Image correction method and system based on graphic recognition

Abstract

The present invention discloses an image correction method and system based on graphic recognition, which is suitable for operation assistance in repairing the surface image of damaged objects, especially for use in fields such as cultural relics investigation. By scanning the structure and image of the damaged and missing object, a twin digital object is established, and the twin digital object is identified and judged. The area range of the missing surface image and the image category and image content of the image in the complete area are determined, and the content is retrieved and matched according to the category, and associated with the edge content of the missing area. The highly matched content is selected to supplement and correct the missing content, and operation assistance is provided to the repair personnel in turn, which facilitates the efficient content possibility screening and optimizes the correction operation efficiency.

Description

Technical Field

[0001] The present invention relates to the field related to graphic recognition and correction, and in particular to an image correction method and system based on graphic recognition. Background Art

[0002] During the process of cultural relics investigation and restoration, cultural relics may be damaged or missing due to various historical reasons, resulting in the structure of the cultural relics being incomplete, and relevant professionals are needed to restore them. The main difficulty in restoration lies in the restoration of surface content, especially for cultural relics with images, texts and other contents. During the restoration process, it is necessary to judge the possibility and rationality of the original content based on the remaining content in order to restore it.

[0003] In the existing technology, due to the importance of cultural relics, manual repair is usually adopted. Professionals repair the damaged content during the process of repairing the cultural relic structure, but this is limited to the partially damaged content. In addition, when the cultural relic is in a serious state of damage, the entire splicing and restoration process is more tedious and lengthy, consuming a lot of energy. At the same time, the damage may be further increased due to repeated movement of cultural relic fragments during the restoration attempt. Summary of the Invention

[0004] The object of the present invention is to provide an image correction method and system based on pattern recognition to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] An image correction method based on pattern recognition, comprising:

[0007] Synchronize the data acquisition end to obtain sensor scanning data, perform object restoration modeling based on multiple sets of sensor scanning data, and obtain a twin digital object including an overlay map layer; the sensor scanning data includes radar point cloud data and surface image data;

[0008] Performing image layer judgment based on the twin digital object to identify and mark image missing areas on the object surface, and performing planarization processing on image data of non-missing areas to generate corresponding image data, wherein the image data includes the image missing area mark;

[0009] Retrieve the image category and image content based on the image data, and perform feature matching and content matching based on the edge content of the missing image area in sequence to obtain a corrected image corresponding to the missing image area, wherein each missing image area corresponds to one or more corrected images;

[0010] The plurality of corrected images are sorted in descending order based on the matching rates of feature matching and content matching, and an interactive queue of image missing areas is generated based on the sorting results to output the image data accordingly.

[0011] As a further solution of the present invention, the steps of retrieving image category and image content based on the image data, and sequentially performing feature matching and content matching based on edge content of the missing image area to obtain a corrected image corresponding to the missing image area specifically include:

[0012] Extracting features of the image in the image data, identifying and determining text features and graphic features in the image, and respectively retrieving and obtaining feature types corresponding to the text features and graphic features, wherein the feature types are used to represent the style categories of the text and image;

[0013] If there is a cropped image object at the edge of the missing image area, the cropped image is obtained, feature matching is performed based on a feature database of feature types corresponding to text features and graphic features, and multiple matching image objects are selected in sequence based on the matching rate as the corrected image, where the image objects are used to represent content elements under the feature type, including text and basic graphics;

[0014] After the cropped image is matched, the image data is understood in terms of content elements and meaning, and the associated text content and graphic elements are matched, and a corrected image is generated based on the text content and graphic elements.

[0015] As a further solution of the present invention, the step of performing image layer judgment based on the twin digital object to identify and mark the image missing area on the object surface specifically includes:

[0016] Performing a color continuity evaluation on the surface of the twin digital object, generating an evenly spaced evaluation point grid with a preset evaluation accuracy, and calculating the color span value between the evaluation points. The color span value is used to represent the interval between two color corresponding points on the same evaluation palette;

[0017] If the color span value reaches the preset evaluation standard, a range mark is performed on the continuous area on the surface of the object based on the color value of the evaluation point to generate an abnormal area to be screened;

[0018] A structural evaluation is performed on the surface of the object in the abnormal area to be screened. If the object structure in the abnormal area to be screened is manifested as regional subsidence or abnormal surface luminosity, then the abnormal area to be screened is an image missing area. The regional subsidence is used to characterize the missing and damaged object structure, and the abnormal surface luminosity is used to characterize the weathering and corrosion of the object surface structure.

[0019] As a further solution of the present invention, the present invention also includes an auxiliary step of restoring the fragmented object, specifically including:

[0020] When the object structure is characterized as a region sinking, a gap space model of the sinking region is obtained, and the surface side of the object is marked;

[0021] Obtaining fragment space models of multiple fragment objects, performing accommodation evaluation on the fragment space models based on the gap space model, determining whether the subsidence gap can accommodate the fragment space model, establishing an accommodation list based on the determination result, and arranging the surface side of the fragment object and the subsidence gap to overlap when performing the accommodation evaluation;

[0022] A cross-matching verification is performed on several corrected images based on the object surface image content of the fragment objects in the accommodation list to screen out double-matched corrected images.

[0023] As a further solution of the present invention, the step of planarizing the image data of the non-deleted area specifically includes:

[0024] Establish a surface mesh, divide the corresponding surface of the object into several triangular meshes, and perform surface discretization processing;

[0025] Performing a planar projection on each of the triangular meshes in the three-dimensional surface to obtain an initial unfolded image, and gradually unfolding the adjacent triangular meshes to maintain the adjacent relationship of the meshes;

[0026] An objective function is defined to generate a conversion objective function for preserving the side length and angle of the triangular mesh, and scaling optimization is performed on the initial unfolded graph based on the objective function.

[0027] The embodiment of the present invention aims to provide an image correction system based on pattern recognition, comprising:

[0028] An object restoration module is used to synchronize with the data acquisition end to obtain sensor scanning data, perform object restoration modeling based on multiple sets of sensor scanning data, and obtain a twin digital object including an overlay map layer; the sensor scanning data includes radar point cloud data and surface image data;

[0029] A missing image location module is configured to perform image layer determination based on the twin digital object to identify and mark image missing areas on the object surface, and perform planarization processing on the image data of the non-missing areas to generate corresponding image data, the image data including the image missing area marks;

[0030] a content retrieval module, configured to retrieve image categories and image contents based on the image data, and perform feature matching and content matching based on the edge content of the image missing region, to obtain corrected images corresponding to the image missing region, wherein each image missing region corresponds to one or more corrected images;

[0031] The correction feedback module is used to sort the multiple corrected images in descending order based on the matching rates of feature matching and content matching, and generate an interactive queue of image missing areas based on the sorting results to output the image data accordingly.

[0032] As a further solution of the present invention: the content retrieval module includes:

[0033] a feature resolution unit for extracting features from the image in the image data, identifying and determining text features and graphic features in the image, and respectively retrieving and obtaining feature types corresponding to the text features and graphic features, wherein the feature types are used to characterize the style categories of the text and image;

[0034] a cropping correction unit, configured to, if a cropped image object exists at the edge of the missing image area, obtain the cropped image, perform feature matching based on a feature database of feature types corresponding to text features and graphic features, and select a plurality of matching image objects in sequence as correction images based on a matching rate, wherein the image objects are used to represent content elements of the feature type, including text and basic graphics;

[0035] The correction verification unit is used to understand the content elements and meaning of the image data after the cropped image is matched, match the associated text content and graphic elements, and generate a corrected image based on the text content and graphic elements.

[0036] As a further solution of the present invention: the missing location module includes:

[0037] Performing a color continuity evaluation on the surface of the twin digital object, generating an evenly spaced evaluation point grid with a preset evaluation accuracy, and calculating the color span value between the evaluation points. The color span value is used to represent the interval between two color corresponding points on the same evaluation palette;

[0038] If the color span value reaches the preset evaluation standard, a range mark is performed on the continuous area on the surface of the object based on the color value of the evaluation point to generate an abnormal area to be screened;

[0039] A structural evaluation is performed on the surface of the object in the abnormal area to be screened. If the object structure in the abnormal area to be screened is manifested as regional subsidence or abnormal surface luminosity, then the abnormal area to be screened is an image missing area. The regional subsidence is used to characterize the missing and damaged object structure, and the abnormal surface luminosity is used to characterize the weathering and corrosion of the object surface structure.

[0040] As a further solution of the present invention, a fragment reset module is also included, specifically including:

[0041] a sinking marking unit, configured to obtain a gap space model of the sinking area and mark the surface side of the object when the object structure is characterized as a sinking area;

[0042] an accommodation judgment unit, configured to obtain a fragment space model of a plurality of fragment objects, perform accommodation evaluation on the fragment space model based on the gap space model, determine whether the subsidence gap can accommodate the fragment space model, establish an accommodation list based on the judgment result, and when performing the accommodation evaluation, arrange the fragment objects to overlap with the surface side of the object of the subsidence gap;

[0043] The cross matching unit is used to perform cross matching verification on a plurality of corrected images based on the object surface image content of the fragment objects in the accommodation list to screen out double-matched corrected images.

[0044] As a further solution of the present invention: the missing location module further includes:

[0045] The meshing unit is used to establish a surface mesh, divide the corresponding surface of the object surface into several triangular meshes, and perform surface discretization processing;

[0046] A projection processing unit, configured to perform planar projection on each of the triangular meshes in the three-dimensional surface to obtain an initial unfolded image, and to gradually unfold adjacent triangular meshes to maintain a mesh adjacency relationship;

[0047] The scaling optimization unit is used to define an objective function to generate a conversion objective function for preserving the side length and angle of the triangular mesh, and perform scaling optimization on the initial expanded graphic based on the objective function.

[0048] Compared with the prior art, the beneficial effects of the present invention are: it is suitable for operational assistance in repairing the surface images of damaged objects, especially in the fields of cultural relics investigation, by scanning the structure and image of damaged and missing objects, establishing a twin digital object, identifying and judging the twin digital object, determining the area range of missing surface images and the image category and image content of the image in the complete area, and retrieving and matching content according to the category, associating it with the edge content of the missing area, selecting highly matched content to supplement and correct the missing content, and providing operational assistance to repair personnel in turn, facilitating efficient content possibility screening, and optimizing the efficiency of correction operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 The figure is a flowchart of an image correction method based on graphic recognition.

[0050] Figure 2 The figure is a flowchart of the auxiliary restoration step in an image correction method based on pattern recognition.

[0051] Figure 3 This is a block diagram of an image correction system based on graphic recognition. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0053] The specific implementation of the present invention is described in detail below with reference to specific embodiments.

[0054] like Figure 1 The image correction method based on pattern recognition provided in one embodiment of the present invention includes the following steps:

[0055] S10, synchronizing the data acquisition end to acquire sensor scanning data, performing object restoration modeling based on multiple sets of sensor scanning data, and acquiring a twin digital object including an overlay map layer, wherein the sensor scanning data includes radar point cloud data and surface image data;

[0056] S20, performing image layer determination based on the twin digital object to identify and mark image missing areas on the object surface, and performing planarization processing on image data of non-missing areas to generate corresponding image data, wherein the image data includes the image missing area mark;

[0057] S30, searching for image category and image content based on the image data, and sequentially performing feature matching and content matching based on edge content of the image missing region to obtain a corrected image corresponding to the image missing region, wherein each image missing region corresponds to one or more corrected images;

[0058] S40 , sorting the plurality of corrected images in descending order based on the matching rates of feature matching and content matching, and generating an interactive queue of image missing areas based on the sorting results to output the image data accordingly.

[0059] In this embodiment, a method for image correction based on graphic recognition is provided, which is suitable for assisting in repairing the surface image of damaged objects, especially in the field of cultural relics investigation. By scanning the structure and image of the damaged or missing object, a twin digital object is established, the twin digital object is identified and judged, the area range of the missing surface image and the image category and image content of the image in the complete area are determined, and the content is retrieved and matched according to the category, and associated with the edge content of the missing area, and the highly matched content is selected to supplement and correct the missing content, thereby providing operational assistance to the repair personnel, facilitating efficient content possibility screening, and optimizing the efficiency of the correction operation. In the process of cultural relics investigation and restoration, cultural relics may be damaged or missing due to various historical reasons, so that the structure of the cultural relics is no longer complete and needs to be restored by relevant professionals. The main difficulty in restoration lies in the restoration of surface content, especially for cultural relics with images, texts, etc. In the restoration process, it is necessary to judge the possibility and rationality of the original content based on the remaining content in order to restore it. The auxiliary implementation process adopted here is: first, the data acquisition end obtains the sensor scanning data for object restoration modeling, including radar point cloud data and image data, to realize structural modeling and surface mapping coverage respectively. After obtaining the twin digital object, the missing area of the surface layer of the twin digital object is judged, which usually includes two parts, one is caused by structural loss, and the other is caused by surface corrosion; the edge of the missing area and the content of the complete area can be retrieved from the database through graphic recognition to judge the dynasty, format, specific use and meaning, and common components of the text and content in the image, so that the matching range can be further narrowed. At this time, the partial incomplete content at the edge of the missing area can be matched, and the content with matching features can be selected as the correction content. For the completely correct part, the possibility can be evaluated by content recognition. For example, when it is text content, the possibility is matched according to the expression relevance of the complete part content and the common expression method under the current content format, and multiple groups of suitable content are selected as the correction standby content to assist relevant personnel in repair.

[0060] As another preferred embodiment of the present invention, the steps of retrieving image category and image content based on the image data, and sequentially performing feature matching and content matching based on edge content of the missing image area to obtain a corrected image corresponding to the missing image area specifically include:

[0061] Extracting features of the image in the image data, identifying and determining text features and graphic features in the image, and respectively retrieving and obtaining feature types corresponding to the text features and graphic features, wherein the feature types are used to represent the style categories of the text and image;

[0062] If there is a cropped image object at the edge of the missing image area, the cropped image is obtained, feature matching is performed based on a feature database of feature types corresponding to text features and graphic features, and multiple matching image objects are selected in sequence based on the matching rate as the corrected image, where the image objects are used to represent content elements under the feature type, including text and basic graphics;

[0063] After the cropped image is matched, the image data is understood in terms of content elements and meaning, and the associated text content and graphic elements are matched, and a corrected image is generated based on the text content and graphic elements.

[0064] In this embodiment, the process of feature matching and content matching is further explained. The feature type here is the content format expressed in the aforementioned embodiment, which depends on many factors such as the different production era background, usage purpose and the specific type of the object. A corresponding data matching library can be established based on historical archaeological records and related historical materials. After matching appropriate content according to features, in scenarios with a lot of missing content, there may be multiple matching structures. Therefore, further screening is required through the understanding of the content elements and meaning composition of the image data, such as whether the meaning expressed by the text combination is consistent with the overall content atmosphere.

[0065] As another preferred embodiment of the present invention, the step of performing image layer judgment based on the twin digital object to identify and mark the image missing area on the object surface specifically includes:

[0066] Performing a color continuity evaluation on the surface of the twin digital object, generating an evenly spaced evaluation point grid with a preset evaluation accuracy, and calculating the color span value between the evaluation points. The color span value is used to represent the interval between two color corresponding points on the same evaluation palette;

[0067] If the color span value reaches the preset evaluation standard, a range mark is performed on the continuous area on the surface of the object based on the color value of the evaluation point to generate an abnormal area to be screened;

[0068] A structural evaluation is performed on the surface of the object in the abnormal area to be screened. If the object structure in the abnormal area to be screened is manifested as regional subsidence or abnormal surface luminosity, then the abnormal area to be screened is an image missing area. The regional subsidence is used to characterize the missing and damaged object structure, and the abnormal surface luminosity is used to characterize the weathering and corrosion of the object surface structure.

[0069] In this embodiment, the steps for determining the missing image area are described, which include two aspects: color continuity judgment and structural judgment. Multiple area ranges can be determined based on the step-by-step changes in color, but the area range may be a conventional color content expression area, so it is necessary to make a judgment based on the actual state of the object. The main causes of image loss include structural damage and loss and surface weathering and corrosion. Therefore, for the former, it is possible to determine that the area of the object has sunk compared to the surrounding area (such as grooves caused by porcelain chipping, and a gap caused by breakage). For the latter, it is common on objects that are easily corroded and damaged, such as paper, bamboo slips, etc., which will become rough and cause surface content loss due to liquid immersion.

[0070] like Figure 2 As shown, as another preferred embodiment of the present invention, it also includes an auxiliary step of restoring the fragmented object, specifically including:

[0071] S51, when the object structure is characterized as a region sinking, obtaining a gap space model of the sinking region and marking the surface side of the object;

[0072] S52, obtaining fragment space models of multiple fragment objects, performing accommodation evaluation on the fragment space models based on the gap space model, determining whether the subsidence gap can accommodate the fragment space model, establishing an accommodation list based on the determination result, and setting the fragment objects to overlap with the surface side of the object in the subsidence gap when performing the accommodation evaluation;

[0073] S53 , performing cross-matching verification on the plurality of corrected images based on the object surface image contents of the fragment objects in the accommodation list to screen out double-matched corrected images.

[0074] In this embodiment, the restoration assistance of fragmented objects is used when there are many smaller fragment objects. By analyzing the size, depth and shape of the gap in the sunken area of the more complete object, the smaller fragment object is embedded and simulated to determine whether it may be part of the missing content in the gap. If it can be accommodated in terms of structure and volume, its surface content features are further cross-matched and verified with the aforementioned matched corrected image, thereby realizing two-way verification and screening of the fragment objects and matching content.

[0075] As another preferred embodiment of the present invention, the step of performing planarization processing on the image data of the non-deleted area specifically includes:

[0076] Establish a surface mesh, divide the corresponding surface of the object into several triangular meshes, and perform surface discretization processing;

[0077] Performing a planar projection on each of the triangular meshes in the three-dimensional surface to obtain an initial unfolded image, and gradually unfolding the adjacent triangular meshes to maintain the adjacent relationship of the meshes;

[0078] An objective function is defined to generate a conversion objective function for preserving the side length and angle of the triangular mesh, and scaling optimization is performed on the initial unfolded graph based on the objective function.

[0079] In this embodiment, before performing the corrected image matching, the content image of the object surface needs to be flattened, because the text image content covered on the curved surface is different from the flat content and there will be a certain degree of stretching and alienation. If it is directly retrieved based on its graphic features, it will be inaccurate. Therefore, it is necessary to use the objective function to optimize the image structure by discretizing the surface features.

[0080] like Figure 3 As shown, the present invention also provides an image correction system based on pattern recognition, which includes:

[0081] The object restoration module 100 is used to synchronize with the data acquisition end to obtain sensor scanning data, perform object restoration modeling based on multiple sets of sensor scanning data, and obtain a twin digital object including an overlay map layer; the sensor scanning data includes radar point cloud data and surface image data;

[0082] A missing image location module 200 is configured to perform image layer determination based on the twin digital object to identify and mark image missing areas on the object surface, and perform planarization processing on image data of non-missing areas to generate corresponding image data, the image data including the image missing area markers;

[0083] a content retrieval module 300 configured to retrieve image categories and image contents based on the image data, and perform feature matching and content matching based on the edge content of the image missing region, to obtain corrected images corresponding to the image missing region, wherein each image missing region corresponds to one or more corrected images;

[0084] The correction feedback module 400 is used to sort the multiple corrected images in descending order based on the matching rates of feature matching and content matching, and generate an interactive queue of image missing areas based on the sorting results to output the image data accordingly.

[0085] As another preferred embodiment of the present invention, the content retrieval module includes:

[0086] a feature resolution unit for extracting features from the image in the image data, identifying and determining text features and graphic features in the image, and respectively retrieving and obtaining feature types corresponding to the text features and graphic features, wherein the feature types are used to characterize the style categories of the text and image;

[0087] a cropping correction unit, configured to, if a cropped image object exists at the edge of the missing image area, obtain the cropped image, perform feature matching based on a feature database of feature types corresponding to text features and graphic features, and select a plurality of matching image objects in sequence as correction images based on a matching rate, wherein the image objects are used to represent content elements of the feature type, including text and basic graphics;

[0088] The correction verification unit is used to understand the content elements and meaning of the image data after the cropped image is matched, match the associated text content and graphic elements, and generate a corrected image based on the text content and graphic elements.

[0089] As another preferred embodiment of the present invention, the missing location module includes:

[0090] Performing a color continuity evaluation on the surface of the twin digital object, generating an evenly spaced evaluation point grid with a preset evaluation accuracy, and calculating the color span value between the evaluation points. The color span value is used to represent the interval between two color corresponding points on the same evaluation palette;

[0091] If the color span value reaches the preset evaluation standard, a range mark is performed on the continuous area on the surface of the object based on the color value of the evaluation point to generate an abnormal area to be screened;

[0092] A structural evaluation is performed on the surface of the object in the abnormal area to be screened. If the object structure in the abnormal area to be screened is manifested as regional subsidence or abnormal surface luminosity, then the abnormal area to be screened is an image missing area. The regional subsidence is used to characterize the missing and damaged object structure, and the abnormal surface luminosity is used to characterize the weathering and corrosion of the object surface structure.

[0093] As another preferred embodiment of the present invention, a fragment reset module is further included, specifically comprising:

[0094] a sinking marking unit, configured to obtain a gap space model of the sinking area and mark the surface side of the object when the object structure is characterized as a sinking area;

[0095] an accommodation judgment unit, configured to obtain a fragment space model of a plurality of fragment objects, perform accommodation evaluation on the fragment space model based on the gap space model, determine whether the subsidence gap can accommodate the fragment space model, establish an accommodation list based on the judgment result, and when performing the accommodation evaluation, arrange the fragment objects to overlap with the surface side of the object of the subsidence gap;

[0096] The cross matching unit is used to perform cross matching verification on a plurality of corrected images based on the object surface image content of the fragment objects in the accommodation list to screen out double-matched corrected images.

[0097] As another preferred embodiment of the present invention, the missing location module further includes:

[0098] The meshing unit is used to establish a surface mesh, divide the corresponding surface of the object surface into several triangular meshes, and perform surface discretization processing;

[0099] A projection processing unit, configured to perform planar projection on each of the triangular meshes in the three-dimensional surface to obtain an initial unfolded image, and to gradually unfold adjacent triangular meshes to maintain a mesh adjacency relationship;

[0100] The scaling optimization unit is used to define an objective function to generate a conversion objective function for preserving the side length and angle of the triangular mesh, and perform scaling optimization on the initial expanded graphic based on the objective function.

[0101] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When executed, the program can include the processes of the above-described method embodiments. Any reference to memory, storage, database, or other media used in the various embodiments provided herein may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct RAMbus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM).

[0102] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the disclosure in the specification and examples. This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered merely as exemplary, and the true scope and spirit of the present disclosure are indicated by the claims.

[0103] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes can be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. An image correction method based on pattern recognition, characterized in that: Include: Synchronize the data acquisition end to obtain sensor scanning data, perform object restoration modeling based on multiple sets of sensor scanning data, and obtain a twin digital object including an overlay map layer; the sensor scanning data includes radar point cloud data and surface image data; Performing image layer judgment based on the twin digital object to identify and mark image missing areas on the object surface, and performing planarization processing on image data of non-missing areas to generate corresponding image data, wherein the image data includes the image missing area mark; Retrieve the image category and image content based on the image data, and perform feature matching and content matching based on the edge content of the missing image area in sequence to obtain a corrected image corresponding to the missing image area, wherein each missing image area corresponds to one or more corrected images; sorting the plurality of corrected images in descending order based on the matching rates of feature matching and content matching, and generating an interactive queue of image missing areas based on the sorting results to output the image data accordingly; The steps of retrieving the image category and image content based on the image data, and sequentially performing feature matching and content matching based on the edge content of the image missing area to obtain a corrected image corresponding to the image missing area specifically include: Extracting features of the image in the image data, identifying and determining text features and graphic features in the image, and respectively retrieving and obtaining feature types corresponding to the text features and graphic features, wherein the feature types are used to represent the style categories of the text and image; If there is a cropped image object at the edge of the missing image area, the cropped image is obtained, feature matching is performed based on a feature database of feature types corresponding to text features and graphic features, and multiple matching image objects are selected in sequence based on the matching rate as the corrected image, where the image objects are used to represent content elements under the feature type, including text and basic graphics; After the cropped image is matched, the image data is analyzed for content elements and meaning, and associated text content and graphic elements are matched, and a corrected image is generated based on the text content and graphic elements; The step of performing image layer judgment based on the twin digital object to identify and mark the image missing area on the object surface specifically includes: Performing a color continuity evaluation on the surface of the twin digital object, generating an evenly spaced evaluation point grid with a preset evaluation accuracy, and calculating the color span value between adjacent evaluation points. The color span value is used to represent the interval between two color corresponding points on the same evaluation palette; If the color span value reaches the preset evaluation standard, a range mark is performed on the continuous area on the surface of the object based on the color value of the evaluation point to generate an abnormal area to be screened; Performing a structural evaluation on the surface of the object in the abnormal area to be screened; if the object structure in the abnormal area to be screened is characterized by regional subsidence or abnormal surface luminosity, the abnormal area to be screened is an image missing area, wherein the regional subsidence is used to indicate damage to the object structure, and the abnormal surface luminosity is used to indicate weathering and corrosion of the object surface structure; It also includes auxiliary steps for restoring fragmented objects, including: When the object structure is characterized as a region sinking, a gap space model of the sinking region is obtained, and the surface side of the object is marked; Obtaining a fragment space model of a plurality of fragment objects, performing an accommodation evaluation on the fragment space model based on the gap space model, determining whether the sunken gap can accommodate the fragment object, establishing an accommodation list based on the determination result, and arranging the fragment object to overlap with the object surface side of the sunken gap when performing the accommodation evaluation; A cross-matching verification is performed on several corrected images based on the object surface image content of the fragment objects in the accommodation list to screen out double-matched corrected images.

2. The image correction method based on pattern recognition according to claim 1, characterized in that: The step of performing planarization processing on the image data of the non-missing area specifically includes: Establish a surface mesh, divide the corresponding surface of the object into several triangular meshes, and perform surface discretization processing; Performing a planar projection on each of the triangular meshes in the three-dimensional surface to obtain an initial unfolded image, and gradually unfolding the adjacent triangular meshes to maintain the adjacent relationship of the meshes; An objective function is defined to generate a conversion objective function for preserving the side length and angle of the triangular mesh, and scaling optimization is performed on the initial unfolded graph based on the objective function.

3. An image correction system based on pattern recognition, characterized in that: Include: An object restoration module is used to synchronize with the data acquisition end to obtain sensor scanning data, perform object restoration modeling based on multiple sets of sensor scanning data, and obtain a twin digital object including an overlay map layer; the sensor scanning data includes radar point cloud data and surface image data; A missing image location module is configured to perform image layer determination based on the twin digital object to identify and mark image missing areas on the object surface, and perform planarization processing on the image data of the non-missing areas to generate corresponding image data, the image data including the image missing area marks; a content retrieval module, configured to retrieve image categories and image contents based on the image data, and perform feature matching and content matching based on the edge content of the image missing region, to obtain corrected images corresponding to the image missing region, wherein each image missing region corresponds to one or more corrected images; A correction feedback module is used to sort the plurality of corrected images in descending order based on the matching rates of feature matching and content matching, and to generate an interactive queue of image missing areas based on the sorting results to output the image data accordingly; The content retrieval module includes: a feature resolution unit for extracting features from the image in the image data, identifying and determining text features and graphic features in the image, and respectively retrieving and obtaining feature types corresponding to the text features and graphic features, wherein the feature types are used to characterize the style categories of the text and image; a cropping correction unit, configured to, if a cropped image object exists at the edge of the missing image area, obtain the cropped image, perform feature matching based on a feature database of feature types corresponding to text features and graphic features, and select a plurality of matching image objects in sequence as correction images based on a matching rate, wherein the image objects are used to represent content elements of the feature type, including text and basic graphics; a correction verification unit, configured to understand the content elements and meaning of the image data after the cropped image is matched, match the associated text content and graphic elements, and generate a corrected image based on the text content and graphic elements; The missing location module includes: Performing a color continuity evaluation on the surface of the twin digital object, generating an evenly spaced evaluation point grid with a preset evaluation accuracy, and calculating the color span value between adjacent evaluation points. The color span value is used to represent the interval between two color corresponding points on the same evaluation palette; If the color span value reaches the preset evaluation standard, a range mark is performed on the continuous area on the surface of the object based on the color value of the evaluation point to generate an abnormal area to be screened; Performing a structural evaluation on the surface of the object in the abnormal area to be screened; if the object structure in the abnormal area to be screened is characterized by regional subsidence or abnormal surface luminosity, the abnormal area to be screened is an image missing area, wherein the regional subsidence is used to indicate damage to the object structure, and the abnormal surface luminosity is used to indicate weathering and corrosion of the object surface structure; It also includes a fragment reset module, specifically including: a sinking marking unit, configured to obtain a gap space model of the sinking area and mark the surface side of the object when the object structure is characterized as a sinking area; an accommodation judgment unit, configured to obtain a fragment space model of a plurality of fragment objects, perform accommodation evaluation on the fragment space model based on the gap space model, judge whether the sunken gap can accommodate the fragment objects, establish an accommodation list based on the judgment result, and when performing the accommodation evaluation, arrange the fragment objects to overlap with the object surface side of the sunken gap; The cross matching unit is used to perform cross matching verification on a plurality of corrected images based on the object surface image content of the fragment objects in the accommodation list to screen out double-matched corrected images.

4. The image correction system based on pattern recognition according to claim 3, characterized in that: The missing location module also includes: The meshing unit is used to establish a surface mesh, divide the corresponding surface of the object surface into several triangular meshes, and perform surface discretization processing; A projection processing unit, configured to perform planar projection on each of the triangular meshes in the three-dimensional surface to obtain an initial unfolded image, and to gradually unfold adjacent triangular meshes to maintain a mesh adjacency relationship; The scaling optimization unit is used to define an objective function to generate a conversion objective function for preserving the side length and angle of the triangular mesh, and perform scaling optimization on the initial expanded graphic based on the objective function.

Images