Intelligent detection method for building surface defects robot

Abstract

The application discloses a building surface defect robot intelligent detection method and relates to the technical field of building defect detection. The method comprises the following steps: pixel projection consistency analysis is performed on images collected by a robot for long-distance detection of high-rise buildings; in combination with the offset relationship between pixel grid distribution and spatial projection position, it is determined whether surface defect cracks and pixel grid misplacement lead to the condition that cracks are distributed in a point shape, and a crack continuity offset judgment parameter set is generated; in the case that cracks are distributed in a point shape, discrete form analysis is performed on the point crack area, and discrete structure characteristics are determined in combination with point distribution density, spatial trend consistency and adjacent interval change. The application solves the problem that cracks cannot be identified due to pixel misplacement in the point shape in long-distance detection of high-rise buildings, realizes crack reconstruction and strategy adaptive adjustment based on discrete structure characteristics, and improves detection continuity and reliability.

Description

Technical Field

[0001] This invention relates to the field of building defect detection technology, and more specifically to a robotic intelligent detection method for building surface defects. Background Technology

[0002] Intelligent robotic inspection of surface defects on high-rise buildings refers to a technological approach that utilizes a robotic platform equipped with image acquisition equipment and an intelligent processing system to automatically inspect the facades of high-rise buildings. This involves acquiring image or video data of the building surface and combining it with image processing and intelligent recognition algorithms to detect and identify defects such as cracks, peeling, and dirt. This technology typically relies on drones, wall-climbing robots, or other high-altitude operation robots as carriers. These robots collect data from multiple angles and scales along a pre-set or autonomously planned path, transmitting the raw image data to a local processing unit or a remote server for processing. The specific implementation generally includes image acquisition, data transmission, image preprocessing, feature extraction, and defect recognition. Image preprocessing primarily involves denoising, enhancing, and correcting the acquired data to improve the accuracy of subsequent analysis. The feature extraction and recognition stage analyzes the building surface condition using traditional image analysis methods or deep learning-based models to automatically identify and classify different types of defects. Simultaneously, it generates annotation information or inspection reports based on the detection results and uses a communication module to store, display, and remotely interact with the data, thus forming a complete intelligent inspection process for surface defects on high-rise buildings.

[0003] The existing technology has the following shortcomings:

[0004] When performing intelligent robotic detection of surface defects on high-rise buildings, if the robot acquires images of the building surface at a distance, surface defects and cracks may spatially misalign with the pixel grid during the imaging process. This causes the cracks to appear as discontinuous point distributions rather than continuous structures in the image. This is because the actual width of the cracks in space is close to or lower than the spatial resolution of the imaging system during long-distance imaging. As a result, the crack signal is dispersed and mapped to multiple non-continuous pixel positions during discrete sampling, thus destroying its original continuous geometric features. Existing technologies cannot adjust image reconstruction or connection strategies based on the discrete structural features of surface defects and cracks in the case of point distribution caused by misalignment between the cracks and the pixel grid during long-distance acquisition of high-rise building images. This leads to the point information of cracks being misjudged as random noise and filtered out, resulting in cracks not being effectively identified, missed detections, and reduced overall detection reliability.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a robotic intelligent detection method for surface defects in buildings, in order to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a robotic intelligent detection method for surface defects in buildings, specifically comprising the following steps: S1. Use a robot to perform pixel projection consistency analysis on images acquired from high-rise buildings at a distance. Combine the offset relationship between pixel grid distribution and spatial projection position to determine whether there are surface defects and cracks that are misaligned with the pixel grid, resulting in cracks being distributed in a point-like manner. Generate a set of crack continuity offset judgment parameters. S2. When cracks are distributed in a point-like pattern, perform discrete morphological analysis on the point-like crack region, combine the density of point distribution, spatial consistency and adjacent interval changes to determine discrete structural features, and generate a set of spatial correlation parameters. S3. Based on the set of spatial correlation parameters, the point cracks are subjected to directional constraint connection processing. The discrete points are connected segment by segment according to the spatial tendency consistency to form the initial shape of the crack trajectory. S4. Based on the initial shape of the crack trajectory and combined with the set of spatial correlation parameters, the crack trajectory is continuously corrected. The discontinuous positions of the fracture location and trajectory direction are adjusted segment by segment to form a continuous crack trajectory. S5. Combining discrete structural features with continuous crack trajectories, the crack continuity offset judgment parameter set, spatial correlation parameter set, and continuity correction parameter are adjusted in a coordinated manner, and the image reconstruction or connection strategy is adjusted according to the discrete structural features.

[0008] Preferably, S1 specifically includes the following steps: S101. Using a robot to detect and collect images of high-rise buildings from a distance, perform pixel projection consistency analysis. By establishing the correspondence between pixel grid distribution and spatial projection position, determine the position of each pixel in the pixel grid distribution and its corresponding spatial projection position. S102. Based on the correspondence between pixel grid distribution and spatial projection position, calculate the difference vector between the position coordinates of the pixel corresponding to the surface defect crack in the pixel grid distribution and the corresponding coordinates of the spatial projection position as the position offset. Calculate the change in the directional angle between adjacent pixels based on the directional distribution of the position offset. Calculate the change in the offset amplitude between adjacent pixels based on the numerical change of the position offset. When the change in the directional angle between adjacent pixels is less than a preset angle threshold and the offset amplitude shows a continuous trend, it is determined to be a continuous distribution. When the change in the directional angle between adjacent pixels is greater than or equal to a preset angle threshold or the offset amplitude shows a discrete jump, it is determined to be a situation where the surface defect crack is misaligned with the pixel grid, resulting in a point-like distribution of the crack. S103. When it is determined that there are surface defects and cracks and pixel mesh misalignment causing the cracks to be distributed in a point-like manner, the position offset, direction distribution and spatial adjacency relationship of the corresponding pixel points of the surface defects and cracks are jointly calculated to generate a set of crack continuity offset judgment parameters.

[0009] Preferably, S103 is as follows: When it is determined that there are surface defects and cracks and pixel grid misalignment causing the cracks to be distributed in a point-like manner, the position offset of the corresponding pixel points of the surface defects and cracks is grouped and processed. According to the spatial adjacency relationship, the pixel points corresponding to the position offset within the preset distance threshold range are divided into multiple adjacency sets, and the directional distribution of the position offset within each adjacency set is calculated. For the directional distribution of pixels in each adjacency set, adjacent pixel pairs are selected according to spatial adjacency relationship, the change in directional angle between adjacent pixel pairs is calculated, and the proportion of pixel pairs with directional angle changes less than a preset angle threshold is counted. At the same time, the amplitude change of position offset within the adjacency set is calculated to obtain the distribution of offset difference between adjacent pixel pairs. By combining the statistical results of the change in directional angle, the distribution of the difference in position offset, and the spatial adjacency relationship, the pixels in the adjacency set are jointly calculated. Pairs of pixels whose change in directional angle is less than a preset angle threshold and whose offset difference is continuously changing are combined to generate a set of parameters for determining the continuity offset of cracks.

[0010] Preferably, S2 specifically includes the following steps: S201. When cracks are distributed in a point-like manner, based on the crack continuity offset judgment parameter set, the position offset of the corresponding pixel points of the surface defect crack is screened, and the pixel points whose position offset direction changes are greater than or equal to the preset angle threshold or whose offset amplitude changes in a discrete jump are extracted as point-like crack regions. Discrete morphology analysis is performed on the point-like crack regions to extract the spatial position coordinates of the pixel points within the point-like crack regions. A fixed spatial range neighborhood is constructed with each pixel point as the center, the number of pixel points in the neighborhood is counted and normalized, and the normalized value is used as the point distribution density. S202. Based on the density of point distribution, the pixels in the point-like crack area are sorted according to their spatial coordinates. For the sorted adjacent pixel pairs, the corresponding position offset direction is extracted and the directional angle between adjacent pixel pairs is calculated. A set of directional angles is constructed based on the directional angles of all adjacent pixel pairs, and the directional angle set is statistically analyzed to characterize the spatial consistency. At the same time, the spatial distance between adjacent pixel pairs is calculated and the difference is calculated to characterize the change in adjacent interval. S203. Based on the density of point distribution, spatial consistency, and changes in adjacent intervals, perform joint calculations on pixels within the point-like crack area. Combine pixel pairs with directional angles less than a preset angle threshold and continuously changing adjacent intervals to determine discrete structural features. Generate a set of spatial association parameters based on the spatial connection relationship and directional angle change relationship between pixels.

[0011] Preferably, S203 specifically refers to: By combining the density of point distribution, spatial consistency, and changes in adjacent intervals, a set of pixel pairs is constructed for pixels within the point crack area according to their spatial adjacency, and the directional angle and adjacent interval changes are extracted for each pixel pair in the set. For a set of pixel pairs, pixel pairs with a direction angle less than the preset angle threshold are selected based on the comparison result between the direction angle and the preset angle threshold. In the selection result, the pixel pairs are grouped according to the continuity of the change in adjacent intervals to form a set of pixel pairs that satisfy the direction constraint and the interval change constraint. For the set of pixel pairs after grouping, each pixel pair is sequentially connected according to spatial adjacency to construct a pixel pair connection path. The directional angle distribution, adjacent interval change distribution, and connection order of pixel pairs within each connection path are encoded. The directional angle distribution data, adjacent interval change data, and connection path index are combined to generate a set of spatial association parameters.

[0012] Preferably, S3 is as follows: Based on the set of spatial correlation parameters, the connection path index, directional angle distribution data and adjacent interval change data of the pixels in the point crack area are extracted. The pixels are then grouped according to the connection path index, and the pixels in each group are sorted according to their spatial coordinates to form a discrete point sequence to be connected. For the discrete point sequence, directional constraint connection processing is implemented for point cracks. The connection direction between adjacent discrete points is filtered according to the directional angle distribution data. Adjacent discrete points with directional angles less than a preset angle threshold are retained as valid connection relationships. The valid connection relationships are connected segment by segment according to spatial tendency consistency. A continuous connection segment is formed by establishing a connection relationship between the current discrete point and the adjacent discrete points that meet the directional constraints. Around the continuous connecting segments formed by segment-by-segment connection, the connection order between the continuous connecting segments is constrained according to the adjacent interval change data. The continuous connecting segments with the interval change difference between adjacent discrete points less than the preset interval threshold are sequentially spliced ​​together, and the continuous connecting segments with the interval change difference greater than or equal to the preset interval threshold are segmented to form a continuous connecting path of point cracks and construct the initial shape of crack trajectory.

[0013] Preferably, S4 is as follows: Based on the initial shape of the crack trajectory, the spatial coordinates of each trajectory point in the crack trajectory are extracted. Combined with the directional angle distribution data and adjacent interval change data in the spatial correlation parameter set, the continuity correction process of the crack trajectory is performed. By calculating the difference between the directional angle and the adjacent interval change between adjacent trajectory points, the positions where the directional angle is greater than or equal to a preset angle threshold or the difference between adjacent interval changes is greater than or equal to a preset interval threshold are marked as the fracture position and the position where the trajectory direction change is discontinuous. For locations where the fracture location and trajectory direction are discontinuous, the continuity of the crack trajectory is corrected by combining a set of spatial correlation parameters. By selecting adjacent trajectory points on both sides of the fracture location as adjustment starting points, the connection direction between adjacent trajectory points is constrained based on the directional angle distribution data, and the trajectory points between fracture locations are adjusted segment by segment according to the spatial position relationship. Transitional connection segments are formed by inserting compensation trajectory points or adjusting the spatial position of the original trajectory points. Around the transition connection segment formed by the segmented adjustment, the crack trajectory is continuously corrected. The connection order between the transition connection segment and the original trajectory segment is constrained according to the adjacent interval change data. The trajectory segments with adjacent interval change difference less than the preset interval threshold are sequentially spliced ​​together, and the positions with adjacent interval change difference greater than or equal to the preset interval threshold are segmented to form a continuous crack trajectory.

[0014] Preferably, S5 is as follows: By combining discrete structural features with continuous crack trajectories, a correspondence is constructed between the crack continuity migration judgment parameter set, the spatial correlation parameter set, and the continuity correction parameter. By matching the point distribution density, spatial tendency consistency, and adjacent interval changes in the discrete structural features with the spatial position coordinates of each trajectory point in the continuous crack trajectory, the crack continuity migration judgment parameter set, the spatial correlation parameter set, and the continuity correction parameter are aligned. Based on the correspondence construction results, the set of parameters for determining crack continuity migration, the set of spatial correlation parameters, and the continuity correction parameters are adjusted in a coordinated manner. By synchronously traversing each parameter according to the spatial position order, and based on the difference between the direction distribution change and the interval change in the discrete structural features, the direction angle data, interval change data, and trajectory correction data in each parameter are adjusted in a consistent manner to form the parameter set after the coordinated adjustment. By combining discrete structural features, the parameter set after linkage adjustment is constrained. Based on the density of point distribution, spatial consistency, and changes in adjacent intervals, the image reconstruction or connection strategy is adjusted. By reconstructing the connection path, connection order, and trajectory construction process of pixels within the crack area, the robot detection of surface defects of high-rise buildings is completed.

[0015] The technical effects and advantages provided by the present invention in the above technical solution are as follows: 1. This invention introduces pixel projection consistency analysis and crack continuity offset determination mechanisms into images acquired from long-distance detection of high-altitude buildings, achieving accurate identification of crack and pixel grid misalignment phenomena. It effectively separates crack information, which is originally presented as discrete point distribution during imaging, from noise interference. By constructing multi-dimensional discrete structural features such as point distribution density, spatial consistency, and adjacent interval changes through discrete morphology analysis, disordered discrete points are transformed into a structured expression with spatial correlation. On this basis, further structural reconstruction of point cracks is achieved through directional constraint connection and segmented connection processing, restoring the crack from a discrete state to a trajectory shape with continuous geometric features, thus improving the ability to identify and express minute cracks under long-distance imaging conditions.

[0016] 2. After constructing the initial shape of the crack trajectory, this invention refines the direction abrupt changes and interval anomalies in the trajectory through continuity correction processing and segmented adjustment mechanism of fracture position. Furthermore, by combining discrete structural features with continuous crack trajectories, it links and adjusts the crack continuity offset judgment parameter set, spatial correlation parameter set, and continuity correction parameter, thereby achieving dynamic coupling control between the parameter layer and the structural layer. On this basis, it drives the adaptive adjustment of image reconstruction or connection strategy, so that the connection path and connection sequence can match the actual spatial distribution characteristics of the crack in real time, avoiding the misjudgment of point cracks as random noise and filtering them out. Overall, it improves the integrity, continuity, and reliability of crack detection, while enhancing the robustness and adaptability in complex imaging environments. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0018] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0019] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0020] This invention provides, for example Figure 1 The intelligent robotic detection method for building surface defects shown includes the following steps: S1. Use a robot to perform pixel projection consistency analysis on images acquired from high-rise buildings at a distance. Combine the offset relationship between pixel grid distribution and spatial projection position to determine whether there are surface defects and cracks that are misaligned with the pixel grid, resulting in cracks being distributed in a point-like manner. Generate a set of crack continuity offset judgment parameters. In this embodiment, S1 specifically includes the following steps: S101. Using a robot to detect and collect images of high-rise buildings from a distance, perform pixel projection consistency analysis. By establishing the correspondence between pixel grid distribution and spatial projection position, determine the position of each pixel in the pixel grid distribution and its corresponding spatial projection position. In the process of intelligent robotic inspection of surface defects of high-rise buildings, when using a robot equipped with imaging equipment to collect images of high-rise buildings from a distance, the mapping relationship between pixel coordinates and actual spatial positions can be established by calibrating the internal and external parameters of the imaging equipment, thereby achieving pixel projection consistency analysis. In practice, the process begins by obtaining the focal length, principal point position, and distortion parameters of the imaging device through camera calibration. This, combined with the robot's pose information during the detection process, determines the spatial observation position corresponding to each frame. The image is then divided into a regular pixel grid distribution. For example, a 1920×1080 resolution image is considered a two-dimensional grid composed of equally spaced pixel units, and the row and column positions of each pixel unit in the image are recorded. Based on this, each pixel in the pixel grid distribution is mapped to its spatial projection position on the surface of a high-altitude building according to imaging geometry. For instance, in a detection scenario 30 meters from the building surface, a crack several centimeters wide may correspond to multiple discrete pixels in the image. By analyzing the distribution of these pixels in the image and their corresponding spatial projection positions, the offset relationship between the pixel grid distribution and the spatial projection position can be calculated. Furthermore, by comparing consecutive frames, it can be observed whether the distribution of cracks at the pixel level exhibits a continuous changing trend or shows discrete jumps, thus providing a basis for subsequent judgment on whether the cracks are distributed in a point-like manner due to sampling.

[0021] The robot is used to carry imaging equipment and perform inspection tasks on high-altitude buildings. High-altitude buildings refer to the exterior structures of buildings located in high-altitude environments. Long-distance detection and image acquisition means that image data is acquired under conditions where there is a large spatial distance between the imaging equipment and the building surface. Pixel grid distribution refers to the set of pixels in the image arranged at fixed intervals. Spatial projection position refers to the corresponding position of each pixel on the actual building surface through the imaging relationship. Pixel projection consistency analysis refers to judging whether there is an image offset by analyzing whether the correspondence between the pixel grid distribution and the spatial projection position remains stable. The correspondence between the pixel grid distribution and the spatial projection position is used to describe the mapping relationship between the position of the pixel in the image and the actual spatial position. When the correspondence changes inconsistently in a local area, it indicates that there may be a misalignment between the crack structure and the pixel grid in the corresponding area of ​​the image. By analyzing this correspondence, basic data support can be provided for identifying the discrete distribution state of cracks.

[0022] S102. Based on the correspondence between pixel grid distribution and spatial projection position, calculate the difference vector between the position coordinates of the pixel corresponding to the surface defect crack in the pixel grid distribution and the corresponding coordinates of the spatial projection position as the position offset. Calculate the change in the directional angle between adjacent pixels based on the directional distribution of the position offset. Calculate the change in the offset amplitude between adjacent pixels based on the numerical change of the position offset. When the change in the directional angle between adjacent pixels is less than a preset angle threshold and the offset amplitude shows a continuous trend, it is determined to be a continuous distribution. When the change in the directional angle between adjacent pixels is greater than or equal to a preset angle threshold or the offset amplitude shows a discrete jump, it is determined to be a situation where the surface defect crack is misaligned with the pixel grid, resulting in a point-like distribution of the crack. Based on the correspondence between pixel grid distribution and spatial projection positions, a pixel-by-pixel analysis is performed on the identified surface defect crack regions. The position coordinates of each crack pixel in the image are paired with its corresponding spatial projection position. A difference vector is formed by calculating the spatial offset between the two, and the length and direction of this difference vector are used as the basic data for the position offset. Furthermore, the pixels continuously distributed along the crack path are sorted. For example, in a set of actual detection data, a crack corresponds to a set of pixels in the image, whose pixel coordinates change row by row along the vertical axis of the image. The corresponding spatial projection positions are approximately linearly distributed on the actual building surface. By comparing the direction of the difference vectors of adjacent pixels pair by pair, the direction can be calculated. For angular variation, when the directional angle changes of ten consecutive pixels in the detection data remain within a small range and the offset amplitude gradually increases or decreases, it can be determined that the crack is continuously distributed. Conversely, in another set of long-distance detection data, because the crack width is lower than the imaging resolution, the crack appears as discrete points in the image. The calculated difference vector between adjacent pixels shows obvious fluctuations in direction, and the offset amplitude shows irregular jumps between adjacent points. By statistically analyzing these changes and comparing them with a preset angle threshold, when the directional angle change exceeds the threshold or the offset amplitude change shows discontinuous fluctuations, it can be determined that the crack is distributed in a point-like manner under the current imaging conditions, thus distinguishing between continuous cracks and discrete cracks, providing a basis for subsequent processing.

[0023] Surface defect cracks refer to elongated structural anomalies existing on the outer surface of high-rise buildings. The position coordinates in the pixel grid distribution refer to the row and column positions of the crack pixels in the image. The spatial projection position coordinates refer to the spatial position of the crack pixels mapped onto the actual building surface. The difference vector is used to describe the offset relationship between the pixel position and the spatial projection position. The position offset reflects the direction and magnitude information of the difference vector. The directional distribution of the position offset reflects the extension direction of the crack in space. The change in the directional angle between adjacent pixels is used to describe the degree of change of the crack direction within a local range. The numerical change of the position offset is used to characterize the continuous change of the crack in space. The change in the offset amplitude between adjacent pixels reflects the stability of the crack in space. The preset angle threshold is used to limit the judgment range of directional change. A continuous trend in the offset amplitude indicates that the crack has stable extension characteristics in space. A continuous distribution indicates that the crack maintains a coherent shape in the image. A discrete jump in the offset amplitude indicates that the crack has irregular changes during spatial mapping. The misalignment of the surface defect crack with the pixel grid, resulting in a point-like distribution of the crack, indicates a discrete expression state of the crack due to sampling limitations during the imaging process.

[0024] S103. When it is determined that there are surface defects and cracks and pixel mesh misalignment causing the cracks to be distributed in a point-like manner, the position offset, direction distribution and spatial adjacency relationship of the corresponding pixel points of the surface defects and cracks are jointly calculated to generate a set of crack continuity offset judgment parameters.

[0025] In this embodiment, S103 specifically refers to: When it is determined that there are surface defects and cracks and pixel grid misalignment causing the cracks to be distributed in a point-like manner, the position offset of the corresponding pixel points of the surface defects and cracks is grouped and processed. According to the spatial adjacency relationship, the pixel points corresponding to the position offset within the preset distance threshold range are divided into multiple adjacency sets, and the directional distribution of the position offset within each adjacency set is calculated. For the directional distribution of pixels in each adjacency set, adjacent pixel pairs are selected according to spatial adjacency relationship, the change in directional angle between adjacent pixel pairs is calculated, and the proportion of pixel pairs with directional angle changes less than a preset angle threshold is counted. At the same time, the amplitude change of position offset within the adjacency set is calculated to obtain the distribution of offset difference between adjacent pixel pairs. By combining the statistical results of the change in directional angle, the distribution of the difference in position offset, and the spatial adjacency relationship, the pixels in the adjacency set are jointly calculated. Pairs of pixels whose change in directional angle is less than a preset angle threshold and whose offset difference is continuously changing are combined to generate a set of parameters for determining the continuity offset of cracks.

[0026] In the process of intelligent robotic detection of surface defects on high-rise buildings, after identifying cracks as point-like distributions, the positional offsets of the corresponding pixels can be grouped. Specifically, the positional offset of each pixel is first mapped to a unified spatial coordinate system, and a connection graph is constructed according to the spatial adjacency relationship between pixels. By traversing point by point, pixels that meet the spatial distance constraints in the neighborhood are searched starting from the current pixel. Pixels that meet the spatial distance constraints are grouped into the same set. This process is repeated until all pixels are divided, thus forming multiple adjacency sets. Within each adjacency set, the direction of the positional offset of all pixels is normalized, and the distribution density of each direction in the set is statistically analyzed. By comparing the concentration of the directional distribution density, the dominant direction is determined, thereby obtaining the directional distribution characteristics of the positional offset. This process can divide discrete crack points into regions with local consistency.

[0027] After completing the adjacency set partitioning, further comparative analysis of the pixels within the adjacency set can be performed. Specifically, a list of pixel pairs is established according to spatial adjacency relationships. The positional offset direction of each pair of pixels is extracted, and the directional difference between the two is calculated. By traversing all pixel pairs, the directional differences are classified and statistically analyzed according to a preset angle range to obtain the proportion of pixel pairs with consistent directions. At the same time, the positional offsets corresponding to the pixels are compared pair by pair, and the trend of offset change between adjacent pixels is recorded. The offset changes are classified according to increasing, decreasing, and irregular changes, and offset difference distribution data is formed. Through this process, both directional change characteristics and amplitude change characteristics can be obtained simultaneously, thus providing dual evidence for crack continuity analysis.

[0028] After obtaining the statistical results of the direction change and the distribution of the offset difference, the pixels in the adjacent set can be jointly processed. Specifically, the pixel pairs with stable direction changes and continuous offset changes are filtered, and the filtered pixels are connected segment by segment according to the spatial adjacency order to form a point set path. At the same time, the direction distribution information, offset change information and adjacency relationship information of each point set path are recorded. This information is organized and stored in a unified manner to construct a set of parameters for determining the continuity offset of the crack. This process realizes the structured transformation from discrete points to continuous paths, making the crack features traceable.

[0029] The position offset of the pixel corresponding to the surface defect crack represents the spatial difference between the pixel's position in the pixel grid distribution and its spatial projection position. Grouping processing means dividing the position offsets into sets according to spatial adjacency relationships. Spatial adjacency relationships represent the proximity connection between pixels in the spatial coordinate system. The preset distance threshold range is used to limit the range of adjacency relationship judgment. Position offsets that meet the preset distance threshold range indicate that there is spatial correlation between pixels. Multiple adjacency sets represent multiple local regions formed by the crack in space. The directional distribution of the position offsets represents the degree of concentration of the offset direction in space.

[0030] Adjacent pixel pairs represent combinations of pixels that are directly connected in spatial adjacency. The change in the directional angle between adjacent pixel pairs is used to characterize the degree of directional change. A preset angle threshold is used to distinguish between directional consistency and directional change states. The proportion of pixel pairs represents the proportion of directional consistency in the whole. The amplitude change calculation represents the analysis process of the trend of positional offset change. The distribution of offset difference between adjacent pixel pairs represents the spatial distribution of amplitude change. A continuous change in offset difference indicates that the offset change has a consistent trend. The crack continuity offset judgment parameter set represents a data set composed of directional distribution, amplitude change, and spatial adjacency, used to describe the spatial characteristics of cracks transforming from a discrete state to a continuous state.

[0031] S2. When cracks are distributed in a point-like pattern, perform discrete morphological analysis on the point-like crack region, combine the density of point distribution, spatial consistency and adjacent interval changes to determine discrete structural features, and generate a set of spatial correlation parameters. In this embodiment, S2 specifically includes the following steps: S201. When cracks are distributed in a point-like manner, based on the crack continuity offset judgment parameter set, the position offset of the corresponding pixel points of the surface defect crack is screened, and the pixel points whose position offset direction changes are greater than or equal to the preset angle threshold or whose offset amplitude changes in a discrete jump are extracted as point-like crack regions. Discrete morphology analysis is performed on the point-like crack regions to extract the spatial position coordinates of the pixel points within the point-like crack regions. A fixed spatial range neighborhood is constructed with each pixel point as the center, the number of pixel points in the neighborhood is counted and normalized, and the normalized value is used as the point distribution density. When a point-like distribution trend of cracks is identified through preprocessing, the positional offset of the corresponding pixels of the surface defect cracks can be filtered based on the crack continuity offset judgment parameter set. Specifically, the positional offset of each pixel is used as input data, and its directional change and offset amplitude change are compared point by point. By comparing with a preset angle threshold, pixels with large directional changes and pixels with discontinuous offset amplitude changes are extracted and regarded as point-like crack regions. Then, discrete morphology analysis is performed on the region, specifically, the spatial coordinates of all relevant pixels are extracted, and a local neighborhood region is constructed in space with each pixel as the center. The number of pixels contained in the neighborhood is counted, and the statistical results are normalized to obtain a value reflecting the degree of local point aggregation. For example, in an actual detection, there is a crack region in the image acquired at a distance, and its pixels are scattered in space. Through neighborhood statistics, it is found that the number of pixels in some areas is significantly higher than that in the surrounding areas. After normalization, the density value of the corresponding area is higher, while the value of the sparse area is lower. This can effectively distinguish the spatial distribution differences of crack points and provide a data foundation for subsequent structural analysis.

[0032] The set of parameters for determining crack continuity offset describes the continuous changes of cracks during spatial mapping. The position offset of the pixel corresponding to the surface defect crack represents the spatial offset relationship between the pixel's position in the pixel grid and its spatial projection position. The preset angle threshold is used to limit the judgment range of directional changes. Discrete jumps in the offset amplitude indicate that the offset changes between adjacent pixels lack continuity. The point-like crack region represents the set of pixels with discrete distribution characteristics extracted under the screening conditions. Discrete morphology analysis represents the process of spatial structure analysis of this set of pixels. The fixed spatial range neighborhood represents the local analysis area defined in space with a single pixel as the center. Normalization processing represents the process of mapping the statistical results of different neighborhoods to a unified scale range. The normalized value is used to characterize the relative relationship between different regions. The density of point distribution represents the aggregation of pixels within a unit space range, which reflects the distribution characteristics of crack points in space.

[0033] S202. Based on the density of point distribution, the pixels in the point-like crack area are sorted according to their spatial coordinates. For the sorted adjacent pixel pairs, the corresponding position offset direction is extracted and the directional angle between adjacent pixel pairs is calculated. A set of directional angles is constructed based on the directional angles of all adjacent pixel pairs, and the directional angle set is statistically analyzed to characterize the spatial consistency. At the same time, the spatial distance between adjacent pixel pairs is calculated and the difference is calculated to characterize the change in adjacent interval. After obtaining the density of the point distribution, the pixels within the point-like crack region can be sorted according to their spatial coordinates. Specifically, a main direction can be selected as the sorting reference direction, for example, sorting them in ascending order based on the projection values ​​of their spatial positions onto a coordinate axis, thus forming an ordered sequence from the originally disordered pixels. Then, for each sorted pair of adjacent pixels, the positional offset direction of each pixel is extracted, and the directions of adjacent pixel pairs are compared one by one. The angle between the directions is calculated to reflect the change in direction. Simultaneously, the angles of all adjacent pixel pairs are uniformly recorded and a set of direction angles is constructed. The dataset is subjected to statistical analysis, such as the range of the concentrated distribution of the statistical direction angle and the proportion of deviation from the range, to characterize the overall consistency of the direction. At the same time, the spatial distance between adjacent pixel pairs after sorting is calculated, and the difference between adjacent distances is calculated for each pair to reflect the change in adjacent intervals. For example, in actual detection, a group of crack points forms an approximately linear arrangement after sorting, and the distance between adjacent points changes relatively smoothly with small difference changes. However, in the discrete point region, the distance between adjacent points shows obvious fluctuations. This comparison can distinguish between continuous cracks and discrete cracks, thus providing a basis for subsequent structural feature analysis.

[0034] The sorted adjacent pixel pairs represent the combination of two adjacent pixels after spatial sorting. The positional offset direction represents the directional information of the pixel during spatial offset. The directional angle between adjacent pixel pairs is used to characterize the degree of difference between the two directions. The directional angle set represents the data set composed of the directional angles of all adjacent pixel pairs. Statistical analysis represents the process of analyzing the concentration and dispersion of the data distribution in this set. Spatial tendency consistency indicates whether the overall distribution of directional angles is concentrated within a certain range. The spatial distance between adjacent pixel pairs represents the distance value between two adjacent pixels in spatial coordinates. Difference calculation represents the calculation of the change in adjacent distances. Adjacent interval change represents the distribution characteristics of these distance changes in space, which is used to reflect the continuity and stability of the crack points in space.

[0035] S203. Based on the density of point distribution, spatial consistency, and changes in adjacent intervals, perform joint calculations on pixels within the point-like crack area. Combine pixel pairs with directional angles less than a preset angle threshold and continuously changing adjacent intervals to determine discrete structural features. Generate a set of spatial association parameters based on the spatial connection relationship and directional angle change relationship between pixels.

[0036] In this embodiment, S203 specifically refers to: By combining the density of point distribution, spatial consistency, and changes in adjacent intervals, a set of pixel pairs is constructed for pixels within the point crack area according to their spatial adjacency, and the directional angle and adjacent interval changes are extracted for each pixel pair in the set. For a set of pixel pairs, pixel pairs with a direction angle less than the preset angle threshold are selected based on the comparison result between the direction angle and the preset angle threshold. In the selection result, the pixel pairs are grouped according to the continuity of the change in adjacent intervals to form a set of pixel pairs that satisfy the direction constraint and the interval change constraint. For the set of pixel pairs after grouping, each pixel pair is sequentially connected according to spatial adjacency to construct a pixel pair connection path. The directional angle distribution, adjacent interval change distribution, and connection order of pixel pairs within each connection path are encoded. The directional angle distribution data, adjacent interval change data, and connection path index are combined to generate a set of spatial association parameters.

[0037] After obtaining the density of point distribution, spatial consistency, and changes in adjacent intervals, a set of pixel pairs can be constructed for pixels within the point-like crack area according to their spatial adjacency. Specifically, with each pixel as the spatial center, a neighborhood search is performed in the spatial coordinates according to a preset adjacency search radius. All pixels within the search range are connected to the center pixel, and all connections are traversed. Pixels that meet the adjacency conditions are paired to form a set of pixel pairs. During the construction of pixel pairs, the corresponding positional offset direction is extracted for each pixel pair, and the directional angle is calculated based on the difference between the two offset directions. At the same time, the spatial distance between the two pixels is recorded, and the changes in adjacent intervals are calculated based on the sorting relationship, thus forming a set of pixel pairs containing directional information and interval change information. This process realizes the structural transformation from a single point to an associated unit.

[0038] After constructing the set of pixel pairs, the set can be filtered and grouped. Specifically, the set of pixel pairs is first traversed, and the directional angle of each pixel pair is compared with a preset angle threshold. Pixel pairs with directional angles within the threshold range are retained, while those exceeding the threshold range are removed. Then, the retained pixel pairs are arranged in spatial order, and the adjacent interval changes of the arranged pixel pairs are compared sequentially. Pixel pairs with adjacent interval changes within a continuous range are grouped into the same group, while pixel pairs with significant fluctuations in interval changes are grouped into different groups. This forms multiple sets of pixel pairs that satisfy both directional and interval change constraints. This process achieves stable structural filtering of discrete point pairs.

[0039] After grouping, a connection path construction process can be performed on the pixel pairs within each group. Specifically, starting from any pixel pair within the group, adjacent pixel pairs with common pixels are found based on spatial adjacency, and then connected one by one according to the spatial extension direction to form a continuous pixel pair connection path. During the path construction process, the pixel pairs in each path are numbered according to the connection order, and the direction angle and adjacent interval change data of each pixel pair are recorded simultaneously. At the same time, the distribution statistics of the direction angle and adjacent interval change are performed on the entire path, and the path order information and statistical information are uniformly encoded. The distribution of direction angle, interval change distribution and path order encoding are integrated into structured data, thereby generating a set of spatial association parameters. This process realizes the transformation from point pair structure to path structure.

[0040] The density of pixel distribution is used to characterize the aggregation state of pixels within a local spatial range. Spatial tendency consistency is used to characterize whether the extension direction of pixels in space is concentrated. Adjacent interval variation is used to characterize whether the distance variation between pixels has a stable trend. Spatial adjacency relationship represents the relationship between pixels in spatial coordinates that satisfy the neighborhood connection condition. Pixel pair set represents the basic analysis unit formed by the combination of pixels that satisfy the adjacency relationship. Directional angle and adjacent interval variation data are used to characterize the directional and distance relationships between pixel pairs. Preset angle threshold is used to limit the range of directional variation. Continuity of adjacent interval variation is used to characterize whether the distance variation remains stable. Grouping processing is used to structurally divide pixel pairs. Directional constraint is used to limit the range of directional variation. Interval variation constraint is used to limit the stability of distance variation.

[0041] The pixel-pair connection path represents a continuous structure formed by connecting multiple pixel pairs according to spatial adjacency. The directional angle distribution within the connection path represents the overall distribution of directional changes in the path, the adjacent interval change distribution represents the overall distance change in the path, and the connection order represents the arrangement order of pixel pairs in the path. The encoding process represents the transformation of the directional angle distribution, the interval change distribution, and the connection order into a unified data representation. The directional angle distribution data represents the path directional change characteristics, the adjacent interval change data represents the path distance change characteristics, the connection path index is used to distinguish different path structures, and the spatial association parameter set represents a data set composed of path structure information, direction information, and interval change information, used to describe the structural relationship of cracks in space.

[0042] S3. Based on the set of spatial correlation parameters, the point cracks are subjected to directional constraint connection processing. The discrete points are connected segment by segment according to the spatial tendency consistency to form the initial shape of the crack trajectory. In this embodiment, S3 specifically refers to: Based on the set of spatial correlation parameters, the connection path index, directional angle distribution data and adjacent interval change data of the pixels in the point crack area are extracted. The pixels are then grouped according to the connection path index, and the pixels in each group are sorted according to their spatial coordinates to form a discrete point sequence to be connected. For the discrete point sequence, directional constraint connection processing is implemented for point cracks. The connection direction between adjacent discrete points is filtered according to the directional angle distribution data. Adjacent discrete points with directional angles less than a preset angle threshold are retained as valid connection relationships. The valid connection relationships are connected segment by segment according to spatial tendency consistency. A continuous connection segment is formed by establishing a connection relationship between the current discrete point and the adjacent discrete points that meet the directional constraints. Around the continuous connecting segments formed by segment-by-segment connection, the connection order between the continuous connecting segments is constrained according to the adjacent interval change data. The continuous connecting segments with the interval change difference between adjacent discrete points less than the preset interval threshold are sequentially spliced ​​together, and the continuous connecting segments with the interval change difference greater than or equal to the preset interval threshold are segmented to form a continuous connecting path of point cracks and construct the initial shape of crack trajectory.

[0043] In the process of intelligent robotic detection of surface defects in high-rise buildings, the pixels within the point-like crack area can first be organized based on the set of spatial correlation parameters. Specifically, the connection path index in the set of spatial correlation parameters is used as the grouping basis to mark all pixels and divide them into multiple pixel sets according to the same connection path index. Within each pixel set, the pixels are sorted according to the projection value of the spatial position coordinates in the main direction, reconstructing the discretely distributed pixels into an ordered arrangement structure. During the sorting process, the arrangement order is determined by comparing the spatial position differences of adjacent pixels point by point, so that the pixels form a unidirectional sequence along the spatial extension direction. In the actual detection scenario, after the discrete points of a certain crack area are divided by the path index, they exhibit the characteristic of being gradually arranged along a certain direction in each set, thus forming a sequence of discrete points to be connected.

[0044] After constructing the discrete point sequence, directional constraint connection processing can be applied to the point cracks. Specifically, the current discrete point is selected from the starting position of the discrete point sequence, the positional offset direction between the current discrete point and its adjacent candidate discrete points is extracted, and the corresponding directional angle distribution data is obtained from the spatial association parameter set. The directional angle between the current discrete point and the candidate discrete points is compared with a preset angle threshold, and only candidate discrete points with directional angles less than the preset angle threshold are retained as adjacent discrete points that satisfy the directional constraint. Then, starting from the current discrete point, adjacent discrete points that satisfy the directional constraint are selected one by one according to the spatial position order for connection, and the connected discrete points are used as the new current discrete points to continue the connection expansion, thereby forming a continuous connection segment that extends gradually along a unified direction. This process ensures that the connection process proceeds along a stable direction.

[0045] After obtaining the continuous connection segments, connection order constraint processing can be performed around them. Specifically, the spatial distance between adjacent discrete points within the continuous connection segment is extracted, and the difference in adjacent interval changes is calculated according to the connection order. By comparing the relationship between the difference in interval changes and a preset interval threshold segment by segment, adjacent connection segments with an interval change difference less than the preset interval threshold are spliced ​​together according to their spatial position. The positions with an interval change difference greater than or equal to the preset interval threshold are used as segmentation boundaries, dividing the continuous connection segment into multiple sub-paths. Subsequently, each sub-path is reordered according to its spatial position and connected sequentially to form a complete path structure. In actual detection, when there is a sudden change in local intervals in the crack area, the corresponding position is divided into different paths, thereby avoiding erroneous connections that cross abnormal intervals.

[0046] After completing the path splicing and segmentation, trajectory construction can be performed on each path. Specifically, the coordinates of all discrete points in the path are integrated according to the connection order, and the spatial position of the discrete points in the path is interpolated to transform the path from a discrete point sequence into a continuous trajectory expression. At the same time, the overall consistency of the path direction is corrected to ensure that the trajectory extends smoothly in the local area, thereby forming the initial shape of the crack trajectory. This trajectory can reflect the basic direction and structural characteristics of the crack in space.

[0047] The spatial association parameter set is used to record the path affiliation, direction change characteristics, and interval change characteristics between pixels. The dotted crack region represents the set of discrete pixels after screening. The connection path index is used to identify the connection path category to which the pixel belongs. The direction angle distribution data is used to describe the range of direction change between pixels. The adjacent interval change data is used to describe the distance change between pixels. The grouping process represents the division of the pixel set according to the path index. The discrete point sequence to be connected represents the sorted set of pixels. The direction constraint connection process represents the process of screening connection relationships based on the direction angle. The preset angle threshold is used to limit the range of direction change.

[0048] Adjacent discrete points represent combinations of pixels arranged adjacently in a sorted sequence; effective connection relationships represent pixel connection relationships that satisfy directional constraints; spatial tendency consistency represents the characteristic that discrete points maintain a consistent extension direction in space; segmented connection represents the process of gradually establishing connection relationships in sequence; the current discrete point represents the pixel being processed during the connection process; adjacent discrete points that satisfy directional constraints represent candidate connection points after directional filtering; continuous connection segments represent continuous structures composed of multiple connection relationships; constraint processing represents the process of restricting the connection order and connection range; the difference in interval variation between adjacent discrete points is used to characterize the magnitude of distance variation; a preset interval threshold is used to divide the boundaries between connections and segments; the continuous connection path of point-like cracks represents the path structure formed by connection processing; and the initial shape of the crack trajectory represents the spatial expression of cracks formed through path integration.

[0049] S4. Based on the initial shape of the crack trajectory and combined with the set of spatial correlation parameters, the crack trajectory is continuously corrected. The discontinuous positions of the fracture location and trajectory direction are adjusted segment by segment to form a continuous crack trajectory. In this embodiment, S4 specifically refers to: Based on the initial shape of the crack trajectory, the spatial coordinates of each trajectory point in the crack trajectory are extracted. Combined with the directional angle distribution data and adjacent interval change data in the spatial correlation parameter set, the continuity correction process of the crack trajectory is performed. By calculating the difference between the directional angle and the adjacent interval change between adjacent trajectory points, the positions where the directional angle is greater than or equal to a preset angle threshold or the difference between adjacent interval changes is greater than or equal to a preset interval threshold are marked as the fracture position and the position where the trajectory direction change is discontinuous. For locations where the fracture location and trajectory direction are discontinuous, the continuity of the crack trajectory is corrected by combining a set of spatial correlation parameters. By selecting adjacent trajectory points on both sides of the fracture location as adjustment starting points, the connection direction between adjacent trajectory points is constrained based on the directional angle distribution data, and the trajectory points between fracture locations are adjusted segment by segment according to the spatial position relationship. Transitional connection segments are formed by inserting compensation trajectory points or adjusting the spatial position of the original trajectory points. Around the transition connection segment formed by the segmented adjustment, the crack trajectory is continuously corrected. The connection order between the transition connection segment and the original trajectory segment is constrained according to the adjacent interval change data. The trajectory segments with adjacent interval change difference less than the preset interval threshold are sequentially spliced ​​together, and the positions with adjacent interval change difference greater than or equal to the preset interval threshold are segmented to form a continuous crack trajectory.

[0050] After the initial shape of the crack trajectory is formed, the trajectory can be represented as a set of trajectory points arranged in spatial order. The spatial coordinates of each trajectory point are extracted point by point. Based on the spatial relationship between adjacent trajectory points, the change in connection direction and the difference in adjacent interval change are calculated. By traversing all adjacent trajectory points, the positions where the direction change exceeds the limit or the interval change shows abnormal fluctuations are marked. In actual detection scenarios, an initial trajectory formed by connecting discrete points may have obvious turns or sudden increases in interval in local areas. By analyzing the trajectory point sequence segment by segment, these abnormal positions can be accurately located, thereby determining the fracture position and the discontinuity of trajectory direction change, providing a positioning basis for subsequent correction processing.

[0051] For identified fracture locations and discontinuous trajectory direction changes, adjacent trajectory points on both sides of the fracture location can be used as adjustment starting points. Combining this with the directional change distribution information recorded in the spatial correlation parameter set, the connection directions between adjustment starting points are screened. During the screening process, candidate connection directions are compared one by one, retaining only connection paths where the directional change is within a stable range. After determining the connection direction, the trajectory points within the fracture area are adjusted segment by segment according to spatial relationships. Specifically, adjacent trajectory points are searched point by point along the connection direction, and trajectory points that meet the directional constraints are connected sequentially. When the original trajectory points cannot meet the connection conditions, compensation trajectory points are inserted between adjacent trajectory points. The position of the compensation trajectory points is determined by interpolation based on the spatial positions of the trajectory points on both sides, thus forming a continuous transition structure.

[0052] After the transition connection segment is formed, the connection order of the entire trajectory can be reorganized. By extracting the difference in the adjacent interval changes between the transition connection segment and the original trajectory segment, the connection relationship between each connection segment is judged segment by segment. Connection segments with stable interval changes are directly spliced ​​in spatial order. The positions where the interval changes show obvious deviations are used as path boundary points for segmentation. During the path reconstruction process, each path segment is reordered according to its spatial position and the connection relationship is re-established. In actual data processing, the change in the trajectory point spacing in continuous crack areas is relatively smooth, and the complete structure can be restored by sequential splicing. However, in areas with abnormal intervals, segmentation can avoid cross-regional misconnection.

[0053] After completing path splicing and segmentation, trajectory continuity processing can be performed on each path. The trajectory points in the path are organized in a unified manner according to the connection order, and the spatial positions between adjacent trajectory points are interpolated segment by segment to form a continuous connection between discrete trajectory points. At the same time, the trajectory direction is locally smoothed to make the trajectory appear as a continuous extension in space. After this processing, the original trajectory with breaks or abrupt changes in direction can be transformed into a continuous crack trajectory with consistent direction, thus completing the transformation of the initial form of the crack trajectory into a continuous crack trajectory.

[0054] The initial shape of the crack trajectory represents the initial trajectory structure formed by connecting discrete points. The spatial coordinates of each trajectory point in the crack trajectory are used to describe the spatial distribution of the trajectory. The continuity correction process represents the process of reconstructing discontinuous parts of the trajectory. The directional angle greater than or equal to a preset angle threshold is used to identify positions where the directional change exceeds the limit range. The difference between adjacent intervals greater than or equal to a preset interval threshold is used to identify positions where the distance changes abnormally. The break position represents the area in the trajectory where the connection is interrupted. The position where the trajectory direction changes discontinuously represents the area where the trajectory direction changes abruptly. The adjacent trajectory points on both sides of the break position represent the boundary points of the correction process. The adjustment starting point represents the starting position of the correction operation. Inserting compensation trajectory points represents the trajectory connection by adding new points. The spatial position of the original trajectory points represents the spatial coordinates of the existing trajectory points. The transition connection segment represents the transition structure formed by connecting the broken areas. The connection order between the transition connection segment and the original trajectory segment represents the arrangement relationship between different trajectory segments. The trajectory segment where the difference between adjacent intervals is less than a preset interval threshold represents the area that can be directly spliced. The position where the difference between adjacent intervals is greater than or equal to a preset interval threshold represents the area that needs to be segmented. The continuous crack trajectory represents the complete crack trajectory structure formed after correction.

[0055] S5. Combining discrete structural features with continuous crack trajectories, the set of crack continuity offset judgment parameters, the set of spatial correlation parameters, and the continuity correction parameters are adjusted in a coordinated manner. Based on the discrete structural features, the image reconstruction or connection strategy is adjusted to complete the robot detection of surface defects of high-rise buildings.

[0056] In this embodiment, S5 specifically refers to: By combining discrete structural features with continuous crack trajectories, a correspondence is constructed between the crack continuity migration judgment parameter set, the spatial correlation parameter set, and the continuity correction parameter. By matching the point distribution density, spatial tendency consistency, and adjacent interval changes in the discrete structural features with the spatial position coordinates of each trajectory point in the continuous crack trajectory, the crack continuity migration judgment parameter set, the spatial correlation parameter set, and the continuity correction parameter are aligned. Based on the correspondence construction results, the set of parameters for determining crack continuity migration, the set of spatial correlation parameters, and the continuity correction parameters are adjusted in a coordinated manner. By synchronously traversing each parameter according to the spatial position order, and based on the difference between the direction distribution change and the interval change in the discrete structural features, the direction angle data, interval change data, and trajectory correction data in each parameter are adjusted in a consistent manner to form the parameter set after the coordinated adjustment. By combining discrete structural features, the parameter set after linkage adjustment is constrained. Based on the density of point distribution, spatial consistency, and changes in adjacent intervals, the image reconstruction or connection strategy is adjusted. By reconstructing the connection path, connection order, and trajectory construction process of pixels within the crack area, the robot detection of surface defects of high-rise buildings is completed.

[0057] After the discrete structural features and continuous crack trajectories are formed, point-by-point matching processing can be performed on the two types of data based on unified spatial coordinates. Specifically, the spatial position of each trajectory point in the continuous crack trajectory is used as the index reference. Within the corresponding spatial range, the density of point distribution, spatial consistency, and adjacent interval changes in the discrete structural features are searched. The searched data are mapped to the corresponding trajectory points according to spatial adjacency relationships. The crack continuity offset judgment parameter set, spatial association parameter set, and continuity correction parameter are synchronously rearranged so that each parameter has a consistent spatial index position in the trajectory point sequence. In actual detection, the crack trajectory extends along the building facade, and the discrete point distribution characteristics corresponding to different positions are different. Through this mapping process, each trajectory point can correspond to a complete set of parameters, thereby completing the correspondence construction and position alignment processing.

[0058] After establishing the correspondence, the set of parameters for determining crack continuity offset, the set of spatial correlation parameters, and the continuity correction parameters can be adjusted in a coordinated manner. Specifically, the system traverses the spatial sequence of the continuous crack trajectory point by point, extracts the difference between the directional distribution change and the change in adjacent intervals at each trajectory point, and uses these changes as the basis for adjustment. The directional angle data in each parameter set is corrected for directional offset, the interval change data is normalized, and the trajectory correction amount in the continuity correction parameters is synchronously corrected, so that different parameters at the same spatial location maintain a consistent trend in change. In actual data processing, when the directional change of a certain trajectory segment is relatively concentrated, the directional angle data at the corresponding location is compressed to a concentrated range, and the interval change data converges synchronously, thus forming the parameter set after the coordinated adjustment.

[0059] After the linkage adjustment is completed, constraint-driven adjustments can be performed on the image reconstruction or connection strategy based on discrete structural features. Specifically, connection priority regions are divided according to the density of point distribution. Connection paths are established preferentially in dense regions, and the connection range is restricted in low-density regions. Connection directions are selected based on spatial consistency. Connection directions are kept stable in regions with consistent directions, and cross-directional connections are restricted in regions with changing directions. The connection order is determined based on the difference in adjacent interval changes. Continuous connections are performed in regions with gentle interval changes, and breakpoint segmentation is performed in regions with abnormal interval changes. By applying these constraint rules to the construction process of pixel connection paths and connection order, dynamic adjustment of the connection strategy is achieved.

[0060] During the adjustment of the connection strategy, the connection paths of pixels within the crack area can be reconstructed. Specifically, the original connection paths are checked segment by segment according to the adjusted parameter set. Connections that do not meet the direction constraints or interval constraints are disconnected, and adjacent pixels that meet the constraints are searched again for replacement connections. At the same time, the connection order is reordered to ensure that the path extension direction is consistent with the spatial tendency. After the path reconstruction is completed, the trajectory construction process is updated synchronously to ensure that the trajectory generation process is consistent with the connection strategy, thereby forming a new crack connection structure.

[0061] Discrete structural features and continuous crack trajectories represent the fusion of discrete distribution information and continuous structural information. The crack continuity offset judgment parameter set describes the continuous offset state of the crack during the imaging process. The spatial correlation parameter set describes the spatial connection relationship between pixels. The continuity correction parameter describes the change information during the trajectory adjustment process. The correspondence construction represents the establishment of the mapping relationship between different data in a unified spatial coordinate. The point distribution density is used to characterize the pixel aggregation in a local area. Spatial tendency consistency is used to describe the degree of directional consistency. Adjacent interval change is used to describe the distance change characteristics. Position alignment processing represents the unified arrangement of different parameters in spatial position. Linkage adjustment represents the synchronous adjustment process of multiple parameters in the same spatial position. Direction angle data, interval change data, and trajectory correction data are used to describe the change characteristics of direction, distance, and trajectory, respectively. Consistency adjustment represents the unification of the change trends of different parameters. Image reconstruction represents the process of forming a continuous structure through connection and reconstruction. Connection strategy represents the control rules for connection path and connection order. Connection path represents the path structure formed by pixel connection. Connection order represents the arrangement rules in the path construction process. Trajectory construction process represents the process of generating crack trajectories from pixels. Reconstruction processing represents the process of reorganizing the original connection relationship.

[0062] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions according to the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means (e.g., infrared, wireless, microwave, etc.). A computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0063] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0064] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0065] In the several embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0066] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0067] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0068] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A robotic intelligent detection method for surface defects in buildings, characterized in that, Specifically, the steps include: S1. Use a robot to perform pixel projection consistency analysis on images acquired from high-rise buildings at a distance. Combine the offset relationship between pixel grid distribution and spatial projection position to determine whether there are surface defects and cracks that are misaligned with the pixel grid, resulting in cracks being distributed in a point-like manner. Generate a set of crack continuity offset judgment parameters. S2. When cracks are distributed in a point-like pattern, perform discrete morphological analysis on the point-like crack region, combine the density of point distribution, spatial consistency and adjacent interval changes to determine discrete structural features, and generate a set of spatial correlation parameters. S3. Based on the set of spatial correlation parameters, the point cracks are subjected to directional constraint connection processing. The discrete points are connected segment by segment according to the spatial tendency consistency to form the initial shape of the crack trajectory. S4. Based on the initial shape of the crack trajectory and combined with the set of spatial correlation parameters, the crack trajectory is continuously corrected. The discontinuous positions of the fracture location and trajectory direction are adjusted segment by segment to form a continuous crack trajectory. S5. Combining discrete structural features with continuous crack trajectories, the crack continuity offset judgment parameter set, spatial correlation parameter set, and continuity correction parameter are adjusted in a coordinated manner, and the image reconstruction or connection strategy is adjusted according to the discrete structural features.

2. The intelligent robotic detection method for building surface defects according to claim 1, characterized in that, S1 specifically includes the following steps: S101. Using a robot to detect and collect images of high-rise buildings from a distance, perform pixel projection consistency analysis. By establishing the correspondence between pixel grid distribution and spatial projection position, determine the position of each pixel in the pixel grid distribution and its corresponding spatial projection position. S102. Based on the correspondence between pixel grid distribution and spatial projection position, calculate the difference vector between the position coordinates of the pixel corresponding to the surface defect crack in the pixel grid distribution and the corresponding coordinates of the spatial projection position as the position offset. Calculate the change in the directional angle between adjacent pixels based on the directional distribution of the position offset. Calculate the change in the offset amplitude between adjacent pixels based on the numerical change of the position offset. When the change in the directional angle between adjacent pixels is less than a preset angle threshold and the offset amplitude shows a continuous trend, it is determined to be a continuous distribution. When the change in the directional angle between adjacent pixels is greater than or equal to a preset angle threshold or the offset amplitude shows a discrete jump, it is determined to be a situation where the surface defect crack is misaligned with the pixel grid, resulting in a point-like distribution of the crack. S103. When it is determined that there are surface defects and cracks and pixel mesh misalignment causing the cracks to be distributed in a point-like manner, the position offset, direction distribution and spatial adjacency relationship of the corresponding pixel points of the surface defects and cracks are jointly calculated to generate a set of crack continuity offset judgment parameters.

3. The intelligent robotic detection method for building surface defects according to claim 2, characterized in that, S103 specifically refers to: When it is determined that there are surface defects and cracks and pixel grid misalignment causing the cracks to be distributed in a point-like manner, the position offset of the corresponding pixel points of the surface defects and cracks is grouped and processed. According to the spatial adjacency relationship, the pixel points corresponding to the position offset within the preset distance threshold range are divided into multiple adjacency sets, and the directional distribution of the position offset within each adjacency set is calculated. For the directional distribution of pixels in each adjacency set, adjacent pixel pairs are selected according to spatial adjacency relationship, the change in directional angle between adjacent pixel pairs is calculated, and the proportion of pixel pairs with directional angle changes less than a preset angle threshold is counted. At the same time, the amplitude change of position offset within the adjacency set is calculated to obtain the distribution of offset difference between adjacent pixel pairs. By combining the statistical results of the change in directional angle, the distribution of the difference in position offset, and the spatial adjacency relationship, the pixels in the adjacency set are jointly calculated. Pairs of pixels whose change in directional angle is less than a preset angle threshold and whose offset difference is continuously changing are combined to generate a set of parameters for determining the continuity offset of cracks.

4. The intelligent robotic detection method for building surface defects according to claim 1, characterized in that, S2 specifically includes the following steps: S201. When cracks are distributed in a point-like manner, based on the crack continuity offset judgment parameter set, the position offset of the corresponding pixel points of the surface defect crack is screened, and the pixel points whose position offset direction changes are greater than or equal to the preset angle threshold or whose offset amplitude changes in a discrete jump are extracted as point-like crack regions. Discrete morphology analysis is performed on the point-like crack regions to extract the spatial position coordinates of the pixel points within the point-like crack regions. A fixed spatial range neighborhood is constructed with each pixel point as the center, the number of pixel points in the neighborhood is counted and normalized, and the normalized value is used as the point distribution density. S202. Based on the density of point distribution, the pixels in the point-like crack area are sorted according to their spatial coordinates. For the sorted adjacent pixel pairs, the corresponding position offset direction is extracted and the directional angle between adjacent pixel pairs is calculated. A set of directional angles is constructed based on the directional angles of all adjacent pixel pairs, and the directional angle set is statistically analyzed to characterize the spatial consistency. At the same time, the spatial distance between adjacent pixel pairs is calculated and the difference is calculated to characterize the change in adjacent interval. S203. Based on the density of point distribution, spatial consistency, and changes in adjacent intervals, perform joint calculations on pixels within the point-like crack area. Combine pixel pairs with directional angles less than a preset angle threshold and continuously changing adjacent intervals to determine discrete structural features. Generate a set of spatial association parameters based on the spatial connection relationship and directional angle change relationship between pixels.

5. The intelligent robotic detection method for building surface defects according to claim 4, characterized in that, S203 specifically refers to: By combining the density of point distribution, spatial consistency, and changes in adjacent intervals, a set of pixel pairs is constructed for pixels within the point crack area according to their spatial adjacency, and the directional angle and adjacent interval changes are extracted for each pixel pair in the set. For a set of pixel pairs, pixel pairs with a direction angle less than the preset angle threshold are selected based on the comparison result between the direction angle and the preset angle threshold. In the selection result, the pixel pairs are grouped according to the continuity of the change in adjacent intervals to form a set of pixel pairs that satisfy the direction constraint and the interval change constraint. For the set of pixel pairs after grouping, each pixel pair is sequentially connected according to spatial adjacency to construct a pixel pair connection path. The directional angle distribution, adjacent interval change distribution, and connection order of pixel pairs within each connection path are encoded. The directional angle distribution data, adjacent interval change data, and connection path index are combined to generate a set of spatial association parameters.

6. The intelligent robotic detection method for building surface defects according to claim 1, characterized in that, S3 specifically refers to: Based on the set of spatial correlation parameters, the connection path index, directional angle distribution data and adjacent interval change data of the pixels in the point crack area are extracted. The pixels are then grouped according to the connection path index, and the pixels in each group are sorted according to their spatial coordinates to form a discrete point sequence to be connected. For the discrete point sequence, directional constraint connection processing is implemented for point cracks. The connection direction between adjacent discrete points is filtered according to the directional angle distribution data. Adjacent discrete points with directional angles less than a preset angle threshold are retained as valid connection relationships. The valid connection relationships are connected segment by segment according to spatial tendency consistency. A continuous connection segment is formed by establishing a connection relationship between the current discrete point and the adjacent discrete points that meet the directional constraints. Around the continuous connecting segments formed by segment-by-segment connection, the connection order between the continuous connecting segments is constrained according to the adjacent interval change data. The continuous connecting segments with the interval change difference between adjacent discrete points less than the preset interval threshold are sequentially spliced ​​together, and the continuous connecting segments with the interval change difference greater than or equal to the preset interval threshold are segmented to form a continuous connecting path of point cracks and construct the initial shape of crack trajectory.

7. The intelligent robotic detection method for building surface defects according to claim 1, characterized in that, S4 specifically refers to: Based on the initial shape of the crack trajectory, the spatial coordinates of each trajectory point in the crack trajectory are extracted. Combined with the directional angle distribution data and adjacent interval change data in the spatial correlation parameter set, the continuity correction process of the crack trajectory is performed. By calculating the difference between the directional angle and the adjacent interval change between adjacent trajectory points, the positions where the directional angle is greater than or equal to a preset angle threshold or the difference between adjacent interval changes is greater than or equal to a preset interval threshold are marked as the fracture position and the position where the trajectory direction change is discontinuous. For locations where the fracture location and trajectory direction are discontinuous, the continuity of the crack trajectory is corrected by combining a set of spatial correlation parameters. By selecting adjacent trajectory points on both sides of the fracture location as adjustment starting points, the connection direction between adjacent trajectory points is constrained based on the directional angle distribution data, and the trajectory points between fracture locations are adjusted segment by segment according to the spatial position relationship. Transitional connection segments are formed by inserting compensation trajectory points or adjusting the spatial position of the original trajectory points. Around the transition connection segment formed by the segmented adjustment, the crack trajectory is continuously corrected. The connection order between the transition connection segment and the original trajectory segment is constrained according to the adjacent interval change data. The trajectory segments with adjacent interval change difference less than the preset interval threshold are sequentially spliced ​​together, and the positions with adjacent interval change difference greater than or equal to the preset interval threshold are segmented to form a continuous crack trajectory.

8. The intelligent robotic detection method for building surface defects according to claim 1, characterized in that, S5 specifically refers to: By combining discrete structural features with continuous crack trajectories, a correspondence is constructed between the crack continuity migration judgment parameter set, the spatial correlation parameter set, and the continuity correction parameter. By matching the point distribution density, spatial tendency consistency, and adjacent interval changes in the discrete structural features with the spatial position coordinates of each trajectory point in the continuous crack trajectory, the crack continuity migration judgment parameter set, the spatial correlation parameter set, and the continuity correction parameter are aligned. Based on the correspondence construction results, the set of parameters for determining crack continuity migration, the set of spatial correlation parameters, and the continuity correction parameters are adjusted in a coordinated manner. By synchronously traversing each parameter according to the spatial position order, and based on the difference between the direction distribution change and the interval change in the discrete structural features, the direction angle data, interval change data, and trajectory correction data in each parameter are adjusted in a consistent manner to form the parameter set after the coordinated adjustment. By combining discrete structural features, the parameter set after linkage adjustment is constrained. Based on the density of point distribution, spatial consistency, and changes in adjacent intervals, the image reconstruction or connection strategy is adjusted. By reconstructing the connection path, connection order, and trajectory construction process of pixels within the crack area, the robot detection of surface defects of high-rise buildings is completed.

Images