CN114509018A - Full-field real-time bridge deflection measurement method - Google Patents

Abstract

The invention provides a full-field real-time bridge deflection measurement method, which comprises the following steps: measurement preparation, full-field scale factor determination, image acquisition, full-field image displacement calculation and full-field deflection calculation. The technical solution claimed in the present application calculates the calibration coefficient from the image displacement to the actual deflection/displacement of each pixel in the image under the condition of imaging the oblique optical axis of the camera (that is, the effect of imaging near large and far small) (about 100 10,000 points), quickly achieve high-precision image matching (real-time matching calculation for millions of points), so as to calculate the full-field bridge deflection in real time.

Description

A full-field real-time bridge deflection measurement method

technical field

The invention relates to the technical field of bridge detection, in particular to a full-field real-time bridge deflection measurement method.

Background technique

The bridge deflection detection method based on machine vision is mainly divided into three steps: 1) Calculate the proportional relationship between the image displacement and the actual deflection/displacement; 2) Compare the images before and after deformation, and calculate the pixel displacement in the deformed image; 3) Combine the ratio The relationship between the coefficient and the image displacement, and finally the actual deflection/displacement is calculated.

Existing common bridge deflection detection technologies based on machine vision are roughly divided into two categories: 1) Single-point detection technology, that is, there is only one point to be measured in the camera's field of view. This method is easier to implement, because there is only one point to be measured, and the field of view can be adjusted to a very small (generally less than 2 cm). The actual deflection/displacement scaling factor is the same. And because the field of view is small enough, when the bridge is deformed, the image displacement must be large, and a variety of image processing methods can be used to extract the image pixel displacement. 2) Multi-point detection technology, that is, there are multiple points to be measured in the camera field of view at the same time. Due to the vigorous development of digital image correlation technology, multi-point measurement technology is an advanced measurement method emerging in recent years. During multi-point measurement, due to the need to capture multiple measurement points at the same time, the camera has a large field of view, and the objects captured are close to large and far small, which involves two key issues: the calculation of the ratio/calibration coefficient of different measurement points, the comparison of Matching calculations for small image displacements (less than 1 pixel). For the calibration problem, the solution in the existing literature is to use auxiliary equipment such as inclinometers and rangefinders to measure specific auxiliary parameters for each measurement point, and then calculate the proportional coefficient of each measurement point in turn. For the matching problem, a high-precision digital image matching algorithm is applied to each measurement point, and the image displacement of each measurement point can be measured in turn.

At present, the prior art has the following shortcomings or deficiencies:

(1) The existing single-point detection technology requires a very large focal length of the lens due to the small field of view, which makes it difficult for the camera to focus. Experienced operators are required to debug the equipment during on-site measurement. Generally speaking, when measuring the deflection of large buildings such as bridges, it is necessary to detect multiple points or the entire surface at the same time. This kind of equipment based on single-point detection technology needs to erect multiple devices at the same time, which is not only difficult to erect but also requires multiple It is difficult to analyze the overall deformation trend of large structures such as bridges as a whole.

(2) The existing multi-point detection technology generally refers to a limited number of measurement points of several key parts (such as measuring the 1/4 span, mid-span, and 3/4 span positions on a bridge). With the advent of the era of big data, it is difficult to meet the existing detection needs with only a few key locations of data. If the multi-point detection technology is directly used for the real-time measurement of the bridge's full-field deflection, there are two main insurmountable deficiencies. Less than one: about calibration. The optical center of the camera and the measured point on the bridge are generally imaged at an oblique optical axis, that is, the bridge imaged in the camera has the effect of being near big and far small, and each pixel point has a different scale factor from the pixel displacement conversion to the actual deflection/displacement . The classic single-point calibration scheme needs to measure the distance from each point to be measured to the optical center of the camera, and use the inclinometer to measure the vertical angle to the horizontal ground, and then apply these auxiliary parameters to calculate the specific ratio of a single measurement point. coefficient. In the whole field measurement, it is assumed that the resolution of the camera is 5 million pixels, and the bridge plane to be measured accounts for 1/3 of the image, which is about 1.67 million pixels. realistic. Less than two: about matching. The deflection measurement of bridges is practical, and real-time in-situ measurement is required. The matching algorithm of the images before and after deformation in the multi-point detection technology generally uses the relatively mature digital image correlation method to achieve high-precision matching. However, in the system based on multi-point detection technology, there are often only a few points (generally no more than 10) to be measured, and the calculation amount is not large, and real-time calculation can be realized. There are 1.67 million pixels. Here, it is necessary to optimize the image matching algorithm of the whole field to achieve a super fast algorithm that does not lose matching accuracy.

The information disclosed in this Background section is only for enhancement of understanding of the general background of the invention and should not be taken as an acknowledgement or any form of suggestion that this information forms the prior art already known to a person skilled in the art.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a full-field real-time bridge deflection measurement method, which calculates the displacement of each pixel in the image from image to The calibration coefficient of actual deflection/displacement (about one million points) can quickly realize high-precision image matching (real-time matching calculation of one million points), so as to calculate the deflection of the whole bridge in real time.

In order to achieve the above object, the present invention adopts the following technical solutions:

The present invention provides a full-field real-time bridge deflection measurement method, which comprises the following steps:

S1: Measurement preparation: Select a fixed-focus lens with a focal length suitable for measurement, install the camera at the set position, adjust the lens focal length and aperture, so that the camera can capture the full-field image of the bridge to be measured;

S2: Determination of the scale factor of the whole field: First specify the zone of interest on the bridge, measure the distance from several points on the bridge to the camera with a laser rangefinder, and then derive each point on the zone of interest according to the geometric relationship of the points on the bridge surface. The distance from each measurement point to the camera, and finally all the scale factors on the band of interest are calculated by the single-point calibration method of the oblique optical axis;

S3: Image acquisition: the images before and after the deformation of the bridge are collected during the loading process;

S4: full-field image displacement calculation: use fast digital image correlation algorithm matching analysis to calculate the image displacement of each pixel corresponding to the band of interest of the bridge;

S5: Full-field deflection calculation: Convert the image displacement and its scale factor of each point on the belt of interest to its actual deflection/displacement.

Preferably, in step S2, a laser rangefinder is used to measure the distances from at least three points on the bridge to the camera.

Preferably, in step S2, the calculation formula for calculating all the scale factors on the band of interest by using the oblique optical axis single-point calibration method is as follows:

In the formula: L is the distance from the single point to be measured to the optical center of the camera, K _SF is the proportional coefficient from the pixel displacement of the image to the actual physical displacement, x is the abscissa of the point to be measured in the image, and x _c is the horizontal axis of the image center. Coordinate position, y is the ordinate of the point to be measured in the image, _yc is the ordinate position of the image center, l _ps is the physical size of the camera pixel, f is the focal length of the lens, and β is the vertical angle between the camera and the horizontal ground.

Preferably, the calculation method of the distance L from the single point to be measured to the optical center of the camera is as follows:

The world coordinate system O _c and the image coordinate system o are respectively established. First, a series of identification points P _n are selected on the strip area to be measured, and the pixel points P _n corresponding to the identification points are obtained on the image collected by the camera to obtain Its image coordinates (x _i , y _i ), transform the image coordinates to the three-dimensional coordinates in the image coordinate system as p _i ′((x _i -x ₀ )l _ps ,(y _i -y ₀ )l _ps ,f) , where (x ₀ , y ₀ ) is the image center coordinate, _lps is the actual physical size of a single pixel, and f is the camera focal length; use a laser _rangefinder to measure the distance Li between each calibration point and the camera optical center, Then the magnification M _i of the selected identification point from the three-dimensional coordinates of the image to the three-dimensional coordinates of the world can be calculated by the following formula:

According to the principle of similar triangles, the three-dimensional world coordinates of each identification point in the world coordinate system can be expressed as:

P _i =p′ _i ·M _i

In order to obtain the magnification M of all points on the area to be measured, the three-dimensional space straight line equation of the measured strip area is approximately fitted in the world coordinate system, and its expression is set as:

Among them, (x _b , y _b , z _b ) is the mean value of the coordinates of all the selected points; (d ₁ , d ₂ , d ₃ ) is the direction vector of the three-dimensional straight line in space; in order to obtain the unknown direction vector in the equation, use Singular value decomposition method calculation;

After obtaining the space equation of the bridge surface to be measured, the distance L _Q from any point Q on the bridge (image coordinates q(x, y)) to the optical center of the camera can be obtained; the three-dimensional coordinates of this point in the world coordinate system are:

Q=((x-x0)l _ps , (yy ₀ )l _ps , f)·M

Putting it into the equation of the line can solve M _Q , and then the world coordinates of the point Q can be calculated, then:

Preferably, the calculation method using singular value decomposition is specifically: after singular value decomposition is performed on the matrix formed by the standardized coordinates of all points, the left singular vector corresponding to the largest singular value is the direction vector.

Adopt above-mentioned technical scheme, the present invention has following beneficial effect:

1) Easy to use: The main component is a camera and has a large field of view, so there is no need to control and install each measurement separately, so as to achieve "visible measurement".

2) Full-field measurement: Each pixel in the image captured by the camera can be accurately calibrated without using auxiliary equipment, that is, all pixels can be converted into actual displacement/deflection.

3) Real-time measurement: An efficient and robust initial value transfer strategy ensures the speed of accurate image matching operations for massive data.

Description of drawings

In order to illustrate the specific embodiments of the present invention or the technical solutions in the prior art more clearly, the following will briefly introduce the accompanying drawings used in the description of the specific embodiments or the prior art. Obviously, the accompanying drawings in the following description The drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without creative efforts.

Fig. 1 is the overall flow chart of the whole field real-time bridge deflection measurement method of the present invention;

Fig. 2 is the schematic diagram that the present invention calculates each measuring point to the optical center distance L of camera;

FIG. 3 is a flow chart of the initial parameter transmission strategy of the present invention.

Detailed ways

The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are a part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only used to illustrate and explain the present invention, but not to limit the present invention.

1, the present invention provides a full-field real-time bridge deflection measurement method, which includes the following steps:

S1: Measurement preparation: select a fixed-focus lens with a focal length suitable for measurement, install camera 3 at the set position, adjust the focal length and aperture of the lens, so that camera 3 can capture the full-field image of the bridge to be measured;

S2: Determination of the scale factor of the whole field: First specify the zone of interest on the bridge 1, use the laser rangefinder 2 to measure the distance from several points on the bridge to the camera 3, and then derive the sense of The distance from each measurement point on the belt of interest to the camera 3, and finally all the scale factors on the belt of interest are calculated by using the single-point calibration method of the oblique optical axis;

S3: Image acquisition: the images before and after the deformation of the bridge are collected during the loading process;

S4: full-field image displacement calculation: use fast digital image correlation algorithm matching analysis to calculate the image displacement of each pixel corresponding to the band of interest of the bridge;

S5: Full-field deflection calculation: Convert the image displacement and its scale factor of each point on the belt of interest to its actual deflection/displacement. The computer 4 is used to process the data measured by the laser rangefinder 2 and obtain the distance from any point on the target line to the target surface of the camera 3 through straight line fitting. Obtain deflection information at any point.

Preferably, in step S2, a laser rangefinder is used to measure the distances from at least three points on the bridge to the camera.

Preferably, in step S2, the calculation formula for calculating all the scale factors on the band of interest by using the oblique optical axis single-point calibration method is as follows:

In the formula: L is the distance from the single point to be measured to the optical center of the camera, K _SF is the proportional coefficient from the pixel displacement of the image to the actual physical displacement, x is the abscissa of the point to be measured in the image, and x _c is the horizontal axis of the image center. Coordinate position, y is the ordinate of the point to be measured in the image, _yc is the ordinate position of the image center, l _ps is the physical size of the camera pixel, f is the focal length of the lens, and β is the vertical angle between the camera and the horizontal ground.

Preferably, the calculation method of the distance L from the single point to be measured to the optical center of the camera is as follows:

2, respectively establish the world coordinate system O _c and the image coordinate system o, first select a series of identification points P _n on the band area to be measured, and obtain the corresponding identification points on the image collected by the camera. Pixel point p _n , get its image coordinates (x _i , y _i ), transform the image coordinates into three-dimensional coordinates in the image coordinate system as p _i ′((x _i -x ₀ )l _ps ,(y _i -y ₀ )l _ps ,f), where (x ₀ , y ₀ ) is the image center coordinate, l _ps is the actual physical size of a single pixel, and f is the camera focal length; use a laser rangefinder to measure the distance between each calibration point and the camera’s optical center The distance L _i between the selected identification points can be calculated by the following formula _:

According to the principle of similar triangles, the three-dimensional world coordinates of each identification point in the world coordinate system can be expressed as:

P _i =p′ _i ·M _i

In order to obtain the magnification M of all points on the area to be measured, the three-dimensional space straight line equation of the measured strip area is approximately fitted in the world coordinate system, and its expression is set as:

Among them, (x _b , y _b , z _b ) is the mean value of the coordinates of all the selected points; (d ₁ , d ₂ , d ₃ ) is the direction vector of the three-dimensional straight line in space; in order to obtain the unknown direction vector in the equation, use Singular value decomposition method calculation;

After obtaining the space equation of the bridge surface to be measured, the distance L _Q from any point Q on the bridge (image coordinates q(x, y)) to the optical center of the camera can be obtained; the three-dimensional coordinates of this point in the world coordinate system are:

Q=((x-x0)l _ps , (yy ₀ )l _ps , f)·M

Putting it into the equation of the line can solve M _Q , and then the world coordinates of the point Q can be calculated, then:

Preferably, the calculation method using singular value decomposition is specifically: after singular value decomposition is performed on the matrix formed by the standardized coordinates of all points, the left singular vector corresponding to the largest singular value is the direction vector.

Combined with Figure 3, the classical high-precision digital image correlation method (DIC) can perform real-time high-precision matching for each measurement point individually. The core of this method is an iterative numerical method, that is, the more accurate the initial value, the less iterations, and the faster the calculation. According to the characteristics of the real-time measurement of the full-field deflection, the invention provides an initial value transmission scheme that conforms to the actual full-field deflection measurement.

First, select several points with high average gray gradient from the area to be tested as the calculation seed points; then, accurately match the calculation reference image and the first image after deformation (ie, the first two frames) of these seed points initial displacement. After obtaining the initial displacement of the seed point, the initial displacement of the remaining calculated points except the seed point can be obtained from their neighboring points according to the classical DIC scheme. Finally, for subsequent images beyond the first two frames, the initial displacement value of each point is directly determined using the displacement obtained in the first two images. By using the method of motion estimation and transferring the initial displacements in time sequence, the initial estimation required for the fast matching calculation of each point can be obtained efficiently and reliably. In most cases, this method is very robust and efficient. The initial parameter transfer strategy is described in Figure 3.

In real-time deflection measurement, the acquisition time interval between two adjacent frames is very short. Therefore, the displacement of corresponding points in three adjacent image frames can be approximately regarded as a linear function with respect to time. Denote the displacement of a point in the (n-2)th frame and the (n-1)th frame image as (u _n-2 ,v _n-2 ) and (u _n-1 ,v _n-1 ), respectively. The displacement (u _n , v _n ) of the same point in n frames can be approximated as:

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, but not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: The technical solutions described in the foregoing embodiments can still be modified, or some or all of the technical features thereof can be equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the technical solutions of the embodiments of the present invention. scope.

Claims (5)

1. A full-field real-time bridge deflection measurement method is characterized by comprising the following steps:

s1: preparation of measurement: selecting a fixed-focus lens with a focal length suitable for measurement, installing a camera at a set position, and adjusting the focal length and the aperture of the lens to enable the camera to acquire a full-field image of the bridge to be measured;

s2: determining a full-field scale factor: firstly, an interesting belt on a bridge is appointed, a laser range finder is used for measuring the distance from a plurality of points on the bridge to a camera, then the distance from each measuring point on the interesting belt to the camera is deduced according to the geometric relation of the points on the surface of the bridge, and finally all scale factors on the interesting belt are calculated by utilizing an oblique optical axis single-point calibration method;

s3: image acquisition: acquiring images before and after the deformation of the bridge in the loading process;

s4: full-field image displacement calculation: calculating the image displacement of each corresponding pixel on the bridge interest zone by using a rapid digital image correlation algorithm matching analysis;

s5: and (3) calculating the full-field deflection: the image displacement of each point on the strip of interest and its scaling factor are converted to its actual deflection/displacement.

2. The full-field real-time bridge deflection measuring method according to claim 1, wherein in step S2, the distances from at least three points on the bridge to the camera are measured by a laser range finder.

3. The full-field real-time bridge deflection measurement method according to claim 1, wherein in step S2, the calculation formula for calculating all the scale factors on the interest zone by using the oblique optical axis single-point calibration method is as follows:

in the formula: l is the distance from the single point to be measured to the optical center of the camera, K_SFIs a proportionality coefficient from image pixel displacement to actual physical displacement, x is the abscissa of the point to be measured in the image, x_cIs the horizontal coordinate position of the center of the image, y is the vertical coordinate of the point to be measured in the image, y_cIs the image center ordinate position,/_psIs the physical size of the pixel of the camera, f is the focal length of the lens, and beta is the vertical included angle between the camera and the horizontal ground.

4. The full-field real-time bridge deflection measurement method according to claim 3, wherein the distance L from the single point to be measured to the optical center of the camera is calculated as follows:

respectively establishing a world coordinate system O_cAnd an image coordinate system o, firstly selecting a series of identification points P on the strip-shaped area to be detected_nAnd acquiring pixel points p corresponding to the identification points on the image acquired by the camera_nTo obtain the image coordinates (x) thereof_i,y_i) Transforming the image coordinate into three-dimensional coordinate p in the image coordinate system_i′((x_i-x₀)l_ps,(y_i-y₀)l_psF) wherein (x)₀,y₀) As the image center coordinates,/_psAs a single pixelF is the camera focal length; measuring the distance L between each calibration point and the optical center of the camera by using a laser range finder_iThen the magnification M from the image three-dimensional coordinate to the world three-dimensional coordinate of the selected identification point_iCan be calculated by the following formula:

according to the principle of similar triangles, the three-dimensional world coordinates of each identification point in the world coordinate system can be expressed as follows:

P_i＝p′_i·M_i

in order to obtain the amplification factor M of all points on the region to be measured, a three-dimensional space linear equation of the measured banded region is approximately fitted in a world coordinate system, and the expression is set as follows:

wherein (x)_b,y_b,z_b) The mean value of the coordinates of all the selected identification points; (d)₁,d₂,d₃) The direction vector of the space three-dimensional straight line is shown; in order to obtain unknown direction vectors in the equation, calculating by adopting a singular value decomposition method;

after obtaining the space equation of the surface of the bridge to be measured, the distance L from any point Q (image coordinate Q (x, y)) on the bridge to the optical center of the camera can be obtained_Q(ii) a The three-dimensional coordinates of the point in the world coordinate system are:

Q＝((x-x₀)l_ps，(y-y₀)l_ps，f)·M

the M can be solved by substituting the linear equation into the linear equation_QThen, the world coordinates of point Q can be calculated, then:

5. the full-field real-time bridge deflection measurement method according to claim 4, wherein the calculation method adopting singular value decomposition specifically comprises: after singular value decomposition is carried out on a matrix formed by the standardized coordinates of all the points, the left singular vector corresponding to the maximum singular value is the direction vector.

Images