Method and device for monitoring and demonstrating weathering of immovable cultural relics and storage medium

Abstract

The application discloses a weathering monitoring and demonstration method and device for immovable cultural relics and a storage medium, and relates to the technical field of weathering detection. First, a hierarchical data structure of a reference panoramic image and a local enhanced image is constructed, which not only retains the original optical details and data integrity of the immovable cultural relics, but also reduces the high computing cost and storage management complexity of large-scale tile preprocessing. Then, by performing cross-time-point image registration and establishing an accurate coordinate mapping relationship, combined with pixel-level difference calculation, the shooting angle and position deviation can be effectively eliminated. The generated disease change heat map has higher spatial resolution and monitoring sensitivity. Finally, an index association table is used to realize the dynamic linkage of panoramic overview and local details. In this way, the weathering detection accuracy of the immovable cultural relics is improved.

Description

Technical Field

[0001] This application relates to the field of weathering detection technology, specifically to a method, device, and storage medium for weathering monitoring and demonstration of immovable cultural relics. Background Technology

[0002] A time-domain file is a sequential data file that uses time as its core dimension and associates multiple perspectives and orientations. Referring to the specific content of CN114820575A, "Image Verification Method, Apparatus, Computer Equipment and Storage Medium," for example, the three-dimensional target object mentioned in paragraphs

[0043] -

[0044] of the specification, a time-domain file can refer to a three-dimensional target object file containing time-dimensional information. A time-domain file can record the dynamic characteristics of the target object at different points in time (or time periods), and also associates information from multiple perspectives and orientations. In the time dimension, a time-domain file contains continuous or discrete time nodes, which can reflect the dynamic changes of the target object over time. In the visual dimension, time-domain data at the same point in time can correspond to different observation perspectives, recording the time-domain performance of the target object under different observation angles.

[0003] Immovable cultural relics (such as grottoes, murals, and ancient buildings) are exposed to the natural environment for extended periods, making them highly susceptible to surface weathering and peeling due to factors such as temperature and humidity changes, wind erosion, and biological attack. Traditional methods for monitoring the weathering of cultural relics mainly rely on regular manual inspections or monitoring with fixed sensors. Manual inspections are highly subjective, inefficient, and fail to detect subtle changes; while fixed sensors, although capable of collecting environmental data, cannot directly reflect the specific evolution of damage to the surface morphology of the cultural relics.

[0004] Existing digital monitoring methods often only demonstrate changes through simple before-and-after image comparisons, lacking effective organization and accurate registration of multi-temporal and multi-view image data. This results in low accuracy in identifying damage across time points and makes it difficult to intuitively locate localized damage areas from a panoramic perspective. Furthermore, the storage and retrieval of monitoring data often lack effective correlation mechanisms, making it impossible to achieve heat map overlay display and data verification of damage changes, severely impacting the scientific rigor and intuitiveness of cultural relic weathering assessments. Summary of the Invention

[0005] The purpose of this application is to provide a method, device, and storage medium for monitoring and demonstrating the weathering of immovable cultural relics, in order to solve the problem of low accuracy in traditional weathering detection of immovable cultural relics.

[0006] To achieve the above objectives, the first aspect of this application provides a method for monitoring and demonstrating the weathering of immovable cultural relics, comprising: Baseline panoramic images of immovable cultural relics at different time points were collected, as well as locally enhanced images of the areas of concern for the damage to the immovable cultural relics. Based on the baseline panoramic image and the local enhanced image, a hierarchical data structure containing multiple time points and multiple viewpoints is constructed and packaged to generate a temporal file. The hierarchical data structure establishes the spatial relationship between the local enhanced image and the baseline panoramic image through an index association table. The local enhanced image is mounted on the baseline panoramic image as an independent nested layer. Based on the time domain file, perform cross-time point image registration, establish coordinate mapping relationship between images at different time points, calculate pixel-level differences based on the registered image data, and generate a heat map of disease changes characterizing weathering changes. When the time-domain file is loaded on the terminal device, in response to the operation command for the reference panoramic image, the index association table is called to retrieve and display the corresponding local enhanced image, and the disease change heat map is overlaid and displayed.

[0007] The second aspect of this application provides a weathering monitoring and demonstration device for immovable cultural relics, comprising: The acquisition module is used to acquire baseline panoramic images of immovable cultural relics at different time points, as well as to acquire local enhanced images of the disease-prone areas of the immovable cultural relics. The construction module is used to construct a hierarchical data structure containing multi-time point layers and multi-viewpoint layers based on the reference panoramic image and the local enhanced image. The hierarchical data structure establishes the spatial relationship between the local enhanced image and the reference panoramic image through an index association table. The local enhanced image is mounted on the reference panoramic image as an independent nested layer. The registration module is used to perform cross-time point image registration based on the hierarchical data structure, establish coordinate mapping relationship between images at different time points, calculate pixel-level differences based on the registered image data, and generate a heat map of disease changes characterizing weathering changes. The loading module is used to load the hierarchical data structure on the terminal device, respond to the operation command for the reference panoramic image, call the index association table to retrieve and display the corresponding local enhanced image, and overlay the heat map of disease changes.

[0008] A third aspect of this application provides a computer-readable storage medium storing a program that can be loaded by a processor and executed as described above regarding the weathering monitoring and demonstration method for immovable cultural relics.

[0009] The beneficial effects of this application are: This application first constructs a hierarchical data structure of baseline panoramic imagery and locally enhanced imagery, abandoning the traditional multi-resolution pyramid tile slicing method. It directly utilizes the original imagery as a nested hierarchy, preserving the original optical details and data integrity of immovable cultural relics while reducing the high computational cost and storage management complexity of large-scale tile preprocessing. Then, by performing cross-time point image registration and establishing precise coordinate mapping relationships, combined with pixel-level difference calculations, it effectively eliminates shooting angle and positional deviations. The generated heatmap of damage changes can intuitively and quantitatively characterize minute changes on the surface of cultural relics, such as weathering, salting, or crack expansion, with higher spatial resolution and monitoring sensitivity. Finally, an indexed association table is used to achieve dynamic linkage between the panoramic overview and local details. When users browse the baseline panoramic imagery on their terminals, they can retrieve and overlay locally enhanced imagery and damage heatmaps as needed. The hidden entry point design ensures the smoothness of the macro-narrative while meeting the needs for immersive exploration and expert assessment of key areas of damage, achieving a multi-dimensional demonstration effect where the main storyline is controllable and details are explorable.

[0010] Other features and advantages of this application will be described in detail in the following detailed description section. Attached Figure Description

[0011] Figure 1 This is a flowchart illustrating a weathering monitoring and demonstration method for immovable cultural relics provided in the embodiments of this application; Figure 2 This is a schematic diagram illustrating a hierarchical data structure provided in an embodiment of this application; Figure 3 This is a flowchart illustrating a method for generating a heatmap of disease changes provided in an embodiment of this application. Figure 4 This is a schematic diagram of the structure of a weathering monitoring and demonstration device for immovable cultural relics provided in the embodiments of this application. Detailed Implementation

[0012] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0013] In the description of this application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "a plurality of" means two or more, unless otherwise explicitly specified. Details are set forth in the following description for illustrative purposes. It should be understood that those skilled in the art will recognize that this application can be implemented without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid unnecessarily obscuring the description of this application. Therefore, this application is not intended to be limited to the embodiments shown, but rather to be consistent with the broadest scope of the principles and features disclosed herein.

[0014] Figure 1 This is a flowchart illustrating a method for monitoring and demonstrating the weathering of immovable cultural relics provided in an embodiment of this application. Figure 1 As shown, this weathering monitoring and demonstration method may include steps 101-104, which will be described in detail below.

[0015] Step 101: Collect baseline panoramic images of immovable cultural relics at different time points, and collect local enhanced images of the areas of concern for the damage to immovable cultural relics.

[0016] A baseline panoramic image refers to an ultra-high-resolution still image of an immovable cultural relic, captured at a specific point in time, showing its overall appearance or a large area. This image serves as a base map or background for subsequent change analysis. As an example, images of the same immovable cultural relic can be collected at multiple different monitoring points according to the required monitoring cycle (e.g., daily / weekly / monthly / quarterly / yearly), ensuring the temporal continuity of the data and providing a basis for subsequent cross-time point weathering change comparisons. For instance, using a camera / drone equipped with a wide-angle lens or panoramic gimbal, the immovable cultural relic can be photographed at different points in time to collect a baseline panoramic image. This baseline panoramic image needs to cover the entire appearance of the relic for macroscopic monitoring.

[0017] The Region of Interest (ROI) refers to a localized area on an immovable cultural relic where weathering damage exists or where there is a potential risk of such damage; it is the focus of monitoring. Enhanced localized images are ultra-high-resolution static close-up images with higher pixel density taken specifically for the ROI. They enhance and precisely record the details of the ROI within the baseline panoramic image, capturing subtle weathering features. Compared to the baseline panoramic image, enhanced localized images have higher pixel resolution, a more focused shooting angle, and clearer details of the damage. For specific areas on the surface of immovable cultural relics known or suspected of having damage (such as cracks, peeling, or fading), i.e., ROIs, high-resolution macro lenses or high-magnification zoom devices can be used for close-up photography. During acquisition, the spatial location information of this area within the panoramic image (such as latitude and longitude, coordinates, or relative position) must be recorded.

[0018] Both panoramic and locally enhanced images are captured using ultra-high resolution static images (pixel counts of 80 million, 100 million, or 150 million pixels are available) to ensure detailed resolution and capture early, minute weathering changes such as micro-cracks and micro-salting. Furthermore, it can optionally acquire equipment attitude / position information (such as inertial measurement units (IMUs) and electronic compasses) and environmental monitoring data (temperature, humidity, carbon dioxide, and particulate matter) simultaneously, providing supplementary data for subsequent analysis of weathering changes and correlation assessments.

[0019] By acquiring baseline panoramic images and locally enhanced images, a complementary macroscopic panorama and microscopic details can be achieved. Ultra-high pixel density ensures image detail resolution, enabling the identification of early weathering signals that are difficult to detect with the naked eye, such as micro-cracks, micro-salting, and slight pigment fading. This ensures both the breadth of the monitoring range and the accuracy of damage identification. Time-segmented acquisition provides a temporal data foundation for subsequent dynamic change monitoring. The entire process employs a non-contact acquisition method using static image capture, without any physical contact or chemical intervention, adhering to the principle of "minimal intervention" in the protection of immovable cultural relics.

[0020] Step 102: Based on the baseline panoramic image and the local enhanced image, construct a hierarchical data structure containing multiple time points and multiple viewpoints, and package it to generate a temporal domain file.

[0021] The hierarchical data structure, centered on a "time-viewpoint layer-nested layer," organizes ultra-high-resolution images in three dimensions: temporal, spatial, and detailed. Unlike traditional tile pyramids and composite image collages, this hierarchical data structure is a non-tile, nested data organization that preserves the original pixels. It establishes spatial relationships between locally enhanced images and baseline panoramic images through an indexed association table, with the locally enhanced images serving as independent nested layers mounted on top of the baseline panoramic images. The indexed association table acts as a spatial mapping index connecting the baseline panoramic images and locally enhanced images, recording key information such as spatial coordinates, feature matching, and geometric transformations.

[0022] As an example, a hierarchical structure can be constructed first. Based on the chronological order of image acquisition, the baseline panoramic images from different time points are constructed into multi-time-point layers to ensure the temporal dimension of the data is distinct. Within the same time-point layer, based on the shooting angle, locally enhanced images from different time angles (or different magnifications) are constructed into multi-viewpoint layers to adapt to the spatial three-dimensionality of immovable cultural relics. Under the baseline panoramic image of each viewpoint layer, locally enhanced images of the corresponding areas of concern regarding disease are attached, forming independent nested layers to achieve spatial correlation between "panoramic" and "local".

[0023] A precise spatial mapping relationship between the local enhanced image and the baseline panoramic image is established through an indexed association table. This table records information such as the spatial coordinate range of the local enhanced image within the baseline panoramic image, feature points of overlapping areas, and geometric transformation parameters, achieving a one-to-one correspondence between the two. The local enhanced image is mounted as an independent nested layer on top of the baseline panoramic image. The local enhanced image retains complete original pixel data, is not stitched or merged with the panoramic image, and does not depend on the parent image, ensuring data independence and traceability.

[0024] Then, the constructed hierarchical data structure, indexed association tables, and raw image metadata (acquisition time, equipment, environmental data, etc.) are packaged together to generate a single time-domain file, realizing integrated encapsulation, storage, and transmission of monitoring data. The time-domain file is the time-domain monitoring demonstration package, which includes an integrated data file containing hierarchical data structure, indexed association tables, raw images, and metadata. Its core features are "time-domain nature, completeness, and portability," and it can be directly loaded and used on different terminal devices.

[0025] This application does not employ tile slicing technology, eliminating the need for pre-compression and pre-slicing of ultra-high-resolution images. It preserves the original pixel data intact, making the images auditable original evidence. This meets the requirements for preserving the evidence chain and tracing versions of immovable cultural relics, solving the problem of traditional tile technology where original data loss makes them unsuitable as evidence. By mounting local details as independent nested levels rather than directly stitching them into the panoramic image, data redundancy and loading burden are reduced. Through an indexed association table, the system can instantly locate and retrieve local details of specific areas. This allows for rapid association with the panoramic image and separate extraction and analysis, improving data access efficiency. Packaging all data into a single temporal file solves the transmission difficulties and terminal incompatibility issues caused by the scattered storage of ultra-high-resolution images, index data, and metadata. It can be directly loaded on different terminal devices, meeting the design requirements of multi-terminal collaboration.

[0026] Step 103: Perform cross-time point image registration based on the time domain file, establish coordinate mapping relationship between images at different time points, calculate pixel-level differences based on the registered image data, and generate a heat map of disease changes characterizing weathering changes.

[0027] Cross-temporal image registration refers to the spatial alignment of images of the same immovable cultural relic taken at different times, from the same perspective / area, to eliminate geometric deviations during the shooting process and establish accurate coordinate mapping relationships. This is a prerequisite and key to achieving cross-temporal comparison of weathering changes. The coordinate mapping relationship records the correspondence between pixel coordinates / spatial coordinates of images at different times, ensuring that the pixels of images from different times can correspond one-to-one after registration, achieving accurate difference comparison. Specifically, the temporal domain files can be decoupled, and the baseline panoramic image, locally enhanced image, and index association table at different time points can be extracted to ensure the integrity of the data source for registration analysis. Image alignment algorithms are used to perform geometric correction on images of the same area at different times, eliminating errors caused by shooting angles and equipment displacement. The transformation matrix between the two time point images is calculated to ensure a one-to-one correspondence between pixels, establishing coordinate mapping relationships between images at different times.

[0028] Pixel-level difference refers to the difference between two corresponding pixels on a tree branch, reflecting the degree of weathering change. For example, based on registered image data, pixel-level difference detection and calculation are performed on areas of concern for disease. Simultaneously, multiple indicator difference detections can be integrated, such as calculating structural similarity differences and color differences, reducing false alarms caused by lighting changes and shooting noise. The pixel-level difference calculation results are then visualized to generate a heat map of disease changes. The heat map of disease changes is a graphical monitoring result that visualizes pixel-level difference results, intuitively displaying the spatial distribution, development trend, and severity of weathering changes in immovable cultural relics in the form of thermal colors. It serves as the core visualization carrier for subsequent terminal demonstrations. For example, color depth / gamut changes can represent the location, range, and severity of weathering changes (e.g., red represents severe weathering, yellow represents moderate weathering, and green represents slight weathering / no change), achieving an intuitive and quantitative display of weathering changes.

[0029] By achieving precise cross-time point registration, image deviations caused by factors such as shooting angle, equipment posture, and displacement are eliminated, ensuring strict comparability of images from different time points. This solves the problem of difficulty in maintaining a consistent perspective across multiple time points, making it difficult to distinguish between changing signals and acquisition noise. Pixel-level difference calculations are performed based on the registered images, using quantified pixel values ​​and differences to characterize weathering changes, replacing traditional manual visual judgment and experience-based assessments, thus reducing observer bias. Pixel-level refined analysis can identify early, subtle weathering changes that are difficult to detect in baseline panoramic images, such as the propagation of micro-cracks, the advancement of micro-saltification, and slight pigment fading, providing data support for early warning and timely intervention in cultural relic protection. Abstract pixel-level difference data is transformed into intuitive heat maps of damage changes, enabling cultural relic protection personnel to quickly and clearly grasp the location, scope, and severity of weathering changes without requiring specialized image analysis knowledge, thus improving the practicality and dissemination of monitoring results.

[0030] Step 104: Load the time-domain file on the terminal device, respond to the operation command for the reference panoramic image, call the index association table to retrieve and display the corresponding local enhanced image, and overlay and display the heat map of disease changes.

[0031] Terminal devices refer to various devices capable of loading and parsing time-domain files and enabling image display and interaction. Terminal devices can include, but are not limited to, personal computers (PCs), mobile devices, and extended reality (XR) devices (Augmented Reality (AR), Mixed Reality (MR), and Virtual Reality (VR)). There are no device type restrictions, aligning with the design concept of multi-terminal collaboration. The generated time-domain files and heatmaps of disease changes can be directly loaded onto any terminal device, which can adaptively parse hierarchical data and index association tables based on its own performance. First, a baseline panoramic image is displayed on the terminal device's interface. Any monitoring point can be selected as the baseline, serving as the overall display base map for the immovable cultural relic, achieving a panoramic overview of the immovable cultural relic's appearance.

[0032] Real-time response is used for operational commands on the baseline panoramic image, accurately locating any area of ​​interest for disease within the baseline panoramic image. Operational commands are the interaction signals between the user and the terminal device, and are the core triggering condition for retrieving locally enhanced images. Figure 2 This is a schematic diagram illustrating a hierarchical data structure provided in an embodiment of this application. For example... Figure 2 As shown, based on the location of the operation command, the system automatically calls the index association table to quickly match the corresponding local enhanced image. This image is then retrieved and displayed on the terminal device's interface, enabling quick navigation from panoramic views to detailed local images. For example, trigger methods for "Easter egg" style detail entry points can be set, including clicks, touch input, staring, gestures, and voice commands. These Easter egg style detail entry points are interactive areas (areas of concern for disease) on the baseline panoramic image. Users can trigger these areas with operation commands to jump to the corresponding local enhanced image and view weathering changes in detail, achieving traceable interaction from overview to detail, and from panoramic to local.

[0033] The system precisely overlays corresponding heatmaps of weathering changes onto the retrieved and displayed localized enhanced images. The heatmaps and enhanced images achieve pixel-level spatial alignment, visually showcasing the details of weathering changes in the area of ​​concern. It also supports cross-time point comparison overlay, slider switching, and flashing comparison, enabling dynamic demonstrations of weathering changes. This replaces the traditional method of on-site inspection and comparison with paper reports, significantly improving the efficiency of on-site interpretation and expert consultation. The entire interactive process is simple and intuitive, requiring no professional knowledge of image analysis or data processing. Even grassroots cultural relic protection staff can quickly learn to use it, solving the problems of complex operation and high professional threshold of traditional cultural relic digitization technologies. All images displayed on the terminal device are original ultra-high pixel data, and the heatmaps are quantitative analysis results. They can be traced back to the original data collection through an indexed table. All displayed content can serve as verifiable monitoring evidence, supporting remote expert consultations, academic review, and project acceptance.

[0034] This application first constructs a hierarchical data structure of baseline panoramic imagery and locally enhanced imagery, abandoning the traditional multi-resolution pyramid tile slicing method. It directly utilizes the original imagery as a nested hierarchy, preserving the original optical details and data integrity of immovable cultural relics while reducing the high computational cost and storage management complexity of large-scale tile preprocessing. Then, by performing cross-time point image registration and establishing precise coordinate mapping relationships, combined with pixel-level difference calculations, it effectively eliminates shooting angle and positional deviations. The generated heatmap of damage changes can intuitively and quantitatively characterize minute changes such as weathering, salting, or crack expansion on the surface of cultural relics, possessing higher spatial resolution and monitoring sensitivity. Finally, an indexed association table is used to achieve dynamic linkage between the panoramic overview and local details. When users browse the baseline panoramic imagery on their terminals, they can retrieve and overlay locally enhanced imagery and damage heatmaps as needed. The hidden entry point design ensures the smoothness of the macro-narrative while meeting the needs for immersive exploration and expert assessment of key damage areas, achieving a multi-dimensional demonstration effect where the main storyline is controllable and details are explorable.

[0035] In step 101, at the same monitoring time point, multiple multi-view still photos with a preset overlap rate are acquired and then stitched together to form a reference panoramic image. The preset overlap rate is a pre-defined proportion of common areas between adjacent / related images to achieve panoramic stitching of multi-view photos, feature matching between locally enhanced images and the reference panoramic image. It is a key parameter ensuring image stitching accuracy and spatial correlation accuracy. Preferably, the preset overlap rate can be greater than or equal to 30% to ensure that adjacent view photos have sufficient common feature areas. Panoramic stitching is a technical process that generates a single complete ultra-high-resolution reference panoramic image by processing multiple multi-view ultra-high-resolution still photos with a preset overlap rate acquired at the same time point through feature matching, geometric transformation, and seamless fusion. For example, Scale-Invariant Feature Transform (SIFT) or Speeded-Up Robust Features (SURF) feature matching combined with global optimization can form a high-resolution and distortion-free reference panoramic image.

[0036] Areas of concern regarding disease can be determined through a semi-automatic mode with manual marking or an automatic mode with AI-assisted recommendations, in order to accurately locate areas with weathering risks or changes that require key monitoring.

[0037] In one example, in response to a labeling instruction for a region of concern regarding disease in a baseline panoramic image, the region of concern to be collected is determined. The labeling instruction is an interactive operation signal that triggers manual labeling to determine the region of concern. For example, in response to a labeling instruction (such as box selection, point selection, circle selection, etc.) issued by a staff member on a terminal device for a baseline panoramic image initially generated at the current monitoring time, the area pointed to by the labeling instruction is directly identified as the region of concern to be collected, adapting to scenarios where immovable cultural relics already have clearly defined disease areas and require targeted monitoring.

[0038] In another example, the disease-related areas of interest to be collected can be determined based on the initial disease distribution heatmap generated at the previous monitoring time point. The initial disease distribution heatmap is a visual difference map generated by cross-time registration and pixel-level difference between the baseline panoramic images of the previous and current monitoring time points, and serves as the basis for intelligently identifying disease-related areas of interest.

[0039] Specifically, the baseline panoramic image from the previous monitoring time point is registered with the baseline panoramic image from the current monitoring time point to eliminate geometric deviations caused by factors such as shooting angle and equipment posture, establishing a precise coordinate mapping relationship between the two and achieving pixel-level spatial alignment. Then, the difference data between the baseline panoramic images from the previous and current monitoring time points (such as calculating the difference in pixel grayscale values ​​or feature vectors) is calculated to generate an initial heatmap of disease distribution, quantifying the degree of difference in pixel grayscale, color, texture, and other features between the two images. Next, areas in the initial heatmap of disease distribution whose change values ​​exceed a preset threshold are identified as areas of concern for disease. The preset threshold is a quantitative judgment value for weathering changes pre-set based on the material of the immovable cultural relic (such as stone, murals, painted sculptures), weathering type (such as salt precipitation, cracks, hollowing), and historical weathering rate. It is used to screen areas of significant change in the initial heatmap of disease distribution, achieving intelligent definition of areas of concern for disease, and adapting to scenarios involving the identification of potential disease areas and the capture of early, subtle weathering changes in immovable cultural relics.

[0040] Then, for the area of ​​concern regarding the disease, a high-resolution macro lens or high-magnification zoom device is used to acquire still photographs containing local details, forming a locally enhanced image. This locally enhanced image and the baseline panoramic image retain an overlapping area for feature matching, and the extent of this overlapping area is greater than a predetermined proportion of the total area of ​​the locally enhanced image. This predetermined proportion is the minimum percentage of the total area of ​​the locally enhanced image that constitutes the overlapping area, ensuring feature matching and spatial correlation between the locally enhanced image and the baseline panoramic image; the predetermined overlap rate is greater than this predetermined proportion.

[0041] High-pixel-density localized imagery is used only in areas of concern regarding disease, reducing invalid acquisition of unaffected areas of cultural relics and minimizing the storage and processing costs of massive amounts of ultra-high-pixel data. This allows for more focused monitoring of core weathering changes. The high-pixel-density characteristics of locally enhanced images clearly record minute weathering details such as microcracks, salt crystallization, and pigment micro-scale formation in affected areas, providing a refined data source for subsequent pixel-level difference detection and enabling precise quantification of weathering changes. A larger-than-set overlap is maintained between locally enhanced images and baseline panoramic images, providing sufficient common feature areas for subsequent indexing and association table construction (feature matching to establish spatial association) and cross-time point registration (feature matching to achieve precise alignment). This reduces registration deviations and spatial association failures caused by insufficient feature points, ensuring the accuracy and reliability of subsequent data processing. Locally enhanced images from different monitoring time points adhere to the acquisition specifications of targeting affected areas and retaining a set overlap ratio, ensuring consistent acquisition standards across time points. This addresses the core pain point of difficulty in maintaining consistency across multiple time points, leading to indistinguishability between change signals and acquisition noise, and provides a matching, high-quality data source for cross-time point weathering change comparison.

[0042] In step 102, the baseline panoramic image and the locally enhanced image acquired at the same monitoring time point are first treated as independent time-point data layers. A time-point data layer is an independent and complete data unit composed of the baseline panoramic image and all corresponding locally enhanced images at the same monitoring time point; it is the basic building block of multi-time-point data. Time-point data layers from different monitoring time points are stacked sequentially to form a multi-time-point layer. The multi-time-point layer is a hierarchical structure formed by vertically stacking all time-point data layers according to their acquisition time sequence. The independence between different time-point data layers reduces data cross-contamination and preserves the original acquisition attributes of each time-point image.

[0043] Within each time-point data layer, a reference panoramic image serves as the root node. Corresponding locally enhanced images are then attached as child nodes through spatial coordinate mapping, forming a multi-view layer. The root node, within the single-time-point data layer, is the reference panoramic image set as the spatial basis and is the core node of the multi-view layer. All locally enhanced images are spatially attached to it as the reference. Child nodes are locally enhanced images attached below the root node within the single-time-point data layer, each uniquely corresponding to a disease-related area of ​​interest / shooting angle. The multi-view layer, within the single-time-point data layer, is a tree-like hierarchical structure formed by the root node and child nodes through spatial coordinate mapping, serving as the spatial dimension carrier of the hierarchical data structure.

[0044] For example, the baseline panoramic image within the data layer at that time point can be set as the root node, serving as the core foundation for the spatial dimension at that time point, carrying the mounting and spatial positioning of all local enhanced images. Local enhanced images acquired within the data layer for different disease-related areas of interest and different shooting angles are each treated as independent child nodes, with each child node uniquely corresponding to a disease-related area of ​​interest / shooting angle. Then, feature matching is performed through overlapping areas to calculate the spatial coordinate mapping relationship between the local enhanced images (child nodes) and the baseline panoramic image (root node). Based on this relationship, all child nodes are mounted one by one to the spatial position corresponding to the root node, achieving precise spatial association between the panoramic root node and local child nodes, forming a multi-view layer within a single time point.

[0045] Next, a unique identifier is assigned to each root node and child node, and an index association table is established between the root nodes and child nodes. The unique identifier is a globally unique identity assigned to the root node and child nodes, used for precise positioning of any image in the hierarchical data structure, and serves as the index basis for the index association table. The index association table stores the spatial location parameters of the child nodes relative to the root node and the original pixel storage path. The spatial location parameters are the spatial positioning information of the child node (locally enhanced image) relative to the root node (baseline panoramic image), including coordinate range, geometric transformation matrix, and coordinates of feature points in overlapping areas. This is the core data for achieving precise spatial association between the two and enabling the terminal to quickly retrieve local images. The original pixel storage path is the physical / logical storage address of the original pixel data of the child node's locally enhanced image, recorded in the index association table, ensuring that the terminal can directly access the original pixel data without compression or resolution conversion. Finally, the completed multi-time point layer, multi-view layer, unique identifier mapping relationship, index association table, and metadata of each time point image are packaged together to generate an integrated time-domain file, achieving structured encapsulation and unified management of all monitoring data.

[0046] The hierarchical structure is constructed without employing multi-resolution pyramid tile slicing technology, eliminating the need for large-scale image pre-slicing and pre-compression processing, and avoiding the need to maintain tile version consistency across time points. This significantly reduces the processing, storage, and cross-device synchronization costs of ultra-high-resolution images, making it suitable for the technical and operational realities of cultural heritage preservation units. All data is constructed into a standardized hierarchical structure and packaged into an integrated time-domain file, which can be directly parsed and loaded on different terminal devices. Terminal devices only need to read the index association table to achieve rapid image retrieval and display, improving the multi-device compatibility of monitoring data. The hierarchical data structure provides a time-ordered, spatially correlated standardized data source for cross-time point registration, and a highly efficient index foundation for rapid retrieval and accurate overlay for terminal interactive demonstrations. From a data structure perspective, this ensures the accuracy, efficiency, and reliability of all subsequent analysis and demonstration steps.

[0047] In step 103, firstly, for the baseline panoramic image in the hierarchical data structure, a pre-trained convolutional neural network is used to extract rigid and volatile feature regions of immovable cultural relics. Rigid feature regions are areas in the image where the structure of the cultural relic is stable, its geometric position is fixed, and it is unaffected by lighting / environment; these are reliable bases for registration, such as stone outlines, structural edges, and fixed carving textures. Volatile feature regions are unstable areas in the image that change with the environment, such as light and shadow, dust, occlusion, vegetation, and reflections; these are actively excluded during registration. The pre-trained convolutional neural network is a deep learning model used to automatically distinguish between rigid and volatile regions.

[0048] Then, feature descriptors are generated based on rigid feature regions, and feature matching is performed between images at different time points using a nearest neighbor search algorithm. Feature descriptors are digitized vectors representing the texture structure surrounding feature points, used for cross-time point image matching. Extracting feature points only from rigid feature regions eliminates interference from volatile regions. The nearest neighbor search algorithm quickly finds the most similar feature point pairs, achieving coarse matching between cross-time point images. Using the nearest neighbor search algorithm, feature descriptors from the current time point are matched with those from historical time points, establishing an initial correspondence between feature points.

[0049] Next, the random sample consensus algorithm is used to eliminate mismatched point pairs, and the homography transformation matrix is ​​calculated based on the remaining inliers. Geometric alignment is then performed on image data from different time points to establish a sub-pixel-level coordinate mapping relationship. The random sample consensus algorithm automatically eliminates mismatched points and retains correctly matched inliers, improving registration accuracy. The homography transformation matrix is ​​a mathematical matrix describing the geometric projection relationship between two images, used to achieve image spatial alignment. The sub-pixel-level coordinate mapping relationship achieves a spatial correspondence accuracy of less than one pixel, which is the core accuracy guarantee for ultra-high pixel monitoring. After panoramic registration, the locally enhanced images are automatically linked and aligned, ensuring that details in the diseased area are not shifted and that there are no comparison errors.

[0050] Based on coordinate mapping, pixel-level alignment can be performed on locally enhanced images at different time points. Then, the color space difference value and gradient direction histogram difference value are calculated for each pixel. The color space difference value represents the pixel difference in the color channels of images from different time points, reflecting fading, discoloration, and salt precipitation caused by weathering. The gradient direction histogram difference value represents the differences in texture structure and edge contours of images from different time points, reflecting crack propagation, surface peeling, and structural deformation.

[0051] Next, the color space difference values ​​and gradient direction histogram difference values ​​are input into a preset weathering quantization model, which outputs the damage confidence score for each pixel. The weathering quantization model is a quantization model that integrates color difference and gradient difference to output pixel-level damage confidence scores. The damage confidence score can be a quantization value between 0 and 1, representing the degree of confidence / intensity of change in weathering damage occurring at that pixel location.

[0052] Finally, based on the disease confidence level, a heatmap of disease changes is generated using bilinear interpolation and pseudo-color mapping. Here, the hue of the color represents the disease type, and the saturation represents the severity of the disease. Bilinear interpolation ensures a smooth, jagged transition in the heatmap, enhancing visualization. Pseudo-color mapping maps the disease confidence level to a continuous color image, visually displaying the location and extent of weathering. This step is based entirely on calculations using original pixels, without compression or downsampling, ensuring the data can serve as legal evidence and long-term archives. Furthermore, the algorithm is robust and suitable for complex on-site environments. Its resistance to changes in lighting, orientation shifts, and dust interference makes it suitable for use in real-world applications at grottoes, ancient buildings, stone carvings, and other immovable cultural relics.

[0053] Figure 3 This is a flowchart illustrating a method for generating a heatmap of disease changes provided in an embodiment of this application. Figure 3 As shown, the input consists of temporal file image data from two different monitoring time points (e.g., time T1 and time T2). Time T1 serves as the historical reference time point, and time T2 serves as the current monitoring time point. The hierarchical data structure image from the two different monitoring time points includes complete information such as the reference panoramic image and local enhanced image for each time point, index association table, and raw data.

[0054] The algorithm sequentially executes two core steps: cross-time point image registration and coordinate mapping, and pixel-level difference calculation. For example, for the baseline panoramic images at times T1 and T2, rigid feature regions are extracted using a pre-trained convolutional neural network (to exclude interference from volatile environments). Feature descriptors are generated based on these rigid regions, and feature matching is performed using nearest neighbor search. Then, a random sampling consensus algorithm is used to eliminate mismatched point pairs. The homography transformation matrix is ​​calculated based on the remaining interior points, and the images at the two times are geometrically aligned to establish a sub-pixel-level coordinate mapping relationship. This coordinate mapping relationship is synchronously transferred to the corresponding locally enhanced images, achieving pixel-level linkage alignment of the locally enhanced images. This eliminates image deviations at times T1 and T2 caused by factors such as shooting angle, equipment posture, and ambient lighting, ensuring strict comparability between the images at the two times.

[0055] Based on the registered coordinate mapping, pixel-level alignment is performed on the locally enhanced images at times T1 and T2. Two core differences are calculated pixel-by-pixel: color space difference values ​​(representing surface weathering such as fading, salting out, and discoloration) and gradient direction histogram difference values ​​(representing structural weathering such as crack propagation, structural deformation, and flaking). These two types of difference values ​​are input into a pre-defined weathering quantification model, outputting the damage confidence score for each pixel (a quantization value of 0-1, with higher values ​​indicating more significant weathering changes). Through dual-dimensional difference fusion, a comprehensive, objective, and quantitative detection of weathering changes in immovable cultural relics is achieved, covering both surface color changes and structural morphological changes. This significantly reduces the false alarm rate of single-difference algorithms and replaces subjective human judgment with damage confidence scores, achieving standardization and auditability of monitoring results. Finally, based on pixel-level damage confidence scores, a smooth transition of the heatmap is achieved through bilinear interpolation, followed by pseudo-color mapping to generate a semantic heatmap, outputting a heatmap of damage changes. In this way, the precise alignment, quantitative detection, and visual presentation of weathering changes of immovable cultural relics are fully realized, providing core data support for subsequent terminal interactive demonstrations and cultural relic protection decisions.

[0056] In step 104, a three-dimensional spherical coordinate system is first constructed in the rendering engine of the terminal device. The reference panoramic image is then mapped onto the inner surface of the three-dimensional spherical coordinate system to generate a virtual display space that can be roamed omnidirectionally. The three-dimensional spherical coordinate system is a three-dimensional spherical space centered on the observer, used to carry the reference panoramic image and realize a virtual display space that can be roamed 360° omnidirectionally. It is the core carrier of immersive panoramic display. The rendering engine is a software module in the terminal device used for graphics rendering, spatial mapping, and interactive response. It supports operations such as three-dimensional scene construction, image projection, and layer overlay, and is compatible with multiple terminals such as PC, mobile, and XR. The center of the three-dimensional spherical coordinate system is the observation origin, and the radius of the sphere is adapted to the display resolution of the terminal. The reference panoramic image in the hierarchical data structure is accurately mapped onto the inner surface of the three-dimensional spherical coordinate system through an equidistant cylindrical projection / equiangular projection algorithm to generate a virtual display space that can be roamed 360° omnidirectionally. This allows users to browse the overall appearance of immovable cultural relics from any angle and field of view. The omnidirectional virtual exhibition space is an interactive space formed by mapping a baseline panoramic image onto a three-dimensional sphere. It allows users to browse the entire cultural relic from any angle and at any zoom level, breaking through the perspective limitations of traditional planar panoramas.

[0057] The system acquires user-inputted viewing direction and angle parameters in real time. Based on these parameters, it determines the target region in a 3D spherical coordinate system and retrieves the corresponding enhanced image and associated disease change heatmap of the target region from a hierarchical data structure using an indexed association table. The viewing direction parameter represents the user's current viewing angle, including azimuth (horizontal rotation angle) and pitch (vertical rotation angle), used to locate the target region in the 3D spherical coordinate system. The field of view parameter represents the zoom level of the current image, determining the display range of the terminal viewport; a smaller field of view results in a higher magnification and clearer details. The target region is the spatial coordinate range of the target region corresponding to the current viewport, calculated in the 3D spherical coordinate system based on the viewing direction and angle parameters.

[0058] Next, perspective correction transformation is performed on the enhanced local image to make it conform to the display boundary of the field of view. Then, based on the Alpha blending algorithm, a heatmap of artifact changes is overlaid as a semi-transparent layer on the perspective-corrected enhanced local image. Perspective correction transformation is an algorithm that converts a spherical projection of a local image into a planar perspective image that conforms to the current field of view. This eliminates visual distortion caused by spherical mapping and ensures the realism of the enhanced local image. The Alpha blending algorithm is used to achieve the semi-transparent overlay of the two layers. By adjusting the Alpha channel value (transparency), the heatmap of artifact changes is overlaid semi-transparently on the enhanced local image while preserving the visibility of the underlying artifact details.

[0059] Finally, interactive controls are rendered synchronously in the overlay display area. These interactive controls are operable user interface (UI) components rendered in the overlay display area, used to trigger queries for disease parameters, comparisons of time-series changes using sliders, or the addition of augmented reality annotations. They are the core entry point for interactive analysis of monitoring data. For example, clicking a control allows viewing quantitative parameters such as disease confidence level, rate of change, and historical monitoring data for the target area, triggering disease parameter queries. Dragging the slider switches between images and heatmaps from different monitoring points, visually displaying the time-series development of the disease and enabling comparisons of time-series changes using sliders. Users can also add AR annotations, notes, and repair suggestions to the overlay layer for expert consultations and protection plan development.

[0060] The construction of a three-dimensional spherical coordinate system allows users to enjoy omnidirectional navigation from a first-person perspective, breaking through the perspective limitations of traditional planar images and enhancing the realism and immersive experience of cultural relic displays. Through perspective correction and alpha blending algorithms, the geometric consistency between locally enhanced images and baseline panoramic images is ensured, while simultaneously allowing the heatmap of defects to be intuitively overlaid on the real scene, avoiding information occlusion and improving the accuracy of defect identification. A real-time data retrieval mechanism based on an indexed association table enables seamless switching between panoramic browsing, local focus, and defect viewing. Combined with slider comparison and time-series playback functions, it supports dynamic analysis of the evolution of cultural relic conditions. The spatial binding design of interactive controls and rendered content allows users to complete defect queries and annotations without leaving their current perspective, conforming to the operating habits of mobile devices and AR devices, and lowering the barrier to entry. Integrating geometrically aligned image data, quantitative defect models, and an interactive annotation system into a unified platform provides an integrated solution for cultural relic protection, combining "visual display + quantitative analysis + decision support."

[0061] In this embodiment of the application, before the steps of calling the index association table to retrieve and display the corresponding local enhanced image, and overlaying the heat map of disease changes, a rollback strategy can also be executed. The rollback strategy refers to a fault-tolerant optimization operation performed to avoid display deviations and misjudgments of information when the registration accuracy between the local enhanced image and the reference panoramic image is not up to standard.

[0062] Specifically, the registration mapping parameters between the local enhanced image and the reference panoramic image are first obtained. These parameters are the output of the registration process and represent the core parameters characterizing the spatial registration relationship between the local enhanced image and the reference panoramic image. They may include the homography transformation matrix, feature point matching pair coordinates, spatial position mapping relationships, etc., and are pre-stored in an indexed association table. Then, based on the registration mapping parameters, the reprojection error and inlier rate are calculated as accuracy metrics, serving as the quantitative basis for determining whether the registration accuracy meets the standards. The reprojection error is one of the core quantitative indicators of registration accuracy. It refers to the pixel distance deviation between the projected points and the actual matching points after projecting the feature points of the local enhanced image onto the reference panoramic image using the registration mapping parameters. A larger value indicates a larger registration deviation and lower accuracy. The inlier rate is another core quantitative indicator of registration accuracy. It refers to the proportion of correctly matched point pairs (inliers) selected during the registration process to the total number of initial feature matching point pairs, expressed as a percentage / decimal. A smaller value indicates more mismatched points and lower registration accuracy.

[0063] Set preset error thresholds and preset inlier rate thresholds, and compare the reprojection error and inlier rate with the preset error thresholds and inlier rate thresholds, respectively. The preset error threshold is a critical value for reprojection error pre-set based on the accuracy requirements of immovable cultural relic monitoring and the needs of terminal display. If the actual reprojection error exceeds this value, the registration accuracy is determined to be substandard. The preset inlier rate threshold is a critical value for inlier rate pre-set based on the reliability requirements of feature matching. If the actual inlier rate is lower than this value, the registration accuracy is determined to be substandard.

[0064] If the reprojection error is less than or equal to a preset error threshold and the inlier rate is greater than or equal to a preset inlier rate threshold, the current registration accuracy is deemed satisfactory. Conversely, if the reprojection error is greater than a preset error threshold or the inlier rate is less than a preset inlier rate threshold, the current registration accuracy is deemed unsatisfactory, and a rollback strategy is triggered and executed. The rollback strategy may include at least one of the following: stopping the overlay display and only displaying the baseline panoramic image; switching to full-screen independent display of the local enhanced image; reducing the transparency of the overlay and fusion; and outputting accuracy prompt information. After triggering rollback, at least one rollback operation is executed based on the terminal device performance and usage scenario requirements, specifically including the following operations: Stopping the overlay display of the disease change heatmap and only displaying the baseline panoramic image on the terminal interface. Switching to full-screen independent display mode of the current local enhanced image, temporarily not overlaying the heatmap. Reducing the overlay and fusion transparency of the disease change heatmap to reduce visual interference from the deviated image. Outputting accuracy prompt information in a prominent position on the terminal interface. Accuracy alerts are text / icon prompts displayed on the terminal interface when registration accuracy is substandard. They are used to remind users that there is a registration deviation in the current disease change heatmap, to prevent users from misinterpreting monitoring data, and to ensure the rigor of data use.

[0065] The registration mapping parameters are directly retrieved from the indexed association table, enabling rapid execution of accuracy verification without repetitive calculations, thus ensuring smooth terminal interaction. The rollback strategy does not modify the hierarchical data structure or the original monitoring data; it only optimizes the terminal display for fault tolerance, preserving the integrity of the original data. As a prerequisite for step 104, this completes the terminal display process into a coherent workflow of "accuracy verification - compliant display / fault-tolerant rollback - interactive analysis," enhancing the engineering sophistication and practicality of the entire monitoring method.

[0066] In this embodiment, the weathering monitoring and demonstration method for immovable cultural relics may further include calculating feature summaries of the local enhanced image and the heat map of disease changes. Taking hash fingerprints as an example, the hash fingerprints are bound and stored with the acquisition timestamp to construct a verification evidence chain for the monitoring data. Specifically, based on a preset perceptual hash algorithm, the low-frequency domain feature vector of the local enhanced image and the color moment feature vector of the heat map of disease changes can be extracted respectively. A cryptographic hash function is used to perform a nonlinear mapping on the feature vectors to generate a fixed-length digital summary as the hash fingerprint. The hash fingerprint is hierarchically combined with the corresponding acquisition timestamp and monitoring device identifier through a Merkle tree structure to generate a data verification root hash. The data verification root hash is uploaded to a distributed ledger or read-only database for solidified storage, forming an irreversible and traceable monitoring data verification evidence chain.

[0067] Because immovable cultural relics (such as grottoes and cliff carvings) are often located in remote areas with poor network signals (restricted sites), we don't need to upload massive amounts of image data in real time. Instead, we only need to upload tiny hash fingerprints to a distributed ledger (blockchain) in the cloud. For example, at the data collection site (such as inside a cave), edge computing devices (such as portable workstations) immediately use hash algorithms like SHA-256 to calculate a file digest (i.e., a hash fingerprint) after generating a time-domain file. This fingerprint is extremely short (usually 64 characters), but it uniquely represents all the image data at that moment. Even if the on-site data is later accidentally modified or damaged, simply recalculating the hash value and comparing it with the original hash value recorded in the ledger can immediately detect whether the data has been tampered with, thus achieving objective solidification of early data.

[0068] The hash value of the time-domain file is encrypted using a private key to generate a digital signature, which is then appended to the file header. The public key is made public to any terminal that needs to view the data (such as research institutes and museums). In industrial deployments, data may pass through multiple hands (collectors, data centers, researchers, and restorers). The digital signature ensures that the data indeed comes from authorized collection equipment and is not a malicious file forged by hackers. Before loading the time-domain file, the terminal device decrypts the signature using the public key and compares it with the hash value calculated for the current file. If they do not match, it indicates that the file has been corrupted or tampered with during transmission, and the system will automatically issue an alert or refuse to load the file.

[0069] The system binds user identities (such as tourists, junior researchers, and chief experts) to their digital certificates (public keys). An access control list is maintained, specifying the data levels accessible to different identities. This balances long-term data storage with multi-role collaboration. In this way, tourists / the public may only see the baseline panoramic image, not the high-precision, locally enhanced image (protecting the details of cultural relics from misuse). Restorers can view heatmaps of damage changes but cannot modify the original images. Administrators have the highest privileges. This management system is implemented through code, preventing unauthorized operations and adhering to the seriousness of cultural relic management.

[0070] Each time-domain file generated during a monitoring period (e.g., annually) has a unique hash fingerprint and records a timestamp. The system maintains a version chain. If anomalies are found in recent data during an analysis (e.g., misregistration due to equipment failure), or if it's necessary to compare with data from 10 years ago, users can select a historical time point via a slider or command. By comparing the hash fingerprint, the system quickly locates and restores to that specific historical version, ensuring data traceability and recoverability.

[0071] Figure 4 This is a schematic diagram of the structure of a weathering monitoring and demonstration device for immovable cultural relics provided in an embodiment of this application. Figure 4 As shown, the weathering monitoring and demonstration device 400 for immovable cultural relics may include a data acquisition module 401, a construction module 402, a registration module 403, and a loading module 404.

[0072] The acquisition module 401 is used to acquire baseline panoramic images of immovable cultural relics at different time points, as well as to acquire local enhanced images of the disease-affected areas of the immovable cultural relics.

[0073] The construction module 402 is used to construct a hierarchical data structure containing multiple time points and multiple viewpoints based on the reference panoramic image and the local enhanced image. The hierarchical data structure establishes the spatial relationship between the local enhanced image and the reference panoramic image through an index association table. The local enhanced image is mounted on the reference panoramic image as an independent nested layer.

[0074] The registration module 403 is used to perform cross-time point image registration based on the hierarchical data structure, establish coordinate mapping relationship between images at different time points, calculate pixel-level differences based on the registered image data, and generate a heat map of disease changes characterizing weathering changes.

[0075] The loading module 404 is used to load the hierarchical data structure into the terminal device, and in response to the operation command for the reference panoramic image, call the index association table to retrieve and display the corresponding local enhanced image, and overlay the heat map of disease changes. The acquisition module 401, construction module 402, registration module 403 and loading module 404 can be used to execute steps 101-104 in the embodiments of the above-mentioned weathering monitoring and demonstration method for immovable cultural relics. For the specific implementation of these modules and more details, please refer to the corresponding method section, which will not be elaborated here.

[0076] This application also provides a computer-readable storage medium storing a program that can be loaded by a processor and executed as any of the weathering monitoring and demonstration methods for immovable cultural relics in this application.

[0077] Those skilled in the art will understand that all or part of the functions of the various methods in the above embodiments can be implemented by hardware or by computer programs. When all or part of the functions in the above embodiments are implemented by computer programs, the program can be stored in a computer-readable storage medium, which may include: read-only memory, random access memory, disk, optical disk, hard disk, etc., and the program is executed by a computer to achieve the above functions. For example, the program can be stored in the memory of a device, and when the program in the memory is executed by the processor, all or part of the above functions can be achieved. In addition, when all or part of the functions in the above embodiments are implemented by computer programs, the program can also be stored in a server, another computer, disk, optical disk, flash drive, or external hard drive, etc., and can be downloaded or copied to the memory of a local device, or the system of the local device can be updated. When the program in the memory is executed by the processor, all or part of the functions in the above embodiments can be achieved.

[0078] The above examples illustrate this application only to aid understanding and are not intended to limit its scope. Those skilled in the art to which this application pertains can make various simple deductions, modifications, or substitutions based on the ideas presented.

Claims

1. A method for monitoring and demonstrating the weathering of immovable cultural relics, characterized in that, include: Baseline panoramic images of immovable cultural relics at different time points were collected, as well as locally enhanced images of the areas of concern for the damage to the immovable cultural relics. Based on the baseline panoramic image and the local enhanced image, a hierarchical data structure containing multiple time points and multiple viewpoints is constructed and packaged to generate a temporal file. The hierarchical data structure establishes the spatial relationship between the local enhanced image and the baseline panoramic image through an index association table. The local enhanced image is mounted on the baseline panoramic image as an independent nested layer. Based on the time domain file, perform cross-time point image registration, establish coordinate mapping relationship between images at different time points, calculate pixel-level differences based on the registered image data, and generate a heat map of disease changes characterizing weathering changes. When the time-domain file is loaded on the terminal device, in response to the operation command for the reference panoramic image, the index association table is called to retrieve and display the corresponding local enhanced image, and the disease change heat map is overlaid and displayed.

2. The method for monitoring and demonstrating the weathering of immovable cultural relics according to claim 1, characterized in that, The acquisition of baseline panoramic images of immovable cultural relics at different time points, and the acquisition of locally enhanced images of areas of concern related to the deterioration of the immovable cultural relics, includes: At the same monitoring time point, multiple multi-view static photos with a preset overlap rate are collected and then stitched together to form the reference panoramic image. In response to the annotation instruction for the disease-related area of ​​interest in the baseline panoramic image, or based on the initial disease distribution heat map generated at the previous monitoring time point, the disease-related area of ​​interest to be collected is determined; For the area of ​​concern related to the disease, static photographs containing local details are collected to form the local enhanced image; The local enhanced image and the reference panoramic image retain an overlapping region for feature matching, and the range of the overlapping region is greater than a set proportion of the total area of ​​the local enhanced image.

3. The method for monitoring and demonstrating the weathering of immovable cultural relics according to claim 2, characterized in that, The initial disease distribution heatmap generated based on the previous monitoring time point determines the disease interest area to be collected, including: Register the reference panoramic image from the previous monitoring time point with the reference panoramic image from the current monitoring time point; Calculate the difference data between the reference panoramic image at the previous monitoring time point and the current monitoring time point to generate the initial disease distribution heat map; The regions in the initial disease distribution heatmap where the change value exceeds a preset threshold are identified as the disease concern areas.

4. The method for monitoring and demonstrating the weathering of immovable cultural relics according to claim 1, characterized in that, The data structure constructed based on the baseline panoramic image and the local enhanced image is a hierarchical data structure containing multiple time-point layers and multiple viewpoint layers, including: The baseline panoramic image and the local enhanced image collected at the same monitoring time point are used as independent time point data layers, and the time point data layers of different monitoring time points are stacked in time sequence to form the multi-time point layer; Within each time point data layer, the reference panoramic image is used as the root node, and the corresponding local enhanced image is mounted as a child node through spatial coordinate mapping to form a multi-view layer. A unique identifier is assigned to each root node and child node, and an index association table is established between the root node and child nodes. The index association table stores the spatial position parameters of the child node relative to the root node and the original pixel storage path.

5. The method for monitoring and demonstrating the weathering of immovable cultural relics according to claim 1, characterized in that, The step of performing cross-time point image registration based on the time domain file to establish coordinate mapping relationships between images at different time points includes: For the baseline panoramic image in the hierarchical data structure, a pre-trained convolutional neural network is used to extract the rigid feature regions and volatile feature regions of the immovable cultural relic. Based on the rigid feature region, feature descriptors are generated, and feature matching is performed between images at different time points using the nearest neighbor search algorithm. The random sampling consensus algorithm is used to remove mismatched point pairs, and the homography transformation matrix is ​​calculated based on the remaining interior points. Geometric alignment is then performed on image data at different time points to establish sub-pixel level coordinate mapping relationships.

6. The method for monitoring and demonstrating the weathering of immovable cultural relics according to claim 1, characterized in that, Based on the registered image data, pixel-level differences are calculated to generate a heat map representing weathering changes, including: Based on the coordinate mapping relationship, the local enhanced images at different time points are aligned at the pixel level; Calculate the color space difference value and gradient direction histogram difference value for each pixel; The color space difference value and the gradient direction histogram difference value are input into the preset weathering quantization model, and the disease confidence of each pixel is output. Based on the disease confidence level, a heat map of disease changes is generated using bilinear interpolation and pseudo-color mapping, where the hue of the color represents the disease type and the saturation of the color represents the severity of the disease.

7. The method for monitoring and demonstrating the weathering of immovable cultural relics according to claim 1, characterized in that, The step of loading the time-domain file on the terminal device, responding to the operation command for the reference panoramic image, calling the index association table to retrieve and display the corresponding local enhanced image, and overlaying the disease change heatmap, includes: A three-dimensional spherical coordinate system is constructed in the rendering engine of the terminal device, and the reference panoramic image is mapped to the inner surface of the three-dimensional spherical coordinate system to generate a virtual display space that can be roamed omnidirectionally. Based on the user-input line-of-sight and field-of-view parameters, the target region is determined in the three-dimensional spherical coordinate system, and the corresponding local enhanced image and associated disease change heat map of the target region are retrieved from the hierarchical data structure according to the index association table. A perspective correction transformation is performed on the local enhanced image to make it fit the display boundary of the field of view, and the lesion change heat map is superimposed on the perspective-corrected local enhanced image as a semi-transparent layer based on the Alpha blending algorithm. Interactive controls are rendered synchronously in the overlay display area. These interactive controls are used to trigger queries of disease parameters, comparisons of time-series changes in sliders, or the addition of augmented reality annotations.

8. The method for monitoring and demonstrating the weathering of immovable cultural relics according to claim 1, characterized in that, Before the steps of calling the index association table to retrieve and display the corresponding local enhanced image, and overlaying the heat map of disease changes, the method further includes: Obtain the registration mapping parameters between the local enhanced image and the reference panoramic image; Based on the registration mapping parameters, the reprojection error and inlier rate are calculated as accuracy measurement parameters. If the reprojection error is greater than a preset error threshold, or the inlier rate is less than a preset inlier rate threshold, then a rollback strategy is executed. The fallback strategy includes at least one of the following: stopping the overlay display and displaying only the baseline panoramic image, switching to full-screen independent display of the local enhanced image, reducing the transparency of the overlay and fusion, and outputting accuracy prompt information.

9. A weathering monitoring and demonstration device for immovable cultural relics, characterized in that, include: The acquisition module is used to acquire baseline panoramic images of immovable cultural relics at different time points, as well as to acquire local enhanced images of the disease-prone areas of the immovable cultural relics. The construction module is used to construct a hierarchical data structure containing multi-time point layers and multi-viewpoint layers based on the reference panoramic image and the local enhanced image. The hierarchical data structure establishes the spatial relationship between the local enhanced image and the reference panoramic image through an index association table. The local enhanced image is mounted on the reference panoramic image as an independent nested layer. The registration module is used to perform cross-time point image registration based on the hierarchical data structure, establish coordinate mapping relationship between images at different time points, calculate pixel-level differences based on the registered image data, and generate a heat map of disease changes characterizing weathering changes. The loading module is used to load the hierarchical data structure on the terminal device, respond to the operation command for the reference panoramic image, call the index association table to retrieve and display the corresponding local enhanced image, and overlay the heat map of disease changes.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that can be loaded by a processor and executed as described in any one of claims 1 to 8, a method for monitoring and demonstrating the weathering of immovable cultural relics.

Images