Antique cultural relic incomplete pattern intelligent reconstruction method and system based on multi-modal feature fusion
By building a cultural relic pattern database and utilizing multimodal feature fusion technology, intelligent reconstruction of antique cultural relic patterns is achieved, solving the problem of low digitization efficiency caused by worn and damaged patterns, and improving the efficiency and accuracy of cultural relic digitization.
Patent Information
- Application Number
- CN202510828596.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-20
- Publication Date
- 2025-10-03
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In the existing technology, the patterns of antique cultural relics are worn and incomplete, resulting in low efficiency of digitization work. It requires professional knowledge and experience, and it is difficult to reconstruct the patterns efficiently.
Build a database of cultural relics patterns, extract and associate feature information, perform retrieval, matching and reconstruction through multimodal feature fusion, and use an online expert database to confirm a small number of patterns.
It improves the efficiency of cultural relics digitization, reduces dependence on expert knowledge, and ensures the accuracy and efficiency of pattern reconstruction.
Smart Images
Figure CN120744152A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of digital cultural relics assets, and in particular to a method and system for intelligently reconstructing incomplete patterns of antique cultural relics based on multimodal feature fusion. Background Art
[0002] Simply put, a pattern is a decorative design. Ancient Chinese people used natural scenery as inspiration, employing realistic, freehand, and distorted techniques to create the exquisitely crafted beauty of traditional Chinese clothing, utensils, and architecture. Cultural relic patterns refer to designs within Chinese traditional culture that have been passed down through generations, embodying unique national artistic styles. These patterns primarily originate from nature. Common examples include: animal patterns, such as auspicious beasts like dragons and phoenixes, often used in architectural and artifact decoration; plant patterns, such as floral and bird motifs, which express the beauty of nature; geometric patterns, including combinations of various geometric shapes, which showcase the creativity of the ancients; auspicious patterns, such as patterns symbolizing happiness and prosperity, often found in textiles, embroidery, and ceramics; and historical relics, such as the exquisite patterns on bronze artifacts, which reflect the culture and art of ancient society.
[0003] The digitization of cultural relics involves the holographic recording and transformation of cultural relics. This process, while enabling the salvage and preservation of comprehensive information about cultural relics, lays a solid foundation for sustainable cultural relic conservation and research. It also transforms cultural relic resources from physical to digital, making them the foundation for sustainable exploration, transformation, and utilization in the digital age. Currently, in the digitization of cultural relics, some artifacts suffer from wear, tear, or even loss of patterns due to age, wear, and incompleteness. Reconstructing these patterns requires specialized research experience and knowledge in cultural relics and archaeology, and therefore generally relies on expert guidance, resulting in particularly low efficiency in digitization. Summary of the Invention
[0004] In order to solve the problems existing in the background technology, the present invention proposes a method and system for intelligent reconstruction of incomplete patterns of antique cultural relics based on multimodal feature fusion.
[0005] The intelligent reconstruction method of incomplete patterns of antique cultural relics based on multimodal feature fusion includes the following steps:
[0006] S100, constructing a cultural relic pattern database, including known cultural relic patterns;
[0007] S200, extracting characteristic information from the collected cultural relic patterns, and building a correlation relationship between the characteristic information of each cultural relic pattern;
[0008] S300, obtaining multimodal feature information of the target cultural relic and its incomplete patterns;
[0009] S400, performing cultural relic pattern retrieval and matching through the cultural relic pattern database based on the acquired multimodal feature information and the association relationship between the feature information;
[0010] S500, using the retrieved and matched cultural relic pattern information as a reference pattern for reconstructing the incomplete pattern;
[0011] S600: Reconstruct the pattern according to the reconstructed reference pattern.
[0012] Based on the above, in step S200, the characteristic information of the cultural relic pattern at least includes the pattern type and name, the pattern carrier type and period, the pattern appearance and popular period, the pattern line characteristics, the pattern meaning information, the pattern development and evolution process and the pattern element style information.
[0013] Based on the above, in step S200, the association relationship of the cultural relic pattern feature information at least includes the association relationship between the cultural relic pattern features, the association relationship of the pattern matching and presentation, and the association relationship of the cultural style of the period corresponding to the pattern.
[0014] Based on the above, the multimodal feature information at least includes the multimodal feature information of the cultural relic itself and the multimodal feature information of the pattern on the cultural relic.
[0015] Based on the above, in step S500 and step S600, if the same cultural relic pattern is retrieved and matched, the proportion of the retrieved and matched cultural relic pattern is adapted, and then the incomplete pattern portion of the target cultural relic is reconstructed based on the retrieved and matched cultural relic pattern.
[0016] Based on the above, in step S500 and step S600, if only similar patterns on cultural relics of the same type from the same period are retrieved and matched, the incomplete pattern part of the target cultural relic is supplemented and reconstructed based on the feature information of the same type of cultural relics and their patterns from the same period, and the line feature information of the incomplete pattern of the target cultural relic.
[0017] Based on the above, an online expert database is constructed, and the supplemented and reconstructed patterns, target cultural relics and their characteristic information, incomplete patterns of target cultural relics and their characteristic information, and retrieved and matched similar cultural relics patterns and their characteristic information are sent to the online expert database, and reviewed and confirmed by experts online.
[0018] A system for intelligently reconstructing incomplete patterns of antique cultural relics based on multimodal feature fusion, comprising a processor, a storage device, and a computer program stored on the storage device and executable on the processor, wherein when the processor executes the computer program, any one of the above-mentioned methods for intelligently reconstructing incomplete patterns of antique cultural relics based on multimodal feature fusion is implemented.
[0019] The present invention has outstanding substantive features and significant progress compared to the prior art. Specifically, the present invention constructs a database of known cultural relic patterns and their characteristics, feature correlations, etc., obtains multimodal feature information of target cultural relics and their patterns, and then searches and matches in the database through these multimodal features, so as to accurately obtain the same or similar patterns, and then reconstruct the incomplete patterns of the target cultural relics based on the obtained patterns. As a result, technicians do not need to rely on particularly professional cultural relics and historical knowledge, and can intelligently reconstruct most of the incomplete patterns. For a small number of reconstructed patterns that cannot be confirmed, the reconstructed patterns and other information can be sent online to experts for confirmation. While ensuring accuracy, efficiency is greatly improved and the use of expert manpower is reduced. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] Figure 1 It is a schematic block diagram of the process of the present invention. DETAILED DESCRIPTION
[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without creative work are within the scope of protection of the present invention.
[0022] like Figure 1 As shown, the method for intelligent reconstruction of incomplete patterns of antique cultural relics based on multimodal feature fusion includes the following steps: S100, constructing a cultural relic pattern database to include known cultural relic patterns; S200, extracting feature information of the included cultural relic patterns, and establishing association relationships for the feature information of each cultural relic pattern; S300, obtaining multimodal feature information of the target cultural relic and its incomplete patterns; S400, searching and matching cultural relic patterns through the cultural relic pattern database based on the obtained multimodal feature information and the association relationship between the feature information; S500, using the retrieved and matched cultural relic pattern information as a reference pattern for reconstructing the incomplete pattern; S600, reconstructing the pattern based on the reconstructed reference pattern.
[0023] In reality, cultural relics themselves have some clear characteristics, such as age characteristics, material characteristics, process characteristics, etc. Through existing technical means such as element analysis such as multi-type spectral scanning, structural analysis such as X-ray diffraction analysis, morphology analysis such as scanning electron microscopy, as well as instruments such as densitometers, microhardness testers, surface tension meters, three-dimensional micromorphometers, colorimeters, ultrasonic flaw detectors, X-ray flaw detectors, CT scanners, etc. used to detect the density, porosity, strength, surface energy, surface roughness, micromorphology, color difference, internal structure, etc. of cultural relics, the multimodal characteristics of the cultural relics themselves can be detected, analyzed and obtained. The patterns on cultural relics also have some specific characteristics, such as the type of pattern, the time of appearance, the period of popularity, the meaning of the pattern, the line characteristics of the pattern, the development and evolution of the pattern and the style information of the pattern elements. The types of patterns include geometric patterns such as cloud and thunder patterns, animal patterns such as chi patterns, flower and bird patterns such as round flower patterns, auspicious patterns such as ruyi patterns, human patterns such as flying patterns, and object patterns such as landscape patterns, etc. In addition, the patterns also have some distinct period characteristics, which help to narrow the scope. For example, the patterns on bronzes in the Shang and Zhou dynasties were mostly animal patterns such as taotie patterns, while animal patterns such as pangui patterns were mainly popular in the late Spring and Autumn Period and Warring States Period. After acquiring the multimodal features of existing cultural relics and their patterns, a cultural relic pattern database is constructed to store the multimodal features of known cultural relics and their patterns, and to construct relevant correlation relationships for these multimodal features, such as at least the correlation relationships between cultural relic pattern features, the correlation relationships between pattern matching and presentation, and the correlation relationships between cultural styles of the period corresponding to the patterns. There are correlation relationships between cultural relic pattern features, such as line trends, etc. If some lines of the incomplete pattern are obviously geometric patterns or animal patterns, then flower and bird patterns, auspicious patterns and other types of patterns can be excluded, which not only helps to narrow the scope, but also can be used to identify the correlation relationships between the lines. Reconstruction and simulation are carried out based on the research on trends. In addition, the characteristics of cultural relics, such as trace characteristics and element characteristics, are helpful to determine the type of patterns. Since the patterns on cultural relics are generally not a single type, but a combination of multiple types, for cultural relic fragments or incomplete patterns, the types of the incomplete patterns can be narrowed down through the known patterns on them. Therefore, by associating the matching presentation of cultural relic patterns, it can help to narrow the scope of recognition of incomplete patterns. In addition, according to the age, material, craftsmanship and other characteristics of the cultural relics, it is helpful to limit the cultural style of the patterns in this period, which helps to analyze the types of incomplete patterns.
[0024] After obtaining the multimodal features of the target cultural relic and its incomplete patterns, these features are input into the cultural relic pattern database. Since the multimodal features are configured with correlations, these known multiple features can be used for retrieval and matching, thereby matching the same cultural relic and pattern or some similar cultural relic and pattern. The more feature information of the target cultural relic and incomplete pattern is obtained, the more accurate and unique the matching result will be. For the unique matching result, the pattern of the retrieval result is proportionally adapted according to the size, proportion, etc. of the target cultural relic and its incomplete pattern, and then the incomplete pattern on the target cultural relic can be reconstructed based on the retrieval result. In reality, since many features of the target cultural relic and incomplete pattern are very clear, usually the vast majority of cultural relic and pattern results can be accurately matched through retrieval. This can greatly reduce the dependence on experts and expert knowledge and improve the digitization efficiency of cultural relic digital assets.
[0025] For a small number of cases with multiple similar matching results, that is, only similar patterns on similar cultural relics from the same period are retrieved, the one with the highest similarity can be selected as the reconstruction reference, and then the incomplete pattern part of the target cultural relic can be supplemented and reconstructed based on the characteristic information of similar cultural relics and their patterns from the same period, as well as the line characteristic information of the incomplete pattern of the target cultural relic, the cultural style of the period, etc.
[0026] Preferably, an online expert database is constructed as an online direct channel between digital cultural relics and cultural relic experts. Since the patterns reconstructed by reference to similar results may contain errors, the reconstructed patterns are sent to the online expert database together with the target cultural relics and their characteristic information, the incomplete patterns of the target cultural relics and their characteristic information, and the retrieved and matched similar cultural relic patterns and their characteristic information. After the experts understand the known characteristic information online, they review and confirm the reconstructed patterns. This not only improves the accuracy of pattern reconstruction, but also avoids a large amount of expert time.
[0027] A system for intelligently reconstructing incomplete patterns of antique cultural relics based on multimodal feature fusion, comprising a processor, a storage device, and a computer program stored on the storage device and executable on the processor, wherein when the processor executes the computer program, any one of the above-mentioned methods for intelligently reconstructing incomplete patterns of antique cultural relics based on multimodal feature fusion is implemented.
[0028] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.
Claims
1. An intelligent reconstruction method for incomplete patterns of antique cultural relics based on multimodal feature fusion, characterized by: Including steps: S100, constructing a cultural relic pattern database, including known cultural relic patterns; S200, extracting characteristic information from the collected cultural relic patterns, and building a correlation relationship between the characteristic information of each cultural relic pattern; S300, obtaining multimodal feature information of the target cultural relic and its incomplete patterns; S400, performing cultural relic pattern retrieval and matching through the cultural relic pattern database based on the acquired multimodal feature information and the association relationship between the feature information; S500, using the retrieved and matched cultural relic pattern information as a reference pattern for reconstructing the incomplete pattern; S600: Reconstruct the pattern according to the reconstructed reference pattern.
2. The method for intelligent reconstruction of incomplete patterns of antique cultural relics based on multimodal feature fusion according to claim 1 is characterized by: In step S200, the characteristic information of the cultural relic pattern includes at least the pattern type and name, the pattern carrier type and period, the pattern appearance and popular period, the pattern line characteristics, the pattern meaning information, the pattern development and evolution process and the pattern element style information.
3. The method for intelligent reconstruction of incomplete patterns of antique cultural relics based on multimodal feature fusion according to claim 1 is characterized by: In step S200 , the association relationship of the cultural relic pattern feature information at least includes the association relationship between the cultural relic pattern features, the association relationship of the pattern matching and presentation, and the association relationship of the cultural style of the period corresponding to the pattern.
4. The method for intelligent reconstruction of incomplete patterns of antique cultural relics based on multimodal feature fusion according to claim 1 is characterized by: The multimodal feature information at least includes multimodal feature information of the cultural relic itself and multimodal feature information of the pattern on the cultural relic.
5. The method for intelligent reconstruction of incomplete patterns of antique cultural relics based on multimodal feature fusion according to claim 1 is characterized by: In step S500 and step S600, if the same cultural relic pattern is retrieved and matched, the proportion of the retrieved and matched cultural relic pattern is adapted, and then the incomplete pattern portion of the target cultural relic is reconstructed based on the retrieved and matched cultural relic pattern.
6. The method for intelligent reconstruction of incomplete patterns of antique cultural relics based on multimodal feature fusion according to claim 1 is characterized by: In step S500 and step S600, if only similar patterns on cultural relics of the same type and from the same period are retrieved and matched, the incomplete pattern part of the target cultural relic is supplemented and reconstructed based on the feature information of the same type of cultural relics and their patterns from the same period, and the line feature information of the incomplete pattern of the target cultural relic.
7. The method for intelligent reconstruction of incomplete patterns of antique cultural relics based on multimodal feature fusion according to claim 6 is characterized by: Build an online expert database, send the supplemented and reconstructed patterns, target cultural relics and their characteristic information, incomplete patterns of target cultural relics and their characteristic information, and retrieved and matched similar cultural relics patterns and their characteristic information to the online expert database, and have experts review and confirm them online.
8. A system for intelligently reconstructing incomplete patterns of antique artifacts based on multimodal feature fusion, comprising a processor, a storage device, and a computer program stored on the storage device and executable on the processor, characterized in that: When the processor executes the computer program, it implements the intelligent reconstruction method of incomplete patterns of antique cultural relics based on multimodal feature fusion according to any one of claims 1 to 7.