Arc-shaped beam identification method based on line geometric characteristic extension

Abstract

The invention relates to the technical field of arc-shaped beam identification, in particular to an arc-shaped beam identification method based on line geometric characteristic extension, which comprises the following steps: generating point cloud data including wall surface, ceiling and ground building structures through laser scanning, carrying out noise reduction and filtering processing, constructing an accurate three-dimensional model, and extracting arc-shaped beam lines by adopting an edge detection algorithm to obtain a three-dimensional model; and geometric parameters such as curvature, radian, arc length and the like are normalized, so that the consistency of geometric recognition is improved. Through the preset arc parameter library, the scheme can accurately extract and mark the arc, and the problem that the arc beam is confused with other structures is effectively avoided. In addition, the convolutional neural network is introduced for intelligent classification and prediction, complex geometric features are captured through multilayer convolution, and the recognition precision and robustness are improved. And finally, supplementing a missing arc-shaped beam structure through a geometric extension algorithm, and mapping the missing arc-shaped beam structure to the three-dimensional model to realize complete visualization, thereby solving the problem of missing visual information of the arc-shaped beam in a traditional method.

Description

Technical Field

[0001] The invention relates to the technical field of arc-beam recognition, and in particular to an arc-beam recognition method based on the extension of line geometric characteristics. Background Art

[0002] With the rapid development of the construction industry, the widespread use of non-standard geometric forms such as curved beams in complex building structures has led to increasingly higher requirements for their design, detection and construction accuracy. As an important element in buildings with load-bearing and decorative functions, the curved beam structure has a complex shape. Identifying and detecting the position of the curved beam is of reference value to help construction workers with construction design. However, traditional methods for identifying and detecting curved beams usually rely on manual measurement and comparison of design drawings, which are inefficient and easily affected by human factors.

[0003] Existing methods have the following limitations when dealing with curved beams with complex curvature changes: Insufficient recognition accuracy of complex geometric shapes: Curved beams usually have continuous and smooth curvature changes and are often connected to other surrounding structures. Traditional geometric feature extraction methods are difficult to accurately distinguish curved beams from other adjacent structures. In three-dimensional space, the local information of curved beams is relatively complex, and existing line extraction algorithms often cannot effectively process partially obscured or missing geometric information, resulting in the inability to visualize the complete curved beam structure. Summary of the invention

[0004] To solve the above problems, the present invention provides a curved beam recognition method based on the extension of line geometric characteristics, which significantly improves the recognition accuracy and geometric integrity of curved beams through technical means such as laser scanning, geometric feature extraction, intelligent classification and geometric extension.

[0005] To achieve the above object, the technical solution adopted by the present invention is:

[0006] A method for identifying a curved beam based on the extension of line geometric characteristics comprises the following steps:

[0007] Scanning the interior of the building and obtaining an indoor three-dimensional model, extracting edge lines from the indoor three-dimensional model and segmenting monomer lines, and normalizing the monomer lines to obtain standardized line parameters;

[0008] By traversing the preset arc parameter library, the corresponding monomer lines whose standardized line parameters meet the preset arc parameters are selected and marked as arcs to be processed, wherein the preset arc parameters include curvature, arc range and arc length;

[0009] Taking the arc to be processed as the center, obtain the monomer lines connecting the arc to be processed and combine them with the arc to be processed to build a beam recognition geometric model, identify and classify the beam recognition geometric model based on the convolutional neural network, and output the point probability distribution of the arc to be processed and the arc beam;

[0010] Based on the probability distribution of the matching points between the arc to be processed and the curved beam, the arc to be processed is supplemented and extended in the beam recognition geometric model to obtain the extended arc and remap it in the three-dimensional model for visualization and the curved beam position is calibrated by the extended arc.

[0011] Furthermore, scanning the interior of the building and acquiring the indoor three-dimensional model includes:

[0012] The laser scanning equipment is used to perform an all-round scan of the interior of the building to obtain point cloud data including the wall, ceiling and ground structure, and the point cloud data is subjected to noise reduction and filtering to generate a three-dimensional indoor model.

[0013] Furthermore, the extraction of edge lines and segmentation of monomer lines from the indoor three-dimensional model includes the following steps:

[0014] Applying edge detection algorithm to the generated indoor 3D model, identifying edge lines of building structures such as walls, ceilings and floors from the 3D model as initial lines;

[0015] Splits the initial line into individual lines based on the curvature and inflection points of the line.

[0016] Furthermore, the process of taking the arc to be processed as the center, acquiring the monomer lines connecting the periphery of the arc to be processed and combining the monomer lines with the arc to be processed to construct the beam recognition geometric model includes the following steps:

[0017] Taking the geometric center of the arc to be processed as the base point, a spatial index algorithm is used to search in three-dimensional space for neighboring monomer lines whose distance to the arc to be processed is less than a preset distance threshold;

[0018] The neighboring monomer lines are geometrically combined with the arcs to be processed, and the connections of the monomer lines are smoothed by the spline interpolation algorithm;

[0019] The connection errors between lines are dynamically adjusted through the error correction algorithm to generate a beam recognition geometric model.

[0020] Furthermore, the method of identifying and classifying the beam recognition geometric model based on a convolutional neural network and outputting the point probability distribution of the arc to be processed and the matching point of the arc-shaped beam comprises the following steps:

[0021] The recognized geometric model is input into the convolutional neural network model, and the spatial connection relationship between the single line and its adjacent lines in the geometric model is extracted through the convolution layer to generate a geometric model feature map;

[0022] Perform multi-layer convolution and pooling operations on the extracted geometric model feature map to fuse local geometric features at different scales to obtain a multi-scale fused geometric feature map;

[0023] The multi-scale fused geometric feature map is input into the fully connected layer, and the geometric features are classified by the Softmax activation function. The probability distribution of the matching between the arc to be processed and the curved beam structure is calculated, and the probability value of the point where each arc to be processed matches the curved beam in the beam recognition geometric model is output.

[0024] Furthermore, the formula of the convolutional neural network is as follows:

[0025]

[0026] Among them, z i，j，k is the probability distribution of the kth curved beam or non-curved structure at the (i, j) position of the final output; X i+m，j+n is the parameter set of the geometric feature graph at position (i+m, j+n), including the standardized geometric parameters of the arc to be processed and its surrounding monomer lines; W m，n，k is the weight of the convolution kernel at position (m,n) corresponding to the kth curved beam or non-curved structure; b k is the bias term of the kth category; Softmax is the activation function; M and N are the height and width of the convolution kernel respectively.

[0027] Furthermore, the loss function formula of the convolutional neural network is as follows:

[0028]

[0029] Among them, L is the total loss function; N is the total number of samples; y i is the true label of the i-th sample; p(y i ) is the predicted probability of the convolutional neural network for the i-th sample; is the regularization term, W k is the weight matrix of the k-th convolution kernel; λ is the regularization coefficient; K is the total number of layers of the convolutional neural network.

[0030] Furthermore, the method of supplementing and extending the arc to be processed in the beam recognition geometric model based on the probability distribution of points matching the arc to be processed with the curved beam comprises the following steps:

[0031] The probability mean of the probability distribution of the points matching the arc beam in the window is calculated by sliding the preset window size;

[0032] Determine the extension path and end point based on the probability mean and the preset probability threshold;

[0033] The arc to be processed is supplemented and extended according to the extended path and end point.

[0034] Furthermore, the process of obtaining the extended arc and remapping it in the three-dimensional model to perform visualization and calibrating the arc beam position by the extended arc includes the following steps:

[0035] The arc data generated by the supplementary extension is mapped back to the global coordinates of the indoor three-dimensional model according to the spatial coordinate system;

[0036] Based on the mapped extended arc, a visualization algorithm is applied in the three-dimensional model to convert the arc into a geometric display model through a graphics rendering engine;

[0037] Semantically calibrate the specific position of the curved beam in the geometric display model.

[0038] The beneficial effect of the present invention is that the present invention performs an all-round scanning of the interior of the building by laser scanning, generates point cloud data including the wall, ceiling and ground building structure, and obtains an accurate three-dimensional model through noise reduction and filtering. Compared with the traditional method of relying on manual measurement and design drawing comparison, the three-dimensional scanning technology can efficiently and accurately obtain the geometric information of the building structure, avoid the recognition error caused by human factors, and ensure the data accuracy. On the basis of the three-dimensional model, the edge detection algorithm is used to extract the initial lines of the curved beam, and the normalization is performed based on its geometric parameters such as curvature, radian and arc length. By standardizing the geometric parameters, this scheme ensures that the geometric features of the curved beam and other adjacent structures can be uniformly processed, effectively improves the geometric consistency in the recognition process, and overcomes the parameter mismatch problem under complex geometric forms. By establishing a preset arc parameter library, this scheme can traverse and extract line features that meet the preset parameters and mark the arcs to be processed. This step can effectively avoid the defects of the traditional method in that the curved beam is not clearly identified when it is adjacent to other building structures, and realizes accurate recognition and separation of the curved beam through the comparison and precise screening of geometric parameters. The convolutional neural network (CNN) model is further introduced, and the matching degree between the arc to be processed and the curved beam is intelligently classified and predicted in combination with the geometric model features. Through multi-layer convolution operations and multi-scale feature fusion, the neural network model can better capture the feature changes in complex geometric forms, improve the accuracy and robustness of recognition, and significantly outperform the traditional fixed geometric rule algorithm. After the arc to be processed is matched with the curved beam, this scheme performs geometric extension based on the probability distribution of the convolutional neural network output to deal with the partial occlusion or missing problem of the curved beam in three-dimensional space, generate extended arcs, and remap them to the three-dimensional model to achieve visualization of the complete curved beam geometric structure, thereby overcoming the problem of incomplete curved beam recognition information in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a flow chart of the steps of a method for identifying an arc beam based on the extension of line geometric characteristics in the present invention.

[0040] Figure 2 It is a flowchart of the steps of obtaining a single line connecting the periphery of an arc to be processed and combining it with the arc to be processed to construct a beam recognition geometric model in the present invention. DETAILED DESCRIPTION

[0041] See also Figure 1-2 As shown, the present invention relates to a method for identifying a curved beam based on the extension of line geometric characteristics, comprising the following steps:

[0042] Scanning the interior of the building and obtaining an indoor three-dimensional model, extracting edge lines from the indoor three-dimensional model and segmenting monomer lines, and normalizing the monomer lines to obtain standardized line parameters;

[0043] By traversing the preset arc parameter library, the corresponding monomer lines whose standardized line parameters meet the preset arc parameters are selected and marked as arcs to be processed, wherein the preset arc parameters include curvature, arc range and arc length;

[0044] Taking the arc to be processed as the center, obtain the monomer lines connecting the arc to be processed and combine them with the arc to be processed to build a beam recognition geometric model, identify and classify the beam recognition geometric model based on the convolutional neural network, and output the point probability distribution of the arc to be processed and the arc beam;

[0045] Based on the probability distribution of the matching points between the arc to be processed and the curved beam, the arc to be processed is supplemented and extended in the beam recognition geometric model to obtain the extended arc and remap it in the three-dimensional model for visualization and the curved beam position is calibrated by the extended arc.

[0046] In some embodiments, first, three-dimensional point cloud data of the building interior is obtained by a laser scanning device. The point cloud data is presented in the form of discrete coordinates in space and contains all geometric information of the surface of the building structure. In order to ensure the high accuracy of the three-dimensional model, the present invention adopts a point cloud denoising algorithm based on density and distance (such as a local weighted average denoising method based on K nearest neighbors) to remove noise points, and applies a Gaussian filter algorithm to smooth the point cloud data to retain the main geometric features. This preprocessing step lays a data foundation for subsequent line extraction and analysis. On the processed three-dimensional model, the Canny edge detection algorithm is applied to extract the edge lines of the building structure. The Canny algorithm identifies areas with significant changes in curvature by calculating image gradients, and combines dual threshold technology to ensure the accuracy and robustness of the detection results. The extracted initial line information is processed by a subsequent line segmentation algorithm to segment the complex building structure curve into several monomer lines. This segmentation process is based on the change of geometric features, using the curvature change rate as a judgment basis. When the curvature change rate exceeds a preset threshold, the line is segmented into different monomer lines. In order to ensure the uniformity of geometric information at different spatial scales, the present invention uses a normalization algorithm to standardize the geometric characteristics (such as curvature, radian and arc length) of each monomer line. The system can map geometric features at different scales to the same distribution, so that subsequent processing can be carried out in a unified geometric space. The line parameters after standardization are input into the arc parameter library in the system. The arc parameter library stores the geometric characteristics of common arc beams. The present invention traverses the parameter library through a brute force search algorithm to find arc features that match the arc to be processed and mark them as arcs to be processed. After the arc marking is completed, the system further retrieves the monomer lines adjacent to the arc to be processed through a spatial indexing algorithm, such as a KD tree or an octree. The spatial indexing algorithm can efficiently handle the neighbor search problem in high-dimensional space and ensure the spatial correlation between the arc to be processed and its adjacent lines. Next, the system combines these adjacent lines with the arc to be processed to construct a beam recognition geometric model. During the combination process, the spline interpolation algorithm is used to smooth the connection points of adjacent lines to ensure the continuity of the geometric model in curvature and spatial position, and avoid geometric mutations or discontinuities. Based on the constructed geometric model, the system introduces a convolutional neural network (CNN) to intelligently classify the geometric model. CNN first extracts the local features of the geometric model, and uses the convolution kernel to slide on the feature map to calculate the geometric features of the local area. Through the superposition of multiple convolutional layers, CNN can extract the spatial connection relationship and local geometric features of the geometric model layer by layer to generate a high-dimensional feature map. Subsequently, the system reduces the dimension of the features through a pooling operation, generates a multi-scale fusion feature map, and calculates the probability distribution of the matching between the arc to be processed and the curved beam through a Softmax classifier. Based on the probability distribution output by the convolutional neural network, the present invention adopts a geometric extension algorithm based on spline curves to supplement and extend the arc to be processed.Specifically, the system uses the high-probability points identified by the convolutional neural network as the starting point of the extension, and generates new extended line segments through spline interpolation or Bezier curves. This process ensures that the extended lines remain smooth and continuous, and conform to the actual geometric characteristics of the curved beams in the building. After the extension is completed, the system remaps the extended arcs to the 3D model. The mapping process uses a spatial coordinate mapping algorithm to ensure that the extended geometric information is accurately aligned with the global model of the building structure. After the 3D model is updated, the system uses visual rendering technology to accurately display the extended curved beam geometry model, which is convenient for subsequent architectural design and personnel identification reference.

[0047] Furthermore, scanning the interior of the building and acquiring the indoor three-dimensional model includes:

[0048] The laser scanning equipment is used to perform an all-round scan of the interior of the building to obtain point cloud data including the wall, ceiling and ground structure, and the point cloud data is subjected to noise reduction and filtering to generate a three-dimensional indoor model.

[0049] It should be noted that a laser scanning device is used to perform a full-scale scan of the interior of the building. The device generates complete three-dimensional point cloud data by emitting laser pulses and measuring the position coordinates of each point in the building based on its reflection time. The point cloud data can fully capture the geometric form of each surface in the room, including detailed information of the building structure such as walls, ceilings, and floors. The resolution of laser scanning can be accurate to the millimeter level, so it can capture very subtle geometric changes on the surface of the building, which is particularly important for the subsequent recognition and geometric processing of curved beams. After the point cloud data is collected, the system performs a series of preprocessing operations on the original point cloud data. Due to the complex scanning environment in the building, noise and redundant information are inevitably introduced during the data collection process, such as invalid data points caused by light interference, scanning equipment errors, or reflective objects. To this end, the system first applies a point cloud denoising algorithm to the point cloud data, which identifies and filters out abnormal data points by calculating the distance difference between each data point and its neighboring points. Specifically, the system uses a statistical analysis-based method to analyze the neighborhood of each data point, mark points that are inconsistent with the neighborhood distance distribution as noise and remove them, thereby improving the overall quality of the point cloud data. After the noise reduction is completed, the system applies filtering to the noise-reduced point cloud data. The filtering operation further eliminates high-frequency errors in the point cloud data through smoothing, while retaining key geometric features. The commonly used filtering methods are Gaussian filtering or mean filtering, which reduces the impact of random errors by weighted averaging adjacent data points in the point cloud and ensures the smoothness and continuity of the data. The filtered point cloud data can more accurately reflect the true geometric structure of the building, which is convenient for subsequent three-dimensional model generation and geometric feature extraction. The system performs geometric fitting on the point cloud data after noise reduction and filtering to generate a three-dimensional model of the building interior. The point cloud data contains all the surface information in the building, so the system fits these discrete point clouds into a continuous geometric surface model through geometric analysis of the data. In this process, surface reconstruction algorithms, such as triangular mesh generation algorithms (such as Delaunay triangulation), are used to connect the data points in the point cloud into several triangular facets to construct a complete three-dimensional model. This model can accurately express the geometric forms of the walls, ceilings, floors and other structures in the building interior.

[0050] Furthermore, the extraction of edge lines and segmentation of monomer lines from the indoor three-dimensional model includes the following steps:

[0051] Applying edge detection algorithm to the generated indoor 3D model, identifying edge lines of building structures such as walls, ceilings and floors from the 3D model as initial lines;

[0052] Splits the initial line into individual lines based on the curvature and inflection points of the line.

[0053] In some embodiments, first, based on the generated indoor three-dimensional model, the system applies an edge detection algorithm. The core purpose of this algorithm is to extract significant geometric edge lines in the building from complex three-dimensional data, especially the edges of large-area structures such as walls, ceilings, and floors. In order to ensure the accuracy of edge detection, the system usually adopts gradient-based edge detection technology (such as Canny edge detection algorithm) to identify areas with large curvature changes by calculating the rate of change of the model surface normal vector. These areas often correspond to geometric boundaries in the building structure, such as corners, the connection between columns and walls, and the corners of the ceiling. The multi-level threshold technology of Canny edge detection ensures noise suppression and captures the main boundaries in the building geometry. After completing edge detection, the system identifies and extracts these edge lines and defines them as initial lines. At this time, the initial lines may contain complex geometric information. Due to the diversity of building structures, these lines usually have a large range of curvature changes. For example, the connection between the wall and the ceiling may contain multiple corners. These geometric features will cause the curvature of the lines to change unevenly, further increasing the complexity of data processing. In order to analyze and process these complex initial lines more accurately, the present invention segments the extracted initial lines through curvature analysis and inflection point recognition algorithms. First, the system calculates the curvature value of each initial line at different positions. The curvature value reflects the degree of curvature of the line in three-dimensional space. Usually, in the geometric structure of the building, the place where the curvature changes greatly often corresponds to the significant geometric features of the structure, such as corners or edge inflection points. The system locates the key geometric points of the line by analyzing the curvature change law, that is, the position where the curvature change exceeds the preset threshold. Combined with the analysis results of the curvature change, the inflection points in the initial line are identified. These inflection points usually correspond to the geometric corners in the building structure, such as the junction of the wall and the ceiling. In order to ensure the accuracy of the segmentation, the system will use each inflection point as the basis for segmenting the line. Specifically, when the curvature change of the line reaches a certain degree or an obvious inflection point appears, the system will segment the initial line at this position. In this way, the relatively complex initial line is divided into a plurality of relatively simple monomer lines. The geometric characteristics of each monomer line are relatively simple, usually with uniform curvature and stable geometric form. The purpose of this is to refine the complex building geometry into more manageable basic units, which is convenient for subsequent geometric analysis, classification and processing. For some curved beam structures, the segmented monomer lines can more clearly express their local geometric characteristics, such as the curvature and radian of the curved beam.

[0054] Furthermore, the process of taking the arc to be processed as the center, acquiring the monomer lines connecting the periphery of the arc to be processed and combining the monomer lines with the arc to be processed to construct the beam recognition geometric model includes the following steps:

[0055] Taking the geometric center of the arc to be processed as the base point, a spatial index algorithm is used to search in three-dimensional space for neighboring monomer lines whose distance to the arc to be processed is less than a preset distance threshold;

[0056] The neighboring monomer lines are geometrically combined with the arcs to be processed, and the connections of the monomer lines are smoothed by the spline interpolation algorithm;

[0057] The connection errors between lines are dynamically adjusted through the error correction algorithm to generate a beam recognition geometric model.

[0058] In some embodiments, first, after completing the identification of the arc to be processed, the system uses the geometric center of the arc to be processed as the base point to search for other lines in the three-dimensional model through a spatial index algorithm. Specifically, the spatial index algorithm often uses efficient data structures such as KD trees or octrees to process neighboring query problems in high-dimensional space. Through these algorithms, the system can quickly determine the monomer lines whose distance from the arc to be processed is less than a preset threshold, and these lines are considered to be candidate lines with geometric association with the arc to be processed. In actual operation, the geometric center of the arc to be processed is usually calculated based on its spatial position and geometric characteristics. Through the execution of the spatial index algorithm, the system can identify the lines that are geometrically closest to the center point in three-dimensional space, and these lines may be located in other parts of the arc beam or on the architectural elements connecting the arcs. In order to ensure the accuracy of the index result, the selection of the preset distance threshold is crucial. Too large a threshold may introduce irrelevant geometric lines, while too small a threshold may lead to omissions. Through experiments, the present invention sets a suitable distance threshold according to the geometric complexity of the building in practical applications. Next, the system combines the neighboring monomer lines obtained by the spatial index algorithm with the arc to be processed to form a preliminary geometric model. In order to ensure the continuity and smoothness between the lines, the present invention uses a spline interpolation algorithm to smooth the connection of the lines. Spline interpolation is a mathematical method commonly used for curve fitting. By calculating the smooth curve between the interpolation points, it avoids the sudden change or discontinuity at the connection of the lines. In this process, the spline interpolation algorithm will consider the geometric parameters of the lines, including curvature, arc length, etc., to generate a smooth transition curve so that the combined geometric model conforms to the actual arc beam structure. In actual architectural applications, the geometric characteristics of the arc beam are complex, and the curvature may change with the spatial position. Therefore, in the process of spline interpolation, it is necessary to ensure that the interpolation result is not only visually smooth, but also mathematically ensure the differentiability of the curve. This means that the derivative change of the interpolation curve should also be smoothly transitioned to avoid unreasonable sharp changes in geometry. In addition, when processing a long arc combination, the present invention will adaptively adjust the number of interpolation points according to the complexity of the lines to ensure the balance between the accuracy of the interpolation curve and the computational efficiency. After completing the preliminary combination and interpolation of the geometric model, the system dynamically adjusts the geometric errors between the lines through the error correction algorithm. Since the lines in the building model may be affected by factors such as scanning accuracy and data acquisition errors, it is inevitable that there will be small geometric errors in the line combination process. In order to ensure the accuracy of the final beam recognition geometric model, the present invention introduces an error correction mechanism. The error correction algorithm automatically adjusts the connection position and angle of the lines by analyzing the spatial deviation between the lines to keep them consistent with the actual building structure.During the error correction process, the system will first calculate the offset and angle difference between the lines, and then use a series of geometric transformation operations (such as translation, rotation, etc.) to accurately align the endpoints of the lines with the geometric positions of the arcs to be processed. At the same time, the error correction algorithm will also determine whether the overall curvature of the lines needs to be adjusted based on the actual building geometry. For small errors, the system will adopt a minimum correction strategy to maintain the integrity of the original data as much as possible; for large errors, the system will gradually eliminate significant errors between lines through iterative correction methods.

[0059] Furthermore, the method of identifying and classifying the beam recognition geometric model based on a convolutional neural network and outputting the point probability distribution of the arc to be processed and the matching point of the arc-shaped beam comprises the following steps:

[0060] The recognized geometric model is input into the convolutional neural network model, and the spatial connection relationship between the single line and its adjacent lines in the geometric model is extracted through the convolution layer to generate a geometric model feature map;

[0061] Perform multi-layer convolution and pooling operations on the extracted geometric model feature map to fuse local geometric features at different scales to obtain a multi-scale fused geometric feature map;

[0062] The multi-scale fused geometric feature map is input into the fully connected layer, and the geometric features are classified by the Softmax activation function. The probability distribution of the matching between the arc to be processed and the curved beam structure is calculated, and the probability value of the point where each arc to be processed matches the curved beam in the beam recognition geometric model is output.

[0063] In some embodiments, first, the constructed beam recognition geometric model is used as input and imported into a convolutional neural network (CNN) for processing. The geometric model contains the geometric information of the arc to be processed and its adjacent lines. The task of the convolutional neural network is to identify the matching degree between the arc to be processed and the arc beam structure by analyzing these geometric data. In this process, the convolutional neural network extracts features of the input geometric model through a convolutional layer. The convolutional layer scans the local area of ​​the entire geometric model by sliding a window, and extracts the spatial connection relationship and geometric characteristics of the area, such as the angle, curvature change, and spatial distance between adjacent lines. The role of the convolution kernel is to capture local features in the geometric model through weight learning. These features can reflect the interrelationship between lines and the continuity of geometric forms. After the convolution operation, the network generates a series of feature maps, which represent the feature expressions of the geometric model in different parts. In order to enhance the hierarchical expression of features, the system inputs the feature maps into multiple convolutional layers for further processing. Each layer of convolution can capture geometric information at different scales. For example, shallower convolution layers mainly capture local details of lines, such as curvature mutations and angle features, while deeper convolution layers can capture the overall spatial structure and macroscopic morphology. This multi-layer convolution processing method ensures that the convolutional neural network can not only recognize the local details of the geometric model, but also understand its global characteristics. After the convolution operation, the system will perform a pooling operation on the extracted feature map. The pooling operation reduces the size of the feature map by downsampling, thereby reducing the dimension and computational complexity of the data. Commonly used pooling methods include maximum pooling and average pooling. The difference between the two is that the former retains the maximum value feature of the local area, while the latter retains the average value feature of the local area. Through the pooling operation, the convolutional neural network can effectively extract the key features of the geometric model and minimize redundant information. The pooled feature map represents the fusion result of local geometric features at different scales, so that the network can perform a global analysis of the spatial structure of the entire geometric model. Next, the multi-scale fused feature map after multi-layer convolution and pooling operations is input into the fully connected layer. The fully connected layer is the last step in the convolutional neural network, which is used to integrate the high-dimensional features extracted previously and map them to the output space. Specifically, the fully connected layer converts the feature map into a linear combination to obtain an output related to the target task. For the present invention, the output is the probability distribution of the matching between the arc to be processed and the curved beam. In the fully connected layer, the system converts the output result into a normalized probability value through the Softmax activation function. The Softmax function can convert the output of the network into a probability distribution for each category, ensuring that the sum of all probability values ​​is equal to 1. In this way, the system can calculate the probability of matching the curved beam for each arc to be processed. The output probability distribution represents the degree to which the arc to be processed matches the curved beam structure in the geometric model.Specifically, for each arc to be processed, the network outputs a probability value of its belonging to a curved beam. For example, if the probability value of a certain arc to be processed is close to 1, it means that it matches the curved beam very well, while if the probability value is lower, it means that it may belong to other structures. In actual operation, the system will perform further processing based on these probability values, such as screening out arcs with probability values ​​higher than a certain threshold as candidate lines for curved beams, or adjusting the details of the geometric model based on the distribution of probability values.

[0064] Furthermore, the formula of the convolutional neural network is as follows:

[0065]

[0066] Among them, z i，j，k is the probability distribution of the kth curved beam or non-curved structure at the (i, j) position of the final output; X i+m，j+n is the parameter set of the geometric feature graph at position (i+m, j+n), including the standardized geometric parameters of the arc to be processed and its surrounding monomer lines; W m，n，k is the weight of the convolution kernel at position (m,n) corresponding to the kth curved beam or non-curved structure; b k is the bias term of the kth category; Softmax is the activation function; M and N are the height and width of the convolution kernel respectively.

[0067] Specifically, Z i，j，k Represents the final output of the convolutional neural network, that is, the probability distribution value of the k-th type of structure (arc beam or non-arc structure) at position (i, j). This value represents the classification result of the network for the current geometric model at a specific position. Specifically, the network will output the probability that each pixel or feature point belongs to a certain category during the recognition process. i+m，j+n The parameter set contains not only the geometric characteristics of the arc to be processed, but also the standardized geometric parameters of the surrounding adjacent single lines. These parameters may include geometric information such as curvature, radian, arc length, etc., which can be uniformly compared and analyzed in different layers of the network through standardization. The convolution kernel is the core part of the convolutional neural network for extracting features. The weights are continuously adjusted through the training process of the network to capture local patterns in the geometric model. For example, the weights of the convolution kernel can help the network identify specific geometric features in the arc beam, such as continuous curves or the connection relationship between adjacent lines. The bias term is used to adjust the results of the convolution calculation so that the network can more flexibly capture the characteristics of different categories when performing feature extraction and classification. For each category, the network balances the network output by setting different bias values, so that the network can more accurately distinguish between arc beams and other structures. The Softmax function is a commonly used activation function in multi-classification tasks. Its function is to normalize the original output value of the network into a probability distribution, and the sum of the probability values ​​of all output categories is equal to 1.

[0068] Furthermore, the loss function formula of the convolutional neural network is as follows:

[0069]

[0070] Among them, L is the total loss function; N is the total number of samples; y i is the true label of the i-th sample; p(y i ) is the predicted probability of the convolutional neural network for the i-th sample; is the regularization term, W k is the weight matrix of the k-th convolution kernel; λ is the regularization coefficient; K is the total number of layers of the convolutional neural network.

[0071] Specifically, the role of the loss function is to measure the difference between the network's predicted results and the actual results, and use this difference to guide the adjustment of network parameters and optimize the performance of the model. The loss function in the present invention consists of two main parts: classification loss - y i log(p(y i ))+(1-y i )log(1-p(y i )) and the regularization term First, the classification loss is used to measure whether the network's prediction of each sample is accurate. For the arc beam recognition task, each sample represents an arc to be processed, and the network's task is to determine whether this arc belongs to a arc beam. Each sample has a true label (such as "arc beam" or "non-arc beam"), and the network will output a predicted probability value based on the characteristics of the geometric model, indicating the possibility that the arc belongs to the arc beam. The classification loss calculates the network's error by comparing the true label and the predicted probability. If the network's prediction is close to the true label, the loss is small; if the prediction result is significantly different from the actual label, the loss value will increase. The larger the loss, the less accurate the network's prediction. Through this mechanism, the network can continuously adjust its internal parameters to minimize the loss, thereby improving the accuracy of the model. For example, in the arc beam recognition task, the true label of a certain arc is "arc beam". If the network predicts that the arc belongs to "arc beam" with a probability of 90%, then the loss value will be small because the predicted value is close to the true label. But if the network predicts that the probability of the arc is only 20%, the loss value will be large, prompting the network to adjust the weights during training to improve subsequent predictions. Secondly, the regularization term in the loss function is used to prevent the model from overfitting. Overfitting means that the model performs very well on the training data, but performs poorly on new data. This is usually because the model is too complex and too dependent on specific training data. To avoid this, the regularization term limits the weight parameters in the network to prevent them from becoming too large or too complex. The regularization term makes the network smoother during the learning process by adding penalties to the weights, avoiding overfitting the training data. In this way, the network can not only perform well on the training data, but also maintain good predictive ability on unknown new data. The strength of regularization is controlled by a parameter called the regularization coefficient, which determines the strength of the penalty for the weights. If the regularization coefficient is large, the network will be forced to reduce the size of the weights, thereby improving generalization ability; if the coefficient is small, the weights of the network can be adjusted more freely. Through the design of this loss function, the convolutional neural network can effectively learn complex geometric features in the arc beam recognition task and avoid overfitting problems while maintaining high recognition accuracy.

[0072] Furthermore, the method of supplementing and extending the arc to be processed in the beam recognition geometric model based on the probability distribution of points matching the arc to be processed with the curved beam comprises the following steps:

[0073] The probability mean of the probability distribution of the points matching the arc beam in the window is calculated by sliding the preset window size;

[0074] Determine the extension path and end point based on the probability mean and the preset probability threshold;

[0075] The arc to be processed is supplemented and extended according to the extended path and end point.

[0076] Specifically, first, the preset window size is used for sliding, and the probability mean of the probability distribution of the points matching the curved beam in the window is calculated. Specifically, after the model is classified by the convolutional neural network, the probability value of matching with the curved beam has been output for each geometric point. In order to further analyze the local characteristics of these points, the system uses the sliding window technology to perform statistics on the probability distribution of these points. The size of the sliding window is preset and is usually adjusted according to the curvature and geometric characteristics of the curved beam. The role of the window is to capture the characteristic fluctuations in the local area and analyze the overall situation of the matching between the points in each area and the curved beam. In the sliding window, the system calculates the average of the probability values ​​of all points in the area, which is the probability mean. In this way, the system can better understand the overall characteristics of the local area, especially when the matching probability of some points may be low or have abnormal fluctuations. The probability mean provides a smooth and global perspective. This operation is particularly important for processing curved beams with complex curvature changes, because the probability value of a single point may not be sufficient to fully reflect the geometric morphology of the area. By smoothing the probability value, the system can avoid interference from local abnormal points, thereby obtaining more reliable matching results. Next, based on the probability mean and the preset probability threshold, the system determines the extension path and the end point. In the actual building model, the curved beam may be missing some data due to occlusion, scanning error or incomplete modeling. In order to automatically complete these missing parts, the system first compares the probability mean calculated by the sliding window with the preset probability threshold. The probability threshold is an important criterion used by the system to determine whether a point can be considered as belonging to the curved beam. If the probability mean of a certain area is higher than the threshold, it means that the point in the area has a high degree of match with the curved beam. The system can consider that the area is part of the curved beam and is suitable for extension. The determination of the extension path is based on the curvature characteristics of the curved beam. The system analyzes the geometric curvature changes of the arc to be processed and calculates the extension direction and the extension path of the arc. In this process, the system combines the existing geometric information to calculate how the extended part should maintain the curvature continuity with the known part. For example, if the existing arc shows a large curvature change, the system will ensure that the extended part transitions in space in a smooth manner to avoid abrupt geometric turns. The end point of the extension path is determined based on further analysis of the probability value. As the arc extends, the system will gradually reduce the probability mean in the sliding window until the mean approaches the preset probability threshold lower limit. At this point, the system will consider that the extended part of the arc beam has reached the end and determine the end point. Accurate determination of the end point is crucial to ensuring the geometric integrity of the arc beam. The system uses this mechanism to avoid the extended part being too long or too short, ensuring that the completed arc beam can be seamlessly connected to the existing structure. Finally, the system supplements and extends the arc to be processed based on the extended path and end point. In this step, the system uses a geometric interpolation algorithm (such as spline interpolation or Bezier curve interpolation) to generate smooth extended lines.The interpolation algorithm uses multiple key points on the extended path to generate a smooth curve that conforms to the curvature characteristics of the curved beam. This algorithm ensures that the newly generated curve not only matches the existing curved beam geometry, but also exhibits a natural transition in space.

[0077] Furthermore, the process of obtaining the extended arc and remapping it in the three-dimensional model to perform visualization and calibrating the arc beam position by the extended arc includes the following steps:

[0078] The arc data generated by the supplementary extension is mapped back to the global coordinates of the indoor three-dimensional model according to the spatial coordinate system;

[0079] Based on the mapped extended arc, a visualization algorithm is applied in the three-dimensional model to convert the arc into a geometric display model through a graphics rendering engine;

[0080] Semantically calibrate the specific position of the curved beam in the geometric display model.

[0081] In some embodiments, first, the system integrates the extended part into the global coordinates of the indoor three-dimensional model by supplementing the arc data generated by the extension in accordance with the spatial coordinate system mapping method. Specifically, in the process of generating the extended arc, the system uses an interpolation algorithm (such as spline interpolation or Bezier interpolation) to supplement the missing part of the curved beam. These supplemented arc data exist in the form of a local coordinate system. In order to accurately integrate them back into the three-dimensional building model, the system needs to perform a global coordinate conversion on these data. When mapping, the system first determines the relationship between the global coordinate system in the three-dimensional model and the local coordinate system of the extended arc. This relationship is usually described by a rotation matrix and a translation vector to ensure that the extended arc can appear at the correct position of the three-dimensional model. In a building, any offset of the three-dimensional coordinates will affect the integrity of the geometric structure, so the geometric accuracy must be maintained during the mapping process to ensure that the extended arc is completely consistent with the other parts of the original curved beam in position, direction and proportion. After the mapping is completed, the system processes the geometric information of the extended arc in the three-dimensional model. Specifically, the system converts these geometric data into a displayable three-dimensional geometric model through a graphics rendering engine. Graphic rendering is the core of the entire visualization process. It converts the mathematical expression of the extended arc into a visual geometric object. The rendering engine is responsible for processing the three-dimensional geometric form, lighting, material and other characteristics of the arc to ensure that the structure of the curved beam is realistically displayed in the three-dimensional view. During the visualization rendering process, the system generates a visual model of the arc according to the curvature, arc length and spatial position of the curved beam, and integrates it with the original building model. The rendering engine ensures that the display effect of the extended arc is consistent with the surrounding building structure through efficient graphics calculation to avoid geometric distortion or display errors. For example, if the extended curved beam has a more complex curvature change, the rendering engine can accurately reflect its smooth transition curve while ensuring its visual consistency in three-dimensional space. In addition, in order to improve the accuracy and authenticity of visualization, the system can introduce detail enhancement technology in the rendering process. For example, by analyzing the curvature of the arc, the display accuracy of the arc is appropriately adjusted to ensure that the key geometric characteristics can be captured visually. After the rendering is completed, the user can view the curved beam from different angles in the three-dimensional view to confirm whether its geometric features meet the design requirements. Finally, the system semantically calibrates the specific position of the extended curved beam in the 3D geometric model. Semantic calibration refers to matching the curved beam in the geometric model with its function and meaning in the actual building. Through semantic calibration, the system can establish a clear association between geometric objects (such as curved beams) and their architectural structural roles, which facilitates subsequent construction design and structural analysis.

[0082] The above implementation modes are merely descriptions of the preferred implementation modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary engineering and technical personnel in the field shall fall within the protection scope determined by the claims of the present invention.

Claims

1. A method for identifying curved beams based on the extension of line geometric characteristics, characterized in that: The following steps are involved: Scanning the interior of the building and obtaining an indoor three-dimensional model, extracting edge lines from the indoor three-dimensional model and segmenting monomer lines, and normalizing the monomer lines to obtain standardized line parameters; By traversing the preset arc parameter library, the corresponding monomer lines whose standardized line parameters meet the preset arc parameters are selected and marked as arcs to be processed, wherein the preset arc parameters include curvature, arc range and arc length; Taking the arc to be processed as the center, obtain the monomer lines connecting the arc to be processed and combine them with the arc to be processed to build a beam recognition geometric model, identify and classify the beam recognition geometric model based on the convolutional neural network, and output the point probability distribution of the arc to be processed and the arc beam; Based on the probability distribution of the matching points between the arc to be processed and the curved beam, the arc to be processed is supplemented and extended in the beam recognition geometric model to obtain the extended arc and remap it in the three-dimensional model for visualization and the curved beam position is calibrated by the extended arc.

2. The arc beam identification method based on line geometric characteristic extension according to claim 1, characterized in that: The scanning of the interior of the building and obtaining the indoor three-dimensional model comprises: The laser scanning equipment is used to perform an all-round scan of the interior of the building to obtain point cloud data including the wall, ceiling and ground structure, and the point cloud data is subjected to noise reduction and filtering to generate a three-dimensional indoor model.

3. The arc beam identification method based on line geometric characteristic extension according to claim 1, characterized in that: The method of extracting edge lines from the indoor three-dimensional model and segmenting the individual lines comprises the following steps: Applying edge detection algorithm to the generated indoor 3D model, identifying edge lines of building structures such as walls, ceilings and floors from the 3D model as initial lines; Splits the initial line into individual lines based on the curvature and inflection points of the line.

4. The arc beam identification method based on line geometric characteristic extension according to claim 1, characterized in that: The method of taking the arc to be processed as the center, obtaining the monomer lines connecting the periphery of the arc to be processed and combining them with the arc to be processed to construct the beam recognition geometric model comprises the following steps: Taking the geometric center of the arc to be processed as the base point, a spatial index algorithm is used to search in three-dimensional space for neighboring monomer lines whose distance to the arc to be processed is less than a preset distance threshold; The neighboring monomer lines are geometrically combined with the arcs to be processed, and the connections of the monomer lines are smoothed by the spline interpolation algorithm; The connection errors between lines are dynamically adjusted through the error correction algorithm to generate a beam recognition geometric model.

5. The arc beam identification method based on line geometric characteristic extension according to claim 1, characterized in that: The method of identifying and classifying the beam recognition geometric model based on the convolutional neural network and outputting the point probability distribution of the arc to be processed and the arc-shaped beam matching includes the following steps: The recognized geometric model is input into the convolutional neural network model, and the spatial connection relationship between the single line and its adjacent lines in the geometric model is extracted through the convolution layer to generate a geometric model feature map; Perform multi-layer convolution and pooling operations on the extracted geometric model feature map to fuse local geometric features at different scales to obtain a multi-scale fused geometric feature map; The multi-scale fused geometric feature map is input into the fully connected layer, and the geometric features are classified by the Softmax activation function. The probability distribution of the matching between the arc to be processed and the curved beam structure is calculated, and the probability value of the point where each arc to be processed matches the curved beam in the beam recognition geometric model is output.

6. The arc beam identification method based on line geometric characteristic extension according to claim 5, characterized in that: The formula of the convolutional neural network is as follows: Among them, Z i，j，k is the probability distribution of the kth curved beam or non-curved structure at the (i, j) position of the final output; X i+m，j+n is the parameter set of the geometric feature graph at position (i+m, j+n), including the standardized geometric parameters of the arc to be processed and its surrounding monomer lines; W m，n，k is the weight of the convolution kernel at position (m, n) corresponding to the kth curved beam or non-curved structure; b k is the bias term of the kth category; Softmax is the activation function; M and N are the height and width of the convolution kernel respectively.

7. The arc beam identification method based on line geometric characteristic extension according to claim 1, characterized in that: The loss function formula of the convolutional neural network is as follows: Among them, L is the total loss function; N is the total number of samples; y i is the true label of the i-th sample; p(y i ) is the predicted probability of the convolutional neural network for the i-th sample; is the regularization term, W k is the weight matrix of the k-th convolution kernel; λ is the regularization coefficient; K is the total number of layers of the convolutional neural network.

8. The arc beam identification method based on line geometric characteristic extension according to claim 1, characterized in that: The method of supplementing and extending the arc to be processed in the beam recognition geometric model based on the point probability distribution matching the arc to be processed and the curved beam comprises the following steps: The probability mean of the probability distribution of the points matching the arc beam in the window is calculated by sliding the preset window size; Determine the extension path and end point based on the probability mean and the preset probability threshold; The arc to be processed is supplemented and extended according to the extended path and end point.

9. The arc beam identification method based on line geometric characteristic extension according to claim 8, characterized in that: The steps of obtaining the extended arc and remapping it in the three-dimensional model to perform visualization and calibrate the arc beam position by the extended arc include the following steps: The arc data generated by the supplementary extension is mapped back to the global coordinates of the indoor three-dimensional model according to the spatial coordinate system; Based on the mapped extended arc, a visualization algorithm is applied in the three-dimensional model to convert the arc into a geometric display model through a graphics rendering engine; Semantically calibrate the specific position of the curved beam in the geometric display model.

Images