Method and system for collecting data of historic building structure through laser scanning

Abstract

The invention provides a laser scanning historic building structure data acquisition method and system, relates to the technical field of laser scanning, and can calculate the scanning range of a scanner in real time by obtaining a preset key monitoring area in a historic building and the real-time spatial position and attitude information of the laser scanner. On the basis, a three-dimensional bounding box intersection algorithm is adopted to carry out real-time space geometric matching calculation on the scanning range and the key monitoring area, so that whether an overlapping relation exists or not is judged. And intelligently marking the collected point cloud data as high-value data or common-value data according to the overlapping relationship. According to the method, the high-value data is written into the high-priority cache queue, the common-value data is written into the common-priority cache queue, and differentiated cache management and unloading strategies are executed on the data in the two queues, so that the dynamic monitoring task of the historic building can be effectively carried out; and the problem that the model loses the analysis value due to data missing or timestamp disorder is avoided.

Description

Technical Field

[0001] This application relates to the field of laser scanning technology, and more specifically, to a method and system for acquiring data on the structure of ancient buildings using laser scanning. Background Technology

[0002] In the digital preservation and structural analysis of ancient buildings, 3D laser scanning technology is widely used to acquire high-precision data of the building's surface. This technology generates a large amount of point cloud data by having a laser scanner emit a laser beam and receive reflected signals. This point cloud data is typically transmitted in real-time via data cable to a portable workstation in the field for processing and storage. To handle the large amount of data generated by the scanner, the workstation is usually equipped with a high-speed cache, which quickly writes the raw point cloud data stream to the cache, and then the background program gradually transfers it to the main storage device. This process performs stably in routine scanning tasks.

[0003] However, when scanning ancient buildings with extremely complex structures, especially wooden structures such as the carvings on brackets, beams, and corbels, the scanning precision needs to be adjusted to an extremely high level to accurately create a model reflecting the fine structure. This leads to a sharp increase in the point cloud density collected per unit area, and an explosive increase in the data flow rate output from the scanner. Existing data transmission and caching systems face enormous pressure; the cache fills up much faster than the data is transferred to the main storage device, and the system quickly becomes congested. To avoid data overflow and loss, the acquisition software can only frequently pause the scanning operation, waiting for the cache to be cleared before continuing. This stop-and-go acquisition method not only greatly prolongs the on-site operation time, but more seriously, for tasks that require creating models that reflect structural changes at different points in time, this interruption disrupts the temporal continuity of the data, rendering the final model worthless for analysis.

[0004] Further analysis revealed that existing caching methods treat all collected point cloud data as equally important, employing a simple "first-in, first-out" (FIFO) sequential writing and transfer strategy. This strategy results in a large amount of low-value data, such as simple, repetitive descriptions of flat wall surfaces, occupying valuable cache space and write bandwidth. Meanwhile, high-value data describing crucial structural details, such as intricate wood carvings, is forced to wait due to a full cache, even increasing the risk of being discarded. This problem is particularly pronounced when the task objective is to build a model reflecting structural changes. For example, to monitor minute deformations of key beams, continuous scanning at different time points is required. In this case, point cloud data reflecting the deformation area is the most valuable. However, existing caching mechanisms cannot identify this crucial data and may delay processing high-value data due to being busy transferring low-value data. This leads to an inability to capture the complete process of structural changes in a timely manner, ultimately resulting in data gaps or incorrect timestamps when comparing data before and after model building. Consequently, the health status of ancient buildings cannot be accurately assessed, rendering the dynamic monitoring task meaningless. Summary of the Invention

[0005] This application provides a laser scanning method and system for acquiring structural data of ancient buildings, aiming to solve the problems in the prior art where, when acquiring detailed structural data of ancient buildings, the surge in data volume leads to caching system blockage, data transmission interruption, and the inability to effectively distinguish between high-value and low-value data, thereby affecting data acquisition efficiency and data continuity, and ultimately rendering dynamic monitoring tasks meaningless.

[0006] On the one hand, this application provides a method for acquiring structural data of ancient buildings using laser scanning, including: The key monitoring areas in the ancient building were identified, and the real-time spatial position and attitude information of the laser scanner were obtained. The scanning range of the laser scanner is calculated in real time based on real-time spatial position and attitude information; The scanning range of the laser scanner and the key monitoring area are matched in real time to obtain the overlap relationship. The real-time spatial geometric matching calculation adopts the three-dimensional bounding box intersection algorithm. If there is an overlap, the point cloud data collected by the laser scanner will be marked as high-value data; if there is no overlap, the point cloud data collected by the laser scanner will be marked as ordinary-value data. Point cloud data marked as high-value data is written to a high-priority cache queue, and point cloud data marked as ordinary-value data is written to an ordinary-priority cache queue. Differentiated cache management and transfer strategies are applied to the data in the high-priority cache queue and the ordinary-priority cache queue.

[0007] Through this technical solution, this application can differentiate point cloud data according to its value, prioritize the storage and transmission of high-value data, effectively solve the problems of caching system blockage and data transmission interruption in the prior art, improve data acquisition efficiency and data continuity, and ensure the effectiveness of dynamic monitoring tasks.

[0008] Optionally, the scanning range of the laser scanner and the key monitoring area are subjected to real-time spatial geometric matching calculation to obtain the overlap relationship. The steps of this real-time spatial geometric matching calculation using a three-dimensional bounding box intersection algorithm include: Calculate the three-dimensional bounding box of the scanning range of the laser scanner; The three-dimensional bounding box of the laser scanner's scanning range is used to determine spatial overlap with the preset three-dimensional bounding box of each key monitoring area. If the projections of the two overlap on the X, Y, and Z coordinate axes, then there is an overlap relationship.

[0009] This application employs a three-dimensional bounding box intersection algorithm, which can efficiently and accurately determine the overlap between the scanning range and the key monitoring area, providing a reliable basis for subsequent data value marking and improving the efficiency and accuracy of matching calculations.

[0010] Optionally, the steps of implementing differentiated cache management and rollover strategies for data in high-priority cache queues and normal-priority cache queues include: Grant the highest write and read permissions to the high-priority cache queue, and transfer the data in the priority queue to the main storage device; When cache space is scarce, storage space is released from the ordinary priority cache queue first. The operation of releasing space includes pausing the writing of new data to the queue, overwriting the historical data in the queue, and downgrading the data in the queue to any of the secondary storage devices. The system dynamically monitors the occupancy status of high-priority cache queues and ordinary-priority cache queues, as well as the load on the main storage device, and adjusts the write bandwidth allocation strategy of the two queues in real time. When there is data waiting to be written in the high-priority cache queue, its write bandwidth is guaranteed first, and the write operation of the ordinary-priority cache queue is interrupted if necessary.

[0011] Through this technical solution, this application ensures that high-value data can be written and transferred first by means of refined cache management and transfer strategies, and prioritizes the processing of high-value data when cache resources are scarce. This effectively avoids the risk of high-value data being lost due to cache overflow, and further improves the reliability and efficiency of data collection.

[0012] Optionally, the step of writing point cloud data marked as high-value data to a high-priority cache queue includes: Real-time analysis and processing of point cloud data marked as high-value data to identify the existence of redundant point cloud clusters; If there are no redundant point cloud clusters, the point cloud data marked as high-value data will be written to a high-priority cache queue; If redundant point cloud clusters exist, compress the information and generate feature summaries for the redundant point cloud clusters to obtain refined data packets, and write the refined data packets into a high-priority cache queue.

[0013] This technical solution enables the identification and processing of redundant point cloud clusters before high-value data is written to the cache, which can further optimize the storage efficiency of high-value data, reduce unnecessary storage overhead, and ensure that the cache space is used more effectively.

[0014] Optionally, the step of performing real-time analysis and processing on point cloud data marked as high-value data to identify whether redundant point cloud clusters exist includes: Calculate the set of points within its preset radius neighborhood based on the point cloud data; Calculate the local 3D variance and mean normal vector of the point set; When the number of points exceeds a preset threshold, the 3D variance is lower than a preset variance threshold, and the angle between the normal vectors is less than a preset angle threshold, it is determined to be a redundant point cloud cluster.

[0015] This technical solution combines multiple geometric features such as the number of point sets, three-dimensional variance, and the angle between normal vectors to accurately and effectively identify redundant point cloud clusters in high-value data, providing a reliable basis for subsequent data compression and refinement. Optionally, if redundant point cloud clusters exist, the steps of compressing information and generating feature summaries for the redundant point cloud clusters to obtain refined data packets include: Extract representative points from redundant point cloud clusters. These representative points are either the average coordinates of all points or the original point closest to the center. Calculate and record the local geometric feature parameters of the redundant point cloud cluster to obtain an information summary. The local geometric feature parameters include the average normal vector, local curvature, bounding box size, and original number of points. A refined data package is obtained based on representative points, information summaries, and a small amount of strategically sampled point cloud data.

[0016] Through this technical solution, by extracting representative points, generating information summaries, and performing strategic sampling, this application can effectively compress redundant point cloud clusters while preserving their key geometric features to the maximum extent, ensuring that the refined data packets can accurately reflect the structural information of the original data, thus achieving a balance between data storage efficiency and information integrity.

[0017] Optionally, a real-time spatial geometric matching calculation is performed between the scanning range of the laser scanner and the key monitoring area to obtain the overlap relationship between the scanning range of the laser scanner and the key monitoring area. This real-time spatial geometric matching calculation, prior to the step of using a 3D bounding box intersection algorithm, includes: The reference coordinates of the key monitoring area are updated based on the point cloud data obtained from the partial overview scan of the laser scanner to acquire the key monitoring area. Real-time spatial position and attitude information of the laser scanner is acquired to predict the scanning trajectory of the laser scanner in the future. Geometric analysis is performed between the predicted scan trajectory and the updated baseline coordinates of the key monitoring area to determine whether there is a deviation trend; If a deviation trend exists, a scan path adjustment command is generated. This adjustment command uses a proportional-integral-derivative control algorithm. The input is the deviation amount, and the output is the adjustment amount of the scanner motion parameters, including the adjustment amount of horizontal rotation speed, pitch angle, or scan sector range.

[0018] The scanning path or scanning sector of the laser scanner is adjusted in real time according to the scanning path adjustment command.

[0019] By using this technical solution, before data collection, this application can predict the scanning trajectory and judge the deviation trend in advance, thereby discovering and correcting the deviation of the scanning path in advance, ensuring that the laser scanner is always aligned with the key monitoring area, thus improving the accuracy and efficiency of data collection, reducing invalid scans, and laying the foundation for subsequent high-value data identification.

[0020] Optionally, the step of updating the reference coordinates of the key monitoring area based on the point cloud data obtained from the partial overview scan of the key monitoring area by the laser scanner includes: The point cloud data obtained by the laser scanner's partial overview scan is matched and locally registered with the digital model of the preset key monitoring area. The feature matching and local registration are implemented using the iterative nearest point algorithm. Calculate the offset of the key monitoring area relative to the initial position; The baseline coordinates of the key monitoring area are updated based on this offset.

[0021] Through this technical solution, this application can accurately update the reference coordinates of key monitoring areas by using local overview scanning and iterative nearest point algorithm, ensuring the accuracy of subsequent scanning path prediction and adjustment, and further improving the accuracy and reliability of data acquisition.

[0022] Optionally, the step of performing geometric analysis between the predicted scan trajectory and the updated baseline coordinates of the key monitoring area to determine whether there is a deviation from the trend includes: Calculate the minimum distance between the predicted scan trajectory and the boundary of the key monitoring area; When the minimum distance is less than the preset critical deviation threshold, it is determined that there is a deviation trend.

[0023] This technical solution allows for the calculation and comparison of the minimum distance between the predicted scanning trajectory and the boundary of the key monitoring area with a preset threshold. This enables the rapid and accurate determination of whether there is a deviation trend, providing a quantitative basis for timely generation of scanning path adjustment instructions.

[0024] Secondly, this application also discloses a laser scanning ancient building structure data acquisition system for acquiring ancient building structure data. The system includes: The information acquisition module is used to acquire the key monitoring areas preset in the ancient building and to acquire the real-time spatial position and attitude information of the laser scanner. The scanning range calculation module is used to calculate the scanning range of the laser scanner in real time based on real-time spatial position and attitude information; The overlap relationship judgment module is used to perform real-time spatial geometric matching calculation between the scanning range of the laser scanner and the key monitoring area to obtain the overlap relationship. This real-time spatial geometric matching calculation adopts the three-dimensional bounding box intersection algorithm. The data labeling module is used to label the point cloud data collected by the laser scanner as high-value data if there is an overlap relationship, and to label the point cloud data collected by the laser scanner as ordinary-value data if there is no overlap relationship. The differentiated caching processing module is used to write point cloud data marked as high-value data into a high-priority cache queue and point cloud data marked as ordinary-value data into an ordinary-priority cache queue, and to perform differentiated cache management and transfer strategies on the data in the high-priority cache queue and the ordinary-priority cache queue.

[0025] This application provides a system that enables priority processing of high-value data. Through modular design, it realizes intelligent management of the data acquisition process for ancient building structures, effectively solving the problems of low data acquisition efficiency and poor data continuity in existing technologies, and providing strong support for the digital protection and structural analysis of ancient buildings. Beneficial effects

[0026] The laser scanning method for acquiring structural data of ancient buildings disclosed in this application can calculate the scanning range of the scanner in real time by acquiring the real-time spatial position and attitude information of the preset key monitoring areas in the ancient building and the laser scanner. Based on this, a 3D bounding box intersection algorithm is used to perform real-time spatial geometric matching calculations between the scanning range and the key monitoring areas to determine whether there is an overlap relationship. According to the overlap relationship, the acquired point cloud data is intelligently marked as high-value data or ordinary-value data. Furthermore, this application writes high-value data into a high-priority cache queue and ordinary-value data into a normal-priority cache queue, and performs differentiated cache management and transfer strategies on the data in these two queues.

[0027] By identifying and prioritizing high-value data, the aforementioned technical solution avoids the waste of valuable cache space and write bandwidth caused by a large amount of low-value data, thus significantly improving the efficiency of data acquisition and the continuity of data transmission. When cache space is scarce, storage space is released first from the ordinary priority cache queue, ensuring that high-value data is not lost due to cache overflow. Furthermore, dynamic monitoring of cache queue occupancy and main storage device load, along with real-time adjustments to the write bandwidth allocation strategy, further guarantees the timely writing of high-value data. Therefore, this application overcomes the limitations of the "first-in, first-out" strategy in existing technologies, ensuring that high-value data reflecting key structural details can be acquired promptly and completely when collecting fine-structure data. This provides a reliable data foundation for establishing accurate models reflecting structural changes at different points in time, enabling effective dynamic monitoring of ancient buildings and avoiding the problem of models losing their analytical value due to missing data or incorrect timestamps. Attached Figure Description

[0028] To illustrate this application more clearly, the accompanying drawings used in the embodiments will be briefly described below. Obviously, those skilled in the art can obtain other drawings based on these drawings without any creative effort.

[0029] Figure 1 The following is an exemplary flowchart of a laser scanning method for acquiring structural data of ancient buildings in an embodiment; Figure 2 The diagram illustrates a module configuration block diagram of a laser scanning ancient building structure data acquisition system according to an embodiment.

[0030] Figure reference numerals: 100, Laser scanning ancient building structure data acquisition system; 10, Information acquisition module; 20, Scanning range calculation module; 30, Overlap relationship judgment module; 40, Data marking module; 50, Differentiated caching processing module. Detailed Implementation

[0031] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0032] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0033] Traditional 3D laser scanning technology, when used for the digital preservation and structural analysis of ancient buildings, especially when dealing with extremely complex wooden structures (such as brackets, beams, and carvings on corbels), requires extremely high scanning precision to obtain high-accuracy data. This leads to a dramatic increase in point cloud density per unit area and an explosive growth in data flow rate. Existing data transmission and caching systems face immense pressure; the cache fills up far faster than the data is transferred to the main storage device, quickly causing system congestion. To avoid data overflow and loss, the acquisition software must frequently pause the scanning operation, waiting for the cache to be cleared before resuming. This stop-and-go approach not only significantly extends on-site work time but, more seriously, for tasks requiring the creation of models reflecting structural changes at different points in time, this interruption disrupts the temporal continuity of the data, rendering the final model worthless for analysis. Furthermore, existing caching methods treat all collected point cloud data as equally important, employing a simple "first-in, first-out" sequential writing and transfer strategy. This results in a large amount of low-value data, such as simple and highly repetitive flat wall surfaces, occupying valuable cache space and write bandwidth, while high-value data, such as intricate wood carvings and other key structural details, are forced to wait due to a full cache, and may even increase the risk of being discarded.

[0034] like Figure 1 The diagram illustrates an exemplary method for acquiring data on the structure of ancient buildings using laser scanning. This application proposes a method for acquiring data on the structure of ancient buildings using laser scanning, comprising: S10: Obtain the preset key monitoring areas in the ancient building and obtain the real-time spatial position and attitude information of the laser scanner; S20, calculate the scanning range of the laser scanner in real time based on the real-time spatial position and attitude information; S30, perform real-time spatial geometric matching calculation between the scanning range of the laser scanner and the key monitoring area to obtain the overlap relationship. The real-time spatial geometric matching calculation adopts the three-dimensional bounding box intersection algorithm. S40, if there is an overlap, the point cloud data collected by the laser scanner is marked as high-value data; if there is no overlap, the point cloud data collected by the laser scanner is marked as ordinary-value data. S50: Point cloud data marked as high-value data is written to a high-priority cache queue, and point cloud data marked as ordinary-value data is written to an ordinary-priority cache queue. Differentiated cache management and transfer strategies are executed on the data in the high-priority cache queue and the ordinary-priority cache queue.

[0035] This application effectively solves the problems of low data acquisition efficiency, high risk of losing high-value data, and inability to guarantee data continuity in existing technologies by distinguishing the value of point cloud data and adopting differentiated caching management and transfer strategies, thus significantly improving the efficiency and quality of ancient building structure data acquisition.

[0036] The term "ancient buildings" as used in this application broadly refers to buildings with historical, artistic, and scientific value, characterized by complex structures containing numerous intricate components such as brackets, beams, and corbels. Digital preservation and structural analysis of these ancient buildings require the acquisition of high-precision three-dimensional point cloud data.

[0037] Laser scanners are used to acquire three-dimensional coordinate data of an object's surface. They measure distances and angles by emitting laser beams and receiving reflected signals, thereby generating point cloud data.

[0038] Point cloud data is a dataset consisting of a series of points with three-dimensional coordinates, used to represent the geometric shape and texture information of an object's surface.

[0039] Key monitoring areas refer to specific areas of ancient buildings that require focused attention and high-precision data collection, such as structural components prone to deformation, those with intricate carvings, or those critically load-bearing structures. Data from these areas is essential for the structural analysis and health monitoring of ancient buildings.

[0040] Real-time spatial position and attitude information refers to the instantaneous data of a laser scanner's precise position (such as X, Y, and Z coordinates) and orientation (such as pitch, yaw, and roll angles) in three-dimensional space during the acquisition process. This information is crucial for determining the scanner's scanning range and performing spatial matching calculations.

[0041] The scanning range refers to the spatial area that the laser scanner can effectively cover and collect point cloud data in its current position and orientation.

[0042] The 3D bounding box intersection algorithm is a geometric algorithm used to determine whether two 3D objects spatially overlap. It determines this by calculating whether the minimum axis-aligned bounding boxes (AABBs) of the two objects intersect.

[0043] In its implementation, this application first acquires the pre-defined key monitoring areas within the ancient building and obtains the real-time spatial position and attitude information of the laser scanner. The key monitoring areas can be acquired in various ways. For example, areas requiring special attention can be pre-defined on the digital model of the ancient building through manual annotation and stored as 3D model data or coordinate range data. Another approach is to utilize historical data or expert experience to identify vulnerable, critical load-bearing, or artistically valuable parts of the ancient building and designate them as key monitoring areas. The real-time spatial position and attitude information of the laser scanner can be acquired through an integrated inertial measurement unit (IMU), a global positioning system (GPS), or a vision-based simultaneous localization and mapping (SLAM) system. For instance, an IMU can provide the scanner's angular velocity and acceleration information, and attitude changes can be calculated through integration; GPS can provide the scanner's absolute position information; and a SLAM system can estimate its position and attitude in the environment in real time by analyzing the images or point cloud data acquired by the scanner.

[0044] Subsequently, the scanning range of the laser scanner is calculated in real time based on its real-time spatial position and attitude information. The calculation of the scanning range can be determined based on the laser scanner's hardware parameters (such as maximum scanning distance, horizontal and vertical field of view) and its current spatial position and attitude. For example, a cone-shaped or fan-shaped scanning area model can be constructed based on the scanner's field of view and current position. By using the scanner's real-time position as the apex of the cone or the center of the fan, and combining it with its attitude information (such as pitch and yaw angles), the three-dimensional spatial range that the laser beam can cover can be accurately calculated.

[0045] Next, real-time spatial geometric matching calculations are performed between the scanning range of the laser scanner and the key monitoring areas to determine the overlap relationship. This real-time spatial geometric matching calculation uses a 3D bounding box intersection algorithm. Specifically, a 3D bounding box can be calculated for both the scanning range of the laser scanner and each key monitoring area. For example, the 3D bounding box for the scanning range can be a minimum axis-aligned bounding box (AABB), which completely encompasses the currently reachable scanning area of ​​the scanner. The 3D bounding box for the key monitoring area can also be a predefined AABB, or dynamically calculated based on its geometric model. Then, the existence of a spatial overlap relationship between the two 3D bounding boxes is determined by whether their projections on the X, Y, and Z coordinate axes overlap. If all axial projections overlap, it is determined that the scanning range and the key monitoring area overlap.

[0046] If overlap exists, the point cloud data acquired by the laser scanner is marked as high-value data; if no overlap exists, it is marked as ordinary-value data. This marking process can be performed early in the data acquisition process, such as before the point cloud data stream enters the caching system. Marking can be achieved by adding a metadata tag or flag to the point cloud data packet. For example, a boolean variable can be set to "True" when the data belongs to a high-value region and "False" otherwise.

[0047] Finally, point cloud data marked as high-value data is written to a high-priority cache queue, and point cloud data marked as ordinary-value data is written to an ordinary-priority cache queue. Differentiated cache management and rollover strategies are applied to the data in these two queues. The high-priority and ordinary-priority cache queues can be different regions in memory or two logically separate data structures. For example, they can be implemented using two independent circular buffers or linked lists. Differentiated cache management and rollover strategies can include: allocating more cache space to the high-priority cache queue, higher write and read bandwidth, faster rollover frequency, and prioritizing the protection of high-priority data from being discarded when cache space is limited. For example, when cache space is insufficient, the system can prioritize deleting old data from the ordinary-priority cache queue or downgrading it for rollover to ensure that the high-priority cache queue has sufficient space to receive new data.

[0048] The laser scanning method for acquiring ancient building structure data in this application effectively solves the problems of low data acquisition efficiency, high risk of losing high-value data, and inability to guarantee data continuity in traditional methods by introducing a real-time judgment of the value of point cloud data and a differentiated caching management mechanism.

[0049] Specifically, in traditional methods, all acquired point cloud data is treated as equally important, and a simple "first-in, first-out" strategy is used for caching and transferring. This approach results in a large amount of low-value data, such as simple, repetitive descriptions of flat wall surfaces, occupying valuable cache space and write bandwidth, while high-value data, such as descriptions of intricate wood carvings and other critical structural details, is forced to wait due to a full cache, even increasing the risk of being discarded. When the scanning accuracy is set to extremely high levels, the data flow rate explodes, and the existing caching system quickly becomes congested. The acquisition software has to frequently pause the scanning operation, severely extending the on-site operation time and disrupting the temporal continuity of the data.

[0050] This application acquires key monitoring areas pre-defined within ancient buildings and obtains the spatial position and orientation information of a laser scanner in real time, enabling real-time calculation of the scanner's scanning range. By performing real-time spatial geometric matching calculations between the scanning range and the key monitoring areas (using a 3D bounding box intersection algorithm), it can accurately determine whether the currently collected point cloud data belongs to the key monitoring areas. Based on this determination, the point cloud data is marked as high-value data or ordinary-value data. This real-time data value assessment mechanism is one of the core innovations of this application.

[0051] Subsequently, this application writes point cloud data marked as high-value data into a high-priority cache queue and point cloud data marked as ordinary-value data into an ordinary-priority cache queue. This dual-queue mechanism, combined with differentiated cache management and transfer strategies, enables the system to prioritize and protect high-value data. For example, higher write and read permissions can be assigned to the high-priority cache queue to ensure that its data is preferentially transferred to the main storage device. When cache space is scarce, the system will prioritize releasing storage space from the ordinary-priority cache queue, thereby ensuring the integrity and timeliness of high-value data. This strategy effectively avoids the risk of losing high-value data due to cache overflow and ensures the continuous acquisition of data in critical areas.

[0052] Compared with existing technologies, this application has the following advantages: First, by identifying and labeling high-value data in real time, it avoids the unnecessary occupation of cache resources by low-value data, thus improving cache utilization efficiency. Second, the differentiated cache management and transfer strategy ensures that high-value data is prioritized for processing and storage, significantly reducing the risk of data loss, which is particularly important for ancient building structure monitoring tasks that require high-precision and continuous data. Finally, by optimizing data stream processing, it reduces scanning interruptions caused by cache blocking, thereby shortening on-site operation time and improving overall acquisition efficiency. This application not only improves the quality and efficiency of data acquisition but also provides more reliable and comprehensive data support for the digital protection and structural analysis of ancient buildings.

[0053] Specifically, the above-mentioned steps of performing real-time spatial geometric matching calculations between the scanning range of the laser scanner and the key monitoring area to obtain the overlap relationship, and using a 3D bounding box intersection algorithm for real-time spatial geometric matching calculations, can be further refined into the following operations: First, the 3D spatial bounding box of the laser scanner's scanning range is calculated. This bounding box can be understood as the smallest axis-aligned bounding box (AABB) that completely contains the current scanning area of ​​the laser scanner. This bounding box can be calculated based on the laser scanner's real-time spatial position and orientation information, combined with known parameters such as its scanning angle, scanning distance, and scanning mode. For example, by using the laser scanner's frustum or scanning sector, its effective scanning area in 3D space can be determined, and then the smallest bounding rectangle of this area, i.e., the 3D spatial bounding box, can be calculated. The purpose is to simplify the complex scanning area into a regular geometric shape to facilitate subsequent spatial matching calculations.

[0054] Secondly, spatial overlap is determined by comparing the 3D bounding box of the laser scanner's scanning range with the preset 3D bounding boxes of each key monitoring area. Specifically, if the projections of the two overlap on the X, Y, and Z axes, an overlap relationship is considered to exist. The preset 3D bounding box of the key monitoring area refers to the minimum bounding rectangle pre-calculated and stored during system initialization or preprocessing, based on the geometry and spatial location of the preset key monitoring areas in the ancient building. When determining spatial overlap, the intersection of the projection intervals of the two 3D bounding boxes on the X, Y, and Z axes can be checked. For example, for the X-axis, if the X-axis projection interval [min_x1, max_x1] of the laser scanner bounding box intersects with the X-axis projection interval [min_x2, max_x2] of the key monitoring area bounding box (i.e., max_x1>= min_x2 and max_x2>= min_x1), then an overlap is considered to exist on the X-axis. When the projections of the three coordinate axes overlap, it can be determined that the two three-dimensional bounding boxes overlap in three-dimensional space, thus indicating that the scanning range of the laser scanner overlaps with the key monitoring area.

[0055] This application abstracts the complex scanning range and key monitoring area of ​​a laser scanner into simplified three-dimensional bounding boxes, and performs axial projection overlap determination based on these bounding boxes, achieving efficient and real-time spatial geometric matching calculations. This method avoids the computational overhead of directly performing intersection operations on complex geometries, instead decomposing the overlap determination problem in three-dimensional space into three independent, one-dimensional interval overlap determination problems. It is precisely this simplification that enables the rapid and accurate determination of the overlap between the scanning area and the key monitoring area during the real-time demanding laser scanning data acquisition process.

[0056] By employing the aforementioned technical solution and utilizing a 3D bounding box intersection algorithm for real-time spatial geometric matching calculations, the computational efficiency of overlapping relationship determination can be significantly improved. Compared to directly performing precise geometric intersection calculations on point cloud data or complex geometric models, the 3D bounding box-based determination method requires less computation and has a faster response speed, making it particularly suitable for laser scanning scenarios that require real-time processing of large amounts of spatial data. Furthermore, this method effectively reduces the system's computational resource requirements while ensuring sufficient accuracy, thereby ensuring that the laser scanner can continuously and stably identify and label high-value areas during movement, improving the real-time performance and robustness of the overall data acquisition process.

[0057] This application further proposes the following steps for implementing differentiated cache management and transfer strategies for data in high-priority cache queues and normal-priority cache queues: Grant the highest write and read permissions to the high-priority cache queue, and transfer the data in the priority queue to the main storage device; When cache space is tight, storage space is released from the ordinary priority cache queue first. The operation of releasing space includes pausing the writing of new data to the queue, overwriting the historical data in the queue, and downgrading the data in the queue to any of the auxiliary storage devices. The system dynamically monitors the occupancy status of high-priority cache queues and ordinary-priority cache queues, as well as the load on the main storage device, and adjusts the write bandwidth allocation strategy of the two queues in real time. When there is data waiting to be written in the high-priority cache queue, its write bandwidth is guaranteed first, and the write operation of the ordinary-priority cache queue is interrupted if necessary.

[0058] Specifically, granting the highest write and read permissions to high-priority cache queues can be understood as the system prioritizing requests from these queues during data access operations. For example, in multi-tasking or concurrent processing environments, read and write operations on high-priority queues are allocated higher processor time slices or faster input / output (I / O) channels to ensure their data can be written or read quickly. Simultaneously, data in the priority queues is transferred to main storage, aiming to ensure that high-value data is moved from temporary caches to more stable, larger-capacity main storage media as quickly as possible, thereby reducing the risk of data loss and providing a reliable data source for subsequent data processing and analysis.

[0059] When cache space is scarce, the system prioritizes releasing storage space from the regular priority cache queue. This space release operation can include various specific strategies. For example, it can pause the writing of new ordinary value data to this queue to avoid further consuming limited cache resources; or it can overwrite older historical data in the queue, i.e., using data eviction policies such as First-In-First-Out (FIFO) or Least Recently Used (LRU) to clear old data; or it can degrade the data in the queue to secondary storage devices, such as hard disk arrays, network storage (NAS), or cloud storage, to free up main cache space for high-priority data while still maintaining the accessibility of ordinary value data, although the access speed may be relatively slower. The goal is to maximize the storage needs and processing priority of high-value data when system resources are limited.

[0060] In practical applications, the system dynamically monitors the occupancy status of high-priority and regular-priority cache queues, as well as the load on the main storage device. Occupancy status refers to the ratio of the amount of data stored in the queue to the total capacity, while the main storage device load reflects the activity level and processing capacity of the storage device. Based on this real-time monitoring data, the system adjusts the write bandwidth allocation strategy for the two queues in real time. For example, when data is waiting to be written to the high-priority cache queue, the system prioritizes its write bandwidth, which means allocating more network bandwidth or I / O resources to accelerate the data writing process. If necessary, the system can even interrupt write operations in the regular-priority cache queue, concentrating all available resources on writing high-priority data to ensure the timely processing and storage of critical data. The goal is to continuously optimize resource allocation in a dynamically changing environment, ensuring the processing priority of high-value data.

[0061] This application effectively addresses potential resource contention and data processing efficiency issues in basic solutions by introducing refined cache management and transfer strategies. Specifically, by granting high-priority cache queues the highest write and read permissions, it ensures that high-value point cloud data collected in key monitoring areas is prioritized for system processing, avoiding the risk of being blocked by ordinary value data in a data deluge. When system cache resources face pressure, space is released from ordinary priority cache queues first, and valuable storage resources are freed up for high-value data by pausing writes, overwriting historical data, or downgrading and transferring data, thus ensuring the continuity and integrity of high-value data storage. Furthermore, by dynamically monitoring the cache queue status and main storage device load, and adjusting write bandwidth allocation in real time, the system can flexibly schedule resources according to actual operating conditions, ensuring that high-priority data always has the optimal write channel. Even in extreme cases, the timely processing of high-value data can be guaranteed by interrupting writes to ordinary priority queues, thereby significantly improving the robustness and efficiency of the entire data acquisition system.

[0062] Through the aforementioned technical solution, this application significantly improves the processing efficiency and reliability of high-value data during laser scanning of ancient building structures. Specifically, by prioritizing permissions, intelligent space release, and dynamic bandwidth allocation, it ensures that high-value point cloud data from key monitoring areas of ancient buildings receive the highest priority processing and storage, effectively avoiding data loss or processing delays caused by insufficient cache resources or excessive system load. This refined differential management strategy enables the system to maximize the integrity and timeliness of core data with limited resources, thereby improving the overall quality and efficiency of ancient building structure data acquisition and providing a more reliable data foundation for subsequent structural analysis and protection work.

[0063] In some embodiments, suppose a laser scanner is scanning an ancient building, where a carved window is designated as a key monitoring area. When the laser scanner's scan area overlaps with this carved window area, the acquired point cloud data is marked as high-value data. This high-value data is written to a high-priority cache queue. The system ensures this queue has the highest write privileges, for example, by allocating a dedicated Direct Memory Access (DMA) channel or higher I / O priority. If system cache space becomes strained, for example, if the ordinary-priority cache queue is 80% full, the system immediately initiates a space release strategy. This may include pausing new ordinary-value point cloud data writes to the ordinary-priority cache queue and starting to downgrade historical data in the queue that has exceeded a certain time threshold to secondary storage devices (such as network-attached storage servers) to free up primary cache space. Simultaneously, the system continuously monitors the write load on the primary storage device. If the main storage device's current write bandwidth is close to saturation, but there is still a large amount of high-value data waiting to be transferred in the high-priority cache queue, the system will dynamically adjust the bandwidth allocation, temporarily reduce or even interrupt the transfer operation from the ordinary priority cache queue to the main storage device, and allocate all available bandwidth resources to the high-priority cache queue first, so as to ensure that the high-value point cloud data of the carved window can be quickly and completely transferred to the main storage device, thereby ensuring the real-time performance and reliability of the critical data.

[0064] This application further proposes that the steps for writing point cloud data marked as high-value data into a high-priority cache queue include: Real-time analysis and processing of point cloud data marked as high-value data to identify the existence of redundant point cloud clusters; If there are no redundant point cloud clusters, the point cloud data marked as high-value data will be written to a high-priority cache queue; If redundant point cloud clusters exist, compress the information and generate feature summaries for the redundant point cloud clusters to obtain refined data packets, and write the refined data packets into a high-priority cache queue.

[0065] Specifically, after marking point cloud data as high-value data, it is not immediately written into a high-priority cache queue, but is first subjected to real-time analysis. This analysis aims to identify whether redundant point cloud clusters exist within these high-value data. Redundant point cloud clusters typically refer to spatially highly dense sets of points with repetitive information, such as overscanning on flat surfaces or regions with simple structures.

[0066] If, after real-time analysis and processing, it is determined that there are no redundant point cloud clusters in the current high-value point cloud data, then the point cloud data is considered concise and has a moderate information density, and can be directly written into a high-priority cache queue to ensure its rapid storage and subsequent processing.

[0067] However, if real-time analysis results indicate the existence of redundant point cloud clusters, these clusters need to be optimized. Specifically, redundant point cloud clusters will undergo information compression and feature summarization. Information compression aims to reduce the data volume, while feature summarization extracts and retains key geometric or semantic features of the redundant regions, such as the mean normal vector, local curvature, bounding box size, and original point count. Through these processes, a refined data packet is obtained, which significantly reduces the data volume while retaining the original high-value information. Subsequently, this refined data packet is written to a high-priority cache queue.

[0068] This application effectively solves the problems of wasted cache resources and low data processing efficiency that may exist in the original scheme by introducing a real-time analysis and processing mechanism for redundant point cloud clusters before writing high-value point cloud data into a high-priority cache queue. Specifically, when redundant point cloud clusters exist in the high-value data, information compression and feature summary generation can transform a large amount of repetitive point cloud data into smaller, more refined data packets. This not only reduces the amount of data that needs to be stored and transmitted, but also ensures that the data stored in the high-priority cache queue has higher information density and lower redundancy. As a result, cache space is utilized more effectively, data transmission bandwidth is optimized, and subsequent data processing modules can process this refined data with higher efficiency, thereby improving the performance of the entire laser scanning ancient building structure data acquisition method.

[0069] Through the above technical solution, this application can significantly optimize the storage and management efficiency of high-value point cloud data. By identifying and processing redundant point cloud clusters, it avoids directly writing a large amount of duplicate information into the high-priority cache queue, thereby effectively saving valuable cache space and data transmission bandwidth. In addition, the generation of refined data packets makes subsequent data processing more efficient, reduces unnecessary computational burden, and thus improves the real-time response capability and data processing throughput of the entire ancient building structure data acquisition system. This differentiated processing strategy ensures that the core information of high-value data is stored and utilized preferentially and efficiently, while avoiding system performance bottlenecks caused by data redundancy.

[0070] In some embodiments, a specific example is given below. Suppose a laser scanner is scanning a flat wall of an ancient building; this area is pre-defined as a key monitoring area, and the collected point cloud data is marked as high-value data. Before writing this high-value point cloud data into a high-priority cache queue, the system performs real-time analysis. For example, by calculating the neighborhood features of each point in the point cloud data, such as local 3D variance and mean normal vector, points that are spatially highly concentrated and have little geometric variation can be identified; these points are determined to be redundant point cloud clusters. Once redundant point cloud clusters are identified, the system does not directly store all the original points, but compresses their information, for example, through downsampling or meshing, and generates a feature summary, such as recording key information like the cluster's center point coordinates, mean normal, local curvature, and the number of original points. Finally, a refined data package containing representative points, a feature summary, and a small number of strategically sampled points is generated and written to the high-priority cache queue. If the scanned area is a complex carving or damaged area with drastic changes in point cloud data, there may be no redundant point cloud clusters. In this case, the original high-value point cloud data will be directly written to a high-priority cache queue. In this way, the system can intelligently manage high-value data, ensuring that storage and processing efficiency are maximized while maintaining data integrity.

[0071] This application further proposes steps for real-time analysis and processing of point cloud data marked as high-value data to identify the existence of redundant point cloud clusters, including: Calculate the set of points within its preset radius neighborhood based on the point cloud data; Calculate the local three-dimensional variance and average normal vector of the point set; When the number of points exceeds a preset threshold, the 3D variance is lower than a preset variance threshold, and the angle between the normal vectors is less than a preset angle threshold, it is determined to be a redundant point cloud cluster.

[0072] Specifically, when performing real-time analysis on point cloud data marked as high-value data, the first step is to calculate the point set within a preset radius neighborhood for each point cloud data point. This preset radius neighborhood can be understood as a spherical or cubic region defined in three-dimensional space with the current point as the center, used to define other points adjacent to the current point. By determining the point set within the neighborhood, basic data can be provided for subsequent calculations of local geometric features. The local three-dimensional variance of the point set refers to a comprehensive measure of the dispersion of the point set along the X, Y, and Z coordinate axes, reflecting the density of the point set's distribution in the local space. The smaller the variance, the more concentrated the point set is, and the more likely it is to form a flat or regular region. The average normal vector is the average of the normal vectors of all points in the point set, used to characterize the overall orientation of the local region. In practical applications, when the number of points exceeds a preset threshold, it indicates that the region contains enough points to warrant statistical analysis. When the 3D variance is below a preset variance threshold, it means that the point set is densely distributed in the local space, with a flat or regular shape. Furthermore, when the angle between the normal vectors is less than a preset angle threshold, it indicates that the points within the set have similar surface orientations. When all three conditions are met, the point set can be considered to constitute a redundant point cloud cluster. These thresholds can be flexibly configured according to the actual application scenario and the definition of redundancy to adapt to different ancient building structural features and data acquisition accuracy requirements.

[0073] This application achieves quantitative identification of redundant point cloud clusters by introducing three key indicators: the number of point sets, local 3D variance, and average normal vector, and setting corresponding thresholds. Specifically, the effectiveness of local feature analysis is ensured by calculating the point set within a preset radius neighborhood. The calculation of local 3D variance accurately reflects the geometric flatness or regularity of the point cloud in a local region, while the average normal vector provides information on the consistency of surface orientation. When the point set is sufficient in number, geometrically flat, and has consistent normal directions, these points are likely to represent repetitive or unimportant surface information in ancient architectural structures, such as large areas of walls or floors. The point cloud data in these areas usually have high similarity and are therefore considered redundant. This comprehensive judgment mechanism based on multi-dimensional geometric features can effectively avoid misjudgments that may be caused by a single indicator, thereby more accurately identifying truly redundant point cloud clusters.

[0074] Through the above technical solution, this application provides an objective, quantitative, and efficient method for identifying redundant point cloud clusters. Compared to methods relying on manual judgment or simple sampling, this solution significantly improves the accuracy and automation of redundant point cloud cluster identification by comprehensively considering the number of point sets, local 3D variance, and average normal vector. This not only effectively reduces the computational burden of subsequent data processing, ensuring that only truly valuable, non-redundant high-value data is further refined and stored, but also avoids the loss of important structural information due to misjudgment, thereby improving the efficiency and data quality of the entire laser scanning ancient building structure data acquisition method.

[0075] In some embodiments, it is assumed that a laser scanner is acquiring data on the flat wall of an ancient building. When the system receives point cloud data marked as high-value data, it calculates the point set within a 0.05-meter radius for each point. If a point set contains more than 50 points (a preset number threshold), and its local 3D variance is less than 0.001 square meters (a preset variance threshold), and the angle between the normal vector of all points in the set and the average normal vector is less than 5 degrees (a preset angle threshold), then the point set is identified as a redundant point cloud cluster. For example, the point cloud data of a straight wall typically meets these conditions because the points are densely distributed, the surface is flat, and the normal directions are consistent. Once these redundant point cloud clusters are identified, they will be further compressed and feature summaries generated to reduce storage and transmission overhead while retaining their key geometric information, thereby optimizing the processing flow of high-value data.

[0076] In some embodiments, this application further proposes a method for information compression and feature summary generation of redundant point cloud clusters to obtain refined data packets, thereby effectively reducing the amount of data while ensuring data accuracy.

[0077] If redundant point cloud clusters exist, the steps to compress the information and generate feature summaries from these clusters to obtain refined data packets include: Representative points are extracted from redundant point cloud clusters, and the representative points are either the average coordinates of all points or the original point closest to the center. Calculate and record the local geometric feature parameters of the redundant point cloud cluster to obtain an information digest. The local geometric feature parameters include the average normal vector, local curvature, bounding box size, and original number of points. A refined data package is obtained based on representative points, information summaries, and a small amount of strategically sampled point cloud data.

[0078] Specifically, extracting representative points from redundant point cloud clusters aims to summarize the overall location and shape of the cluster using the fewest possible points. These representative points can be determined in various ways. For example, the average coordinates of all points in the redundant point cloud cluster can be calculated as the representative point, reflecting the geometric center of the cluster. Alternatively, the original point closest to the geometric center of the cluster can be selected as the representative point to ensure that it is a real, existing physical point.

[0079] The process involves calculating and recording the local geometric feature parameters of the redundant point cloud cluster to obtain an information digest. The purpose is to describe the geometric characteristics of the redundant point cloud cluster using structured data, rather than storing all the original points. These local geometric feature parameters include the average normal vector, local curvature, bounding box size, and the number of original points. The average normal vector characterizes the overall orientation and surface smoothness of the point cloud cluster; the local curvature reflects the degree of curvature of the cluster's surface; the bounding box size provides the area occupied by the cluster in three-dimensional space; and the number of original points records the amount of original data before compression, facilitating subsequent data recovery or evaluation. These parameters together constitute the information digest, which can summarize the key geometric information of the redundant point cloud cluster in a compact form.

[0080] In practical applications, a refined data package is obtained based on representative points, an information digest, and a small amount of strategically sampled point cloud data. The refined data package is the final compressed representation of redundant point cloud clusters. Representative points provide the core location information of the cluster; the information digest provides the macroscopic geometric features of the cluster; and the small amount of strategically sampled point cloud data is used to preserve any small but important details that may exist within the cluster, or for local reconstruction and verification in subsequent processing. Through this combination, the refined data package can significantly reduce the amount of data while preserving the structural and geometric information of redundant point cloud clusters to the maximum extent.

[0081] This application effectively solves the problem of information loss or low compression efficiency caused by simple compression in traditional methods by intelligently compressing and generating feature summaries of redundant point cloud clusters. Specifically, firstly, by extracting representative points, a large amount of repetitive point cloud data is simplified into one or a few key points, thereby significantly reducing the data volume. Secondly, by calculating and recording the local geometric feature parameters of redundant point cloud clusters, such as the average normal vector, local curvature, bounding box size, and original number of points, these parameters accurately capture the macroscopic geometric characteristics of redundant regions in the form of structured information summaries, avoiding the redundancy caused by directly storing all points. Finally, by combining representative points, information summaries, and a small amount of strategically sampled point cloud data, a refined data package is constructed. This multi-level representation ensures that sufficient detailed information is retained while minimizing the data volume, to meet the needs of subsequent ancient building structure analysis and modeling.

[0082] Through the above technical solution, this application can significantly improve the storage and transmission efficiency of high-value point cloud data. By performing refined compression and feature summary generation on redundant point cloud clusters, the data volume can be greatly reduced, thereby reducing the occupation of cache space and main storage devices and accelerating data transfer speed. At the same time, since the refined data package contains representative points, key geometric feature parameters, and a small amount of strategically sampled point cloud data, it ensures that while the data volume is reduced, the key structural information and geometric features of ancient buildings are effectively preserved, avoiding the information loss that may be caused by traditional simple compression. This provides a high-quality and efficient data foundation for subsequent structural analysis, monitoring, and digital protection of ancient buildings.

[0083] In some embodiments, assuming that a laser scanner, while scanning a section of straight wall of an ancient building, continuously acquires a large amount of highly similar point cloud data due to minimal changes in scanning angle and distance, this data is identified as a redundant point cloud cluster. Specifically, this redundant point cloud cluster may contain tens of thousands of points, forming an approximately planar region in space.

[0084] At this point, this application will first extract representative points from the redundant point cloud cluster. For example, the average coordinates of these tens of thousands of points can be calculated to obtain a virtual point representing the center of the wall area. Alternatively, the actual data collection point closest to the geometric center of the point cloud cluster can be selected as the representative point.

[0085] Next, the system calculates and records the local geometric feature parameters of the redundant point cloud cluster. For example, it calculates the average normal vector of all points, which indicates the overall orientation of the wall; it calculates the local curvature, which will be very small since the wall is flat, reflecting its flatness; it records the maximum and minimum ranges of the point cloud cluster in the X, Y, and Z directions to obtain its bounding box size; and it records the original number of points in the cluster, for example, 50,000 points. These parameters together constitute a compact information summary.

[0086] Finally, the extracted representative points, the calculated information summary, and a small number (e.g., 50) of strategically sampled point cloud data randomly or uniformly selected from the redundant point cloud cluster are combined to form a refined data package. This refined data package may contain only one representative point, a few geometric feature parameters, and 50 sampling points. Its data volume is much smaller than the original 50,000 points, but it can accurately describe the spatial location, geometric shape, and surface characteristics of the straight wall, thus achieving efficient data compression and information preservation.

[0087] In some embodiments, this application proposes a scheme to dynamically update the reference coordinates of the key monitoring area, predict the scanning trajectory of the laser scanner in real time and judge the deviation trend before performing real-time spatial geometric matching calculation between the scanning range of the laser scanner and the key monitoring area, and then adjust the scanning path or scanning sector of the laser scanner in real time according to the generated scanning path adjustment command, so as to ensure that the laser scanner can focus on the key monitoring area more accurately and efficiently for data acquisition.

[0088] The above-mentioned real-time spatial geometric matching calculation between the scanning range of the laser scanner and the key monitoring area to obtain the overlap relationship includes the following steps before the real-time spatial geometric matching calculation using the three-dimensional bounding box intersection algorithm: The reference coordinates of the key monitoring area are updated based on the point cloud data obtained from the partial overview scan of the laser scanner to acquire the key monitoring area. Real-time spatial position and attitude information of the laser scanner is acquired to predict the scanning trajectory of the laser scanner in the future. Geometric analysis is performed between the predicted scan trajectory and the updated baseline coordinates of the key monitoring area to determine whether there is a deviation trend; If a deviation trend exists, a scanning path adjustment instruction is generated. The adjustment instruction adopts a proportional-integral-derivative control algorithm. The input is the deviation amount, and the output is the adjustment amount of the scanner motion parameters. The adjustment amount of the motion parameters includes the adjustment amount of horizontal rotation speed, the adjustment amount of pitch angle, or the adjustment amount of scanning sector range.

[0089] The scanning path or scanning sector of the laser scanner is adjusted in real time according to the scanning path adjustment command.

[0090] Specifically, updating the baseline coordinates of a key monitoring area based on point cloud data obtained from a partial overview scan using a laser scanner involves performing a rapid, low-resolution overview scan to acquire preliminary point cloud data of the key monitoring area. This point cloud data is then compared with a pre-set digital model or historical data to identify any minor changes or drifts in the spatial location of the key monitoring area, and accordingly corrects its baseline coordinates. The purpose is to ensure that the spatial definition of the key monitoring area remains consistent with the actual physical environment.

[0091] This involves acquiring real-time spatial position and attitude information of the laser scanner and predicting its scanning trajectory in the future. This can be understood as continuously monitoring the laser scanner's precise position and orientation in three-dimensional space using an inertial measurement unit, global positioning system, or other positioning sensors. Based on this real-time data, combined with the scanner's motion model and a pre-set scanning plan, its movement path and scanning coverage area can be predicted in the near future. The purpose is to provide forward-looking information for subsequent deviation assessments and path adjustments.

[0092] In practical applications, geometric analysis is performed between the predicted scan trajectory and the updated baseline coordinates of the critical monitoring area to determine if there is a deviation trend. Specifically, this involves calculating the distance between the predicted trajectory and the boundary of the critical monitoring area, or assessing the degree of overlap between the predicted trajectory and the critical monitoring area. For example, the distance between the centerline of the predicted trajectory and the geometric center of the critical monitoring area can be calculated, or the intersection of the bounding box of the predicted trajectory and the bounding box of the critical monitoring area can be calculated. The purpose is to promptly detect whether the scanner is deviating from or is about to deviate from the expected critical monitoring area.

[0093] In a preferred implementation, if a deviation trend is detected, a scan path adjustment command is generated. This adjustment command employs a proportional-integral-derivative (PID) control algorithm, with the deviation amount as input and the scanner motion parameter adjustment amount as output. This motion parameter adjustment includes horizontal rotation speed adjustment, pitch angle adjustment, or scan sector range adjustment. This means that once a deviation is detected, the system immediately initiates a feedback control mechanism. The PID controller calculates the necessary adjustment amount based on the detected deviation. These adjustments can directly affect the laser scanner's motion actuators, such as changing its horizontal rotation speed, adjusting its pitch angle, or modifying the start and end angles of its scan sectors to guide the scanner back to the correct path. The aim is to achieve precise, smooth, and real-time correction of the scan path.

[0094] Therefore, real-time adjustment of the laser scanner's scanning path or scanning sector according to the scanning path adjustment command means sending the motion parameter adjustment amount calculated by the proportional-integral-derivative controller to the laser scanner's drive system, causing it to immediately execute the corresponding physical motion adjustment. This ensures that the scanner can dynamically correct its scanning behavior to continuously align with the key monitoring area.

[0095] This application effectively addresses the problem of decreased data acquisition efficiency and accuracy caused by drift in the reference coordinates of key monitoring areas or scanner path deviation in traditional methods by introducing preprocessing and proactive adjustment mechanisms before the core real-time spatial geometric matching calculation. First, by performing a local overview scan and updating the reference coordinates of the key monitoring area, the spatial definition of the target area is ensured to be always up-to-date and accurate, avoiding inaccurate matching due to environmental changes or initial setup errors. Second, the scanning trajectory of the laser scanner is predicted in real time and geometrically analyzed against the updated key monitoring area, enabling the system to proactively detect potential deviation trends. This proactive judgment allows the system to intervene before deviations actually occur or at their initial stage. Finally, by employing a proportional-integral-derivative (PI-DE) control algorithm to generate scanning path adjustment commands and applying them to the scanner's motion parameters in real time, precise, smooth, and real-time dynamic correction of the scanning path or scanning sector is achieved. This closed-loop control mechanism ensures that the laser scanner can continuously and stably focus on high-value key monitoring areas, thereby maximizing the capture of effective data.

[0096] Through the aforementioned technical solutions, this application significantly improves the efficiency and data quality of laser scanning data acquisition of ancient building structures. Specifically, by dynamically updating the reference coordinates of key monitoring areas, the problem of inaccurate target area definition due to the passage of time or environmental changes is avoided, ensuring the accuracy of subsequent matching. By predicting the scanning trajectory and promptly detecting deviation trends, the system can proactively intervene, guiding the scanner to the optimal scanning position, thereby reducing invalid scans and increasing the capture ratio of high-value data. Furthermore, the use of a proportional-integral-derivative control algorithm for path adjustment ensures the smoothness and accuracy of the adjustment process, avoiding data interruptions or jitter that may be caused by rough adjustments. Compared to basic solutions that rely solely on real-time matching to filter data, this application, through pre-emptive path optimization and adjustment, ensures the targeting and effectiveness of data acquisition from the source, greatly optimizing the data acquisition process, reducing the complexity of subsequent data processing, and ultimately improving the overall efficiency of ancient building structure data acquisition.

[0097] For example, suppose a laser scanner is collecting data on a specific beam-column structure of an ancient building. First, the system uses the laser scanner to perform a rapid local overview scan, acquiring point cloud data of the beam-column area. This point cloud data is then input into a registration algorithm, such as the iterative nearest-point algorithm, to match it with a pre-established digital model of the beams and columns, calculating the minute offset of the beams and columns relative to initially set reference coordinates. Based on this offset, the system updates the reference coordinates of the beam-column area, ensuring the accuracy of its spatial position.

[0098] Simultaneously, the laser scanner's built-in inertial measurement unit and encoder provide its current spatial position and attitude in real time. Based on this real-time data, combined with the scanner's kinematic model, the system predicts the scanner's trajectory over the next 5 seconds. Subsequently, this predicted trajectory is geometrically analyzed against the newly updated reference coordinates of the beam-column region. For example, the minimum distance between the centerline of the predicted trajectory and the bounding box of the beam-column region is calculated. If this minimum distance is less than a preset critical deviation threshold (e.g., 5 cm), the scanner is determined to have a deviation trend.

[0099] Once a deviation is detected, the system immediately generates a scan path adjustment command. This command uses a proportional-integral-derivative (PID) control algorithm, taking the detected deviation as input. For example, if the scanner's horizontal position deviates 5 centimeters from the center of the target area, the PID controller calculates a corresponding horizontal rotation speed adjustment. This adjustment is then sent to the laser scanner's horizontal rotation motor, causing it to fine-tune its rotation speed, thereby redirecting the scanner's scan path back to the center of the beam and column area. In this way, even with slight external interference or internal drift during scanning, the system can adjust the scanner's movement in real time and precisely, ensuring it remains efficiently focused on the key monitoring area, thus acquiring high-quality structural data of ancient buildings.

[0100] In some embodiments, this application further proposes the step of updating the reference coordinates of the key monitoring area based on the point cloud data obtained from the partial overview scan of the key monitoring area by the laser scanner, which includes: The point cloud data obtained by the laser scanner partial overview scan is matched and locally registered with the digital model of the preset key monitoring area. The feature matching and local registration are implemented using the iterative nearest point algorithm. Calculate the offset of the key monitoring area relative to the initial position; Update the baseline coordinates of the key monitoring area based on the offset.

[0101] Specifically, the point cloud data obtained from a partial overview scan by a laser scanner refers to the three-dimensional point set data collected by the laser scanner during a preliminary or partial scan of an ancient building. This data contains the current geometric information of the key monitoring area. The preset digital model of the key monitoring area refers to a high-precision three-dimensional model of the key monitoring area in the ancient building, pre-established before the data acquisition task begins. This model typically includes information such as geometry and texture, and has known reference coordinates. Feature matching and local registration refer to the process of aligning the real-time acquired point cloud data with the preset digital model in space using algorithms to find the best correspondence between the two. The Iterative Closest Point (ICP) algorithm is a widely used algorithm for point cloud registration. It iteratively calculates the nearest point correspondence between point sets and minimizes the distance between corresponding points, thereby achieving high-precision registration. Offset refers to the amount of translation and rotation in three-dimensional space of the key monitoring area represented by the real-time point cloud data relative to the initial position of its preset digital model after feature matching and local registration. Updating the reference coordinates of the key monitoring area based on the offset means applying the calculated offset to the reference coordinates of the preset digital model to obtain the current accurate reference coordinates of the key monitoring area.

[0102] This application achieves precise identification of the current actual position and orientation of key monitoring areas by performing feature matching and local registration between point cloud data obtained from partial overview scanning with a laser scanner and a digital model of a pre-defined key monitoring area. The use of an iterative nearest-point algorithm for feature matching and local registration significantly improves the alignment accuracy between the point cloud data and the digital model, enabling accurate calculation of the offset of the key monitoring area relative to its initial position. By applying this offset to the reference coordinates of the key monitoring area, its reference coordinates can be updated in real-time and accurately, ensuring that subsequent scanning trajectory prediction and adjustment are based on the most accurate spatial reference.

[0103] The above technical solution significantly improves the accuracy and robustness of benchmark coordinate updates in key monitoring areas. Compared to simply updating coordinates based on sensor data, this solution employs a pre-set digital model and an iterative nearest-point algorithm for fine registration, effectively avoiding the impact of sensor error accumulation and environmental interference on coordinate update accuracy. This ensures that the laser scanner can more accurately predict the scanning trajectory and make more precise path adjustments during subsequent scanning, thereby improving the overall accuracy and reliability of ancient building structure data acquisition.

[0104] For example, suppose that when monitoring a specific beam frame structure of an ancient building, this beam frame structure is defined as a key monitoring area. First, a local overview scan of the beam frame structure is performed using a laser scanner to acquire its current point cloud data. Simultaneously, the system already stores a high-precision digital model of the beam frame structure and its initial reference coordinates. Next, the real-time acquired point cloud data and the preset digital model are input into the registration module, which uses an iterative nearest-point algorithm for feature matching and local registration. For example, the ICP algorithm iterates continuously, finding the nearest point in the digital model for each point in the point cloud data and calculating a transformation matrix (including translation and rotation) to minimize the distance between all corresponding points. When the iteration converges, an accurate transformation matrix is ​​obtained, which represents the offset of the beam frame structure relative to its initial position in the digital model. Finally, this offset is applied to the initial reference coordinates of the beam frame structure's digital model to obtain the accurate reference coordinates of the beam frame structure at the current moment. For example, if the initial reference coordinates are (X0, Y0, Z0), the calculated offsets are (ΔX, ΔY, ΔZ) and the rotation matrix R, then the updated reference coordinates will be (X0+ΔX, Y0+ΔY, Z0+ΔZ) after rotation R.

[0105] Specifically, the step of performing geometric analysis between the predicted scan trajectory and the updated baseline coordinates of the key monitoring area to determine whether there is a deviation from the trend may include the following: Calculate the minimum distance between the predicted scan trajectory and the boundary of the key monitoring area; When the minimum distance is less than the preset critical deviation threshold, it is determined that there is a deviation trend.

[0106] Calculating the minimum distance between the predicted scanning trajectory and the boundary of the key monitoring area refers to determining the closest distance in three-dimensional space between the laser scanner's scanning trajectory and the key monitoring area within a future time period using geometric calculation methods. This calculation can be implemented using various geometric algorithms. For example, it can be based on the distance calculation between discrete sampling points on the predicted scanning trajectory and points or surfaces on the boundary of the key monitoring area, or it can directly calculate the minimum distance between a line segment (or curve) and a polygon (or surface) using functions provided by a spatial geometry library. Its purpose is to quantify the proximity of the scanning trajectory to the target area.

[0107] Furthermore, when the calculated minimum distance is less than a preset critical deviation threshold, the system will determine that a deviation trend exists. The preset critical deviation threshold is a configurable parameter, and its value depends on the actual application scenario's requirements for scanning accuracy and coverage. For example, this threshold can be empirically set or determined through experimental calibration based on the structural complexity of the ancient building, the monitoring accuracy requirements, and the performance parameters of the laser scanner. Its purpose is to provide an objective judgment standard to identify whether the scanning trajectory has deviated to the point where adjustment is necessary.

[0108] This application provides a precise and quantifiable method to determine whether a laser scanner's scanning trajectory exhibits a deviation trend by quantifying and predicting the minimum distance between the scan trajectory and the boundary of the key monitoring area, and comparing it with a preset critical deviation threshold. This method avoids the uncertainty of subjective judgment, making the identification of deviation trends more objective and accurate. Once a deviation is detected, subsequent scan path adjustment commands can be triggered promptly to ensure that the laser scanner remains focused on the key monitoring area, thereby guaranteeing the effective acquisition of high-value data.

[0109] The above technical solution enables accurate and automated judgment of deviation trends in the scanning trajectory of a laser scanner. This judgment mechanism improves the reliability and efficiency of data acquisition, ensuring that the laser scanner can continuously and effectively cover the key monitoring areas of ancient buildings, avoiding the problems of missing key data or redundant acquisition of low-value data due to scanning path deviation. Furthermore, by setting an adjustable critical deviation threshold, the system can flexibly adapt to monitoring tasks with different accuracy requirements, further enhancing the practicality and adaptability of the method.

[0110] On the other hand, such as Figure 2 As shown, an exemplary laser scanning system for acquiring data on ancient building structures is illustrated. This application further proposes a laser scanning system 100 for acquiring data on ancient building structures, which includes: The information acquisition module 10 is used to acquire the key monitoring areas preset in the ancient building and to acquire the real-time spatial position and attitude information of the laser scanner. The scanning range calculation module 20 is used to calculate the scanning range of the laser scanner in real time based on the real-time spatial position and attitude information. The overlap relationship judgment module 30 is used to perform real-time spatial geometric matching calculation between the scanning range of the laser scanner and the key monitoring area to obtain the overlap relationship. The real-time spatial geometric matching calculation adopts the three-dimensional bounding box intersection algorithm. The data labeling module 40 is used to label the point cloud data collected by the laser scanner as high-value data if there is an overlap relationship, and to label the point cloud data collected by the laser scanner as ordinary-value data if there is no overlap relationship. The differentiated caching processing module 50 is used to write point cloud data marked as high-value data into a high-priority cache queue and point cloud data marked as ordinary-value data into an ordinary-priority cache queue, and to perform differentiated cache management and transfer strategies on the data in the high-priority cache queue and the ordinary-priority cache queue.

[0111] This system, through its modular design, enables intelligent identification, classification, and priority processing of point cloud data. This effectively solves the problems of low data acquisition efficiency, high risk of losing high-value data, and inability to guarantee data continuity in traditional methods, significantly improving the efficiency and quality of ancient building structure data acquisition.

[0112] This application's system achieves real-time intelligent identification and classification of point cloud data value by introducing an information acquisition module, a scan range calculation module, an overlap relationship judgment module, and a data tagging module. This mechanism enables the system to distinguish high-value data within key monitoring areas from ordinary-value data in non-key areas. Furthermore, the differentiated caching processing module writes data into high-priority and ordinary-priority cache queues based on data value, and executes differentiated cache management and transfer strategies. For example, high-priority data is given higher write and read permissions, is preferentially transferred to the main storage device, and receives priority protection when the cache is strained. This intelligent cache management method contrasts sharply with the indiscriminate processing methods in existing technologies, greatly improving the processing efficiency and security of high-value data, effectively avoiding data loss and scan interruptions caused by cache overflow, and offering significant advantages, especially for tasks requiring dynamic monitoring of ancient building structures.

[0113] In summary, the system of this application overcomes the bottleneck of existing technologies in processing massive high-precision point cloud data through modular design and intelligent data value identification and differentiated caching management, and significantly improves the efficiency, quality and reliability of ancient building structure data acquisition.

[0114] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for acquiring structural data of ancient buildings using laser scanning, characterized in that, include: The key monitoring areas in the ancient building were identified, and the real-time spatial position and attitude information of the laser scanner were obtained. The scanning range of the laser scanner is calculated in real time based on the real-time spatial position and attitude information; The scanning range of the laser scanner is matched with the key monitoring area in real time to obtain the overlap relationship. The real-time spatial geometric matching calculation adopts the three-dimensional bounding box intersection algorithm. If there is an overlap, the point cloud data collected by the laser scanner will be marked as high-value data; if there is no overlap, the point cloud data collected by the laser scanner will be marked as ordinary-value data. Point cloud data marked as high-value data is written to a high-priority cache queue, and point cloud data marked as ordinary-value data is written to an ordinary-priority cache queue. Differentiated cache management and transfer strategies are applied to the data in the high-priority cache queue and the ordinary-priority cache queue.

2. The laser scanning method for acquiring structural data of ancient buildings according to claim 1, characterized in that, The step of performing real-time spatial geometric matching calculation between the scanning range of the laser scanner and the key monitoring area to obtain the overlap relationship, and the real-time spatial geometric matching calculation using a three-dimensional bounding box intersection algorithm, includes: Calculate the three-dimensional bounding box of the scanning range of the laser scanner; The three-dimensional bounding box of the laser scanner's scanning range is used to determine spatial overlap with the preset three-dimensional bounding box of each key monitoring area. If the projections of the two overlap on the X, Y, and Z coordinate axes, then there is an overlap relationship.

3. The laser scanning method for acquiring structural data of ancient buildings according to claim 1, characterized in that, The step of implementing differentiated cache management and transfer strategies for the data in the high-priority cache queue and the normal-priority cache queue includes: Grant the highest write and read permissions to the high-priority cache queue, and transfer the data in the priority queue to the main storage device; When cache space is tight, storage space is released from the ordinary priority cache queue first. The operation of releasing space includes pausing the writing of new data to the queue, overwriting the historical data in the queue, and downgrading the data in the queue to any of the auxiliary storage devices. The system dynamically monitors the occupancy status of high-priority cache queues and ordinary-priority cache queues, as well as the load on the main storage device, and adjusts the write bandwidth allocation strategy of the two queues in real time. When there is data waiting to be written in the high-priority cache queue, its write bandwidth is guaranteed first, and the write operation of the ordinary-priority cache queue is interrupted if necessary.

4. The laser scanning method for acquiring structural data of ancient buildings according to claim 1, characterized in that, The step of writing the point cloud data marked as high-value data into a high-priority cache queue includes: Real-time analysis and processing of point cloud data marked as high-value data to identify the existence of redundant point cloud clusters; If there are no redundant point cloud clusters, the point cloud data marked as high-value data will be written to a high-priority cache queue; If redundant point cloud clusters exist, compress the information and generate feature summaries for the redundant point cloud clusters to obtain refined data packets, and write the refined data packets into a high-priority cache queue.

5. The laser scanning method for acquiring structural data of ancient buildings according to claim 4, characterized in that, The step of performing real-time analysis and processing of point cloud data marked as high-value data to identify whether redundant point cloud clusters exist includes: Calculate the set of points within its preset radius neighborhood based on the point cloud data; Calculate the local three-dimensional variance and average normal vector of the point set; When the number of points exceeds a preset threshold, the 3D variance is lower than a preset variance threshold, and the angle between the normal vectors is less than a preset angle threshold, it is determined to be a redundant point cloud cluster.

6. The laser scanning method for acquiring structural data of ancient buildings according to claim 4, characterized in that, If redundant point cloud clusters exist, the steps of compressing information and generating feature summaries for the redundant point cloud clusters to obtain refined data packets include: Representative points are extracted from redundant point cloud clusters, and the representative points are either the average coordinates of all points or the original point closest to the center. Calculate and record the local geometric feature parameters of the redundant point cloud cluster to obtain an information digest. The local geometric feature parameters include the average normal vector, local curvature, bounding box size, and original number of points. Based on the representative points, information summary, and a small amount of strategically sampled point cloud data, a refined data package is obtained.

7. The laser scanning method for acquiring structural data of ancient buildings according to claim 1, characterized in that, The step of performing real-time spatial geometric matching calculations between the scanning range of the laser scanner and the key monitoring area to obtain the overlap relationship, and prior to the step of using a 3D bounding box intersection algorithm for real-time spatial geometric matching calculations, includes: The reference coordinates of the key monitoring area are updated based on the point cloud data obtained from the partial overview scan of the laser scanner to acquire the key monitoring area. Real-time spatial position and attitude information of the laser scanner is acquired to predict the scanning trajectory of the laser scanner in the future. Geometric analysis is performed between the predicted scan trajectory and the updated baseline coordinates of the key monitoring area to determine whether there is a deviation trend; If there is a deviation trend, a scanning path adjustment instruction is generated. The adjustment instruction adopts a proportional-integral-derivative control algorithm. The input is the deviation amount, and the output is the adjustment amount of the scanner motion parameters. The adjustment amount of the motion parameters includes the adjustment amount of horizontal rotation speed, the adjustment amount of pitch angle, or the adjustment amount of scanning sector range. The scanning path or scanning sector of the laser scanner is adjusted in real time according to the scanning path adjustment command.

8. The laser scanning method for acquiring structural data of ancient buildings according to claim 7, characterized in that, The step of updating the reference coordinates of the key monitoring area based on the point cloud data obtained from the partial overview scan of the key monitoring area by the laser scanner includes: The point cloud data obtained by the laser scanner partial overview scan is matched and locally registered with the digital model of the preset key monitoring area. The feature matching and local registration are implemented using the iterative nearest point algorithm. Calculate the offset of the key monitoring area relative to the initial position; Update the baseline coordinates of the key monitoring area based on the offset.

9. The laser scanning method for acquiring structural data of ancient buildings according to claim 7, characterized in that, The step of performing geometric analysis between the predicted scan trajectory and the updated reference coordinates of the key monitoring area to determine whether there is a deviation trend includes: Calculate the minimum distance between the predicted scan trajectory and the boundary of the key monitoring area; When the minimum distance is less than the preset critical deviation threshold, it is determined that there is a deviation trend.

10. A laser scanning system for acquiring structural data of ancient buildings, used to acquire structural data of ancient buildings, characterized in that, The system includes: The information acquisition module is used to acquire the key monitoring areas preset in the ancient building and to acquire the real-time spatial position and attitude information of the laser scanner. The scanning range calculation module is used to calculate the scanning range of the laser scanner in real time based on the real-time spatial position and attitude information. The overlap relationship determination module is used to perform real-time spatial geometric matching calculation between the scanning range of the laser scanner and the key monitoring area to obtain the overlap relationship. The real-time spatial geometric matching calculation adopts the three-dimensional bounding box intersection algorithm. The data labeling module is used to label the point cloud data collected by the laser scanner as high-value data if there is an overlap relationship, and to label the point cloud data collected by the laser scanner as ordinary-value data if there is no overlap relationship. The differentiated caching processing module is used to write point cloud data marked as high-value data into a high-priority cache queue and point cloud data marked as ordinary-value data into an ordinary-priority cache queue, and to perform differentiated cache management and transfer strategies on the data in the high-priority cache queue and the ordinary-priority cache queue.

Images