CN115238368B - Automated modeling method and medium for bridge pier drawing recognition based on computer vision - Google Patents

Abstract

The invention discloses a pier drawing identification automatic modeling method and medium based on computer vision, comprising the following steps: acquiring an image of a plane drawing of a pier, and identifying the outlines and marked geometric dimensions of the pier and a reinforcing steel bar; extracting geometric features based on pixels from the identified outline, acquiring the geometric features based on pixels corresponding to the marked geometric dimensions from the outline, and then performing common system unit conversion on all the extracted geometric features based on pixels according to the geometric dimensions based on the common system and the corresponding geometric features based on the pixels to obtain all the geometric features based on the common system; and establishing a three-dimensional finite element model for the pier based on the reinforced concrete material according to the geometric characteristics of the pier and the reinforcing steel bars based on the common system, wherein the three-dimensional finite element model comprises an elastic model and an elastic-plastic model of the pier. The method has the advantages of quick modeling, simple operation, accuracy and reliability, and can be widely applied to seismic response analysis, seismic performance evaluation and design of piers.

Description

Automated modeling method and medium for bridge pier drawing recognition based on computer vision

technical field

The invention belongs to the technical field of automatic intelligent construction and railway engineering application computing, and specifically relates to a computer vision-based automatic modeling method and medium for bridge pier blueprint recognition.

Background technique

In recent years, my country's high-speed railway has developed rapidly. In order to overcome the obstacles caused by different terrains, the existing high-speed railway lines in operation are mainly built in the way of "replacing roads with bridges", and bridges account for more than 10% of the lines. Fifty. With the development of high-speed railways to the central and western regions where earthquakes occur frequently, the seismic performance of high-speed railway girder bridges is an issue worthy of attention. From the perspective of components, bridge piers are the main load-bearing components of bridge structures, and their seismic performance under earthquakes is related to the normal operation of high-speed railway lines. High-speed railway bridges are different from road bridges. The design of high-speed railway bridge piers is controlled by stiffness. It is characterized by large cross-sectional dimensions and high stiffness of bridge components, so as to meet the functional requirements of the disturbance deformation of high-speed railway trains within 2mm. Due to the lack of longitudinal reinforcement of the bridge piers, the possibility of damage to the bridge piers is relatively high during earthquakes. Therefore, it is of great significance to study the seismic performance and failure characteristics of high-speed railway bridge piers. Usually, the analysis of the seismic performance of bridge piers is based on the finite element model. It is necessary to calculate the geometric properties of the bridge piers, such as cross-sectional size, area, and reinforcement layout. Such as round-end solid piers, variable-section hollow piers, rectangular solid piers, etc.

Contents of the invention

The invention provides a computer vision-based automatic modeling method and medium for bridge pier drawing recognition, which can realize bridge pier drawing recognition, fast and automatic generation of a three-dimensional bridge pier finite element model, and nonlinear time-history analysis of bridge piers. The invention has the characteristics of fast modeling, simple operation, accurate and reliable calculation structure, etc., and can be widely used in the seismic response analysis, seismic performance evaluation and design of bridge piers, and has very important engineering applications.

In order to realize the above-mentioned technical purpose, the present invention adopts following technical scheme:

The invention provides a computer vision-based automatic modeling method for pier blueprint recognition, including:

Obtain the image of the plan drawing of the bridge pier, identify the outline of the bridge pier and steel bars and the marked geometric dimensions based on public ownership;

Extract the pixel-based geometric features from the identified pier outline and steel bar outline, and obtain the pixel-based geometric features corresponding to the marked geometric dimensions, and then according to the public ownership-based geometric dimensions and the corresponding pixel-based geometric features, Perform public unit conversion on all the extracted pixel-based geometric features to obtain all public-owned geometric features;

According to the geometric characteristics of bridge piers and steel bars based on public ownership, a three-dimensional finite element model is established for bridge piers based on reinforced concrete materials;

Among them, when establishing the three-dimensional finite element model: use the beam-column element to simulate the elastic mechanical behavior of the bridge pier to establish the elastic model of the bridge pier; use the nonlinear beam-column element to simulate the elastic-plastic mechanical behavior of the bridge pier to establish the elastic-plastic model of the bridge pier.

Further, the annotated geometric dimension based on public ownership refers to the linear length of the pier outline marked in any straight line direction in the pier plan image.

Further, the geometric features include the size, geometric center, area, and moment of inertia of the pier profile, the number of steel bars, the area and distribution position of each steel bar profile, and the geometric features corresponding to the marked geometric dimensions based on public ownership.

Further, an optical character recognition algorithm is used to recognize the public ownership-based geometric size of the mark in the image.

Further, an edge detection algorithm is used to identify the outline of the pier and the outline of the reinforcement.

Further, before identifying the contour and geometric dimensions in the image, the image is first binarized and preprocessed, and the conversion relationship for the public ownership unit conversion of the geometric features is:

为基于像素的几何特征，

为

对应的基于公有制的几何尺寸，

为从

转换为

的转换比例。 In the formula,

is a pixel-based geometric feature,

for

The corresponding geometric dimensions based on public ownership,

for from

converted to

conversion ratio.

Furthermore, before extracting pixel-based geometric features, the image is preprocessed with morphological operations to eliminate noise including the outline of transverse steel bars in the image.

Furthermore, when building a three-dimensional finite element model for a bridge pier based on reinforced concrete materials, in addition to the public-owned geometric features of the pier and steel bars, the material properties of the steel bars and concrete required for the bridge pier, and the load of the pier are also obtained from the user. And the boundary constraints of the bridge pier; the material properties include but not limited to the density, elastic modulus and elastic stiffness of the material; the loads include gravity load, lateral load and earthquake load.

Further, the elastic model of the bridge pier is used for modal analysis and static analysis of the bridge pier; the elastic-plastic model of the bridge pier is used for time-history analysis of the bridge pier.

A computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the computer vision-based automated modeling method for bridge pier blueprint recognition described in any one of the above is implemented.

Compared with prior art, the beneficial effect of the present invention is:

(1) The present invention is based on computer vision and machine learning technology. According to the input pier drawings, it can identify the cross-sectional size of the pier, the size of the steel bar and the coordinate position, and can efficiently calculate the geometric parameters of the structural components required for finite element modeling, and Automatically transfer parameters to the finite element model to realize automatic finite element modeling. This method calculates the structural size and geometric parameters without cumbersome modeling parameter settings, so it has significant advantages in practical engineering applications.

(2) The present invention can realize the marked recognition of the steel bar size of the bridge pier, the calculation of geometric modeling parameters, the calculation, analysis and visualization of finite element modeling.

(3) The present invention realizes the conversion of drawings to finite element models, considers the mechanical behavior of different models, and realizes structural modal response analysis, static force analysis, time history analysis and visualization of results. For the seismic response analysis of the bridge pier model, different seismic waves can be input to analyze and evaluate the seismic performance of the bridge pier. The whole modeling process is simple in form and convenient in calculation, which has important engineering application significance for the seismic design and performance evaluation of bridge piers.

Description of drawings

Fig. 1 is a structural schematic diagram of an automatic modeling method for pier drawing recognition based on computer vision;

Fig. 2 is a schematic diagram of the image pre-processing module, specifically showing the process of the image from input to processing;

Fig. 3 is a schematic diagram of the module principle of text annotation recognition, contour detection and pixel unit conversion into public units, which specifically shows that the embodiment of the present invention uses OCR to extract the dimensions of bridge piers and steel bars in the bridge pier drawings, identifies the detected bridge piers and steel bar outlines, and obtains The transformation between pixel value and public ownership unit is realized;

Figure 4 is a schematic diagram of the principle of the morphological operation and coordinate conversion module, which is mainly to remove redundant lines in the drawing, so as to ensure the accuracy of the extracted pier outline and steel bar geometric features;

Figure 5 is a schematic diagram of the transformation of the pier physical model into a finite element model and the input earthquake motion;

Figure 6 is the finite element modeling and result visualization module.

detailed description

The following is a detailed description of the embodiments of the present invention. This embodiment is carried out based on the technical solution of the present invention, and provides detailed implementation methods and specific operation processes to further explain the technical solution of the present invention.

This embodiment provides a computer vision-based automatic modeling method for bridge pier drawing recognition, as shown in FIG. 1 , including the following steps:

S1: Obtain the image of the plan drawing of the bridge pier, and identify the contours of the bridge pier and steel bars and the marked geometric dimensions based on public ownership. As shown in Figure 2 and 3.

Before identifying the contour and geometric dimensions in the image, the image is binarized and preprocessed: since the image of the pier plan drawing obtained is a color image containing three channels of RGB, it is necessary to convert the RGB image into grayscale Figure, and convert the grayscale image into a binary image from the cv2.threshold function for text recognition (based on public geometric dimensions) and contour detection (contours of bridge piers and steel bars)

The annotated geometric dimension based on public ownership refers to the linear length of the pier outline marked in any straight line direction in the pier plan image. In this embodiment, the dimension in the x-direction of the cross-section of the bridge pier is used as the basis to obtain the conversion relationship between the geometric dimension based on public ownership and the geometric feature based on pixels.

Use edge detection algorithms (such as Canny, Sobel, Prewitt, Roberts, etc.) to identify the outline of the pier and the outline of the reinforcement. This embodiment is based on Canny's edge detection operator. On the basis of the gradient operator, it introduces a calculation strategy that can obtain single-pixel edges with good anti-noise performance and high positioning accuracy, and adds non-maximum suppression and double thresholds. detection, which reduces the probability of false detection and missed detection, and can greatly improve the accuracy and accuracy of contour detection.

This embodiment uses the mature Tesseract-OCR (Optical Character Recognition, optical character recognition) to scan the image, check the characters printed on the drawing, and analyze and process the image file, so as to realize the text recognition algorithm to extract the public-owned geometry of the bid in the image. size.

S2: Extract the pixel-based geometric features from the identified pier outline and steel bar outline, and obtain the pixel-based geometric features corresponding to the marked geometric dimensions, and then according to the public ownership-based geometric dimensions and the corresponding pixel-based geometry Features, perform public unit conversion on all the extracted pixel-based geometric features, and obtain all public-owned geometric features.

Before extracting pixel-based geometric features, the image is preprocessed with morphological operations to remove noise including the outline of transverse steel bars in the image:

Since the contour detection operator can detect the edges of all objects in the image, the detected image still retains some noise, such as straight lines representing horizontal steel bars, so it is necessary to perform morphological operations on the image. As shown in Figure 4, the commonly used morphological operations in OpenCV include erosion and dilation. Expansion is mainly used to deal with image defects and complement them; corrosion is used to deal with image glitches and eliminate redundant lines in the image. The combination of these two basic operations can effectively denoise the image and improve the accuracy of image detection. Therefore, after the contour detection of the pier drawings, the image is first corroded to eliminate redundant lines, and then the image is expanded to restore the defects caused by corrosion, so as to effectively segment the longitudinal reinforcement elements. By calling the contour detection function again, the position of each longitudinal reinforcement can be obtained, and finally the final coordinate position of the longitudinal reinforcement is determined through the relationship between the pixel value and the public unit system. As shown in Figure 4.

The geometric features include the size, geometric center, area, and moment of inertia of the pier profile, the number of steel bars, the area and distribution position of each steel bar profile, and the geometric features corresponding to the marked geometric dimensions based on public ownership.

In order to determine the area, moment of inertia and geometric center of the cross-section of the bridge pier, as well as the distribution position of the longitudinal reinforcement, it is first necessary to identify the cross-sectional profile of the bridge pier. The cv2.findContours() function is used to quickly identify the cross-sectional profile of the bridge pier.

表示x关于c点的k阶矩。对于一幅图像，我们把像素的坐标看成 是一个二维随机变量(x,y)，那么一幅灰度图像可以用二维灰度密度函数来表示，因此可以 用矩来描述灰度图像的特征. 对于二值化图像，空间矩可表示为：

当j, i=0时，

表示轮廓的面积；当j=2, i=0，

表 示关于y轴的惯性矩，当j=0, i=2，

表示关于x轴的惯性矩。相应地，轮廓的几何中心可 表示为

，

。由于计算机识别图纸读取的是二维像素值，因此需要将计算获取 的轮廓面积，惯性矩以及几何中心转化为真实的尺寸参数。 In order to calculate the geometric features of the contour, it is necessary to use the cv2.moment() function to solve the moments of each order of the contour. The explanation of moment is as follows: From the perspective of probability and statistics, moment is a digital characteristic of random variable. Let x be a random variable, c be a constant, and k be a positive integer. but

Represents the kth order moment of x about point c. For an image, we regard the coordinates of the pixels as a two-dimensional random variable (x, y), then a grayscale image can be represented by a two-dimensional grayscale density function, so the grayscale image can be described by moments Features of . For binarized images, the spatial moments can be expressed as:

When j, i=0,

Represents the area of the contour; when j=2, i=0,

Represents the moment of inertia about the y-axis, when j=0, i=2,

Indicates the moment of inertia about the x-axis. Correspondingly, the geometric center of the contour can be expressed as

,

. Since the computer recognizes drawings to read two-dimensional pixel values, it is necessary to convert the calculated contour area, moment of inertia, and geometric center into real size parameters.

In this embodiment, the geometric dimension of the cross-section of the pier in the x direction (that is, the linear distance of the pier profile along the x coordinate axis of the image) can be used to obtain the conversion relationship between the pixel-based geometric features and the real geometry of the pier based on the public-owned geometric dimension, namely :

为基于像素的几何特征，

为

对应的基于公有制的几何尺寸，

为从

转换为

的转换比例。本实施例中，桥墩横截面x方向的几何尺寸

，再将提取到的基于像素的桥墩横截面x方向的几何尺寸（属于几何特征）代入 上述转化关系式中，即可得到从

转换为

的转换比例

。从而可根据提 取到的基于像素的所有几何特征，通过

，计算得到基于公有制的所有几何 特征。 In the formula,

is a pixel-based geometric feature,

for

The corresponding geometric dimensions based on public ownership,

for from

converted to

conversion ratio. In this embodiment, the geometric dimension of the cross-section of the pier in the x direction

, and then substituting the extracted pixel-based geometric dimensions (belonging to geometric features) of the cross-section of the pier in the x direction into the above conversion relation, we can get

converted to

Conversion ratio of

. Therefore, according to all the extracted geometric features based on pixels, through

, calculate all the geometric features based on public ownership.

S3: According to the public ownership-based geometric characteristics of bridge piers and steel bars, establish a three-dimensional finite element model for bridge piers based on reinforced concrete materials; as shown in Figures 5 and 6.

Among them, when establishing the three-dimensional finite element model: use the beam-column element (elasticBeamColumn) to simulate the elastic mechanical behavior of the bridge pier, establish the elastic model of the bridge pier, and consider the elastic deformation of the bridge pier for the aseismic design of the bridge pier. Carry out modal analysis and static analysis; use nonlinear beam-column elements (such as dispBeamColumn, forceBeamColumn) to simulate the elastic-plastic mechanical behavior of bridge piers, and establish elastic-plastic models of bridge piers, which can fully consider the nonlinear behavior of bridge piers under earthquake action, It can be used for seismic response analysis and seismic performance evaluation of bridge piers, and for time-history analysis of bridge piers.

When building a three-dimensional finite element model for a bridge pier based on reinforced concrete materials, in addition to the public-owned geometric features of the pier and steel bars, the material properties of the steel bars and concrete required for the pier, the load of the pier, and the properties of the pier Boundary constraints; the material properties include but not limited to the density, elastic modulus and elastic stiffness of the material; the loads include gravity loads, lateral loads and earthquake loads. Establishing a three-dimensional finite model of a bridge pier based on these acquired data belongs to the prior art, and will not be described in detail in this embodiment.

Based on the above-mentioned embodiment of the automatic modeling method for bridge piers based on computer vision and drawing recognition, the present invention also provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above-mentioned implementation is realized. The automatic modeling method of bridge piers based on computer vision and intelligent recognition of drawings described in the example.

The above embodiments are preferred embodiments of the present application, and those skilled in the art can also perform various transformations or improvements on this basis, and without departing from the general concept of the application, these transformations or improvements should all belong to the present application. within the scope of the application.

Claims (9)

1. The automated modeling method for bridge pier drawing recognition based on computer vision, is characterized in that, comprises: Obtain the image of the plan drawing of the bridge pier, identify the outline of the bridge pier and steel bars and the marked geometric dimensions based on public ownership; Extract the pixel-based geometric features from the identified pier outline and steel bar outline, and obtain the pixel-based geometric features corresponding to the marked geometric dimensions, and then according to the public ownership-based geometric dimensions and the corresponding pixel-based geometric features, Perform public unit conversion on all the extracted pixel-based geometric features to obtain all public-owned geometric features; The geometric features include the size, geometric center, area, moment of inertia of the pier profile, the number of steel bars, the area and distribution position of each steel bar profile, and the geometric features corresponding to the marked geometric dimensions based on public ownership; Among them, the conversion relationship of public ownership unit conversion for geometric features is:

In the formula,

is a pixel-based geometric feature,

for

The corresponding geometric dimensions based on public ownership,

for from

converted to

conversion ratio; According to the geometric characteristics of bridge piers and steel bars based on public ownership, a three-dimensional finite element model is established for bridge piers based on reinforced concrete materials; Among them, when establishing the three-dimensional finite element model: use the beam-column element to simulate the elastic mechanical behavior of the bridge pier to establish the elastic model of the bridge pier; use the nonlinear beam-column element to simulate the elastic-plastic mechanical behavior of the bridge pier to establish the elastic-plastic model of the bridge pier. 2. The method according to claim 1, wherein the marked geometric dimension based on public ownership refers to the linear length of the pier outline marked in the plane image of the pier along any straight line direction. 3. The method according to claim 1, characterized in that an optical character recognition algorithm is used to identify the geometric dimensions based on public ownership of the mark in the image. 4. The method according to claim 1, characterized in that an edge detection algorithm is used to identify the outline of the pier and the outline of the reinforcement. 5. The method according to claim 1, characterized in that, before recognizing the contour and geometric dimensions in the image, the image is first binarized and preprocessed. 6. The method according to claim 1, characterized in that, before extracting the pixel-based geometric features, the image is preprocessed with morphological operations to eliminate noise including the outline of transverse steel bars in the image. 7. The method according to claim 1, characterized in that, when a three-dimensional finite element model is established for a bridge pier based on reinforced concrete materials, in addition to the geometric features based on the public ownership of the pier and the steel bar, the pier is also obtained from the user. The material properties of the required steel bars and concrete, the loads of the piers, and the boundary constraints of the piers; the material properties include but not limited to the density, elastic modulus, and elastic stiffness of the materials; the loads include gravity loads, lateral loads, and earthquakes load. 8. The method according to claim 1, wherein the elastic model of the bridge pier is used for carrying out modal analysis and static analysis to the bridge pier; the elastic-plastic model of the bridge pier is used for carrying out the time history of the bridge pier analyze. 9. A computer-readable storage medium, on which a computer program is stored, wherein the computer program implements the method according to any one of claims 1-8 when executed by a processor.

Images