CN116777848A - Jade ware similarity analysis method and system - Google Patents
Jade ware similarity analysis method and system Download PDFInfo
- Publication number
- CN116777848A CN116777848A CN202310664784.7A CN202310664784A CN116777848A CN 116777848 A CN116777848 A CN 116777848A CN 202310664784 A CN202310664784 A CN 202310664784A CN 116777848 A CN116777848 A CN 116777848A
- Authority
- CN
- China
- Prior art keywords
- jade
- target
- similarity
- data
- type
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 239000010977 jade Substances 0.000 title claims abstract description 181
- 238000004458 analytical method Methods 0.000 title claims abstract description 52
- 230000000877 morphologic effect Effects 0.000 claims abstract description 40
- 238000000034 method Methods 0.000 claims abstract description 33
- 238000013136 deep learning model Methods 0.000 claims abstract description 9
- 230000000295 complement effect Effects 0.000 claims description 14
- 238000000605 extraction Methods 0.000 claims description 7
- 238000012549 training Methods 0.000 claims description 7
- 238000003860 storage Methods 0.000 claims description 5
- 238000010586 diagram Methods 0.000 description 10
- 238000005259 measurement Methods 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 7
- 239000003550 marker Substances 0.000 description 7
- 239000004575 stone Substances 0.000 description 7
- 238000000513 principal component analysis Methods 0.000 description 6
- 238000012545 processing Methods 0.000 description 6
- 238000004590 computer program Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000007619 statistical method Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 230000002708 enhancing effect Effects 0.000 description 2
- QBKSWRVVCFFDOT-UHFFFAOYSA-N gossypol Chemical compound CC(C)C1=C(O)C(O)=C(C=O)C2=C(O)C(C=3C(O)=C4C(C=O)=C(O)C(O)=C(C4=CC=3C)C(C)C)=C(C)C=C21 QBKSWRVVCFFDOT-UHFFFAOYSA-N 0.000 description 2
- 230000014759 maintenance of location Effects 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- QHOPXUFELLHKAS-UHFFFAOYSA-N Thespesin Natural products CC(C)c1c(O)c(O)c2C(O)Oc3c(c(C)cc1c23)-c1c2OC(O)c3c(O)c(O)c(C(C)C)c(cc1C)c23 QHOPXUFELLHKAS-UHFFFAOYSA-N 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000012790 confirmation Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 229930000755 gossypol Natural products 0.000 description 1
- 229950005277 gossypol Drugs 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 230000003562 morphometric effect Effects 0.000 description 1
- 238000013425 morphometry Methods 0.000 description 1
- 239000005416 organic matter Substances 0.000 description 1
- 238000004321 preservation Methods 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 238000005728 strengthening Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- 230000004304 visual acuity Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- 230000016776 visual perception Effects 0.000 description 1
Landscapes
- Image Analysis (AREA)
Abstract
The disclosure provides a jade ware similarity analysis method and system, wherein the method comprises the following steps: acquiring multidimensional form data of the target jade through an image acquisition device and a linear data measuring tool; the multi-dimensional form data are learned and classified by utilizing a preset deep learning model and the existing types in a sample database to obtain the types of the target jade ware; performing shape complementation on the target jade device according to the data of the complete ware belonging to the type of the target jade device in the sample database; and automatically calibrating the complemented target jade device, carrying out geometric analysis according to the calibration result, and calculating the morphological similarity of the target jade device.
Description
Technical Field
The application relates to the field of data analysis, in particular to a jade ware similarity analysis method and system.
Background
At present, in the archaeological field, the comparison of the morphology of the things is still limited to the use of qualitative observation means, and the similarity of the earth-going things on a certain place and the earth-going things on other sites is judged through descriptive language. However, since the morphology of many things is irregular, the criteria for qualitative description also often vary from one researcher to another, which leads to a difference in knowledge of the same things, and the focus of the discussion is also different, and the criteria are different in communication and discussion.
In addition, qualitative comparisons often only describe what researchers themselves consider important parts, while subjectively ignoring other "unimportant" features, the complete morphological information of the object is not actually recorded completely, which may lead to the loss of some important information. The above problems all lead to differences in comparison results, which in turn affect the correct knowledge of cultural origin, communication, propagation.
In the archaeological discovery of the new jade stone era, most of the produced jade stone devices are flat and have a thickness far smaller than the length of the jade stone devices due to the restriction of the human production capacity and the utilization of raw materials at the moment. These devices are easy to damage because the body is often thin. It is often stored in the form of damage in tomb and in the site with the breaker custom. This aspect prevents proper knowledge of these items and also makes subsequent studies related to morphometric measurements difficult, which is disadvantageous in that the digitalized preservation of these items and the morphological comparison of the analysis of the existing items remain mainly described qualitatively and lack uniform or quantitative criteria.
Disclosure of Invention
The embodiment of the disclosure aims to provide a jade ware similarity analysis method and system, which are used for collecting quantized data and carrying out data analysis and comparison to improve the accuracy and precision of jade ware comparison.
In order to achieve the above object, in a first aspect, an embodiment of the present application provides a jade ware similarity analysis method, which includes:
acquiring multidimensional form data of the target jade through an image acquisition device and a linear data measuring tool;
the multi-dimensional form data are learned and classified by utilizing a preset deep learning model and the existing types in a sample database to obtain the types of the target jade ware;
performing shape complementation on the target jade device according to the data of the complete ware belonging to the type of the target jade device in the sample database;
and automatically calibrating the complemented target jade device, carrying out geometric analysis according to the calibration result, and calculating the morphological similarity of the target jade device.
In some possible embodiments, the step of learning and classifying the multi-dimensional morphological data to obtain the type of the target jade using a preset deep learning model and an existing type in a sample database includes:
extracting a feature center of each type in the existing types in the sample database as a target feature;
extracting a characteristic vector of a target jade device, and calculating cosine distances between the jade device to be detected and target characteristics of preset jade device types;
calculating the morphological similarity between the target jade device and the characteristic center of each type according to the cosine distance,
and taking the type with the highest similarity as the type of the target jade ware.
In some possible embodiments, the step of shape complement of the target jade device according to the data of the complete ware belonging to the type of the target jade device in the sample database includes:
calculating the regional similarity between the complete ware of the type of the target jade device and the target jade device;
and when the area similarity is larger than a first preset threshold value, performing shape complementation on the target jade device according to the data of the corresponding complete ware.
In some possible embodiments, the step of performing shape complementation on the target jade device according to the data of the corresponding complete device includes:
pre-training a generated countermeasure network on the data of the complete object, wherein the generated countermeasure network comprises a generator and a discriminator;
restoring the target jade device by using the generator, so that the generation result of the generator can confuse the discriminator;
optimizing the discriminator using the data of the corresponding intact implement such that the discriminator identifies an incomplete portion of the target jade;
and the incomplete part of the target jade device is opposed to the corresponding complete device for shape complementation.
In some possible embodiments, the step of automatically calibrating the target jade after the completion and performing geometric analysis according to the calibration result, and calculating the morphological similarity of the target jade includes:
extracting the outline of the graphic data in the multi-dimensional morphological data of the target jade by using a trained edge extraction network model;
assigning a landmark point to a contour of the target jade machine;
and analyzing the characteristics corresponding to the mark points on the outline of the target jade by using a geometric morphology method, and calculating the morphological similarity of the target jade.
In a second aspect, an embodiment of the present application provides a jade ware similarity analysis system, including:
the data acquisition module is used for acquiring multidimensional form data of the target jade through the image acquisition device and the linear data measuring tool;
the type identification module is used for learning and classifying the multidimensional form data by utilizing a preset deep learning model and the existing types in the sample database to obtain the types of the target jade ware;
the shape complement module is used for carrying out shape complement on the target jade device according to the data of the complete ware belonging to the type of the target jade device in the sample database;
and the similarity analysis module is used for automatically calibrating the completed target jade device, performing geometric analysis according to the calibration result and calculating the morphological similarity of the target jade device.
In some possible embodiments, the type identification module is specifically configured to extract, as the target feature, a feature center of each type in the existing types in the sample database; extracting a characteristic vector of a target jade device, and calculating cosine distances between the jade device to be detected and target characteristics of preset jade device types; and calculating the morphological similarity between the target jade and the characteristic center of each type according to the cosine distance, and taking the type with the highest similarity as the type of the target jade.
In some possible embodiments, the shape complement module is specifically configured to calculate a region similarity between the complete object of the type of the target jade device and the target jade device; when the area similarity is larger than a first preset threshold value, pre-training a generated countermeasure network on the data of the complete object, wherein the generated countermeasure network comprises a generator and a discriminator; restoring the target jade device by using the generator, so that the generation result of the generator can confuse the discriminator; optimizing the discriminator using the data of the corresponding intact implement such that the discriminator identifies an incomplete portion of the target jade; and countering the incomplete part of the target jade with the corresponding complete object, and performing shape complementation.
In some possible embodiments, the similarity analysis module is specifically configured to extract a contour of the graphic data in the multi-dimensional morphological data of the target jade using a trained edge extraction network model; assigning a landmark point to a contour of the target jade machine; and analyzing the characteristics corresponding to the mark points on the outline of the target jade by using a geometric morphology method, and calculating the morphological similarity of the target jade.
In a third aspect, a computer readable storage medium is provided, which when executed by at least one processor causes the at least one processor to perform the jade-like degree analysis method.
The jade ware similarity analysis method and system can acquire and digitize jade ware information as completely as possible, expand a database, facilitate the digitization and informatization retention of ware, acquire information by using an image recognition technology, classify the damaged jade ware, complement the damaged jade ware based on a machine learning principle, analyze the multidimensional digital information by adopting a quantitative statistical method, and finally output a visual result obtained based on complete information analysis, thereby enhancing the similarity comparison and classification precision and accuracy among different cultural relics.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present application, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flow chart of a method of analyzing similarity of a jade stone according to an embodiment of the disclosure.
Fig. 2 is a schematic diagram of a jade machine similarity analysis method according to an embodiment of the disclosure.
Fig. 3 is a schematic diagram of marker points in a method of analyzing similarity of a jade stone according to an embodiment of the disclosure.
Fig. 4 is a schematic diagram of a geometric analysis in a method of similarity analysis of a jade-like device according to an embodiment of the disclosure.
Fig. 5 is a schematic structural view of a jade machine similarity analysis system according to an embodiment of the disclosure.
Detailed Description
The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
Fig. 1 is a flow chart of a method of analyzing similarity of a jade stone according to an embodiment of the disclosure. As shown in FIG. 1, the jade ware similarity analysis method comprises the following steps:
step 101: and acquiring multidimensional morphological data of the target jade through the image acquisition device and the linear data measuring tool. Wherein the image acquisition device includes, but is not limited to, a three-dimensional surface scanner and/or a digital camera. The linear data measuring tool comprises a scale and is mainly used for measuring the length, the width and the thickness of the ancient jade article to be measured.
Step 102: and learning and classifying the multidimensional form data by using a preset deep learning model and the existing types in a sample database to obtain the type of the target jade machine. The sample database refers to a geometric form database of samples, and comprises length, width and thickness, and mark point (multi-dimensional coordinate) data obtained according to a unified standard. The existing types are objects that have been explicitly classified into categories, such as: gossypol, axe, tablet, chisel, knife, etc.
Step 103: and performing shape complementation on the target jade device according to the data of the complete ware belonging to the type of the target jade device in the sample database.
Step 104: and automatically calibrating the complemented target jade device, carrying out geometric analysis according to the calibration result, and calculating the morphological similarity of the target jade device.
The morphological similarity is the overlapping degree of morphological space areas obtained after geometric analysis. The degree of coincidence of this area is the degree of similarity of the different types of jade stones. The comparison of the similarity degree is mainly used for comparing the same type of objects in different heritage points.
The embodiment can acquire and digitize the jade ware information as completely as possible, expand a database, facilitate the digitization and informatization retention of ware, acquire the information by using an image recognition technology, classify the damaged jade ware, complement the damaged jade ware based on a machine learning principle, analyze the multidimensional digital information by adopting a quantitative statistical method, and finally output a visualized result obtained based on complete information analysis, thereby enhancing the similarity comparison and classification precision and accuracy among different cultural relics.
As shown in fig. 2, there are various preferred embodiments of the above-mentioned jade-like degree analysis method:
specifically, step 101 may specifically include: firstly, digitizing the outline of an object by utilizing an image acquisition device, and establishing graphic data sets of the ancient jade ware in the upper, lower, left, right, front and back directions in equipment, namely the reference numeral 1 in fig. 2; and performs measurement of linear variables (e.g., length, width, height, thickness, etc.) and calculation of proportional data (e.g., aspect ratio, length to thickness ratio). The linear variable in this example is 18cm long, 5cm wide and 1cm thick. These graphical data and measurements form the basis of a multi-dimensional morphological database. The data will be stored in the server via the network to participate in the formation of the database, i.e. reference numeral 2 in fig. 2. The database may be called a sample database, which refers to the various information collected. The measurement and scale calculation is part of the database as is the coordinate data obtained after the marker point selection.
During specific operation, erect image acquisition device on mobilizable slide rail, can take a picture to ancient jade ware from each angle. And a scale is attached during photographing, and the scale and the jade ware to be photographed are collected together. The measurement of the linear variable may be performed by means of image recognition. Specific implementation details include: the ancient jade device and the scale are photographed together through a digital camera carried by the equipment, the obtained photo is subjected to image recognition, the outline of the ancient jade device and the scale are respectively distinguished, and the length, the width and the height of the ancient jade device are measured and calculated according to the length of the scale. Based on a deep learning algorithm, the comparison scale is used for identifying, and based on the comparison scale, the length, the width, the thickness and other indexes of the ancient jade ware are measured. Based on the measurement results, the aspect ratio, length-thickness ratio and the like proportional variable were calculated. And uploading the graphic data, the measured value and the calculated variable value to a database (namely a sample database) to serve as basic data.
Specifically, step 102 may specifically include: based on the image recognition principle, the collected graphic data is compared with the existing object types in the database by adopting an ancient jade ware classification method based on deep small sample learning, so that the automatic classification of the newly added objects is realized. For an uncertainty classified object, a determination may be made as to the possible result of the classification of the newly added object (probability of outputting the closest object type), i.e. reference numeral 3 in fig. 2, based on the records already in the database.
The specific calculation flow comprises the following steps: firstly, extracting the characteristic center of each type of the existing organic matter (each type refers to the type characteristic center and cosine distance which are already recorded in the equipment) as a target point; when a new object arrives, extracting the feature vector of the new object, and making cosine distance with all target features; and then normalizing (namely adopting a generalized Prukast method to perform operations such as scaling, rotation, translation and the like on the feature vector of the new object, enabling the feature centers of the new object to coincide and the sum of the cosine distances of the features of each target point to be minimum) to serve as a similarity value from the new object to each type of feature center, giving out the similarity value, representing the similarity value in percentage (namely, having XX% of morphological similarity with the XXX object), and displaying the representation of the possibly-classified object and the new object simultaneously, thereby being convenient for equipment users to observe and judge. The final classification result is judged by a professional and the manually judged classification result is input to the device. The result may be used as training data in a subsequent image recognition classification.
Specifically, step 103 may specifically include: on the basis of the establishment of a sufficiently sufficient database, the defective articles can be recovered, i.e. reference number 4 in fig. 2. When the equipment collects incomplete objects, the similarity of the areas of the incomplete objects and the intact objects in the database can be calculated according to the method of area matching consistency. And when the similarity degree of all areas of the incomplete object and the selected complete object reaches a preset threshold value, the image of the complete object is referred to complete the incomplete part. The depth generation countermeasure network can be adopted to complement. Specifically, the generation of the countermeasure network (including the generator and the discriminator) is first pre-trained on good object data. And then restoring the incomplete ware by using a generator, namely, complementing the incomplete part to form a complete jade ware, so that the generated result can confuse the discriminator, namely, the discriminator recognizes the complemented jade ware as a truly complete jade ware. And optimizing the discriminator by using the selected intact ware data again, and adopting a mode of improving the detail resolving power to enable the discriminator to effectively find the incomplete part of the ware, and finally obtaining a restoration result in a fine adjustment mode through the countermeasure of the incomplete part and the incomplete part. Recoverable imperfections include unfilled corners, imperfections, but symmetrical textures. The restoration result can be incorporated into a digital library on one hand, becomes a training data set for further strengthening machine learning on the basis of verification and confirmation by professional staff, and can be used as a reference in the process of repairing cultural relics on the other hand.
Specifically, the automatic pointing and geometric analysis of the landmark points in step 104 may refer to fig. 3 and icon number 4, respectively. As shown in fig. 3, after a certain number of databases are established, the collected multidimensional morphological data can be called, and the marker point is selected according to the analysis precision requirement. First, a certain number of labels are made on an existing database, and desired object contour information is drawn. Next, the edge extraction network model is trained using the data. According to the number of the input mark points, the trained edge extraction network model is utilized to extract the outline of the graphic data, and then the mark points are distributed on the outline of the whole object. The marker points are typically selected from the points of aliquotation (e.g., trisection, quartering, penta, etc.) on each boundary of the ancient jades outline. Quantitative morphological comparison is carried out by using a geometric morphology method, data dimension reduction is carried out on the extracted data mark point characteristics, main information is extracted for analysis and comparison, namely, principal component analysis is carried out, and analysis results are used for visualizing the similarity or difference degree of different object forms. Wherein the reference numerals 1-21 in fig. 3, i.e. the number of geometrically shaped index points, in this example 21 geometrically shaped index points of the body, stop and handle of the object are shown.
The geometric method is a method for visualizing the similarity degree of the patterns. Firstly, carrying out generalized Pruck transformation on the mark points, namely unifying the feature point data sets to the origin of coordinates, and then carrying out scaling and rotation operation to minimize the sum of distances between corresponding position data points in each data set, thereby realizing the purpose of removing the size influence and only carrying out morphological comparison. Principal component analysis is a statistical means by which a set of variables that may have a correlation is converted into a set of linearly uncorrelated variables by a positive-to-negative transformation, the converted set of variables being called the principal component. The result of the principal component analysis is the same as the data effect expressed by the XY axis, and the distance relation between the mark points of each sample, namely the similarity degree between different samples, can be clearly seen.
As shown in fig. 4, geometric analysis was performed in conjunction with a sufficient number of specimens to visualize the degree of similarity of different cultural identities. Based on the deep neighbor method, the overlapped geometrical form area S_ { overlap } between the different types of the ancient jade articles and the geometrical form area S of each of the different types of the ancient jade articles are calculated 1 ，S 2 And takes the similarity as the morphological similarity between different object typesThe abscissa in fig. 4 refers to principal component 1 and principal component 2, and is the result of performing the dimension reduction operation on the marker points (such as 1-21 marked in fig. 3). Principal component analysis seeks to replace the original index by a new set of mutually independent composite indices that are originally numerous and have some correlation (e.g., 21 marker points here). Principal component analysis is a multivariate statistical method for examining the correlation among a plurality of variables, and for studying how to reveal the internal structure among a plurality of variables by a few principal components, i.e., deriving a few principal components from the original variables so that they retain as much information as possible of the original variables and are uncorrelated with each other.
In this case, there are 21 marker points, and the data of each point represents one dimension. In this example, there are 21 dimensions. However, the human visual perception dimension is only three-dimensional, so that the space expressed by the 21-dimension is projected in the two-dimensional space by a principal component analysis mode, and the obtained two shadows reflecting the most information are principal component 1 (PC 1) and principal component 2 (PC 2). These two principal components have no practical significance and each contains a portion of the total data information. PC1 and PC2 are those two parts that contain the most information.
In the embodiment, through high-efficiency multidimensional morphological data acquisition, the morphological information of the ancient jade ware which is as complete as possible and digital is extracted; based on the image recognition technique, it is determined which object type is possible, i.e. based on a comparison with the existing object types (complete) in the database, what type of shape classification the acquired image data should correspond to. The method comprises a ancient jade ware classification method based on deep small sample learning, and if the ancient jade ware classification method is difficult to judge, the probability of the closest certain ware type is given. Judging whether the scanned object is complete or not according to comparison with the object types in the database, and if the scanned object is determined to be damaged, performing residual part completion (including jade morphological outline and surface texture) on the damaged ancient jade ware. By automatically completing the selection work of the mark points in the geometric morphology analysis, the problems of errors during manual acquisition and low working efficiency during large data volume are avoided, and by using the image recognition technology, linear variable measurement, classification, incomplete part inference, restoration and geometric morphology comparison are automatically completed, so that the analysis precision and efficiency are improved; the method combines the image recognition technology, can be automatically performed after the multi-dimensional data acquisition is completed, integrates a plurality of hardware and software which can be matched for use, and is beneficial to the rapid acquisition of standardized data and the establishment of a database.
Fig. 5 is a block diagram of a jade machine similarity analysis system according to an embodiment of the disclosure. The embodiment shown in fig. 1-4 may be used to explain the present embodiment. As shown in fig. 5, a jade ware similarity analysis system includes:
the data acquisition module 501 is used for acquiring multidimensional form data of the target jade through the image acquisition device and the linear data measuring tool;
the type recognition module 502 is configured to learn and classify the multidimensional form data to obtain a type of the target jade apparatus by using a preset deep learning model and an existing type in a sample database;
a shape complement module 503, configured to perform shape complement on the target jade device according to data of a complete object belonging to the type of the target jade device in the sample database;
and the similarity analysis module 504 is used for automatically calibrating the completed target jade device, performing geometric analysis according to the calibration result, and calculating the morphological similarity of the target jade device.
Preferably, the type identification module 502 is specifically configured to extract a feature center of each type in the existing types in the sample database as a target feature; extracting a characteristic vector of a target jade device, and calculating cosine distances between the jade device to be detected and target characteristics of preset jade device types; and calculating the morphological similarity between the target jade and the characteristic center of each type according to the cosine distance, and taking the type with the highest similarity as the type of the target jade.
The shape complement module 503 is specifically configured to calculate a region similarity, that is, an overlapping proportion of geometric areas, of the complete object of the type of the target jade device and the target jade device; when the area similarity is larger than a first preset threshold value, pre-training a generated countermeasure network on the data of the complete object, wherein the generated countermeasure network comprises a generator and a discriminator; restoring the target jade device by using the generator, so that the generation result of the generator can confuse the discriminator; optimizing the discriminator using the data of the corresponding intact implement such that the discriminator identifies an incomplete portion of the target jade; and countering the incomplete part of the target jade with the corresponding complete object, and performing shape complementation.
The similarity analysis module 504 is specifically configured to extract a contour of graphic data in the multi-dimensional morphological data of the target jade using the trained edge extraction network model; assigning a landmark point to a contour of the target jade machine; and analyzing the characteristics corresponding to the mark points on the outline of the target jade by using a geometric morphology method, and calculating the morphological similarity of the target jade.
In the embodiment, through high-efficiency multidimensional morphological data acquisition, the morphological information of the ancient jade ware which is as complete as possible and digital is extracted; based on the image recognition technology, determining which shape classification the acquired image data should correspond to. Judging whether the scanned object is complete or not according to comparison with the type of the object in the database, and if the object is determined to be damaged, completing the residual part of the damaged ancient jade ware. By automatically completing the selection of the mark points in the geometric analysis, the problems of errors during manual acquisition and low working efficiency during large data volume are avoided. Through the image recognition technology, linear variable measurement, classification, incomplete part inference, restoration, geometric morphology comparison are automatically completed, and analysis precision and efficiency are improved. The method combines the image recognition technology, can be automatically performed after the multi-dimensional data acquisition is completed, integrates a plurality of hardware and software which can be matched for use, and is beneficial to the rapid acquisition of standardized data and the establishment of a database.
Specifically, the present application also provides a computer readable storage medium, where the instructions in the computer readable storage medium, when executed by at least one processor, cause the at least one processor to perform the above-mentioned jade ware similarity analysis method.
The foregoing describes specific embodiments of the present disclosure. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims can be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
Although the application provides method operational steps as an example or a flowchart, more or fewer operational steps may be included based on conventional or non-inventive labor. The order of steps recited in the embodiments is merely one way of performing the order of steps and does not represent a unique order of execution. When implemented by an actual device or client product, the instructions may be executed sequentially or in parallel (e.g., in a parallel processor or multi-threaded processing environment) as shown in the embodiments or figures.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The principles and embodiments of the present application have been described in detail with reference to specific examples, which are provided to facilitate understanding of the method and core ideas of the present application; meanwhile, as those skilled in the art will have variations in the specific embodiments and application scope in accordance with the ideas of the present application, the present description should not be construed as limiting the present application in view of the above.
Claims (10)
1. The jade ware similarity analysis method is characterized by comprising the following steps of:
acquiring multidimensional form data of the target jade through an image acquisition device and a linear data measuring tool;
the multi-dimensional form data are learned and classified by utilizing a preset deep learning model and the existing types in a sample database to obtain the types of the target jade ware;
performing shape complementation on the target jade according to the data of the complete ware belonging to the type of the target jade in the sample database;
and automatically calibrating the complemented target jade device, carrying out geometric analysis according to the calibration result, and calculating the morphological similarity of the target jade device.
2. The method for analyzing the similarity of the jade machine according to claim 1, wherein the step of learning and classifying the multidimensional morphological data to obtain the type of the target jade machine by using a preset deep learning model and the existing types in the sample database comprises the steps of:
extracting a feature center of each type in the existing types in the sample database as a target feature;
extracting a characteristic vector of a target jade device, and calculating cosine distances between the jade device to be detected and target characteristics of preset jade device types;
calculating the similarity between the target jade device and the characteristic center of each type according to the cosine distance,
and taking the type with the highest similarity as the type of the target jade ware.
3. The method for analyzing similarity of jade articles according to claim 2, wherein said step of shape-complementing said target jade articles according to data of complete articles belonging to said target jade article type in said sample database comprises:
calculating the regional similarity between the complete ware of the type of the target jade device and the target jade device;
and when the area similarity is larger than a first preset threshold value, performing shape complementation on the target jade device according to the data of the corresponding complete ware.
4. The method for analyzing similarity of jade articles according to claim 3, wherein said step of shape complement of said target jade articles according to the data of the corresponding complete articles comprises:
pre-training a generated countermeasure network on the data of the complete object, wherein the generated countermeasure network comprises a generator and a discriminator;
restoring the target jade device by using the generator, so that the generation result of the generator can confuse the discriminator;
optimizing the discriminator using the data of the corresponding intact implement such that the discriminator identifies an incomplete portion of the target jade;
and the incomplete part of the target jade device is opposed to the corresponding complete device for shape complementation.
5. The method for analyzing similarity of jade articles according to claim 4, wherein said step of automatically calibrating said target jade articles after completion and performing geometric analysis based on the calibration result, and calculating the similarity of the morphology of said target jade articles comprises:
extracting the outline of the graphic data in the multi-dimensional morphological data of the target jade by using a trained edge extraction network model;
assigning a landmark point to a contour of the target jade machine;
and analyzing the characteristics corresponding to the mark points on the outline of the target jade by using a geometric morphology method, and calculating the morphological similarity of the target jade.
6. A jade-like degree analysis system, comprising:
the data acquisition module is used for acquiring multidimensional form data of the target jade through the image acquisition device and the linear data measuring tool;
the type identification module is used for learning and classifying the multidimensional form data by utilizing a preset deep learning model and the existing types in the sample database to obtain the types of the target jade ware;
the shape complement module is used for carrying out shape complement on the target jade device according to the data of the complete ware belonging to the type of the target jade device in the sample database;
and the similarity analysis module is used for automatically calibrating the completed target jade device, performing geometric analysis according to the calibration result and calculating the morphological similarity of the target jade device.
7. The jade ware similarity analysis system of claim 6, wherein the type recognition module is specifically configured to extract a feature center of each type of the existing types in the sample database as a target feature; extracting a characteristic vector of a target jade device, and calculating cosine distances between the jade device to be detected and target characteristics of preset jade device types; and calculating the morphological similarity between the target jade and the characteristic center of each type according to the cosine distance, and taking the type with the highest similarity as the type of the target jade.
8. The jade-like degree analysis system of claim 7, wherein the shape complement module is specifically configured to calculate a regional similarity of the complete object of the type of the target jade device to the target jade device; when the area similarity is larger than a first preset threshold value, pre-training a generated countermeasure network on the data of the complete object, wherein the generated countermeasure network comprises a generator and a discriminator; restoring the target jade device by using the generator, so that the generation result of the generator can confuse the discriminator; optimizing the discriminator using the data of the corresponding intact implement such that the discriminator identifies an incomplete portion of the target jade; and countering the incomplete part of the target jade with the corresponding complete object, and performing shape complementation.
9. The jade-like degree analysis system of claim 8, wherein the degree of similarity analysis module is specifically configured to extract a contour of graphic data in the multidimensional morphological data of the target jade using a trained edge extraction network model; assigning a landmark point to a contour of the target jade machine; and analyzing the characteristics corresponding to the mark points on the outline of the target jade by using a geometric morphology method, and calculating the morphological similarity of the target jade.
10. A computer-readable storage medium, wherein instructions in the computer-readable storage medium, when executed by at least one processor, cause the at least one processor to perform the jade-like similarity analysis method of any one of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310664784.7A CN116777848A (en) | 2023-06-06 | 2023-06-06 | Jade ware similarity analysis method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310664784.7A CN116777848A (en) | 2023-06-06 | 2023-06-06 | Jade ware similarity analysis method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116777848A true CN116777848A (en) | 2023-09-19 |
Family
ID=88010692
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310664784.7A Pending CN116777848A (en) | 2023-06-06 | 2023-06-06 | Jade ware similarity analysis method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116777848A (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106096028A (en) * | 2016-06-24 | 2016-11-09 | 北京旷视科技有限公司 | Historical relic indexing means based on image recognition and device |
| CN107862668A (en) * | 2017-11-24 | 2018-03-30 | 河海大学 | A kind of cultural relic images restored method based on GNN |
| CN109544578A (en) * | 2018-11-30 | 2019-03-29 | 扬州大学 | Recombination restored method based on the matched three-dimensional fragment of maximum similarity |
| CN109727209A (en) * | 2018-12-13 | 2019-05-07 | 北京爱奇艺科技有限公司 | A kind of method and device of determining incomplete historical relic complete image |
| CN112132200A (en) * | 2020-09-17 | 2020-12-25 | 山东大学 | Lithology identification method and system based on multi-dimensional rock image deep learning |
| CN112241997A (en) * | 2020-09-14 | 2021-01-19 | 西北大学 | Three-dimensional model establishing and repairing method and system based on multi-scale point cloud up-sampling |
| CN113689360A (en) * | 2021-09-30 | 2021-11-23 | 合肥工业大学 | Image restoration method based on generation countermeasure network |
| CN114187178A (en) * | 2021-12-13 | 2022-03-15 | 浙大城市学院 | Porcelain fragment classification and splicing system and method for assisting in cultural relic repair |
| CN114332874A (en) * | 2021-12-30 | 2022-04-12 | 浙江大学 | Calligraphy repairing method based on crowdsourcing |
| CN115439694A (en) * | 2022-09-19 | 2022-12-06 | 南京邮电大学 | High-precision point cloud completion method and device based on deep learning |
| CN115861093A (en) * | 2022-11-21 | 2023-03-28 | 武汉纺织大学 | Ancient damaged textile image restoration method and system based on deep learning |
-
2023
- 2023-06-06 CN CN202310664784.7A patent/CN116777848A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106096028A (en) * | 2016-06-24 | 2016-11-09 | 北京旷视科技有限公司 | Historical relic indexing means based on image recognition and device |
| CN107862668A (en) * | 2017-11-24 | 2018-03-30 | 河海大学 | A kind of cultural relic images restored method based on GNN |
| CN109544578A (en) * | 2018-11-30 | 2019-03-29 | 扬州大学 | Recombination restored method based on the matched three-dimensional fragment of maximum similarity |
| CN109727209A (en) * | 2018-12-13 | 2019-05-07 | 北京爱奇艺科技有限公司 | A kind of method and device of determining incomplete historical relic complete image |
| CN112241997A (en) * | 2020-09-14 | 2021-01-19 | 西北大学 | Three-dimensional model establishing and repairing method and system based on multi-scale point cloud up-sampling |
| CN112132200A (en) * | 2020-09-17 | 2020-12-25 | 山东大学 | Lithology identification method and system based on multi-dimensional rock image deep learning |
| CN113689360A (en) * | 2021-09-30 | 2021-11-23 | 合肥工业大学 | Image restoration method based on generation countermeasure network |
| CN114187178A (en) * | 2021-12-13 | 2022-03-15 | 浙大城市学院 | Porcelain fragment classification and splicing system and method for assisting in cultural relic repair |
| CN114332874A (en) * | 2021-12-30 | 2022-04-12 | 浙江大学 | Calligraphy repairing method based on crowdsourcing |
| CN115439694A (en) * | 2022-09-19 | 2022-12-06 | 南京邮电大学 | High-precision point cloud completion method and device based on deep learning |
| CN115861093A (en) * | 2022-11-21 | 2023-03-28 | 武汉纺织大学 | Ancient damaged textile image restoration method and system based on deep learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Jovančević et al. | 3D point cloud analysis for detection and characterization of defects on airplane exterior surface | |
| Guo et al. | Efficient center voting for object detection and 6D pose estimation in 3D point cloud | |
| Moyano et al. | Semantic interpretation of architectural and archaeological geometries: Point cloud segmentation for HBIM parameterisation | |
| Han et al. | Automated monitoring of operation-level construction progress using 4D BIM and daily site photologs | |
| WO2003088085A1 (en) | Three-dimensional digital library system | |
| CN115546116A (en) | Method and system for extracting and calculating spacing of discontinuous surface of fully-covered rock mass | |
| Liu et al. | Application of three-dimensional laser scanning in the protection of multi-dynasty ceramic fragments | |
| CN112561989B (en) | Recognition method for hoisting object in construction scene | |
| Maiza et al. | Automatic classification of archaeological potsherds | |
| Kampel et al. | Profile-based pottery reconstruction | |
| CN116757713A (en) | Work estimation method, device, equipment and storage medium based on image recognition | |
| CN116777848A (en) | Jade ware similarity analysis method and system | |
| CN109300150B (en) | Hand bone X-ray image texture feature extraction method for bone age assessment | |
| Kashihara | An intelligent computer assistance system for artifact restoration based on genetic algorithms with plane image features | |
| Mara et al. | Ancient coins & ceramics–3D and 2D documentation for preservation and retrieval of lost heritage | |
| Kuçak | The feature extraction from point clouds using geometric features and ransac algorithm | |
| Burton | Automated 3 D reconstruction of neuronal structures from serial sections | |
| Hu et al. | A novel hierarchical object recognition algorithm based on saliency analysis | |
| Pamart et al. | Morphological analysis of shape semantics from curvature-based signatures | |
| Carrilho et al. | Quality assessment for automatic LiDAR data classification methods | |
| Guldur et al. | Laser-based automatic cross-sectional change detection for steel frames | |
| CN114155228B (en) | Method and device for rapidly measuring outline compliance of building material test piece | |
| Ioniță et al. | Automation of the expertise of the Roman Mosaic Arts in Constanta: Analytical and statistical models for a fuzzy inference-based system | |
| Bortolotti et al. | Combined XRD-XRF cluster analysis for automatic chemical and crystallographic surface mappings | |
| Lami et al. | Pottery goes digital. 3D laser scanning technology and the study of archaeological ceramics |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |