Intelligent generation method for deepening drawing of slender steel member of modular building

Abstract

The invention provides an intelligent generation method for a deepening drawing of a long and thin steel member of a modular building. According to the method, a view is generated through geometric operation projection according to three-dimensional geometric information of the slender component, image features of the view of the slender component are extracted, change characteristics are analyzed, determination of a cut-off position and a sectioning position of the slender component is achieved, a corresponding sectioning graph and a cut-off graph of the slender component are generated intelligently, and self-adaptive size marking of all views is completed. Clear and complete expression of thin and long component drawings is achieved, and drawing space is saved.

Description

Technical Field

[0001] The present invention relates to the fields of information technology and architectural detailed design, and in particular to a method for intelligently generating detailed drawings of slender steel components of modular buildings. Background Art

[0002] Modular buildings have been increasingly used in the construction field in recent years due to their significant advantages such as fast construction speed, high project quality, low labor consumption, energy saving and environmental protection.

[0003] In traditional modular building component design, detailed design drawings for slender components often rely on manual drawing, which is not only time-consuming and labor-intensive but also prone to inaccuracies due to human factors. There are also methods that generate drawings directly based on the model's 3D information through geometric calculations and projection. However, for slender components, due to their large aspect ratio, drawings generated directly from 3D projection are difficult to present completely and harmoniously in the drawing. This not only wastes drawing space but also affects the clarity of the drawing. Furthermore, the detailed design of modular building components often requires coordination with mechanical and electrical design. In practical applications, components often require a large number of voids to accommodate the installation of mechanical and electrical equipment and the routing of pipelines. To fully express the detailed features of the component, especially those with voids or special variations, corresponding cross-sectional drawings are required. However, existing methods are unable to intelligently identify these voids and automatically generate corresponding cross-sectional drawings.

[0004] In view of the above problems, it is urgent to propose a new method that can efficiently and intelligently generate detailed drawings of slender steel components of modular buildings. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for intelligently generating detailed drawings of slender steel components of modular buildings to solve the problems existing in the prior art.

[0006] The technical solution adopted to achieve the purpose of the present invention is as follows: a method for intelligently generating detailed drawings of slender steel components of modular buildings, comprising the following steps:

[0007] 1) Perform parametric modeling of slender components of modular buildings.

[0008] 2) Extract the point, line and surface information of the modular building slender component model, and use the projection elimination method to generate the three-view image of the slender component.

[0009] 3) Use the image segmentation method to segment the three-view image into grids in the length direction, calculate the average grayscale value of each segmented area, and form a histogram.

[0010] 4) Generate corresponding sectional views and truncation views according to the sectional position and truncation position. Step 4) includes the following sub-steps.

[0011] 4.1) The Euclidean distance of the histogram is calculated along the length direction to measure the similarity of the histogram. The change in similarity represents the change characteristics of the slender component. The truncation position and sectioning position are determined based on the change characteristics.

[0012] 4.2) Use Hough transform to detect circular cavities in slender components and correct the cutting position.

[0013] 4.3) Cut at all obtained cutting positions to obtain specific cutting images, and use grid segmentation to segment the cutting images and combine the same cutting images with the hierarchical clustering algorithm to obtain the final cutting position.

[0014] 5) Using the line data of the view, the inner layer void information and outer layer component information are divided into separate annotations based on the line connection relationship. The outer layer components use a depth sorting method to cover each layer to determine the required dimensions, and the inner layer voids are dimensioned by combining the graphics fitting optimization algorithm.

[0015] Furthermore, in step 1), the parametric information of the modular building slender components is saved as a JSON file, and parametric modeling of the modular building slender components is implemented based on the IFC standard. The parametric information includes relationship information, attribute information, classification and identification information, geometric shape information, position and orientation information, and topology information. Using this parametric information and the IFC OpenShell library, the slender components are parametrically established using a stretching and solid opening method. A rotation and translation matrix is then constructed using the position information to place all components in their corresponding positions, completing the parametric creation of the modular building slender component model.

[0016] Furthermore, in step 2), all the faces whose normal vectors n are dot-producted with the projection direction l are considered as visible faces if the result is greater than zero. All visible face information is sorted by depth. The sorting method is to take a face i from all faces in turn, and the maximum and minimum values of all vertices of i in the projection direction are defined as a and b respectively. max and a min , compare i with the maximum value b of the vertex direction of all remaining faces max and minimum value b min For comparison, perform the following calculations:

[0017] a min ≥b max , i+1

[0018] b min >a max , i-1

[0019] Sorting in descending order according to the value of i gives the depth sequence of the faces. According to the depth sequence, cover the faces layer by layer from high to low to obtain the projection view of the slender member in a certain direction. Project separately in the x, y, and z directions to obtain the three-view drawings.

[0020] Further, in step 4.2), map the edge points from the image space to the parameter space, and find the center and radius of the circle through the voting mechanism in the parameter space. Calculate each possible (a, b, r) in the parameter space:

[0021]

[0022] where H(a, b, r) is the cumulative value in the parameter space, and γ is the Dirac function used to determine whether a point satisfies the corresponding parameter relationship.

[0023] Further, after performing model sectioning on all obtained section positions and projecting to obtain the section images, divide each section image into a 10×10 grid, use the method of average gray value as the grid feature, and flatten it into a one-dimensional array. Then the i-th section drawing:

[0024] m i =[v1 v2 … v 100

[0025] For these section drawings, calculate the Euclidean distance between each pair to obtain the symmetric distance matrix D. i Based on the distance matrix D, use the hierarchical clustering algorithm to group the images, and use the Ward minimum variance algorithm to calculate the distance between clusters. The method is:

[0026]

[0027] where C i , C j are clusters. μ i , μ j are the center vectors of the clusters, and ||μ i - μ j || 2 is the Euclidean distance between the center vectors. max min max min min Set the dynamic threshold t:

[0031]

[0032]

[0033] Determine the actual required section positions through clustering, avoid the repeated occupation of redundant drawing space by section drawings, and generate corresponding section drawings according to the section positions. ​

[0033] Furthermore, in step 5), the method for determining the marking starting point of the outer component information is as follows: all line information is dot-producted with the marking direction vector, and the result is not equal to zero, and is considered visible. The visible lines are sorted in depth in the direction of the marking information, and a line i is selected from all the lines in turn. The maximum and minimum values of the two vertices of i in the projection direction are defined as a and b respectively. max and a min , the maximum value b of the vertex direction of i and all remaining lines (set as j) max and minimum value b min For comparison, perform the following calculations:

[0034] a min ≥b max , i+1

[0035] b min >a max , i-1

[0036] The depth sequence of the lines is sorted from large to small according to the value of i. According to the depth sequence, the lines are covered layer by layer from high to low, and the result is the information of the lines to be marked.

[0037] Furthermore, in step 5), the inner layer hole information is determined as follows: the main recognition targets are polygons and circles. The recognition method is: in the 3D model, the circle is approximately represented by a polyline, and the polygon is initially determined based on the number of line segments. For other line segment groups, all point information is extracted and the circle is fitted using the least squares method. Assuming that n data points (x i ,y i ), the fitting goal is to find the parameters of the circle (a, b, r) so that the sum of squared errors between the data points and the fitted circle is minimized:

[0038]

[0039] t=0.05r

[0040] The residual threshold t is set to 5% of the radius, that is, if there is a point that is not on the circle and the difference between the distance and the radius is greater than t, it is considered to be other shapes, and this shape is only marked with the outer contour range.

[0041] The technical effect of the present invention is unquestionable: views are generated based on the geometric calculation projection of the three-dimensional geometric information of the slender component, and the image features of the slender component views are extracted and the change characteristics are analyzed, so as to determine the truncation position and section position of the slender component, intelligently generate the corresponding slender component section view and truncation view, and complete the adaptive dimensioning of all views, so as to achieve clear and complete expression of the slender component drawings and save drawing space. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1Flow chart of the intelligent generation method for the detailed drawings of slender steel components in modular buildings;

[0043] Figure 2 Standard box model diagram of modular buildings;

[0044] Figure 3 Image processing method diagram for determining the cutting and truncation positions;

[0045] Figure 4 Hierarchical clustering method diagram for sectional views;

[0046] Figure 5 Adaptive dimensioning method diagram;

[0047] Figure 6 Intelligent generation result of the detailed drawings of slender steel components. Specific implementation manners

[0048] The present invention will be further described below in conjunction with embodiments, but it should not be understood that the above-mentioned subject scope of the present invention is limited to the following embodiments. Without departing from the above-mentioned technical idea of the present invention, various substitutions and changes made according to the common general knowledge and conventional means in the art shall be included within the protection scope of the present invention.

[0049] Embodiment 1:

[0050] In recent years, image processing technology has developed rapidly and is widely used in various professional fields for feature extraction and recognition, extracting key features (such as edges, corners, textures, etc.) from images, and performing pattern recognition for classification, detection, etc. However, this method is less applied in the field of architectural drawing generation.

[0051] In view of the need for intelligent generation of detailed drawings of slender components in modular buildings, this embodiment proposes a method for intelligently identifying the truncation position and cutting position of slender components based on image processing technology, intelligently generating corresponding sectional views and truncation views of slender components, and realizing adaptive dimensioning for all views. Refer to Figure 1 , this embodiment provides an intelligent generation method for the detailed drawings of slender steel components in modular buildings, including the following steps:

[0052] 1) Parametric modeling of slender components of modular buildings. Save the parametric information of slender components of modular buildings as a JSON file, and implement parametric modeling of slender components of modular buildings based on the IFC standard. Among them, the parametric information includes relationship information, attribute information, classification and identification information, geometric shape information, position and orientation information, and topological information. Using the parametric information and based on the ifcopenshell library, adopt the method of stretching and solid cutting to realize the parametric establishment of slender components, and then construct a rotation and translation matrix through the position information to place all components in the corresponding positions, completing the parametric creation of the slender component model of modular buildings.

[0053] 2) Extract the point, line, and surface information of the slender component model of modular buildings, and generate three-view images of the slender components by means of projection and shading elimination. Take the dot product of the normal vector n of all surfaces and the projection direction l, and regard the result greater than zero as the visible surface. Sort the information of all visible surfaces in depth. The sorting method is as follows: Take a surface i from all surfaces in turn, and define the maximum and minimum values of all vertices of i in the projection direction as a max and a min , compare i with the maximum value b max and the minimum value b min of the vertex directions of all the remaining surfaces, and perform the following calculations:

[0054] a min ≥b max , i + 1

[0055] b min >a max , i - 1

[0056] Sort according to the value of i from large to small to obtain the depth sequence of the surfaces. According to the depth sequence, cover the surfaces layer by layer from high to low to obtain a projection view of the slender component in a certain direction. Project separately in the x, y, and z directions to obtain three views.

[0057] 3) Use the method of image segmentation to segment the three-view image in the length direction according to the grid, calculate the average gray value of each segmented area, and form a histogram.

[0058] 4) Generate corresponding sectional views and truncated views according to the cutting position and truncation position. Step 4) includes the following sub-steps.

[0059] 4.1) Calculate the Euclidean distance of the histogram in the length direction to measure the similarity of the histogram. The change in similarity represents the change characteristics of the slender component. Determine the truncation position and cutting position according to the change characteristics.

[0060] 4.2) Use the Hough transform to detect circular holes in slender components and correct the sectioning position. Map the edge points from image space to parameter space, and use a voting mechanism in the parameter space to find the center and radius of the circle. Calculate for each possible (a, b, r) in the parameter space:

[0061]

[0062] Among them, H(a, b, r) is the cumulative value in the parameter space, and δ is the Dirac function, which is used to determine whether the point satisfies the corresponding parameter relationship.

[0063] 4.3) Cut at all obtained section locations to obtain specific section images. Use grid segmentation and hierarchical clustering to merge identical section images to obtain the final section location. Based on all obtained section locations, perform model sectioning and post-projection to obtain section images. Each section image is divided into a 10×10 grid, and the average grayscale value is used as the grid feature. Flatten it into a one-dimensional array. Then, for the i-th section image:

[0064] m i =[v1 v2 … v 100 ]

[0065] These cross-sectional views of m i Calculate the Euclidean distance between each pair to get the symmetric distance matrix D.

[0066]

[0067] Based on the distance matrix D, a hierarchical clustering algorithm is used to group images, and the distance between clusters is calculated using the Ward minimum variance algorithm. The method is:

[0068]

[0069] Among them, C i ,C j is a cluster. i ,μ j is the center vector of the cluster, ||μ i -μ j || 2 is the Euclidean distance between the center vectors.

[0070] Set the dynamic threshold t:

[0071]

[0072] The actual required sectioning position is determined by clustering to avoid the sectioning diagram from repeatedly occupying redundant drawing space, and the corresponding sectioning diagram is generated according to the sectioning position.

[0073] 5) Use the line data information of the view, and label the inner cavity information and the outer component information separately according to the connection relationship of the lines. Among them, for the outer components, the depth sorting method is used to cover layer by layer to determine the dimension information to be labeled, and for the inner cavities, the optimized algorithm of graphic fitting is used to realize the dimension labeling of the internal cavities.

[0074] The method for determining the starting point of the label of the outer component information is as follows: Take the dot product of all line information and the label direction vector, and consider the result not equal to zero as visible. Perform depth sorting on the visible lines in the direction of the label information. Take a line i from all the lines in turn. The maximum and minimum values of the two vertices of i in the projection direction are defined as a max and a min , and compare the maximum value b max and the minimum value b min of the vertex directions of i and all the remaining lines (assumed to be j), and perform the following calculations:

[0075] a min ≥b max , i + 1

[0076] b min >a max , i - 1

[0077] Sort the lines from largest to smallest according to the value of i to obtain the depth sequence of the lines. According to the depth sequence, cover the lines layer by layer from high to low, and the obtained result is the line information to be labeled.

[0078] Furthermore, in step 5), the method for determining the inner cavity information is as follows: The main recognition targets are polygons and circles. The recognition method is: In the 3D model, a circle is approximately represented by a polyline. Initially determine it as a polygon according to the number of line segments. For other line segment groups, extract all point information and perform circle fitting using the least squares method. Assume n data points (x i , y i ), and the fitting target is to find the parameters (a, b, r) of the circle to minimize the sum of the squared errors between the data points and the fitted circle:

[0079]

[0080] t = 0.05r

[0081] Set the residual threshold t to 5% of the radius, that is, if there are points not on the circle and the difference between the distance and the radius is greater than t, then it is considered to be other shapes, and only the external contour range of this shape is labeled.

[0082] Example 2:

[0083] The main content of this embodiment is the same as that of Embodiment 1. Among them, based on JSON data, parametric information is saved in a JSON file, including relationship information (hierarchical information, dependency relationship), attribute information (material, physical properties), classification and identification information (entity type, identifier, name), geometric shape information (length, width, height, cross-section), position and orientation information (member position, rotation direction), topological information (connection relationship, component structure), etc. Using this parametric information and based on the ifcopenshell library, the method of stretching and creating holes in the solid is adopted to realize the parametric establishment of slender members. Then, a rotation and translation matrix is constructed through the position information, and all members are placed at the corresponding positions to complete the parametric creation of the slender member model of the modular building.

[0084] The position and rotation direction are represented as (x, y, z, θ x , θ y , θ z ), where x, y, and z represent the x, y, and z coordinates of the member, and θ x , θ y , θ z represent the angles of rotation around the x, y, and z axes. R y , R y , R z are the rotation and translation matrices corresponding to the x, y, and z axes respectively.

[0085]

[0086] The rotation matrix R is:

[0087] R = R z ·R y ·R x

[0088] The translation matrix T is: The rotation and translation matrix is:

[0089]

[0090] Apply the transformation:

[0091] P' = M · P

[0092] Embodiment 3:

[0093] The main content of this embodiment is the same as that of Embodiment 1. Among them, the point, line, and surface information of the slender member model is extracted, the projection direction is determined, and then the dot product of the normal vector n of all surfaces and the projection direction l is calculated. If the result is greater than zero, it is regarded as a visible surface. The information of all visible surfaces is sorted by depth. The sorting method is to take a surface i from all surfaces in turn, and the maximum and minimum values of all vertices of i in the projection direction are defined as a max and a min, compare i with the maximum value b and the minimum value b of the vertex directions of all the remaining faces (assumed to be j), and perform the following calculations: max and the minimum value b min Compare, and perform the following calculations:

[0094] a min ≥b max , i + 1

[0095] b min >a max , i - 1

[0096] Sort according to the value of i from large to small to obtain the depth sequence of the faces. According to the depth sequence, cover the faces layer by layer from high to low to obtain a projection view of the slender member in a certain direction. Project separately in the x, y, and z directions to obtain three views.

[0097] Example 4:

[0098] The main content of this example is the same as that of Example 1. Among them, according to the generated three-view grayscale image, divide the image into grids in the length direction, and set the grid length to the line width, that is

[0099] g = (linewidth · dpi) / 72

[0100] Among them, g represents the number of pixel points of the grid width, linewidth represents the width of the line segment, the unit is point, and dpi is the value of the image DPI.

[0101] After dividing the image into grids in this direction, calculate the average grayscale value in each grid as the grid feature value v:

[0102]

[0103] Among them, N is the total number of pixels in the grid, and x i represents the grayscale value of the i-th pixel. Each column in the length direction can be represented as a one-dimensional sequence m:

[0104] m = [v1 v2 … v k T

[0105]

[0106] Among them, H is the total number of pixels of the image width.

[0107] Example 5:

[0108] The main content of this example is the same as that of Example 1. Among them, according to the m obtained in step 3), the entire view image can be expressed as:

[0109] [m1 m2 … m n ​​

[0110]

[0111] Where W is the number of pixels that the image has in length.

[0112] Calculate the Euclidean distance between adjacent grids m and obtain the grid change characteristics:

[0113]

[0114] d=[d1d2…d i …d n-1 ]

[0115] The one-dimensional array d representing the characteristic changes of the slender component along the length direction is obtained. The truncation position represents a long section with no obvious change, that is, a continuous zero position, and this position is also used as the pre-selected cutting position. At the same time, d i ≠0 means there is a change, then d i ≠0 as a separation of different characteristic segments along the length direction, if the continuous existence d i If ≠0, it means that there is a curve in this part, which is often a location where a cavity exists. In this case, the gradient change of this section d is calculated, and the minimum gradient value is taken as the cutting position. The gradient calculation method is as follows:

[0116]

[0117] Example 6:

[0118] The main contents of this embodiment are the same as those of embodiment 1. The corresponding sectional and truncation views are generated based on the sectional and truncation positions. The line data information of the view is used to group the lines according to their connection relationships, i.e., line segments with connected endpoints are grouped together. The lines are then divided into inner and outer groups based on the coordinate range of the grouped line segments, including inner layer cavity information and outer layer component information, which are labeled separately. Component information is labeled in four directions: up, down, left, and right. The required information includes: reference point, labeling start point, end point, labeling text height, labeling text position (up / down), etc.

[0119] Example 7:

[0120] This embodiment selects a modular building standard box and uses the slender components therein as an example. Figures 2 to 6 , Figure 6 (a)Ⅰ is the three-dimensional model of the corner column. Figure 6 (b)Ⅰ is the detailed drawing of the corner column. Figure 6 (a)Ⅱ is the three-dimensional model of the bottom frame longitudinal beam, Figure 6 (b)Ⅱ is the detailed drawing of the bottom frame longitudinal beam. Figure 6 (a) III, IV, V, and V are three-dimensional models of steel beam I.Figure 6 (b)Ⅲ is the detailed drawing of steel beam Ⅰ. Figure 6 (a) IV is the three-dimensional model of the steel support, Figure 6 (b) IV is the detailed drawing of the steel support. Figure 6 (a) V is the horizontal three-dimensional model of the steel beam, Figure 6 (b) V is a transverse detail drawing of a steel beam. Experimental results demonstrate that the algorithm for intelligent segmentation and extraction of building interior scene components disclosed in this invention can intelligently identify locations requiring sectioning and truncation based on IFC-standard 3D models of modular slender building components, adaptively generate complete dimensioning, and intelligently produce detail drawings of modular slender building components.

Claims

1. An intelligent generation method for the detailed drawings of slender steel members of modular buildings, characterized in that, The following steps are involved: 1) Parametric modeling of slender components of modular buildings; 2) Extract the point, line, and surface information of the modular building slender component model and generate the three-view image of the slender component using the projection elimination method; 3) Using image segmentation methods to segment the three-view image into grids along the length direction, calculate the average grayscale value of each segmented area, and form a histogram; 4) generating corresponding sectional views and truncation views according to the sectional position and the truncation position; Step 4) includes the following sub-steps; 4.1) The Euclidean distance of the histogram is calculated along the length direction to measure the similarity of the histogram. The change in similarity represents the change characteristics of the slender component. The truncation position and sectioning position are determined based on the change characteristics. 4.2) Using Hough transform to detect circular cavities in slender components for correcting the sectioning position; 4.3) Cut at all obtained cutting positions to obtain specific cutting images, and use grid segmentation and hierarchical clustering algorithm to merge the same cutting images to obtain the final cutting position; 5) Using the line data information of the view, the inner layer void information and outer layer component information are divided into separate annotations according to the connection relationship of the lines; among them, the outer layer components use the depth sorting method to cover layer by layer to determine the size information that needs to be annotated, and the inner layer voids are combined with the graphics fitting optimization algorithm to realize the size annotation of the internal voids.

2. The intelligent generation method of the detailed drawing of the slender steel member of the modular building according to claim 1, characterized in that: In step 1), the parametric information of the modular building slender components is saved as a JSON file, and the parametric modeling of the modular building slender components is realized based on the IFC standard; wherein the parametric information includes relationship information, attribute information, classification and identification information, geometric shape information, position and orientation information, and topology information; using the parametric information and based on the ifcopenshell library, the stretching and solid opening methods are adopted to realize the parametric establishment of the slender components, and then the rotation and translation matrix is constructed according to the position information, and all components are placed in the corresponding positions to complete the parametric creation of the modular building slender component model.

3. The intelligent generation method of the detailed drawing of the slender steel member of the modular building according to claim 1, characterized in that: In step 2), the dot product of the normal vector n of all faces and the projection direction l is calculated. If the result is greater than zero, the face is regarded as a visible face. Then, the information of all visible faces is sorted by depth. The sorting method is as follows: Take a face i from all faces in turn. The maximum and minimum values of all vertices of i in the projection direction are defined as a max and a min . Compare i with the maximum value b max and the minimum value b min of the vertex directions of all the remaining faces, and perform the following calculations: a min ≥b max ,i + 1 b min >a max , i - 1 The depth sequence of the surface is obtained by sorting the i values from large to small. According to the depth sequence, the surface is covered layer by layer from high to low to obtain a projection view of a slender component in a certain direction; the three views are obtained by projecting in the x, y, and z directions respectively.

4. The intelligent generation method of the detailed drawing of the slender steel member of the modular building according to claim 1, wherein: In step 4.2), the edge points are mapped from the image space to the parameter space, and the center and radius of the circle are found through the voting mechanism in the parameter space; for each possible (a, b, r) in the parameter space, the following equations are calculated: Among them, H(a, b, r) is the cumulative value in the parameter space, and δ is the Dirac function, which is used to determine whether the point satisfies the corresponding parameter relationship.

5. The intelligent generation method of the detailed drawing of the slender steel member of the modular building according to claim 1, characterized in that: In step 4.3), the model is sectioned based on all the obtained section positions and then projected to obtain section images. Each section image is divided into a 10×10 grid, and the average gray value method is used as the grid feature. The grid is flattened into a one-dimensional array. Then, the i-th section image is: m i = [v1 v2 … v 100 ​ The m of these cross-sectional views i Calculate the Euclidean distance between each pair to obtain the symmetric distance matrix D, Based on the distance matrix D, a hierarchical clustering algorithm is used to group images, and the distance between clusters is calculated using the Ward minimum variance algorithm. The method is: Among them, C i , C j is a cluster; μ i , μ j is the center vector of the cluster, ||μ i - μ j || 2 is the Euclidean distance between the center vectors. Set the dynamic threshold t: The actual required sectioning position is determined by clustering to avoid the sectioning diagram from repeatedly occupying redundant drawing space, and the corresponding sectioning diagram is generated according to the sectioning position.

6. The intelligent generation method of the detailed drawing of the slender steel member of the modular building according to claim 1, characterized in that: In step 5), the method for determining the starting point of the annotation of the outer component information is as follows: Take the dot product of all line information and the annotation direction vector. A non-zero result is regarded as visible. Sort the visible lines in the depth direction in the direction of the annotation information. Sequentially select a line i from all lines. The maximum and minimum values of the two vertices of line i in the projection direction are defined as a max and a min , compare the maximum value b max and the minimum value b min of the vertex directions of line i and all the remaining lines (denoted as j), and perform the following calculations: a min ≥b max ,i + 1 b min >a max , i-1 The depth sequence of the line is obtained by sorting the i values from large to small. According to the depth sequence, the lines are covered layer by layer from high to low, and the result is the line information that needs to be marked.

7. The intelligent generation method of the detailed drawing of the slender steel member of the modular building according to claim 1, characterized in that: In step 5), the method for determining the inner cavity information is as follows: the main recognition targets are polygons and circles; the recognition method is: in the three-dimensional model, a circle is approximately represented by a polyline. If the number of line segments is preliminarily determined to be a polygon, for other line segment groups, all point information is extracted, and circle fitting is performed using the least squares method. Assume n data points (x i , y i ). The fitting target is to find the parameters (a, b, r) of the circle to minimize the sum of the squared errors between the data points and the fitted circle: t＝0.05r The residual threshold t is set to 5% of the radius, that is, if there is a point that is not on the circle and the difference between the distance and the radius is greater than t, it is considered to be other shapes, and this shape is only marked with the outer contour range.

Images