CN116958454B - Construction contour construction method and device based on graph network - Google Patents
Construction contour construction method and device based on graph network Download PDFInfo
- Publication number
- CN116958454B CN116958454B CN202311221026.4A CN202311221026A CN116958454B CN 116958454 B CN116958454 B CN 116958454B CN 202311221026 A CN202311221026 A CN 202311221026A CN 116958454 B CN116958454 B CN 116958454B
- Authority
- CN
- China
- Prior art keywords
- building
- graph
- line segments
- contour
- line segment
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000010276 construction Methods 0.000 title claims abstract description 28
- 238000013528 artificial neural network Methods 0.000 claims abstract description 53
- 230000011218 segmentation Effects 0.000 claims abstract description 20
- 238000012549 training Methods 0.000 claims description 38
- 238000000034 method Methods 0.000 claims description 18
- 238000012545 processing Methods 0.000 claims description 9
- 238000010606 normalization Methods 0.000 claims description 5
- 238000000547 structure data Methods 0.000 claims description 4
- 238000003860 storage Methods 0.000 description 7
- 230000008901 benefit Effects 0.000 description 6
- 238000004590 computer program Methods 0.000 description 6
- 230000000694 effects Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000004519 manufacturing process Methods 0.000 description 3
- 239000000203 mixture Substances 0.000 description 3
- 230000004075 alteration Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000007792 addition Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 239000000178 monomer Substances 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005215 recombination Methods 0.000 description 1
- 230000006798 recombination Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/13—Architectural design, e.g. computer-aided architectural design [CAAD] related to design of buildings, bridges, landscapes, production plants or roads
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/042—Knowledge-based neural networks; Logical representations of neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T19/00—Manipulating 3D models or images for computer graphics
- G06T19/20—Editing of 3D images, e.g. changing shapes or colours, aligning objects or positioning parts
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/70—Labelling scene content, e.g. deriving syntactic or semantic representations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2210/00—Indexing scheme for image generation or computer graphics
- G06T2210/04—Architectural design, interior design
Abstract
The invention provides a building contour construction method and device based on a graph network. The building contour construction method based on the graph network comprises the following steps: obtaining a single building model based on the obtained image data; performing cross section slicing on the single building model, and obtaining a set building contour picture based on the cross section slicing; identifying segmentation points from the building contour picture, thereby segmenting the continuously closed contour into line segments; constructing the segmentation points and the line segments into a graph structure, and inputting the graph structure into a graph neural network so as to obtain the relationship between the line segment types and the line segments; fitting the line segments according to the relationship between the line segment types and the line segments to obtain a fitting result, and adjusting the positions of the line segment endpoints in the fitting result according to the relationship between the line segments to obtain the building contour. The aim of drawing the building outline rapidly and automatically is achieved.
Description
Technical Field
The invention belongs to the technical field of building three-dimensional reconstruction, and particularly relates to a building contour construction method and device based on a graph network.
Background
In the existing modeling manner, there are two modes for drawing building outlines: one is manually drawn by drawing software, which is very time-consuming; the other is to directly correct the building noise profile obtained directly from the input data at right angles, which is more limited. The prior art has the problems of long time and non-automatic drawing of building outlines.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a building contour construction method and a building contour construction module based on a graph network, which at least partially solve the problems of long building contour drawing time and non-automation in the prior art.
In a first aspect, an embodiment of the present disclosure provides a building contour construction method based on a graph network, including:
obtaining a single building model based on the obtained image data;
performing cross section slicing on the single building model, and obtaining a set building contour picture based on the cross section slicing;
identifying segmentation points from the building contour picture, thereby segmenting the continuously closed contour into line segments;
constructing the segmentation points and the line segments into a graph structure, and inputting the graph structure into a graph neural network so as to obtain the relationship between the line segment types and the line segments;
fitting the line segments according to the relationship between the line segment types and the line segments to obtain a fitting result, and adjusting the positions of the line segment endpoints in the fitting result according to the relationship between the line segments to obtain a building contour;
training the graph neural network to obtain the type of the building contour line segment and the relationship between the line segments, wherein the training comprises the following steps:
constructing a contour line segment into a graph structure data set;
identifying key points for each building contour slice graph, and carrying out normalization processing on coordinates of the key points by using the obtained key points as node characteristics of the graph; the line segment is used as the edge feature of the graph, the length of the line segment and the included angle between the line segment and the horizontal direction are used as two-dimensional features, the marking of two parts of contents is carried out on the outline line segment, and if the line segment is a curve, the marking is 0; if the line segments are straight lines, marking as 0, and if the line segments are vertical, marking as 0; if the line segments intersect and the included angle is not 90 degrees, marking as 1 and attaching the included angle degree; if the line segments are parallel, marking as 2;
the training graph neural network outputs the relationship between the line segment types and the line segments:
inputting the constructed data set into a graph neural network, and extracting characteristics of the graph neural network so as to predict the relation between adjacent line segments;
the obtained line segments and points are constructed into a graph structure, and are used as the input of a graph neural network and output as the relationship between the line segment types and the line segments.
Optionally, the obtaining a single building model based on the obtained image data includes:
the image data of oblique photography is subjected to singulation.
Optionally, the obtaining the set building contour picture based on the cross section slice includes:
and scoring the building contour pictures by using a scoring neural network, and selecting building contour pictures with the score larger than the set score based on the scoring result.
Optionally, scoring neural network training includes:
collecting building crosscut profiles comprising different types of buildings;
marking the score of each building transverse profile slice, marking the continuous noiseless building profile as 1, and representing the good building profile; marking a noisy broken, noisy protruding or recessed building contour as a one-bit decimal from 0 to 1, the higher the value the better the contour quality; marking a noisy and broken and noisy raised or recessed building outline as 0;
and training a building contour scoring network by using the building contour and the corresponding score as training data.
Optionally, the line segment type includes a straight line or a curved line, and the relationship between the line segments includes vertical, intersecting or parallel.
Optionally, the fitting the line segment according to the relationship between the line segment type and the line segment includes:
and if the curve is a straight line, connecting the end points of the line segments, and if the curve is a curve, taking the midpoint of the line segments to draw an arc.
Optionally, the cross section slicing of the single building model includes that the single building model is transversely cut at intervals of 0.1 meter, and intersection lines are obtained by using a plane and the three-dimensional model, so that an outer contour slice sequence of the building is obtained, and the slice sequence is a contour map of the building from top to bottom.
Optionally, constructing the segmentation point and the line segment into a graph structure, inputting the graph structure into a graph neural network, thereby obtaining a training set of the graph neural network in the relationship between the line segment type and the line segment, including:
building contours are randomly obtained through horizontal, horizontal and vertical rules, buildings with random heights are generated through the building contours, and building models are derived;
the building model is used for obtaining model pictures at different angles through virtual camera surrounding, and three-dimensional reconstruction is carried out by using the model pictures and corresponding internal parameters and external parameters of the virtual camera, so as to obtain an oblique photography three-dimensional reconstruction building model;
respectively slicing building contours of the reconstructed building models by using the building models and the oblique photography three-dimensional reconstruction building models to obtain a horizontal and vertical building contour map and a building contour map with noise as paired data;
knowing the position and the connection sequence of each key point in the horizontal and vertical profile, finding the point with the minimum Manhattan distance in the noisy profile to mark as the corresponding key point, and marking by taking the same connection sequence as an edge; meanwhile, the straight line is marked 1, and the curve is marked 0.
Optionally, constructing the segmentation point and the line segment into a graph structure, inputting the graph structure into a graph neural network, thereby obtaining a training set of the graph neural network in the relationship between the line segment type and the line segment, including:
the coordinates of the key points are normalized by taking the key points as node characteristics of the graph; the line segment is used as the edge feature of the graph, the length of the line segment and the included angle between the line segment and the horizontal direction are used as two-dimensional features, the marking of two parts of contents is carried out on the outline line segment, namely the relationship between the line segment type and the line segment, and if the line segment is a curve, the marking is 0; if the line segments are straight lines, marking as 0, and if the line segments are vertical, marking as 0; if the line segments intersect and the included angle is not 90 degrees, the line segments are marked as 1 and the included angle degrees are attached, and if the line segments are parallel, the line segments are marked as 2.
In a second aspect, an embodiment of the present disclosure further provides a building contour construction apparatus based on a graph network, including: the model module is used for obtaining a single building model based on the acquired image data;
the slicing module is used for carrying out cross section slicing on the single building model and obtaining a set building contour picture based on the cross section slicing;
the identification module is used for identifying segmentation points from the building contour picture so as to divide the continuously closed contour into line segments;
the relation module is used for constructing the segmentation points and the line segments into a graph structure, inputting the graph structure into a graph neural network, and thus obtaining the relation between the line segment types and the line segments;
the fitting module is used for fitting the line segments according to the relationship between the line segment types and the line segments so as to obtain a fitting result, and adjusting the positions of the line segment endpoints in the fitting result according to the relationship between the line segments so as to obtain the building outline;
training the graph neural network to obtain the type of the building contour line segment and the relationship between the line segments, wherein the training comprises the following steps:
constructing a contour line segment into a graph structure data set;
identifying key points for each building contour slice graph, and carrying out normalization processing on coordinates of the key points by using the obtained key points as node characteristics of the graph; the line segment is used as the edge feature of the graph, the length of the line segment and the included angle between the line segment and the horizontal direction are used as two-dimensional features, the marking of two parts of contents is carried out on the outline line segment, and if the line segment is a curve, the marking is 0; if the line segments are straight lines, marking as 0, and if the line segments are vertical, marking as 0; if the line segments intersect and the included angle is not 90 degrees, marking as 1 and attaching the included angle degree; if the line segments are parallel, marking as 2;
the training graph neural network outputs the relationship between the line segment types and the line segments:
inputting the constructed data set into a graph neural network, and extracting characteristics of the graph neural network so as to predict the relation between adjacent line segments;
the obtained line segments and points are constructed into a graph structure, and are used as the input of a graph neural network and output as the relationship between the line segment types and the line segments.
According to the building contour construction method and the building contour construction module based on the graph network, which are provided by the invention, the building contour picture is constructed into the graph structure by utilizing the characteristics of the graph neural network, and the relationship between the line segment type and the line segment is obtained more accurately according to the graph characteristics, so that the purpose of drawing the building contour rapidly and automatically is achieved.
Drawings
The foregoing and other objects, features and advantages of the disclosure will be apparent from the following more particular descriptions of exemplary embodiments of the disclosure as illustrated in the accompanying drawings wherein like reference numbers generally represent like parts throughout the exemplary embodiments of the disclosure.
Fig. 1 is a flowchart of a building contour construction method based on a graph network provided in an embodiment of the present disclosure;
FIG. 2 provides a horizontal, vertical profile view for an embodiment of the present disclosure;
fig. 3 is a profile view with noise provided by an embodiment of the present disclosure.
Detailed Description
Embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.
It should be appreciated that the following specific embodiments of the disclosure are described in order to provide a better understanding of the present disclosure, and that other advantages and effects will be apparent to those skilled in the art from the present disclosure. It will be apparent that the described embodiments are merely some, but not all embodiments of the present disclosure. The disclosure may be embodied or practiced in other different specific embodiments, and details within the subject specification may be modified or changed from various points of view and applications without departing from the spirit of the disclosure. It should be noted that the following embodiments and features in the embodiments may be combined with each other without conflict. All other embodiments, which can be made by one of ordinary skill in the art without inventive effort, based on the embodiments in this disclosure are intended to be within the scope of this disclosure.
It is noted that various aspects of the embodiments are described below within the scope of the following claims. It should be apparent that the aspects described herein may be embodied in a wide variety of forms and that any specific structure and/or function described herein is merely illustrative. Based on the present disclosure, one skilled in the art will appreciate that one aspect described herein may be implemented independently of any other aspect, and that two or more of these aspects may be combined in various ways. For example, an apparatus may be implemented and/or a method practiced using any number of the aspects set forth herein. In addition, such apparatus may be implemented and/or such methods practiced using other structure and/or functionality in addition to one or more of the aspects set forth herein.
It should also be noted that the illustrations provided in the following embodiments merely illustrate the basic concepts of the disclosure by way of illustration, and only the components related to the disclosure are shown in the illustrations, rather than being drawn according to the number, shape and size of the components in actual implementation, and the form, number and proportion of the components in actual implementation may be arbitrarily changed, and the layout of the components may be more complex.
In addition, in the following description, specific details are provided in order to provide a thorough understanding of the examples. However, it will be understood by those skilled in the art that the aspects may be practiced without these specific details.
As shown in fig. 1, this embodiment discloses a building contour construction method based on a graph network, including:
obtaining a single building model based on the obtained image data;
performing cross section slicing on the single building model, and obtaining a set building contour picture based on the cross section slicing;
identifying segmentation points from the building contour picture, thereby segmenting the continuously closed contour into line segments;
constructing the segmentation points and the line segments into a graph structure, and inputting the graph structure into a graph neural network so as to obtain the relationship between the line segment types and the line segments;
fitting the line segments according to the relationship between the line segment types and the line segments to obtain a fitting result, and adjusting the positions of the line segment endpoints in the fitting result according to the relationship between the line segments to obtain the building contour.
Optionally, the obtaining a single building model based on the obtained image data includes:
the image data of oblique photography is subjected to singulation.
Optionally, the obtaining the set building contour picture based on the cross section slice includes:
and scoring the building contour pictures by using a scoring neural network, and selecting building contour pictures with the score larger than the set score based on the scoring result.
Optionally, scoring neural network training includes:
collecting building crosscut profiles comprising different types of buildings;
marking the score of each building transverse profile slice, marking the continuous noiseless building profile as 1, and representing the good building profile; marking a noisy broken, noisy protruding or recessed building contour as a one-bit decimal from 0 to 1, the higher the value the better the contour quality; marking a noisy and broken and noisy raised or recessed building outline as 0;
and training a building contour scoring network by using the building contour and the corresponding score as training data.
Optionally, the line segment type includes a straight line or a curved line, and the relationship between the line segments includes vertical, intersecting or parallel.
Optionally, the fitting the line segment according to the relationship between the line segment type and the line segment includes:
and if the curve is a straight line, connecting the end points of the line segments, and if the curve is a curve, taking the midpoint of the line segments to draw an arc.
Optionally, the cross section slicing of the single building model includes that the single building model is transversely cut at intervals of 0.1 meter, and intersection lines are obtained by using a plane and the three-dimensional model, so that an outer contour slice sequence of the building is obtained, and the slice sequence is a contour map of the building from top to bottom.
Optionally, constructing the segmentation point and the line segment into a graph structure, inputting the graph structure into a graph neural network, thereby obtaining a training set of the graph neural network in the relationship between the line segment type and the line segment, including:
building contours are randomly obtained through horizontal, horizontal and vertical rules, buildings with random heights are generated through the building contours, and building models are derived;
the building model is used for obtaining model pictures at different angles through virtual camera surrounding, and three-dimensional reconstruction is carried out by using the model pictures and corresponding internal parameters and external parameters of the virtual camera, so as to obtain an oblique photography three-dimensional reconstruction building model;
respectively slicing building contours of the reconstructed building models by using the building models and the oblique photography three-dimensional reconstruction building models to obtain a horizontal and vertical building contour map and a building contour map with noise as paired data;
knowing the position and the connection sequence of each key point in the horizontal and vertical profile, finding the point with the minimum Manhattan distance in the noisy profile to mark as the corresponding key point, and marking by taking the same connection sequence as an edge; meanwhile, the straight line is marked 1, and the curve is marked 0.
Optionally, constructing the segmentation point and the line segment into a graph structure, inputting the graph structure into a graph neural network, thereby obtaining a training set of the graph neural network in the relationship between the line segment type and the line segment, including:
the coordinates of the key points are normalized by taking the key points as node characteristics of the graph; the line segment is used as the edge feature of the graph, the length of the line segment and the included angle between the line segment and the horizontal direction are used as two-dimensional features, the marking of two parts of contents is carried out on the outline line segment, namely the relationship between the line segment type and the line segment, and if the line segment is a curve, the marking is 0; if the line segments are straight lines, marking as 0, and if the line segments are vertical, marking as 0; if the line segments intersect and the included angle is not 90 degrees, the line segments are marked as 1 and the included angle degrees are attached, and if the line segments are parallel, the line segments are marked as 2.
In a specific application scenario, the building contour construction method based on the graph network comprises the following steps:
step 1: and (3) performing singulation on the oblique photographing data to obtain a single building model.
The oblique photographing data is individualized. A pre-trained point cloud instantiation network may be used to accomplish a personalization task, such as a PiontRend network.
Step 2: and slicing the cross section of the single building model, and obtaining the better building outline in the single building model by a scoring neural network mode.
And slicing the cross section of the single building model, and obtaining the better building outline in the single building model by a way of scoring the neural network. And (3) transversely cutting the monomer building after the step (1) at intervals of 0.1 meter, and solving intersection lines of a plane and the three-dimensional model to obtain an outer contour slice sequence of the building, wherein the slice sequence is a png-format contour map of the building from top to bottom.
A scoring network for training contours comprising
Preparing training sets
Collecting a large number of building transverse profiles, wherein different types of buildings such as a 'mouth' -shaped, a 'convex' -shaped, a 'concave' -shaped, an 'L' -shaped and the like are needed to be contained; while it is desirable to include building contours of different qualities such as continuous and broken, noisy and noiseless, smooth and noisy protrusions or depressions, etc. Marking the score of each building contour slice, marking the continuous noiseless building contour as 1, and representing the good building contour; marking a noisy, or broken, or noisy raised or recessed building contour as a one-digit decimal from 0 to 1, a higher value indicating a better contour quality; the noisy and broken and noisy protrusions or depressions of the building profile are marked 0, indicating the worst building profile.
Training scoring network
A score is calculated for each contour based on its continuity, variability, integrity, etc., with higher scores representing that contour being a better contour throughout the building. A building contour scoring network is trained using a pair of building contours and corresponding scores as training data.
Usage scoring network
Scoring the building contour slices in the step 2 by using a trained network, and selecting a better building contour for segment correction; bad building profiles are discarded.
Step 3: and (3) obtaining segmentation points from the building contour picture obtained in the step (2) by identifying the classification neural network, and dividing the continuously closed contour into line segments.
Step 301: training the key point identification neural network pair to obtain key points in the contour map, and dividing the contour into line segments comprises the following steps:
step 3011: preparing a training set:
building contours are obtained randomly through a horizontal-vertical rule, buildings with random heights are generated in the CityEngine through the building contours, and a CAD building model is derived. And using the building models to obtain pictures with different angles through virtual camera surrounding through an OpenGL library, and using the pictures and corresponding camera internal parameters and external parameters to perform three-dimensional reconstruction to obtain the oblique photography three-dimensional reconstruction building model.
Using the CAD building model and the oblique photography three-dimensional reconstruction building model, the paired building models are respectively subjected to building contour slicing, as shown in fig. 2 and 3, to obtain a horizontal-vertical building contour map and a noisy building contour map as paired data. The position of each key point and the connection sequence of the key points are known in the horizontal and vertical profile, and the key point corresponding to the point marking dimension with the minimum Manhattan distance is found in the noisy profile and marked by taking the same connection sequence as an edge. Meanwhile, the straight line is marked 1, and the curve is marked 0. Thus, the training data set is manufactured.
Step 3012: training key point recognition network
A key point recognition network, such as the first half of the LCNN, is trained using the training data set created in step 3011 to identify key points for each contour slice, thereby dividing a contour into different segments.
Step 302: training the graph neural network to obtain the type of the building contour line segment and the relationship between the line segments, wherein the training comprises the following steps:
step 3021: constructing contour line segments as graph structure datasets
Using the key points obtained in the step 301 as node characteristics of the graph, and carrying out normalization processing (dividing by the maximum value) on the (x, y) coordinates of the key points; the line segment is used as the edge feature of the graph, and the length of the line segment and the included angle between the line segment and the horizontal direction are used as two-dimensional features. Labeling two parts of contents of the contour line segment: relationship between line segment type and line segment. If the line segment is a curve, marking as 0; if the line segment is a straight line, it is marked as 0. If the line segments are vertical, marking as 0; if the line segments intersect and the included angle is not 90 degrees, marking as 1 and attaching the included angle degree; if the line segments are parallel, then the label is 2.
Step 3022: the training graph neural network outputs the relationship between the line segment types and the line segments:
inputting the data constructed in step 3021 into a graph neural network, such as PointGNN, performing graph neural network feature extraction, and dividing the full connection layer into two parts: the relationship between the predicted line segment type and the predicted line segment. Only the relation between adjacent segments is predicted here.
Step 4: the line segments and points obtained in the step 3 are constructed into a graph structure, and are output as the relationship (vertical/intersecting/parallel) between the line segment type (straight line/curve) and the line segment as the input of the graph neural network.
Step 5: fitting the line segment according to the result output in the step 4, connecting the end points of the line segment if the line segment is a straight line, and drawing an arc line by taking the midpoint of the line segment if the line segment is a curve; and adjusting the positions of the line segment endpoints according to the relationship between the line segments to obtain a final output result.
When the line segment type is a straight line, directly connecting two endpoints of the line segment; when the line segment type is an arc, the line segment is taken to focus on drawing the arc. When the relationship between the line segments is vertical, automatically adjusting the end point positions to enable the two line segments to be vertical; when the line segment relation is intersecting non-perpendicular, automatically adjusting the position of the end point according to the predicted angle; when the line segment relationship is parallel, the two line segments are combined, and the starting point of one line segment and the ending point of the other line segment are connected.
The embodiment also discloses a building contour constructing device based on the graph network, which comprises: the model module is used for obtaining a single building model based on the acquired image data;
the slicing module is used for carrying out cross section slicing on the single building model and obtaining a set building contour picture based on the cross section slicing;
the identification module is used for identifying segmentation points from the building contour picture so as to divide the continuously closed contour into line segments;
the relation module is used for constructing the segmentation points and the line segments into a graph structure, inputting the graph structure into a graph neural network, and thus obtaining the relation between the line segment types and the line segments;
and the fitting module is used for fitting the line segments according to the relationship between the line segment types and the line segments so as to obtain a fitting result, and adjusting the positions of the line segment endpoints in the fitting result according to the relationship between the line segments so as to obtain the building contour.
The embodiment discloses an electronic device including a memory and a processor. The memory is for storing non-transitory computer readable instructions. In particular, the memory may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory may include, for example, random Access Memory (RAM) and/or cache memory (cache), and the like. The non-volatile memory may include, for example, read Only Memory (ROM), hard disk, flash memory, and the like.
The processor may be a Central Processing Unit (CPU) or other form of processing unit having data processing and/or instruction execution capabilities, and may control other components in the electronic device to perform the desired functions. In one embodiment of the present disclosure, the processor is configured to execute the computer readable instructions stored in the memory to cause the electronic device to perform all or part of the steps of the graph network-based construction profile construction method of the embodiments of the present disclosure described above.
It should be understood by those skilled in the art that, in order to solve the technical problem of how to obtain a good user experience effect, the present embodiment may also include well-known structures such as a communication bus, an interface, and the like, and these well-known structures are also included in the protection scope of the present disclosure.
In particular, according to embodiments of the present disclosure, the processes described above with reference to flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program embodied on a non-transitory computer readable medium, the computer program comprising program code for performing the method shown in the flow chart. In such an embodiment, the computer program may be downloaded and installed from a network via a communication device, or installed from a storage device, or installed from ROM. All or part of the steps of the graph network-based construction contour construction method of the embodiments of the present disclosure are performed when the computer program is executed by the processing device.
The detailed description of the present embodiment may refer to the corresponding description in the foregoing embodiments, and will not be repeated herein.
A computer-readable storage medium according to an embodiment of the present disclosure has stored thereon non-transitory computer-readable instructions. When executed by a processor, perform all or part of the steps of the graph network-based construction profile construction method of the various embodiments of the disclosure described above.
The computer-readable storage medium described above includes, but is not limited to: optical storage media (e.g., CD-ROM and DVD), magneto-optical storage media (e.g., MO), magnetic storage media (e.g., magnetic tape or removable hard disk), media with built-in rewritable non-volatile memory (e.g., memory card), and media with built-in ROM (e.g., ROM cartridge).
The detailed description of the present embodiment may refer to the corresponding description in the foregoing embodiments, and will not be repeated herein.
The basic principles of the present disclosure have been described above in connection with specific embodiments, however, it should be noted that the advantages, benefits, effects, etc. mentioned in the present disclosure are merely examples and not limiting, and these advantages, benefits, effects, etc. are not to be considered as necessarily possessed by the various embodiments of the present disclosure. Furthermore, the specific details disclosed herein are for purposes of illustration and understanding only, and are not intended to be limiting, since the disclosure is not necessarily limited to practice with the specific details described.
In this disclosure, relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions, and the block diagrams of devices, apparatuses, devices, systems involved in this disclosure are merely illustrative examples and are not intended to require or implicate that connections, arrangements, configurations must be made in the manner shown in the block diagrams. As will be appreciated by one of skill in the art, the devices, apparatuses, devices, systems may be connected, arranged, configured in any manner. Words such as "including," "comprising," "having," and the like are words of openness and mean "including but not limited to," and are used interchangeably therewith. The terms "or" and "as used herein refer to and are used interchangeably with the term" and/or "unless the context clearly indicates otherwise. The term "such as" as used herein refers to, and is used interchangeably with, the phrase "such as, but not limited to.
In addition, as used herein, the use of "or" in the recitation of items beginning with "at least one" indicates a separate recitation, such that recitation of "at least one of A, B or C" for example means a or B or C, or AB or AC or BC, or ABC (i.e., a and B and C). Furthermore, the term "exemplary" does not mean that the described example is preferred or better than other examples.
It is also noted that in the systems and methods of the present disclosure, components or steps may be disassembled and/or assembled. Such decomposition and/or recombination should be considered equivalent to the present disclosure.
Various changes, substitutions, and alterations are possible to the techniques described herein without departing from the teachings of the techniques defined by the appended claims. Furthermore, the scope of the claims of the present disclosure is not limited to the particular aspects of the process, machine, manufacture, composition of matter, means, methods and acts described above. The processes, machines, manufacture, compositions of matter, means, methods, or acts, presently existing or later to be developed that perform substantially the same function or achieve substantially the same result as the corresponding aspects described herein may be utilized. Accordingly, the appended claims are intended to include within their scope such processes, machines, manufacture, compositions of matter, means, methods, or acts.
The previous description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects without departing from the scope of the disclosure. Thus, the present disclosure is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
The foregoing description has been presented for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of the disclosure to the form disclosed herein. Although a number of example aspects and embodiments have been discussed above, a person of ordinary skill in the art will recognize certain variations, modifications, alterations, additions, and subcombinations thereof.
Claims (9)
1. A method for building contour construction based on a graph network, comprising:
obtaining a single building model based on the obtained image data;
performing cross section slicing on the single building model, and obtaining a set building contour picture based on the cross section slicing;
identifying segmentation points from the building contour picture, thereby segmenting the continuously closed contour into line segments;
constructing the segmentation points and the line segments into a graph structure, and inputting the graph structure into a graph neural network so as to obtain the relationship between the line segment types and the line segments;
fitting the line segments according to the relationship between the line segment types and the line segments to obtain a fitting result, and adjusting the positions of the line segment endpoints in the fitting result according to the relationship between the line segments to obtain a building contour;
training the graph neural network to obtain the type of the building contour line segment and the relationship between the line segments, wherein the training comprises the following steps:
constructing a contour line segment into a graph structure data set;
identifying key points for each building contour slice graph, and carrying out normalization processing on coordinates of the key points by using the obtained key points as node characteristics of the graph; the line segment is used as the edge feature of the graph, the length of the line segment and the included angle between the line segment and the horizontal direction are used as two-dimensional features, the marking of two parts of contents is carried out on the outline line segment, and if the line segment is a curve, the marking is 0; if the line segments are straight lines, marking as 0, and if the line segments are vertical, marking as 0; if the line segments intersect and the included angle is not 90 degrees, marking as 1 and attaching the included angle degree; if the line segments are parallel, marking as 2;
the training graph neural network outputs the relationship between the line segment types and the line segments:
inputting the constructed data set into a graph neural network, and extracting characteristics of the graph neural network so as to predict the relation between adjacent line segments;
the obtained line segments and points are constructed into a graph structure, and are used as the input of a graph neural network and output as the relationship between the line segment types and the line segments.
2. The building contour construction method based on a graph network as claimed in claim 1, wherein the obtaining a single building model based on the acquired image data includes:
the image data of oblique photography is subjected to singulation.
3. The method for constructing the building contour based on the graph network according to claim 1 or 2, wherein the obtaining the set building contour picture based on the cross-sectional slice comprises:
and scoring the building contour pictures by using a scoring neural network, and selecting building contour pictures with the score larger than the set score based on the scoring result.
4. A method of building contour construction based on a graph network as claimed in claim 3, wherein scoring neural network training comprises:
collecting building crosscut profiles comprising different types of buildings;
marking the score of each building transverse profile slice, marking the continuous noiseless building profile as 1, and representing the good building profile; marking a noisy broken, noisy protruding or recessed building contour as a one-bit decimal from 0 to 1, the higher the value the better the contour quality; marking a noisy and broken and noisy raised or recessed building outline as 0;
and training a building contour scoring network by using the building contour and the corresponding score as training data.
5. A method of constructing a building contour based on a graph network as claimed in claim 1 or 2, wherein the line segment types include straight lines or curved lines, and the relationships between the line segments include vertical, intersecting or parallel.
6. The method for constructing a building contour based on a graph network according to claim 1 or 2, wherein the fitting the line segments according to the relationship between the line segment type and the line segments comprises:
and if the curve is a straight line, connecting the end points of the line segments, and if the curve is a curve, taking the midpoint of the line segments to draw an arc.
7. The building contour construction method based on the graph network according to claim 1 or 2, wherein the cross section slicing is performed on the single building model, the cross section is performed on the single building model at intervals of 0.1 meter, a plane is used for intersecting with the three-dimensional model, and an outline slicing sequence of the building is obtained, and the slicing sequence is a contour graph of the building from top to bottom.
8. The building contour construction method based on graph network according to claim 1 or 2, wherein constructing the segmentation points and the line segments into graph structures, inputting the graph structures into the graph neural network, thereby obtaining a training set of the graph neural network in the relationship between the line segment types and the line segments, comprises:
building contours are randomly obtained through horizontal, horizontal and vertical rules, buildings with random heights are generated through the building contours, and building models are derived;
the building model is used for obtaining model pictures at different angles through virtual camera surrounding, and three-dimensional reconstruction is carried out by using the model pictures and corresponding internal parameters and external parameters of the virtual camera, so as to obtain an oblique photography three-dimensional reconstruction building model;
respectively slicing building contours of the reconstructed building models by using the building models and the oblique photography three-dimensional reconstruction building models to obtain a horizontal and vertical building contour map and a building contour map with noise as paired data;
knowing the position and the connection sequence of each key point in the horizontal and vertical profile, finding the point with the minimum Manhattan distance in the noisy profile to mark as the corresponding key point, and marking by taking the same connection sequence as an edge; meanwhile, the straight line is marked 1, and the curve is marked 0.
9. A graph network-based building contour construction apparatus, comprising:
the model module is used for obtaining a single building model based on the acquired image data;
the slicing module is used for carrying out cross section slicing on the single building model and obtaining a set building contour picture based on the cross section slicing;
the identification module is used for identifying segmentation points from the building contour picture so as to divide the continuously closed contour into line segments;
the relation module is used for constructing the segmentation points and the line segments into a graph structure, inputting the graph structure into a graph neural network, and thus obtaining the relation between the line segment types and the line segments;
the fitting module is used for fitting the line segments according to the relationship between the line segment types and the line segments so as to obtain a fitting result, and adjusting the positions of the line segment endpoints in the fitting result according to the relationship between the line segments so as to obtain the building outline;
training the graph neural network to obtain the type of the building contour line segment and the relationship between the line segments, wherein the training comprises the following steps:
constructing a contour line segment into a graph structure data set;
identifying key points for each building contour slice graph, and carrying out normalization processing on coordinates of the key points by using the obtained key points as node characteristics of the graph; the line segment is used as the edge feature of the graph, the length of the line segment and the included angle between the line segment and the horizontal direction are used as two-dimensional features, the marking of two parts of contents is carried out on the outline line segment, and if the line segment is a curve, the marking is 0; if the line segments are straight lines, marking as 0, and if the line segments are vertical, marking as 0; if the line segments intersect and the included angle is not 90 degrees, marking as 1 and attaching the included angle degree; if the line segments are parallel, marking as 2;
the training graph neural network outputs the relationship between the line segment types and the line segments:
inputting the constructed data set into a graph neural network, and extracting characteristics of the graph neural network so as to predict the relation between adjacent line segments;
the obtained line segments and points are constructed into a graph structure, and are used as the input of a graph neural network and output as the relationship between the line segment types and the line segments.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311221026.4A CN116958454B (en) | 2023-09-21 | 2023-09-21 | Construction contour construction method and device based on graph network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311221026.4A CN116958454B (en) | 2023-09-21 | 2023-09-21 | Construction contour construction method and device based on graph network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116958454A CN116958454A (en) | 2023-10-27 |
| CN116958454B true CN116958454B (en) | 2023-12-12 |
Family
ID=88453261
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311221026.4A Active CN116958454B (en) | 2023-09-21 | 2023-09-21 | Construction contour construction method and device based on graph network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116958454B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117454495B (en) * | 2023-12-25 | 2024-03-15 | 北京飞渡科技股份有限公司 | CAD vector model generation method and device based on building sketch outline sequence |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111902825A (en) * | 2018-03-23 | 2020-11-06 | 多伦多大学管理委员会 | Polygonal object labeling system and method for training object labeling system |
| CN113160177A (en) * | 2021-04-23 | 2021-07-23 | 杭州电子科技大学 | Plane segmentation method based on superpixel and graph convolution network |
| CN114510772A (en) * | 2022-04-08 | 2022-05-17 | 北京飞渡科技有限公司 | Rapid generation method of building surface contour based on oblique scanning data |
| CN116628834A (en) * | 2023-07-26 | 2023-08-22 | 北京飞渡科技股份有限公司 | Contour segmentation correction method and device based on neural network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TW550521B (en) * | 2002-02-07 | 2003-09-01 | Univ Nat Central | Method for re-building 3D model of house in a semi-automatic manner using edge segments of buildings |
-
2023
- 2023-09-21 CN CN202311221026.4A patent/CN116958454B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111902825A (en) * | 2018-03-23 | 2020-11-06 | 多伦多大学管理委员会 | Polygonal object labeling system and method for training object labeling system |
| CN113160177A (en) * | 2021-04-23 | 2021-07-23 | 杭州电子科技大学 | Plane segmentation method based on superpixel and graph convolution network |
| CN114510772A (en) * | 2022-04-08 | 2022-05-17 | 北京飞渡科技有限公司 | Rapid generation method of building surface contour based on oblique scanning data |
| CN116628834A (en) * | 2023-07-26 | 2023-08-22 | 北京飞渡科技股份有限公司 | Contour segmentation correction method and device based on neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116958454A (en) | 2023-10-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8711143B2 (en) | System and method for interactive image-based modeling of curved surfaces using single-view and multi-view feature curves | |
| WO2020108610A1 (en) | Image processing method, apparatus, computer readable medium and electronic device | |
| CN116958454B (en) | Construction contour construction method and device based on graph network | |
| Liu et al. | Strokeaggregator: Consolidating raw sketches into artist-intended curve drawings | |
| Fan et al. | Robust feature-preserving mesh denoising based on consistent subneighborhoods | |
| CN110060255A (en) | Semantic segmentation is carried out to 2D plan view using classifier pixel-by-pixel | |
| CN107578467B (en) | Three-dimensional modeling method and device for medical instrument | |
| CN113724401B (en) | Three-dimensional model cutting method and device, computer equipment and storage medium | |
| CN116628834B (en) | Contour segmentation correction method and device based on neural network | |
| CN113112600B (en) | Indoor scene three-dimensional modeling method based on structure | |
| Buchanan et al. | Automatic single-view character model reconstruction | |
| Soundararajan et al. | An in-depth study of graph partitioning measures for perceptual organization | |
| Strodthoff et al. | Automatic decomposition of 3D solids into contractible pieces using Reeb graphs | |
| CN112084855B (en) | Outlier elimination method for video stream based on improved RANSAC method | |
| US10282858B2 (en) | Methods and systems for estimating three-dimensional information from two-dimensional concept drawings | |
| CN110222621B (en) | Similar house type detection method, electronic equipment and storage medium | |
| CN114170257A (en) | Curved target contour fitting device, computer equipment and storage medium | |
| González et al. | Simplification method for textured polygonal meshes based on structural appearance | |
| CN113378864A (en) | Method, device and equipment for determining anchor frame parameters and readable storage medium | |
| Han et al. | 3D human model reconstruction from sparse uncalibrated views | |
| KR101811135B1 (en) | Apparatus and method for simplifying model through optimal level-of-detail decision | |
| JP2008533614A (en) | System and method for generating matched contour profiles | |
| Wang et al. | Segmentation-driven feature-preserving mesh denoising | |
| JP6320806B2 (en) | 3D model search method and 3D model search system | |
| CN116883599B (en) | Clothing try-on system based on three-dimensional modeling technology |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |