reflect the specific type information of ancient coins.

[0049] Step 2: Data Preprocessing

[0050] -Image enhancement: Perform data enhancement on images to increase the diversity of the data set. Common enhancement techniques include rotation, translation, scaling, mirroring, etc.

[0051] -Image normalization: Normalize the image pixel values ​​to the range of [0, 1] to improve the convergence speed and stability of the model.

[0052] - Image Segmentation: Segment the image when necessary to separate the coin from the background, ensuring that the model focuses on the coin itself.

[0053] Step 3: Build a Convolutional Neural Network (CNN) model

[0054] -Model architecture design: Design a CNN model architecture suitable for ancient coin image type recognition. Common architectures include multiple convolutional layers, pooling layers, fully connected layers, etc.

[0055] -Convolutional layer: extracts local features of the image and captures important information such as the texture and pattern of the coin.

[0056] -Pooling layer: reduces the dimension of the feature map, reduces the amount of computation, and retains important features.

[0057] -Fully connected layer: Integrate the extracted features to generate the final classification results.

[0058] -Select activation function: Common activation functions include ReLU (Rectified Linear Unit) and Softmax. ReLU is used between convolutional layers and fully connected layers, and Softmax is used in the output layer for classification. -Model compilation: Select a suitable loss function (such as cross entropy loss) and optimizer (such as Adam optimizer) for model compilation.

[0059] Step 4: Model training

[0060] - Training set and validation set division: The labeled data set is divided into training set and validation set. The training set is used for model training, and the validation set is used for model performance evaluation.

[0061] -Model training: Input the training set into the CNN model for training, optimize the model parameters through iteration, and minimize the loss function. During the training process, use the validation set to monitor the performance of the model to prevent overfitting.

[0062] -Hyperparameter tuning: According to the performance of the model on the validation set, adjust the model's hyperparameters (such as learning rate, batch size, convolution kernel size, etc.) to improve the model's recognition accuracy.

[0063] Step 5: Model evaluation and optimization

[0064] -Model evaluation: Evaluate the trained model on an independent test set, calculate accuracy, recall, F1 score and other indicators, and comprehensively evaluate the performance of the model.

[0065] -Model optimization: Optimize the model based on the evaluation results. You can try to adjust the model architecture, add data enhancement methods, introduce regularization techniques (such as Dropout), etc. to improve the generalization ability of the model.

[0066] Step 6: Model Deployment

[0067] -Model saving: Save the trained CNN model in a loadable format (such as H5 format).

[0068] -Model deployment: Deploy the model to the actual application environment, provide an API interface for users to upload ancient coin images, automatically identify the type, and return the identification results.

[0069] -Continuous updates: Regularly update the model, add new image data and labels, and continuously improve the model's recognition capabilities.

[0070] Through the above steps, the convolutional neural network (CNN) can effectively realize the type recognition of ancient coin images and determine the type of ancient coins to be detected. This process includes data collection and preprocessing, model construction and training, model evaluation and optimization, and final model deployment and continuous updating to ensure the efficiency and accuracy of the model in practical applications.

[0071] The main functions and significance of step S201 include:

[0072] 1. Automated identification: Through deep learning models, the types of ancient coins can be automatically identified, reducing dependence on manual experience and improving identification efficiency and accuracy.

[0073] 2. Improve accuracy: The deep learning model can learn the subtle features in the images of ancient coins. Through large amounts of data training, it can improve the accuracy of recognition and avoid misjudgments in manual recognition.

[0074] 3. Standardization processing: Using deep learning models for identification can achieve standardization of ancient coin types and ensure the consistency and reliability of the evaluation process.

[0075] 4. Efficient processing: The deep learning model has a fast processing speed and can process a large number of ancient coin images in a short time, meeting the needs of large-scale evaluation.

[0076] S202, extracting casting identification information from the ancient coin image, and determining the casting bureau of the ancient coin to be evaluated by comparing it with the casting bureau database;

[0077] Specifically, the ancient coin image can be subjected to image enhancement, grayscale conversion and binarization processing to obtain a binarized image;

[0078] Explanation:

[0079] - Image enhancement: Improve image quality and highlight the details of casting logos through image enhancement techniques (such as contrast adjustment, denoising, etc.).

[0080] -Grayscale processing: Convert color images into grayscale images to simplify subsequent processing steps and reduce the amount of calculation.

[0081] - Binarization: Convert the grayscale image into a binary image, retaining only the black and white pixels in the image, highlighting the outline and shape of the casting logo.

[0082] Function and significance:

[0083] - Improved image quality, ensuring accurate extraction of casting identification information in subsequent processing steps.

[0084] -Simplify image processing steps, reduce calculation amount and speed up processing.

[0085] -Highlight the outline and shape of the casting logo to facilitate subsequent edge detection and feature extraction.

[0086] Detect the edge of the casting mark in the binary image, and extract the contour in the image based on the edge detection result;

[0087] Explanation:

[0088] -Edge detection: Use an edge detection algorithm (such as the Canny algorithm) to detect the edges in the binary image and extract the outline of the casting logo.

[0089] -Contour extraction: Based on the edge detection results, the contour information in the image is extracted to retain the shape features of the casting logo.

[0090] Function and significance:

[0091] -Accurately extract the edges and contours of casting marks, laying the foundation for subsequent feature extraction and comparison.

[0092] -Edge detection and contour extraction can effectively reduce background interference and focus on the key features of the casting mark.

[0093] According to the detected contour, the casting mark area is cut out from the binary image to obtain a separate casting mark image;

[0094] Explanation:

[0095] -Contour positioning: Based on the contour information detected in the previous step, locate the specific position of the casting mark in the image.

[0096] -Image cropping: crop the casting logo area from the binary image to generate a separate casting logo image for subsequent processing.

[0097] Function and significance:

[0098] -Extract individual casting logo images, remove background interference, and retain key features.

[0099] -It facilitates the subsequent feature point detection and feature description, and improves the accuracy of comparison.

[0100] Detecting specific key points in the cropped casting mark image, wherein the specific key points describe local features of the image;

[0101] Explanation:

[0102] -Keypoint Detection: Use keypoint detection algorithms (such as SIFT, SURF, etc.) to detect specific keypoints in the cropped casting logo image, which describe the local features of the image.

[0103] Function and significance:

[0104] -Extract local features from casting mark images to provide a basis for subsequent feature description and matching.

[0105] -Key points describe unique features of the image and help distinguish between the logos of different foundries.

[0106] Performing feature description on the detected key points to generate a casting feature vector, wherein the casting feature vector includes texture information of the image area around the key points;

[0107] Explanation:

[0108] - Feature description: Perform feature description on the detected key points and generate feature vectors (such as SIFT feature vectors) containing texture information of the image area around the key points.

[0109] Function and significance:

[0110] -The feature vector contains detailed texture information of the image, which helps in subsequent feature matching.

[0111] -Generate feature vectors to facilitate comparison with features in the foundry database and improve recognition accuracy.

[0112] The feature vector is matched with the identification image features of each foundry in the foundry database to obtain the foundry corresponding to the best match.

[0113] Explanation:

[0114] - Feature matching: Match the generated casting feature vector with the logo image features in the mint database to find the mint logo that is most similar to the ancient coin to be evaluated.

[0115] - Best match: Based on the matching results, select the mint mark with the highest similarity to determine the mint of the ancient coin to be evaluated.

[0116] Function and significance:

[0117] -Through feature matching, the minting bureau origin of ancient coins can be accurately determined, improving the accuracy of evaluation.

[0118] -Automated feature matching process improves evaluation efficiency and reduces manual intervention.

[0119] The main functions and significance of step S202 include:

[0120] 1. Determine the casting source: By identifying the casting mark information of ancient coins, you can accurately determine the source of its casting bureau and help trace the historical background.

[0121] 2. Improve recognition accuracy: The casting mark is an important feature of the ancient coin types. By comparing with the casting bureau database, the recognition accuracy can be improved and misjudgment can be avoided.

[0122] 3. Standardized evaluation process: Through automated image processing and comparison, the evaluation process can be standardized to ensure the consistency and reliability of the evaluation results.

[0123] 4. Efficient processing: Automated extraction and comparison of casting identification information improves evaluation efficiency and meets the needs of large-scale ancient coin evaluation.

[0124] S203, extracting texture features of the surface of the ancient coin, and determining the master mold used for the ancient coin to be evaluated by comparing it with the master mold texture database;

[0125] Specifically, the local binary pattern LBP algorithm can be used to extract the micro texture features of the ancient coin image and generate the mother model feature vector;

[0126] Explanation:

[0127] -LBP algorithm: Local Binary Pattern (LBP) is an algorithm for texture feature extraction. It converts texture information into a binary pattern and generates a feature vector by analyzing the relationship between each pixel in the image and its surrounding pixels.

[0128] Function and significance:

[0129] -LBP algorithm can capture the subtle texture features of ancient coin images. The generated feature vector contains rich texture information, providing an accurate data basis for subsequent comparison.

[0130] -The LBP algorithm is used to extract micro-texture features, which ensures a detailed description of the texture features and helps improve the accuracy of the comparison.

[0131] The principal component analysis (PCA) algorithm is used to reduce the dimension of the mother-model feature vector, and the cosine similarity between the mother-model feature vector and the feature vector in the mother-model texture database is calculated, wherein the feature vectors in the mother-model texture database are extracted by the LBP algorithm and reduced in dimension by the PCA algorithm;

[0132] Explanation:

[0133] -PCA algorithm: Principal component analysis (PCA) is a dimensionality reduction technique that reduces the dimension of the feature vector, retains the most representative part, and reduces the computational complexity.

[0134] - Cosine similarity: Evaluate the similarity between the parent feature vector and the feature vector in the database by calculating the cosine similarity between them.

[0135] Function and significance:

[0136] -The PCA algorithm reduces the dimension of the feature vector, simplifies the calculation process, retains the key features, and improves the matching efficiency.

[0137] -Cosine similarity can effectively evaluate the similarity between feature vectors and accurately identify the most matching parent model features.

[0138] The values ​​of each cosine similarity are sorted, and the mother pattern in the mother pattern texture database corresponding to the feature vector with the highest similarity is determined as the mother pattern used for the ancient coin to be evaluated.

[0139] Explanation:

[0140] -Similarity sorting: Sort the calculated cosine similarities and find the eigenvector with the highest similarity.

[0141] - Master model determination: Determine the master model used for the ancient coin to be evaluated based on the feature vector corresponding to the highest similarity.

[0142] Function and significance:

[0143] -By sorting the similarities, the most matching parent model features can be quickly located to ensure the accuracy and reliability of the evaluation results.

[0144] - The process of determining the master model is automated, reducing manual intervention and improving evaluation efficiency and standardization.

[0145] The main functions and significance of step S203 include:

[0146] 1. Accurately identify the source of the master mold: By extracting and comparing texture features, the master mold used in ancient coins can be accurately identified to help determine its historical background and value.

[0147] 2. Improve evaluation accuracy: Texture feature is one of the important characteristics of ancient coins. Through detailed texture analysis and database comparison, the accuracy of evaluation can be greatly improved.

[0148] 3. Automation and standardization: Using computer technology to extract and compare texture features realizes the automation and standardization of the evaluation process, reducing human intervention and errors.

[0149] 4. Efficiently process a large number of samples: Automated texture feature matching technology can efficiently process a large number of ancient coin samples to meet large-scale evaluation needs.

[0150] S204, extracting the difference features of different batches of sub-coins, and classifying the ancient coins to be evaluated into specific batches of sub-coins; specifically, one implementation method may include:

[0151] Step 1: Preprocess the ancient coin image to remove noise and background interference

[0152] Explanation:

[0153] -Preprocessing: Use image processing algorithms to preprocess ancient coin images, remove noise and background interference, and retain clear coin surface images.

[0154] Specific operations:

[0155] -Apply Gaussian filter to denoise the image.

[0156] -Use image segmentation algorithms (such as GrabCut algorithm) to remove the background and highlight the ancient coin image.

[0157] Step 2: Extract the main feature points in the ancient coin image

[0158] Explanation:

[0159] - Feature point extraction: The main feature points on the surface of ancient coins are extracted through image processing algorithms to obtain important texture and shape information on the coin surface.

[0160] Specific operations:

[0161] -Use SIFT (Scale-Invariant Feature Transform) algorithm to extract key feature points in the image.

[0162] -Generate a descriptor of the feature point and record the local image information of the feature point.

[0163] Step 3: Use clustering algorithm to group the extracted feature points

[0164] Explanation:

[0165] -Feature point grouping: The extracted feature points are grouped according to their characteristics to form a specific feature set for subsequent classification processing.

[0166] Specific operations:

[0167] - Apply K-means clustering algorithm to divide feature points into multiple clusters, each cluster represents a type of feature.

[0168] -Determine the center point of each cluster as the representative of the feature class.

[0169] Step 4: Calculate the statistical characteristics of each feature cluster

[0170] Explanation:

[0171] -Statistical feature calculation: Perform statistical analysis on each feature cluster and calculate its statistical features, such as mean value, variance, etc.

[0172] Specific operations:

[0173] -Calculate statistics such as the number of feature points, average position, average descriptor, etc. for each feature cluster.

[0174] -Record these statistical features as feature vectors of feature clusters.

[0175] Step 5: Compare the feature clusters of the ancient coins to be evaluated with the feature clusters of the known sub-coin batches

[0176] Explanation:

[0177] - Feature cluster comparison: Compare the feature cluster of the ancient coin to be evaluated with the feature clusters of known sub-coin batches in the database to determine the sub-coin batch to which it belongs.

[0178] Specific operations:

[0179] -Use cosine similarity or Euclidean distance to calculate the similarity between the feature cluster of the ancient coin to be evaluated and the feature cluster of the known sub-coin batches.

[0180] -Based on the similarity results, determine the sub-coin batch to which the ancient coin to be evaluated is most likely to belong.

[0181] Step 6: Classify the batches based on the matching results

[0182] Explanation:

[0183] -Batch classification: According to the matching results of feature clusters, the ancient coins to be evaluated are classified into specific sub-coin batches.

[0184] Specific operations:

[0185] -Determine the sub-coin batch to which the ancient coin to be evaluated belongs based on the feature cluster matching results with the highest similarity.

[0186] -Update database records and classify the ancient coins to be evaluated into corresponding batches.

[0187] Through the above specific implementation method, it is possible to accurately classify different batches of ancient coins, provide accurate data support for the intelligent evaluation of ancient coins, and improve the accuracy and reliability of the evaluation. At the same time, through batch classification, it is helpful to better understand the historical background and value of ancient coins, and provide strong support for research and collection. The main functions and significance of step S204 include:

[0188] 1. Accurate batch classification: By extracting differential features, ancient coins can be accurately classified into specific sub-coin batches, which helps to understand their historical background and minting period in more detail.

[0189] 2. Improve evaluation accuracy: Batch classification makes the evaluation of ancient coins more refined, reduces errors, and improves the accuracy of the overall evaluation.

[0190] 3. Systematic management: Batch classification helps to systematically manage and record the information of ancient coins, providing valuable data support for research and collection.

[0191] 4. Value assessment: Different batches of ancient coins may have differences in historical value and market value. Batch classification can enable more accurate value assessment.

[0192] S205, detecting multiple quality attributes of the ancient coin image, quantifying and scoring the multiple quality attributes respectively, and calculating the quality score of the ancient coin according to each quantified score, so as to realize intelligent evaluation of the ancient coin.

[0193] Specifically, the appearance attributes may include: surface characteristics, patina characteristics, notch characteristics, and detail characteristics.

[0194] Surface characteristics refer to the overall state of the surface of ancient coins, including wear, scratches, rust, etc. Surface characteristics reflect the preservation status of ancient coins. Good surface characteristics usually indicate that the ancient coins are well preserved and have an important impact on their appearance. Specific detection methods include:

[0195] 1. Image preprocessing: Use image denoising, grayscale and enhancement processing to improve image quality.

[0196] 2. Edge detection: Use the Canny edge detection algorithm to identify surface scratches and cracks.

[0197] 3. Texture analysis: Surface wear and corrosion are evaluated using the gray-level co-occurrence matrix (GLCM), and surface characteristics are quantified by calculating texture features such as contrast, correlation, energy, and uniformity.

[0198] The patina feature refers to the natural oxide layer formed on the surface of ancient coins, which reflects the history and age of the ancient coins. Uniform and natural patina usually enhances the collection value of ancient coins. Specific detection methods:

[0199] 1. Color histogram analysis: Extract the color histogram of the ancient coin image and analyze the patina color distribution.

[0200] 2. Color clustering: Use the K-means clustering algorithm to distinguish the patina color areas and evaluate the uniformity and naturalness of the patina.

[0201] 3. Texture analysis: The local binary pattern (LBP) method is used to analyze the texture characteristics of the patina.

[0202] The gap feature refers to the gaps and damage on the edge and surface of ancient coins, which affects the overall integrity and appearance of the ancient coins. Specific detection methods:

[0203] 1. Morphological processing: Use dilation and erosion algorithms to identify the gap locations.

[0204] 2. Edge detection: Use SobeI edge detection algorithm to extract the gap contour.

[0205] 3. Gap statistics: calculate the number and area of ​​gaps, and evaluate the impact of gaps on ancient coins.

[0206] Detailed features include the casting texture, clarity of text and patterns on ancient coins. These details reflect the production process and historical value of ancient coins. Specific detection methods:

[0207] 1. High-pass filter: Enhance the detailed features in the ancient coin image.

[0208] 2. Shape analysis: Use the Histogram of Oriented Gradients (HOG) to evaluate the clarity of text and patterns.

[0209] 3. Feature matching: feature matching with known standard patterns to quantify detail clarity.

[0210] According to the preset scoring criteria, the surface characteristics, patina characteristics, notch characteristics and detail characteristics are scored. The scoring range for each phase attribute is 0-10 points, and the specific score is determined based on the test results and scoring criteria.

[0211] Specifically, a calculation formula for the quality score can be:

[0212]

[0213] Among them, S is the appearance score, which is a quantitative evaluation of the overall appearance of the ancient coin; C_w is the weighted appearance score, which combines the scores of surface features, patina features, gap features and detail features; H _ b is the historical background score, which evaluates the historical value of the ancient coins; R is the rarity score, which reflects the rarity of the ancient coins; A is the age score, which is scored according to the age of the ancient coins; T is the known stock score, which evaluates the stock situation of ancient coins on the market; C_d is the market demand score, which reflects the market demand for the ancient coins; K is a comprehensive adjustment factor, which is used to adjust the scoring result to make it more in line with the actual situation.

[0214] The historical background parameter mainly considers the historical era of the coin, the importance of that era, and the role the coin played in it. For example, the same coin, if it is a Kaiyuan Tongbao from the Tang Dynasty, will have a higher score because of the importance of that era in Chinese history. For another example, a special commemorative coin issued by an ancient dynasty will also receive a higher score because of its special significance.

[0215] Rarity can be found in various coin catalogs, auction records, and professional coin journals. It takes into account how rare the coin is and whether there are other coins of the same type circulating in the market.

[0216] The market demand parameter takes into account the current market demand for the coin. It can be determined based on auction prices, search popularity and trading volume on coin trading platforms. The higher the score, the higher the market demand.

[0217] The known number of coins can be determined by reference to historical records, museum collections, and the expertise of numismatic experts. It represents the total known number of coins of that type.

[0218] The scores of H_b, R, A, T, C_d and K are obtained through historical data, literature and market research. Among them, H_b and R can be determined through expert evaluation or reference literature; A is determined according to the manufacturing age of ancient coins and historical literature; T is obtained through market research and existing literature data; C_d is determined through market demand analysis; K is adjusted according to actual needs to make the scoring results more accurate.

[0219] Specifically, a calculation formula for the appearance weighted score can be:

[0220]

[0221] Among them, S_f is the surface feature score, P is the patina score, C_p is the notch score, and D is the overall detail score.

[0222] -Surface feature score: The score can be given based on the surface integrity, cleanliness and gloss of the coin. A score of 0 means that the ancient coin has severe wear, missing or severe surface damage; a score of 10 means that the surface is almost perfect, with no signs of wear and excellent gloss. Other situations are given scores from 1 to 9 according to the corresponding standards.

[0223] - Patina score: 0 points means there is no natural patina and the surface has been artificially treated; 10 points means that the patina is extremely natural, uniform and beautiful, showing the age of the ancient coin. Other situations are also scored according to the corresponding standards.

[0224] -Crack score: 0 points means the coin has a serious chip or is severely damaged; 10 points means the coin has intact edges.

[0225] -Detail score: The score is given based on the subtle details of the coin pattern and text. 0 points means the pattern is blurred and the details are distorted; 10 points means the pattern is clear and the details are distinct, and even the hammer marks during casting can be seen.

[0226] The main functions and significance of step S205 include:

[0227] 1. Comprehensive evaluation: By detecting and quantifying multiple appearance attributes, we can comprehensively evaluate all aspects of ancient coins and provide a comprehensive appearance evaluation.

[0228] 2. Objective quantification: The quantification process of appearance scoring ensures the objectivity and consistency of the evaluation and reduces the influence of human subjective factors.

[0229] 3. Refine the evaluation criteria: Quantify and score multiple appearance attributes separately to make the evaluation criteria more detailed and clear, which will help to more accurately reflect the actual status of ancient coins.

[0230] 4. Improve evaluation efficiency: Utilize automated image processing and scoring algorithms to improve evaluation efficiency and reduce the time and cost of manual evaluation.

[0231] 5. Market reference: The quality score provides a reliable reference for the market value assessment of ancient coins, facilitating value judgment in collection and trading.

[0232] It can be seen that the deep learning model is used to identify the type of ancient coin images to determine the type of ancient coins to be evaluated; the casting identification information in the ancient coin images is extracted, and the casting bureau of the ancient coins to be evaluated is determined by comparing with the casting bureau database; the texture features on the surface of the ancient coins are extracted, and the mother mold used for the ancient coins to be evaluated is determined by comparing with the mother mold texture database; the difference features of different sub-coin batches are extracted, and the ancient coins to be evaluated are classified into specific sub-coin batches; multiple phase attributes of the ancient coin images are detected, and multiple phase attributes are quantitatively scored respectively, and the phase score of the ancient coins is calculated according to each quantitative score to realize the intelligent evaluation of ancient coins, so as to fully realize the type identification, casting bureau determination, mother mold comparison, batch classification and phase evaluation of ancient coins, improve the accuracy and efficiency of ancient coin evaluation, and have broad application prospects.

[0233] Another embodiment of the present invention provides an intelligent evaluation system based on ancient coin images. Figure 3 , the system may include:

[0234] The identification module 301 is used to identify the type of the ancient coin image using a deep learning model to determine the type of the ancient coin to be evaluated;

[0235] The first extraction module 302 is used to extract the casting identification information in the ancient coin image, and determine the casting bureau of the ancient coin to be evaluated by comparing it with the casting bureau database;

[0236] The second extraction module 303 is used to extract the texture features of the surface of the ancient coin and determine the master mold used for the ancient coin to be evaluated by comparing it with the master mold texture database;

[0237] The third extraction module 304 is used to extract the difference characteristics of different sub-coin batches and classify the ancient coins to be evaluated into specific sub-coin batches;

[0238] The detection module 305 is used to detect multiple quality attributes of the ancient coin image, quantify and score the multiple quality attributes respectively, and calculate the quality score of the ancient coin according to each quantitative score to realize intelligent evaluation of the ancient coin.

[0239] It can be seen that the deep learning model is used to identify the type of ancient coin images to determine the type of ancient coins to be evaluated; the casting identification information in the ancient coin images is extracted, and the casting bureau of the ancient coins to be evaluated is determined by comparing with the casting bureau database; the texture features on the surface of the ancient coins are extracted, and the mother mold used for the ancient coins to be evaluated is determined by comparing with the mother mold texture database; the difference features of different sub-coin batches are extracted, and the ancient coins to be evaluated are classified into specific sub-coin batches; multiple phase attributes of the ancient coin images are detected, and multiple phase attributes are quantitatively scored respectively, and the phase score of the ancient coins is calculated according to each quantitative score to realize the intelligent evaluation of ancient coins, so as to fully realize the type identification, casting bureau determination, mother mold comparison, batch classification and phase evaluation of ancient coins, improve the accuracy and efficiency of ancient coin evaluation, and have broad application prospects.

[0240] An embodiment of the present invention further provides a storage medium, in which a computer program is stored, wherein the computer program is configured to execute the steps of any of the above method embodiments when running.

[0241] Specifically, in this embodiment, the above storage medium may be configured to store a computer program for performing the following steps:

[0242] S201, using a deep learning model to identify the type of the ancient coin image to determine the type of the ancient coin to be evaluated;

[0243] S202, extracting casting identification information from the ancient coin image, and determining the casting bureau of the ancient coin to be evaluated by comparing it with the casting bureau database;

[0244] S203, extracting texture features of the surface of the ancient coin, and determining the master mold used for the ancient coin to be evaluated by comparing it with the master mold texture database;

[0245] S204, extracting the difference features of different batches of sub-coins, and classifying the ancient coins to be evaluated into specific batches of sub-coins;

[0246] S205, detecting multiple quality attributes of the ancient coin image, quantifying and scoring the multiple quality attributes respectively, and calculating the quality score of the ancient coin according to each quantified score, so as to realize intelligent evaluation of the ancient coin.

[0247] It can be seen that the deep learning model is used to identify the type of ancient coin images to determine the type of ancient coins to be evaluated; the casting identification information in the ancient coin images is extracted, and the casting bureau of the ancient coins to be evaluated is determined by comparing with the casting bureau database; the texture features on the surface of the ancient coins are extracted, and the mother mold used for the ancient coins to be evaluated is determined by comparing with the mother mold texture database; the difference features of different sub-coin batches are extracted, and the ancient coins to be evaluated are classified into specific sub-coin batches; multiple phase attributes of the ancient coin images are detected, and multiple phase attributes are quantitatively scored respectively, and the phase score of the ancient coins is calculated according to each quantitative score to realize the intelligent evaluation of ancient coins, so as to fully realize the type identification, casting bureau determination, mother mold comparison, batch classification and phase evaluation of ancient coins, improve the accuracy and efficiency of ancient coin evaluation, and have broad application prospects.

[0248] An embodiment of the present invention further provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0249] Specifically, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0250] Specifically, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0251] S201, using a deep learning model to identify the type of the ancient coin image to determine the type of the ancient coin to be evaluated;

[0252] S202, extracting casting identification information from the ancient coin image, and determining the casting bureau of the ancient coin to be evaluated by comparing it with the casting bureau database;

[0253] S203, extracting texture features of the surface of the ancient coin, and determining the master mold used for the ancient coin to be evaluated by comparing it with the master mold texture database;

[0254] S204, extracting the difference features of different batches of sub-coins, and classifying the ancient coins to be evaluated into specific batches of sub-coins;

[0255] S205, detecting multiple quality attributes of the ancient coin image, quantifying and scoring the multiple quality attributes respectively, and calculating the quality score of the ancient coin according to each quantified score, so as to realize intelligent evaluation of the ancient coin.

[0256] It can be seen that the deep learning model is used to identify the type of ancient coin images to determine the type of ancient coins to be evaluated; the casting identification information in the ancient coin images is extracted, and the casting bureau of the ancient coins to be evaluated is determined by comparing with the casting bureau database; the texture features on the surface of the ancient coins are extracted, and the mother mold used for the ancient coins to be evaluated is determined by comparing with the mother mold texture database; the difference features of different sub-coin batches are extracted, and the ancient coins to be evaluated are classified into specific sub-coin batches; multiple phase attributes of the ancient coin images are detected, and multiple phase attributes are quantitatively scored respectively, and the phase score of the ancient coins is calculated according to each quantitative score to realize the intelligent evaluation of ancient coins, so as to fully realize the type identification, casting bureau determination, mother mold comparison, batch classification and phase evaluation of ancient coins, improve the accuracy and efficiency of ancient coin evaluation, and have broad application prospects.

[0257] The above describes in detail the structure, features and effects of the present invention based on the embodiments shown in the drawings. The above is only a preferred embodiment of the present invention, but the present invention is not limited to the scope of implementation shown in the drawings. Any changes made according to the concept of the present invention, or modifications to equivalent embodiments with equivalent changes, which still do not exceed the spirit covered by the description and drawings, should be within the protection scope of the present invention.

Claims

1. An intelligent evaluation method based on ancient coin images, characterized in that: The method comprises: Use deep learning models to identify the types of ancient coin images and determine the types of ancient coins to be evaluated; Extract the casting identification information from the ancient coin image, and determine the casting bureau of the ancient coin to be evaluated by comparing it with the casting bureau database; Extract the texture features of the surface of ancient coins, and determine the master mold used for the ancient coins to be evaluated by comparing with the master mold texture database; Extract the difference features of different batches of sub-coins and classify the ancient coins to be evaluated into specific batches of sub-coins; Detect multiple quality attributes of ancient coin images, quantify and score each of the multiple quality attributes, and calculate the quality score of the ancient coins based on each quantitative score to achieve intelligent evaluation of ancient coins.

2. The method according to claim 1, characterized in that The step of extracting the casting identification information from the ancient coin image and determining the casting bureau of the ancient coin to be evaluated by comparing the casting bureau database includes: Perform image enhancement, grayscale conversion and binarization on the ancient coin image to obtain a binary image; Detect the edge of the casting mark in the binary image, and extract the contour in the image based on the edge detection result; According to the detected contour, the casting mark area is cut out from the binary image to obtain a separate casting mark image; Detecting specific key points in the cropped casting mark image, wherein the specific key points describe local features of the image; Performing feature description on the detected key points to generate a casting feature vector, wherein the casting feature vector includes texture information of the image area around the key points; The feature vector is matched with the identification image features of each foundry in the foundry database to obtain the foundry corresponding to the best match.

3. The method according to claim 2, characterized in that The extracting of the texture features of the surface of the ancient coin and determining the master mold used for the ancient coin to be evaluated by comparing the master mold texture database include: The local binary pattern (LBP) algorithm is used to extract the micro-texture features of ancient coin images and generate the master feature vector. The principal component analysis (PCA) algorithm is used to reduce the dimension of the mother-model feature vector, and the cosine similarity between the mother-model feature vector and the feature vector in the mother-model texture database is calculated, wherein the feature vectors in the mother-model texture database are extracted by the LBP algorithm and reduced in dimension by the PCA algorithm; The values ​​of each cosine similarity are sorted, and the mother pattern in the mother pattern texture database corresponding to the feature vector with the highest similarity is determined as the mother pattern used for the ancient coin to be evaluated.

4. The method according to claim 3, characterized in that The quality attributes include: surface features, patina features, notch features, and detail features; the calculation formula for the quality score is: Among them, S is the quality score, C_w is the weighted appearance score, H_b is the historical background score, R is the rarity score, A is the age score, T is the known stock score, C_d is the market demand score, and K is the comprehensive adjustment factor.

5. The method according to claim 4, characterized in that The calculation formula of the appearance weighted score is: Among them, S_f is the surface feature score, P is the patina score, C_p is the notch score, and D is the overall detail score.

6. An intelligent evaluation system based on ancient coin images, characterized in that: The system comprises: The recognition module is used to use the deep learning model to identify the types of ancient coins images and determine the types of ancient coins to be evaluated; The first extraction module is used to extract the casting identification information in the ancient coin image, and determine the casting bureau of the ancient coin to be evaluated by comparing it with the casting bureau database; The second extraction module is used to extract the texture features of the surface of the ancient coin and determine the master mold used for the ancient coin to be evaluated by comparing it with the master mold texture database; The third extraction module is used to extract the difference features of different sub-coin batches and classify the ancient coins to be evaluated into specific sub-coin batches; The detection module is used to detect multiple quality attributes of ancient coin images, quantify and score the multiple quality attributes respectively, and calculate the quality score of the ancient coins based on each quantitative score to achieve intelligent evaluation of ancient coins.

7. The system according to claim 6, characterized in that The first extraction module is specifically used for: Perform image enhancement, grayscale conversion and binarization on the ancient coin image to obtain a binary image; Detect the edge of the casting mark in the binary image, and extract the contour in the image based on the edge detection result; According to the detected contour, the casting mark area is cut out from the binary image to obtain a separate casting mark image; Detecting specific key points in the cropped casting mark image, wherein the specific key points describe local features of the image; Performing feature description on the detected key points to generate a casting feature vector, wherein the casting feature vector includes texture information of the image area around the key points; The feature vector is matched with the identification image features of each foundry in the foundry database to obtain the foundry corresponding to the best match.

8. The system according to claim 7, characterized in that The second extraction module is specifically used for: The local binary pattern (LBP) algorithm is used to extract the micro-texture features of ancient coin images and generate the master feature vector. The principal component analysis (PCA) algorithm is used to reduce the dimension of the mother-model feature vector, and the cosine similarity between the mother-model feature vector and the feature vector in the mother-model texture database is calculated, wherein the feature vectors in the mother-model texture database are extracted by the LBP algorithm and reduced in dimension by the PCA algorithm; The values ​​of each cosine similarity are sorted, and the mother pattern in the mother pattern texture database corresponding to the feature vector with the highest similarity is determined as the mother pattern used for the ancient coin to be evaluated.

9. A storage medium, characterized in that: The storage medium stores a computer program, wherein the computer program is configured to execute the method according to any one of claims 1 to 5 when executed.

10. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to perform the method according to any one of claims 1 to 5.

Images