An AI-based Cultural Relic Restoration Traceability System and Method

Abstract

The present invention discloses an artificial intelligence-based cultural relic restoration traceability system and method, which relates to the field of traceability technology, and includes a historical event extraction module, an object to be analyzed and restored determination module, an optimal restoration algorithm analysis module, a traceability feature pair construction module, a matching warning module, and a real-time restoration verification module; the historical event extraction module is used to extract historical events of using artificial intelligence for cultural relic restoration; the object to be analyzed and restored determination module is used to determine the object to be analyzed and restored in the subset of restoration events; the optimal restoration algorithm analysis module is used to output the optimal restoration algorithm for the object to be analyzed and restored; the traceability feature pair construction module is used to construct traceability feature pairs for the restoration objects corresponding to each type of cultural relic; the matching warning module is used to, when a newly added cultural relic needs to be restored, search for the type of cultural relic and the types of objects included in the restoration, and perform matching warnings for the traceability feature pairs.

Description

Technical Field

[0001] The present invention relates to the technical field of traceability, and specifically to an artificial intelligence-based cultural relic restoration traceability system and method. Background Art

[0002] Currently, in the analysis process of cultural relic restoration traceability, machine learning technology is often used to complete the cultural relic protection review work more quickly and accurately. Through training algorithms, rules and patterns related to cultural relic protection can be found from the far larger data obtained from previous experiments, and then problems and areas that need to be repaired in cultural relics can be quickly and accurately identified. For example, machine learning methods can identify areas with the same color and texture as the cultural relics, thereby determining where the cultural relic remains are located and finding the normal / abnormal areas of the cultural relics. However, the current situation is that for the same area that needs to be repaired, different mathematical algorithms are applied for the same type of cultural relics or different types of cultural relics. However, the selection of these algorithms generally needs to be compared and screened with other algorithms to be finally determined. In a large number of data restoration events, if this operation is performed on each restoration object, a lot of additional workload will be generated, and the restoration efficiency will be greatly reduced, losing the meaning of using artificial intelligence for cultural relic restoration. Summary of the Invention

[0003] The purpose of the present invention is to provide an artificial intelligence-based cultural relic restoration traceability system and method to solve the problems raised in the prior art.

[0004] To achieve the above purpose, the present invention provides the following technical solution: An artificial intelligence-based cultural relic restoration traceability method, the method comprising the following steps:

[0005] Step S100: Extract historical events of using artificial intelligence for cultural relic restoration, classify and store the historical events by cultural relic type, traverse all historical events of each type of cultural relic restoration, and store historical events with the same type of restoration object as the restoration event subset of the corresponding type of cultural relic; Determine the restoration objects with algorithm differences in the restoration event subset as the restoration objects to be analyzed for the corresponding type of cultural relic;

[0006] Step S200: Based on the restoration objects to be analyzed, search for the restoration process and restoration evaluation results recorded in the historical events of the corresponding type of cultural relic, and output the optimal restoration algorithm for the restoration objects to be analyzed.

[0007] Step S300: Perform traceability marking on the restoration processes of the cultural relics corresponding to the restoration objects to be analyzed with the optimal restoration algorithm recorded and other types of cultural relics respectively, and construct traceability feature pairs for the restoration objects corresponding to each type of cultural relic.

[0008] Step S400: When a newly added cultural relic needs to be restored, search for the type of the cultural relic and the types of restoration objects it contains, perform matching warnings for traceability feature pairs, and perform verification based on the real-time restoration evaluation results.

[0009] Further, determine that the restoration objects with algorithm differences in the subset of restoration events are the restoration objects to be analyzed for the corresponding type of cultural relic, including the following steps:

[0010] The restoration object refers to all restoration objects targeted in the restoration process of the historical event recording algorithm; the restoration order of various restoration objects corresponding to each subset of restoration events is the same in the historical event; when the restoration order of restoration objects is different for the restoration of the same type of cultural relic, store the historical events corresponding to different recording orders as independent subsets of restoration events, and they are parallel subsets corresponding to the same type of cultural relic under the same recorded restoration objects;

[0011] Traverse and search all historical events in the subset of restoration events corresponding to each type of cultural relic, obtain the restoration algorithm applied to each restoration object in any historical event as the target restoration algorithm, and search whether the restoration algorithms of the same restoration object corresponding to the remaining historical events in the same subset of restoration events are the same as the target restoration algorithm. If there are differences, output that there are algorithm differences for the corresponding restoration object, and if they are all the same, output that there are no algorithm differences for the corresponding restoration object.

[0012] Further, step S2 includes the following:

[0013] The restoration process refers to the execution process of first performing disease annotation and data sampling on the cultural relic, and then performing corresponding algorithm restorations on the restoration objects with different annotations in sequence. After each restoration object completes the algorithm restoration process, it is necessary to perform a manual visual compliance judgment. After passing the judgment, perform the algorithm restoration of the next restoration object. When not meeting the standard, return to this restoration object for re-algorithm restoration until the manual visual compliance judgment is passed as the restoration is qualified and then perform the algorithm restoration of the next restoration object; after all restoration objects are executed and judged to be restored qualified, output the restoration evaluation results of the corresponding type of cultural relic;

[0014] The restoration evaluation results record the return execution times and the applied restoration algorithms of each type of cultural relic for each restoration object during the algorithm restoration.

[0015] Capture all types of restoration algorithms recorded in the historical events of the restoration objects to be analyzed in the same subset of restoration events and store them in the first algorithm set. Extract the average return execution times D of the corresponding historical event records for the i-th type of restoration algorithm in the first algorithm set 0i , D 0i =(1 / n i )∑d 0i , where d 0iDenote the return execution times of different historical event records corresponding to the same object to be analyzed for repair in the same subset of repair events and for the i-th type of repair algorithm, n i Denote the total number of historical events corresponding to the same object to be analyzed for repair in the same subset of repair events and for the i-th type of repair algorithm;

[0016] Using the formula: U i = D 0i / [(1 / m)∑(D 0i )];

[0017] Calculate the return rate U of the execution link of each type of repair algorithm for the records of the objects to be analyzed for repair i , i ≤ m, where m represents the total number of types of repair algorithms recorded for the objects to be analyzed for repair in the first algorithm set;

[0018] The smaller the return rate of the execution link, the more effective and accurate the repair algorithm corresponding to the object to be analyzed for repair;

[0019] Mark the repair stage of the historical event record corresponding to the i-th type of repair algorithm. The repair stage refers to the execution cycle from the start of the algorithm repair for the object to be analyzed for repair to the start of the algorithm repair for the next object; Extract the area Q of the image of the object to be analyzed for repair corresponding to the disease annotation of the cultural relic during the repair stage of the i-th type of repair algorithm 1i and the overall image area Q of the cultural relic shown 0i , and obtain the area q of the image marked as effectively repaired for the object to be analyzed for repair during the repair stage corresponding to the i-th type of repair algorithm 1i and the area q of the image marked as repaired multiple times 2i ;

[0020] Effective repair means that the image of the object to be analyzed for repair with disease annotation remains unchanged during the remaining repair stage after the first execution of the algorithm repair during the repair stage. Multiple repairs mean that the image of the object to be analyzed for repair with disease annotation changes during the remaining repair stage after the first execution of the algorithm repair during the repair stage; Using the formula: A i =(Q 1i / Q 0i )×(q 1i / q 2i ); Calculate the repair coefficient A of the i-th type of repair algorithm i ;

[0021] The larger the repair coefficient, the higher the repair efficiency of the repair algorithm corresponding to the disease annotation during the repair stage and the better the overall repair result;

[0022] Extract the repair coefficient A and the return rate U of the execution link of the i-th type of repair algorithm corresponding to each object to be analyzed for repair in the same subset of repair events i and the return rate U of the execution linki , calculate the evaluation index R of each repair object to be analyzed corresponding to the i-th type of repair algorithm in the subset of repair events i , R i = e 1 × A i + e 2 ×(1 / U i );

[0023] Based on the evaluation index, sort the m repair algorithms of the repair object to be analyzed from large to small, and select the repair algorithm ranked first in the sequence as the optimal repair algorithm of the corresponding repair object to be analyzed in the subset of repair events where it is located.

[0024] By analyzing the execution link return rate and repair coefficient of the repair algorithm, it is possible to effectively realize the difficulty of algorithm selection in the process of traceability repair of a large number of cultural relics based on intelligent data, reduce the errors and manpower pressure brought by human judgment of the pros and cons of the algorithm, and improve the precision intelligence and high efficiency and convenience of using artificial intelligence for cultural relic repair.

[0025] Furthermore, step S300 includes:

[0026] Extract the repair object to be analyzed corresponding to the optimal repair algorithm, as well as the subset of repair events and the cultural relic type where the repair object to be analyzed is located;

[0027] Take the cultural relic type as the traceability target of the traceability feature pair, the repair process recorded in the subset of repair events and the repair objects in the process as the traceability clue of the traceability feature pair, and the optimal repair algorithm of the repair object to be analyzed as the traceability result of the traceability feature pair, to form the traceability feature pair of each repair object to be analyzed;

[0028] For other cultural relics not marked as repair objects to be analyzed, construct traceability feature pairs with the cultural relic type as the traceability target and the repair algorithm of the repair object as the traceability result.

[0029] Furthermore, step S400 includes the following:

[0030] When the cultural relic type is the traceability feature pair that does not record the repair object to be analyzed, extract the repair object corresponding to the newly added cultural relic and perform matching warning on the traceability result recorded by the corresponding traceability feature pair;

[0031] When the cultural relic type is the traceability feature pair that records the repair object to be analyzed, determine the type of the newly added cultural relic to lock the traceability target, extract the traceability feature pairs recorded by the same traceability target, and match the traceability feature pairs with the same traceability clue as the final matching feature pair, output the traceability result in the final matching feature pair as the optimal repair algorithm of the repair object included in the corresponding newly added cultural relic, and give a warning;

[0032] Extract all historical events corresponding to the optimal repair algorithm for real-time matching warning, obtain the evaluation index of the repair algorithm implemented for the repair object to be analyzed recorded in the historical events, and select the minimum value of the evaluation index as the evaluation index for implementing the optimal repair algorithm for the corresponding repair object; and calculate the real-time evaluation index. If the real-time evaluation index is greater than or equal to the evaluation index, continue to monitor; if the real-time evaluation index is less than the evaluation index, mark the repair event as a special repair event;

[0033] When the number of records of special repair events is greater than the threshold, output a response signal to give a warning response to the traceability feature pair.

[0034] An artifact restoration traceability system based on artificial intelligence, the system includes a historical event extraction module, a repair object to be analyzed determination module, an optimal repair algorithm analysis module, a traceability feature pair construction module, a matching warning module and a real-time repair verification module;

[0035] The historical event extraction module is used to extract historical events of artifact restoration using artificial intelligence;

[0036] The repair object to be analyzed determination module is used to determine the repair objects with algorithm differences in the repair event subset as the repair objects to be analyzed for the corresponding type of artifacts;

[0037] The optimal repair algorithm analysis module is used to output the optimal repair algorithm for the repair object to be analyzed;

[0038] The traceability feature pair construction module is used to construct the traceability feature pairs for the repair objects corresponding to each type of artifact;

[0039] The matching warning module is used to, when a new artifact needs to be repaired, find the artifact type and the types of repair objects included, and perform matching warning of the traceability feature pairs;

[0040] The real-time repair verification module is used to perform verification based on the real-time repair evaluation results.

[0041] Furthermore, the optimal repair algorithm analysis module includes a repair process determination unit, a repair evaluation result recording unit, an average return execution times calculation unit, an execution link return rate calculation unit, a repair coefficient calculation unit and an evaluation index calculation unit;

[0042] The repair process determination unit is used to determine the overall process of artifact restoration;

[0043] The repair evaluation result recording unit is used to record the return execution times and the applied repair algorithms of each type of artifact when performing algorithm repair on each repair object;

[0044] The average return execution times calculation unit is used to extract the average return execution times of the corresponding historical event records for each type of repair algorithm in the first algorithm set;

[0045] The execution link return rate calculation unit is used to calculate the execution link return rate of each type of repair algorithm for each repair object to be analyzed and repaired;

[0046] The repair coefficient calculation unit is used to calculate the repair coefficient of the corresponding repair algorithm;

[0047] The evaluation index calculation unit is used to calculate the evaluation index of each repair object to be analyzed and repaired corresponding to each type of repair algorithm in the repair event subset. Based on the evaluation index, the repair algorithms of the repair objects to be analyzed and repaired are sorted from large to small, and the repair algorithm with the first sequence is selected as the optimal repair algorithm of the corresponding repair object to be analyzed and repaired in the repair event subset where it is located.

[0048] Further, the traceability feature pair construction module includes a traceability target determination unit, a traceability clue induction unit, a traceability result output unit, and a traceability feature pair output unit;

[0049] The traceability target determination unit is used to use the cultural relic type as the traceability target of the traceability feature pair;

[0050] The traceability clue induction unit is used to use the repair process recorded in the repair event subset and the repair objects in the process as the traceability clues of the traceability feature pair;

[0051] The traceability result output unit is used to use the optimal repair algorithm of the repair object to be analyzed and repaired as the traceability result of the traceability feature pair,

[0052] The traceability feature pair output unit is used to construct a traceability feature pair with the cultural relic type as the traceability target and the repair algorithm of the repair object as the traceability result.

[0053] Compared with the prior art, the beneficial effects of the present invention are:

[0054] 1. The present invention differentiates and divides the repair objects of different types of cultural relics, solving the problem of the complexity and diversity of repair objects in cultural relic repair events and the difficulty of induction; laying a foundation for subsequent data processing and analysis;

[0055] 2. The present invention analyzes the historical execution repair algorithms of the inducted repair objects, and uses data processing and analysis methods to screen out the repair algorithm that can effectively reflect the repair algorithm corresponding to the highest efficiency of the repair object as the repair algorithm of the repair object under the corresponding type and repair process. This makes the scenario matching relatively clear, and after determining the repair algorithm, it can be effectively applied to subsequent additional cultural relic repair events, effectively solving the problem of difficult algorithm selection in the process of traceability repair of a large number of cultural relics based on intelligent data, reducing the error and manpower pressure brought by human judgment of algorithm advantages and disadvantages, and improving the precise intelligence and high efficiency and convenience of applying artificial intelligence to cultural relic repair. Description of the Drawings

[0056] Figure 1 This is a schematic structural diagram of a cultural relic restoration traceability system based on artificial intelligence according to the present invention. Specific implementation manners

[0057] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0058] Embodiment: As Figure 1 shown, the present invention provides a cultural relic restoration traceability system based on artificial intelligence. The system includes a historical event extraction module, an object to be analyzed for restoration determination module, an optimal restoration algorithm analysis module, a traceability feature pair construction module, a matching warning module, and a real-time restoration verification module;

[0059] The historical event extraction module is used to extract historical events of using artificial intelligence for cultural relic restoration;

[0060] The object to be analyzed for restoration determination module is used to determine the objects to be analyzed for restoration of corresponding types of cultural relics with algorithm differences in the sub-set of restoration events;

[0061] The optimal restoration algorithm analysis module is used to output the optimal restoration algorithm for the object to be analyzed for restoration;

[0062] The traceability feature pair construction module is used to construct traceability feature pairs for the restoration objects corresponding to each type of cultural relic;

[0063] The matching warning module is used to search for the cultural relic type and the types of restoration objects included when a new cultural relic needs to be restored, and perform matching warnings for the traceability feature pairs;

[0064] The real-time restoration verification module is used to perform verification based on the real-time restoration evaluation results.

[0065] The optimal restoration algorithm analysis module includes a restoration process determination unit, a restoration evaluation result recording unit, an average return execution times calculation unit, an execution link return rate calculation unit, a restoration coefficient calculation unit, and an evaluation index calculation unit;

[0066] The restoration process determination unit is used to determine the overall process of cultural relic restoration;

[0067] The restoration evaluation result recording unit is used to record the return execution times and the applied restoration algorithms when each type of cultural relic performs algorithm restoration on each restoration object;

[0068] The average return execution times calculation unit is used to extract the average return execution times of the corresponding historical event records for each type of repair algorithm in the first algorithm set;

[0069] The execution link return rate calculation unit is used to calculate the execution link return rate of each type of repair algorithm for each repair object to be analyzed;

[0070] The repair coefficient calculation unit is used to calculate the repair coefficient of the corresponding repair algorithm;

[0071] The evaluation index calculation unit is used to calculate the evaluation index of each repair object to be analyzed corresponding to each type of repair algorithm in the repair event subset. Based on the evaluation index, the repair algorithms for the repair objects to be analyzed are sorted from largest to smallest, and the repair algorithm with the first sequence is selected as the optimal repair algorithm for the corresponding repair object to be analyzed in the repair event subset where it is located.

[0072] The traceability feature pair construction module includes a traceability target determination unit, a traceability clue induction unit, a traceability result output unit, and a traceability feature pair output unit;

[0073] The traceability target determination unit is used to use the cultural relic type as the traceability target of the traceability feature pair;

[0074] The traceability clue induction unit is used to use the repair process recorded in the repair event subset and the repair objects in the process as the traceability clues of the traceability feature pair;

[0075] The traceability result output unit is used to use the optimal repair algorithm of the repair object to be analyzed as the traceability result of the traceability feature pair,

[0076] The traceability feature pair output unit is used to construct a traceability feature pair with the cultural relic type as the traceability target and the repair algorithm of the repair object as the traceability result.

[0077] An artificial intelligence-based cultural relic repair traceability method, the method includes the following steps:

[0078] Step S100: Extract the historical events of using artificial intelligence for cultural relic repair, classify and store the historical events by cultural relic type, traverse all historical events of each type of cultural relic repair, and store the historical events with the same type of repair object as the repair event subset of the corresponding type of cultural relic; Determine the repair objects with algorithm differences in the repair event subset as the repair objects to be analyzed of the corresponding type of cultural relic;

[0079] For example, the cultural relic types can be divided into ancient paintings and ancient porcelain;

[0080] Step S200: Based on the repair object to be analyzed, search for the repair process and repair evaluation results of the historical event records of the corresponding type of cultural relic, and output the optimal repair algorithm of the repair object to be analyzed;

[0081] Step S300: Traceability mark the restoration processes of the cultural relics corresponding to the restoration objects to be analyzed that record the optimal restoration algorithm and other types of cultural relics respectively, and construct traceability feature pairs of the restoration objects corresponding to each type of cultural relic;

[0082] Step S400: When a newly added cultural relic needs to be restored, search for the type of cultural relic and the types of restoration objects it contains, perform matching warnings for traceability feature pairs, and perform verification based on the real-time restoration evaluation results.

[0083] Determine that the restoration objects with algorithm differences in the restoration event subset are the restoration objects to be analyzed for the corresponding type of cultural relic, including the following steps:

[0084] The restoration object refers to all restoration objects targeted in the restoration process of the historical event record algorithm; the restoration order of each type of restoration object corresponding to each restoration event subset is the same in the historical event; when the restoration order of the restoration objects for the same type of cultural relic is different, store the historical events corresponding to different recording orders as independent restoration event subsets, and they are parallel subsets corresponding to the same type of cultural relic under the same recorded restoration objects;

[0085] One type of cultural relic can record multiple restoration event subsets, and all the restoration objects included in each recorded restoration event subset are not completely the same;

[0086] Traverse and search all historical events in the restoration event subset corresponding to each type of cultural relic, obtain the restoration algorithm applied to each restoration object in any historical event as the target restoration algorithm, and check whether the restoration algorithms of the same restoration objects corresponding to the remaining historical events in the same restoration event subset are the same as the target restoration algorithm. If there are differences, output that there are algorithm differences for the corresponding restoration object; if all are the same, output that there are no algorithm differences for the corresponding restoration object.

[0087] Step S2 includes the following:

[0088] The restoration process refers to the execution process of first performing disease annotation and data sampling on the cultural relics, and then performing corresponding algorithm restorations on different restoration objects based on the annotation in sequence. After each restoration object completes the algorithm restoration process, it is necessary to perform a manual visual compliance judgment. After passing the judgment, perform the algorithm restoration of the next restoration object. When the judgment fails, return to this restoration object for re-algorithm restoration until the manual visual compliance judgment is passed for restoration compliance, and then perform the algorithm restoration of the next restoration object; after all restoration objects are completed and the judgment is made as restoration compliant, output the restoration evaluation results of the corresponding type of cultural relic;

[0089] The restoration objects in sequence refer to the restoration strategies corresponding to different types of cultural relics based on disease annotation and data sampling, and each type of cultural relic corresponds to a restoration strategy based on the initial disease annotation and data sampling, which is recorded in the restoration traceability system;

[0090] The restoration evaluation result records the number of return executions and the applied restoration algorithms for each type of cultural relic in the algorithm-based restoration of each restoration object;

[0091] Capture all types of restoration algorithms in the historical event records of the restoration object to be analyzed in the same sub-set of restoration events and store them in the first algorithm set. Extract the average return execution count D of the corresponding historical event records for the i-th type of restoration algorithm in the first algorithm set 0i , D 0i =(1 / n i )∑d 0i , where d 0i represents the return execution count of different historical event records for the same restoration object to be analyzed and the execution of the i-th type of restoration algorithm in the same sub-set of restoration events, and n i represents the total number of historical events corresponding to the same restoration object to be analyzed and the execution of the i-th type of restoration algorithm in the same sub-set of restoration events;

[0092] Use the formula: U i =D 0i / [(1 / m)∑(D 0i )];

[0093] Calculate the execution link return rate U of each restoration object to be analyzed for each type of restoration algorithm i , i≤m, where m represents the total number of types of restoration algorithms recorded for the restoration object to be analyzed in the first algorithm set;

[0094] The sub-set of restoration events will record historical events with the same restoration execution process but different applied restoration algorithms;

[0095] The smaller the execution link return rate, the more effective and accurate the restoration algorithm corresponding to the restoration object to be analyzed;

[0096] Mark the restoration stage of the historical event records corresponding to the i-th type of restoration algorithm. The restoration stage refers to the execution cycle from the start of the algorithm-based restoration of the corresponding restoration object to be analyzed to the start of the algorithm-based restoration of the next restoration object; Extract the image area Q of the restoration object to be analyzed corresponding to the disease annotation of the cultural relic in the restoration stage corresponding to the i-th type of restoration algorithm 1i and the overall image area Q of the cultural relic displayed 0i , and obtain the image area q marked as effectively restored and the image area q marked as repeatedly restored for the restoration object to be analyzed in the restoration stage corresponding to the i-th type of restoration algorithm 1i and the image area q marked as repeatedly restored 2i ;

[0097] Effective repair means that the image of the repair object to be analyzed marked with diseases remains unchanged in the remaining repair stages after the first execution of the algorithm repair in the repair stage. Multiple repairs mean that the image of the repair object marked with diseases changes in the remaining repair stages after the first execution of the algorithm repair in the repair stage. Using the formula: A i =(Q 1i / Q 0i )×(q 1i / q 2i ); Calculate the repair coefficient A i of the i-th type of repair algorithm;

[0098] The image remaining unchanged means that the similarity remains 100%. The image changing means that the similarity changes. The image changing indicates that there is a situation where the repair object is not repaired well when the repair algorithm is first applied in the repair stage, so it needs to be repaired again;

[0099] The larger the repair coefficient, the higher the repair efficiency of the corresponding repair algorithm for the disease annotation in the repair stage and the better the overall repair result;

[0100] Extract the repair coefficient A i of the i-th type of repair algorithm corresponding to each repair object to be analyzed in the same repair event subset and the execution link return rate U i , Calculate the evaluation index R i of each repair object to be analyzed corresponding to the i-th type of repair algorithm in the repair event subset, R i =e 1 ×A i +e 2 ×(1 / U i );

[0101] Based on the evaluation index, sort the m repair algorithms for the repair object to be analyzed from largest to smallest, and select the repair algorithm with the first sequence as the optimal repair algorithm for the corresponding repair object to be analyzed in the repair event subset where it is located.

[0102] As shown in the embodiment: Taking the mural as the cultural relic to be repaired, first perform disease annotation and data sampling on the mural. If the repair object is classified as image shedding and cracks after disease annotation; If the image shedding is repaired first, the historical events record two types of algorithms, namely the mask repair algorithm based on Fourier convolution and the ACP-LaMa algorithm;

[0103] That is, for the cultural relic to be repaired as a mural, the historical event record with the repair process of first repairing the image shedding and then the cracks can be used as the repair event subset, and this subset records two algorithms for the repair object of image shedding;

[0104] Based on a number of historical events corresponding to two types of algorithms, the final result of comparison can be obtained by analyzing the repair coefficient and the return rate of the execution link.

[0105] By analyzing the return rate of the execution link and the repair coefficient of the repair algorithm, it is possible to effectively solve the problem of difficult algorithm selection in the process of traceability repair of a large number of cultural relics based on intelligent data, reduce the errors and manpower pressure caused by human judgment of the advantages and disadvantages of algorithms, and improve the precision, intelligence, efficiency and convenience of using artificial intelligence for cultural relic restoration.

[0106] Step S300 includes:

[0107] Extract the repair objects to be analyzed corresponding to the optimal repair algorithm, as well as the subset of repair events and the types of cultural relics in which the repair objects to be analyzed are located;

[0108] Take the type of cultural relic as the traceability target of the traceability feature pair, the repair process recorded in the subset of repair events and the repair objects in the process as the traceability clue of the traceability feature pair, and the optimal repair algorithm of the repair object to be analyzed as the traceability result of the traceability feature pair, to form the traceability feature pair of each repair object to be analyzed.

[0109] For other cultural relics not marked as repair objects to be analyzed, construct traceability feature pairs with the type of cultural relic as the traceability target and the repair algorithm of the repair object as the traceability result.

[0110] Step S400 includes the following:

[0111] When the type of cultural relic is a traceability feature pair that does not record the repair object to be analyzed, extract the repair objects corresponding to the newly added cultural relics and perform matching warnings on the traceability results recorded by the corresponding traceability feature pairs;

[0112] When the type of cultural relic is a traceability feature pair that records the repair object to be analyzed, determine the type of the newly added cultural relic to lock the traceability target, extract the traceability feature pairs recorded by the same traceability target, and match the traceability feature pairs with the same traceability clue as the final matching feature pair, output the traceability result in the final matching feature pair as the optimal repair algorithm for the repair object included in the corresponding newly added cultural relic, and issue a warning;

[0113] Extract all historical events corresponding to the optimal repair algorithm of the real-time matching warning, obtain the evaluation index of the repair algorithm implemented for the corresponding repair object to be analyzed recorded in the historical events, select the minimum value of the evaluation index as the evaluation index for implementing the optimal repair algorithm for the corresponding repair object; and calculate the real-time evaluation index. If the real-time evaluation index is greater than or equal to the evaluation index, continue to monitor; if the real-time evaluation index is less than the evaluation index, mark the repair event as a special repair event.

[0114] When the number of special repair event records is greater than the threshold, an output response signal is used to give an early warning response to the traceability feature pair. The purpose of the early warning response is to remind the system or the administrator that the optimal repair algorithm for the corresponding repair object may change as the data increases and further updates are required.

[0115] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above-described exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention. Therefore, in any aspect, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be embraced within the present invention. Any reference signs in the claims should not be construed as limiting the claims involved.

Claims

1. A method for tracing the source of cultural relics restoration based on artificial intelligence, characterized in that: The method comprises the following steps: Step S100: extracting historical events of cultural relic restoration using artificial intelligence, classifying and storing the historical events by cultural relic type, traversing all historical events of cultural relic restoration of each type, and storing the historical events that record the same type of restoration objects as a restoration event subset of the corresponding type of cultural relic; determining the restoration objects with algorithm differences in the restoration event subset as the restoration objects to be analyzed for the corresponding type of cultural relic; Step S200: Based on the restoration object to be analyzed, the restoration process and restoration evaluation results of the historical event records of the corresponding type of cultural relics are searched, and the optimal restoration algorithm of the restoration object to be analyzed is output; Step S300: Mark the restoration process of the cultural relics corresponding to the restoration object to be analyzed and recording the optimal restoration algorithm and other types of cultural relics, and construct a traceability feature pair corresponding to the restoration object of each type of cultural relic; Step S400: When a new cultural relic needs to be repaired, the type of cultural relic and the type of repair object are searched, and a matching warning of the traceability feature pair is performed and verified based on the real-time repair evaluation result; The step S400 includes the following: When the type of cultural relic does not record the traceability feature pair of the restoration object to be analyzed, the restoration object corresponding to the newly added cultural relic is extracted to match the traceability result recorded by the corresponding traceability feature pair for early warning; When the type of cultural relic is a traceability feature pair that records the repair object to be analyzed, determine the type of the newly added cultural relic to lock the traceability target, extract the traceability feature pairs recorded with the same traceability target, and match the traceability feature pairs with the same traceability clues as the final matching feature pairs. Output the traceability result in the final matching feature pair as the optimal repair algorithm for the newly added cultural relic containing the repair object, and issue an early warning; Extract all historical events corresponding to the optimal repair algorithm of the real-time matching warning, obtain the evaluation index of the repair algorithm implemented for the repair object to be analyzed recorded in the historical events, select the minimum value of the evaluation index as the evaluation index for the optimal repair algorithm implemented for the corresponding repair object; and calculate the real-time evaluation index. If the real-time evaluation index is greater than or equal to the evaluation index, continue monitoring; if the real-time evaluation index is less than the evaluation index, mark the repair event as a special repair event; When the number of special repair event records is greater than a threshold, a response signal is output to perform an early warning response to the traceability feature pair.

2. The method for tracing the source of cultural relics restoration based on artificial intelligence according to claim 1, characterized in that: The step of determining the restoration objects with algorithm differences in the restoration event subset as restoration objects to be analyzed for the corresponding type of cultural relics comprises the following steps: The restoration objects refer to all restoration objects targeted by the restoration process of the historical event recording algorithm; the restoration order of each type of restoration object corresponding to each restoration event subset is the same in the historical events; when the restoration order of restoration objects of the same type of cultural relics is different, the historical events corresponding to the different recording orders are stored as independent restoration event subsets, and belong to the parallel subsets of the corresponding type of cultural relics under the same restoration object record; Traverse and search for all historical events in the restoration event subset corresponding to each type of cultural relic, obtain the restoration algorithm applied to each restoration object in any historical event as the target restoration algorithm, and find out whether the restoration algorithms corresponding to the same restoration object of the remaining historical events in the same restoration event subset are the same as the target restoration algorithm. If there is a difference, output that there is an algorithm difference for the corresponding restoration object; if all are the same, output that there is no algorithm difference for the corresponding restoration object.

3. The artificial intelligence-based cultural relic restoration and tracing system and method according to claim 1, characterized in that: The step S200 includes the following: The restoration process refers to the process of first marking the defects and sampling the data of the cultural relics, and then performing the corresponding algorithm restoration execution process in sequence based on the marked different restoration objects, and after each restoration object completes the algorithm restoration process, it is necessary to perform manual visual standard judgment, and after the standard is met, the algorithm restoration of the next restoration object is performed, and if the standard is not met, the restoration object is returned to perform algorithm restoration again, until the manual visual standard is judged to be restored to the standard, and the algorithm restoration of the next restoration object is performed; after all restoration objects are executed and judged to be restored to the standard, the restoration evaluation results of the corresponding type of cultural relics are output; The restoration evaluation result records the number of times each type of cultural relic is restored by performing the restoration algorithm on each restoration object and the restoration algorithm applied; Capture all types of repair algorithms of the historical event records of the repair object to be analyzed in the same repair event subset and store them in the first algorithm set. Extract the average return execution times D of the corresponding historical event records for the i-th type of repair algorithm in the first algorithm set. 0i , D 0i =(1 / n i )∑d 0i , where d 0i Indicates the number of times different historical event records corresponding to the execution of the i-th type of repair algorithm in the same repair event subset for the same repair object to be analyzed, n i represents the total number of historical events corresponding to the same repair object to be analyzed and the execution of the i-th type repair algorithm in the same repair event subset; Using the formula: U i =D 0i / [(1 / m)∑(D 0i )]; Calculate the return rate U of each type of repair algorithm executed for each repair object to be analyzed i , i≤m, m represents the total number of repair algorithm types recorded in the first algorithm set for the repair object to be analyzed; Mark the restoration stage of the historical event record corresponding to the i-th type restoration algorithm, where the restoration stage refers to the execution period from the start of the restoration algorithm execution on the corresponding restoration object to be analyzed to the start of the restoration algorithm execution on the next restoration object; extract the image area Q of the restoration object to be analyzed corresponding to the disease annotation of the cultural relics in the restoration stage corresponding to the i-th type restoration algorithm 1i and the overall image area Q of the artifact display 0i , and obtain the image area q marked as valid repair for the repair object to be analyzed in the corresponding repair stage of the i-type repair algorithm 1i and the image area q marked as multiple restorations 2i ; The effective repair means that the image of the repair object to be analyzed and repaired after the first execution of the algorithm in the repair stage remains unchanged in the remaining repair stages, and the multiple repairs mean that the image of the repair object to be analyzed and repaired after the first execution of the algorithm in the repair stage changes in the remaining repair stages; using the formula: A i =(Q 1i / Q 0i )×(q 1i / q 2i ); Calculate the repair coefficient A of the i-th type repair algorithm i ; Extract the repair coefficient A of the i-th type repair algorithm corresponding to each repair object to be analyzed in the same repair event subset i and the execution phase return rate U i , calculate the evaluation index R of the i-th type repair algorithm for each repair object to be analyzed in the repair event subset i , R i =e1×A i +e2×(1 / U i ); Based on the evaluation index, the m repair algorithms of the repair object to be analyzed are sorted from large to small, and the repair algorithm at the first place in the sequence is selected as the optimal repair algorithm corresponding to the repair object to be analyzed in the repair event subset.

4. The artificial intelligence-based cultural relic restoration and tracing system and method according to claim 1, characterized in that: The step S300 includes: Extract and record the restoration object to be analyzed corresponding to the optimal restoration algorithm, as well as the restoration event subset and cultural relic type of the restoration object to be analyzed; The type of cultural relic is used as the traceability target of the traceability feature pair, the restoration process and the restoration object in the process recorded in the restoration event subset are used as the traceability clue of the traceability feature pair, and the optimal restoration algorithm of the restoration object to be analyzed is used as the traceability result of the traceability feature pair, thus forming the traceability feature pair of each restoration object to be analyzed; For other cultural relics that are not marked as restoration objects to be analyzed, traceability feature pairs are constructed with the cultural relic type as the traceability target and the restoration algorithm of the restoration object as the traceability result.

5. An artificial intelligence-based cultural relic restoration and tracing system, using an artificial intelligence-based cultural relic restoration and tracing method as described in any one of claims 1 to 4, characterized in that: The system includes a historical event extraction module, a to-be-analyzed repair object determination module, an optimal repair algorithm analysis module, a traceability feature pair construction module, a matching warning module, and a real-time repair verification module; The historical event extraction module is used to extract historical events of cultural relics restoration using artificial intelligence; The module for determining the restoration object to be analyzed is used to determine the restoration object with algorithm differences in the restoration event subset as the restoration object to be analyzed of the corresponding type of cultural relic; The optimal repair algorithm analysis module is used to output the optimal repair algorithm of the repair object to be analyzed; The traceability feature pair construction module is used to construct a traceability feature pair corresponding to the restoration object of each type of cultural relic; The matching warning module is used to find the type of cultural relics and the types of restoration objects when new cultural relics need to be restored, and to carry out matching warning of traceability feature pairs; The real-time repair verification module is used to perform verification based on the real-time repair evaluation result.

6. The artificial intelligence-based cultural relic restoration and tracing system according to claim 5 is characterized by: The optimal repair algorithm analysis module includes a repair process determination unit, a repair evaluation result recording unit, an average return execution times calculation unit, an execution link return rate calculation unit, a repair coefficient calculation unit and an evaluation index calculation unit; The restoration process determination unit is used to determine the overall process of cultural relic restoration; The restoration evaluation result recording unit is used to record the number of times each type of cultural relic is restored by performing the algorithm restoration on each restoration object and the restoration algorithm applied; The average return execution times calculation unit is used for extracting the average return execution times of corresponding historical event records by each type of repair algorithm in the first algorithm set; The execution link return rate calculation unit is used to calculate the execution link return rate of each type of repair algorithm for each repair object record to be analyzed; The repair coefficient calculation unit is used to calculate the repair coefficient of the corresponding repair algorithm; The evaluation index calculation unit is used to calculate the evaluation index of each type of repair algorithm corresponding to each repair object to be analyzed in the repair event subset. Based on the evaluation index, the repair algorithms of the repair objects to be analyzed are sorted from large to small, and the repair algorithm first in the sequence is selected as the optimal repair algorithm corresponding to the repair object to be analyzed in the repair event subset.

7. The artificial intelligence-based cultural relic restoration and tracing system according to claim 6 is characterized by: The traceability feature pair construction module includes a traceability target determination unit, a traceability clue induction unit, a traceability result output unit and a traceability feature pair output unit; The traceability target determination unit is used to use the cultural relic type as the traceability target of the traceability feature pair; The traceability clue induction unit is used to repair the repair process and the repair object in the process recorded in the repair event subset as the traceability clue of the traceability feature pair; The traceability result output unit is used to use the optimal repair algorithm of the repair object to be analyzed as the traceability result of the traceability feature pair, The traceability feature pair output unit is used to construct a traceability feature pair with the cultural relic type as the traceability target and the restoration algorithm of the restoration object as the traceability result.

Images