CN110146024B - Double-precision displacement measurement method based on self-adaptive search - Google Patents

Abstract

The invention relates to the technical field of ancient building protection, and provides a double-precision displacement measuring method based on self-adaptive search in order to solve the problem that a colored drawing beam is damaged because a target needs to be arranged on the colored drawing beam in the conventional deformation measurement of the colored drawing beam, wherein the double-precision displacement measuring method based on self-adaptive search comprises the following steps: an image acquisition step: acquiring an image of a measured object and generating image information; a storage step: storing the calculation rule; a calculation step: and calculating the image information according to the calculation rule to obtain the displacement.

Description

Double-precision Displacement Measurement Method Based on Adaptive Search

technical field

The invention relates to the technical field of ancient building protection, in particular to a double-precision displacement measurement method based on self-adaptive search.

Background technique

At present, in the deformation measurement of painted beams in ancient buildings, measuring equipment such as levels and total stations are mainly used. When using levels or total stations for measurement, it is necessary to set targets on the painted beams to complete high-precision measurements. . Since the setting of the target on the painted beam requires the target to be in close contact with the painted beam, the surface of the painted beam will be damaged, and the target set on the painted beam will also affect the appearance of the painted beam. Therefore, in order to avoid the problem of damage to the painted beam due to setting targets on the painted beam, the present invention proposes a non-contact measuring method.

Contents of the invention

The invention intends to provide a double-precision displacement measurement method based on self-adaptive search to solve the problem of damage to the painted beam due to the need to set targets on the painted beam in the current deformation measurement of the painted beam.

The basic solution provided by the present invention is: a double-precision displacement measurement method based on adaptive search, comprising the following steps:

Image acquisition step: acquire the image of the object under test and generate image information;

Storage step: store calculation rules;

Calculation steps: calculate the image information according to the calculation rules to obtain the displacement;

Among them: calculation rules include parameter algorithm, coefficient threshold algorithm, size algorithm, local exhaustive search method, gradient method and weight calculation method;

Calculation steps include:

Parameter calculation step: calculate the two image information of the measured object before and after a period of time according to the parameter algorithm to obtain the parameter value;

The coefficient threshold calculation step: calculate the parameter value according to the coefficient threshold algorithm to obtain the coefficient threshold;

Subset selection step: Calculate the size of the subset satisfying the coefficient threshold according to the size algorithm;

Integer pixel displacement search step: used to obtain the initial value of the integer pixel according to the local exhaustive search of the initial calculation subset;

The sub-pixel displacement initial value search step: used to determine the sub-pixel area subset according to the initial value of the integer pixel and the subset selection method, and calculate the sub-pixel area subsets according to the gradient method for the integer pixel initial value points at both ends of the sub-pixel area subset sub-pixel displacement within;

Calculation step of accurate displacement value: performing weight calculation on sub-pixel displacement to obtain accurate displacement value.

As shown in Figure 7, the part inside the dotted line box in the figure is the part of the adaptive subset size calculation method, and the optimal subset size at the point to be measured is selected through multiple cycle calculations.

其中式中D(η)为图像噪声方差，∑∑(f_x)²与∑∑(f_y)²分别为x方向与y方向上的子集灰度梯度平方和系数；f(x,y)为变形前的子集图像。

In the formula, D(η) is the variance of image noise, ∑∑(f _x ) ² and ∑∑(f _y ) ² are the coefficients of the sum of squares of the subset gray gradient in the x direction and y direction respectively; f(x,y ) is the subset image before deformation.

Explanation: In this scheme, the calculation of deformation displacement is based on the DIC data image processing method. In the theoretical system of DIC digital image processing method, it is generally believed that the gray information on the surface of the material will be displaced and deformed synchronously with the material. It is also based on this It is assumed that a mathematical relationship between the measured area before deformation and the deformed image is established. The DIC digital image correlation method obtains the digital image of the surface of the measured plane object before and after deformation through the camera, and then obtains the displacement of each point on the surface of the measured object by matching the corresponding image subsets in the digital image before and after deformation, that is, obtains the displacement of each point on the surface of the measured object through a camera or a video camera. The digital image of the measured object, and then the displacement of the numerical image movement is obtained through the solution method. The solution method firstly selects the size of the subset manually through the selection of the correlation function and the shape function, and then continues to adjust the pixel displacement, sub-pixel It is located in the search, and finally the deformation measurement is completed through the calibration of the pixel displacement.

The working principle and beneficial effects of the basic scheme are: compared with the existing measurement methods, 1. In this scheme, the displacement is obtained by calculating the image information of the painted beam, and the non-contact measurement is realized, which avoids the The problem of damage to the painted beam caused by direct contact with the painted beam during the measurement process;

2. At present, although the adaptive path search method used to calculate the integer pixel displacement overcomes the problems of low search accuracy and low calculation efficiency, its first displacement search point is assumed to start from the origin, so There is a problem that the initial value of the displacement is unknown, so that it is easy to cause the problem of multiple wrong local optimal solutions, thus making the final calculation result inaccurate. In this solution, the local exhaustive search method is used to calculate the entire pixel displacement, and the problem of multiple wrong local optimal solutions is avoided through the determined accurate initial displacement, thereby ensuring the accuracy of the calculation results.

3. Considering that when selecting the calculation subset, in the traditional selection method, the square subsets in the x and y directions of the speckle image are generally selected, as shown in Figure 6. However, for the painted beams of ancient buildings Said that due to the serious difference in the pattern gradient in the x direction and the y direction, the size of the subset in the x direction and the y direction is unreasonable, which not only increases the calculation amount and reduces the calculation speed, but also for the x direction and y direction For an inhomogeneous displacement field with an unreasonable size in the direction, an excessively large subset size will also reduce the calculation accuracy of the points to be measured. Therefore, in this scheme, when determining the size of the subset, the size of the subset calculated by the size algorithm that satisfies the coefficient threshold is selected as the initial size Mx and My of the x-direction and y-direction of the initial subset around the point to be measured, so as to determine the size to be measured The optimal subset size of points, compared with the existing square subset, the self-adaptive self-determination in this scheme is obviously smaller than the calculation amount of the square subset, and the calculation speed is improved without reducing the accuracy.

Preferred solution 1: as a preferred basic solution, it also includes a timing step: sending a start signal according to the stored timing information; a control step: controlling the start of the image acquisition step after receiving the start signal. Beneficial effects: in this solution, the automatic collection of the image of the measured object is realized through the setting of the timing module and the control module, and the operation is convenient.

Preferred option 2: As the preferred option of the basic option, the calculation step also calculates the obtained accurate displacement value and the displacement threshold, and sends an alarm message when the calculated accurate displacement value is greater than or equal to the displacement threshold. Beneficial effects: Considering that the measurement of the deformation of the painted beam is to avoid the problem of fracture or damage of the painted beam due to excessive deformation, in this scheme, by setting the displacement threshold, the accurate displacement value of the painted beam is calculated to be greater than or equal to When the displacement threshold is reached, that is to say, the painted beam is facing the problem of damage, the sending of the alarm information can remind in time to ensure that the staff can take protective measures on the painted beam in time to avoid damage to the painted beam.

The third preferred solution: as a preferred option of the second preferred solution, it also includes an input step: inputting the displacement threshold, and storing the displacement threshold. Beneficial effects: Considering that for different painted beams, the critical point of deformation is different, that is to say, the displacement threshold is different, so an input module is also set in this scheme, which is convenient for the staff to input the displacement threshold, thus ensuring the accuracy of the reminder .

Preferred option 4: As a preferred option of preferred option 2, the displacement threshold includes multiple sets of matching painted beam types and critical thresholds, and also includes an input step: input the painted beam type; matching step: cooperate with the input painted beam type as the current Critical threshold for Painted Beam Displacement Threshold. Beneficial effects: Considering that for different painted beams, the critical point of deformation is different, that is to say, the displacement threshold is different, so the displacement threshold in this scheme includes the type of painted beam and the matching critical threshold, and an input module is also provided to facilitate the work Personnel input the type of painted beam, and the matching module automatically matches the critical threshold of the current painted beam for calculation, thus ensuring the accuracy of the reminder.

Preferred option five: As the preferred option of the basic option, a CCD camera is used in the image acquisition step. Beneficial effect: the CCD camera has the characteristics of small size, light weight, no influence of magnetic field, and anti-vibration and impact.

Preferred option 6: As a preferred option of preferred option 5, in the image acquisition step, the optical axis of the CCD camera is set directly to the centroid of the object to be measured and perpendicular to the pattern surface of the object to be measured. Beneficial effects: In this solution, the optical axis of the image acquisition module is set facing the object to be measured, so as to ensure the accuracy of the acquired image, thereby ensuring the accuracy of the calculated displacement exact value.

The preferred solution 7: as the preferred basic solution, it also includes a processing step: performing edge removal processing on the image information, and performing calculation on the image information after the edge removal processing in the calculation step. Beneficial effects: In this scheme, in order to avoid the impact of the junction of the structure edge area on the displacement recognition result, the generated image information is also processed to remove the edge, and the calculation module calculates the processed image information, thereby improving the final displacement accuracy. value accuracy.

The eighth preferred solution: As a preferred option of the seventh preferred solution, edge removal processing is performed on the left and right edges of the image information in the processing step. Beneficial effects: Considering that after the painted beam is fixed, the left and right ends are usually all fixed on the column, so in this solution, only the left and right edges of the image information are removed, which reduces the amount of processing .

The ninth preferred solution: as a preferred option of the eighth preferred option, it also includes an input step: inputting the edge processing size, and performing edge removal processing on the image information according to the edge processing size in the processing step. Beneficial effects: the setting of the input module is convenient for workers to manually input the size of the edge treatment, and the operation is convenient.

Description of drawings

Fig. 1 is the modular block diagram of the measurement system adopted in the first embodiment of the double-precision displacement measurement method based on adaptive search in the present invention;

Fig. 2 is test loading device and measuring device in embodiment one;

Figure 3(a) is the initial working condition of Hexi painted beam;

Figure 3(b) is the final working condition of Hexi painted beam;

Figure 4 is the pixel calibration data of Hexi color painting beam;

Figure 5 shows the DIC deflection identification results of the three monitoring and measuring points of Xi Caihua Liang under working condition 5;

FIG. 6 is a schematic diagram of adaptive subset size parameters;

Fig. 7 is the whole pixel and sub-pixel search subset adaptive program selection process;

Figure 8 is a schematic diagram of the search process;

Figure 9 is a schematic diagram of the gradient method;

Figure 10(a) is the calculation result of the sub-pixel displacement with the initial value point of the whole pixel being 0 under the influence of noise;

Figure 10(b) is the calculation result of the sub-pixel displacement with the initial value point of the whole pixel being 1 under the influence of noise;

Fig. 11 is a schematic diagram of fitting the noise standard deviation SD and the weight coefficient a at eight sub-pixel displacement points;

Fig. 12 is the cubic spline interpolation curve of sub-pixel displacement and weight coefficient under different noise conditions;

Figure 13 is a comparison of the accuracy of the integer pixel displacement calculation method;

Figure 14 shows the double-precision algorithm and stability verification;

Figure 15 is the test beam pattern of Hexi color painting;

Figure 16 is a schematic diagram of test loading.

Detailed ways

The following is further described in detail through specific implementation methods:

The reference signs in the accompanying drawings of the specification include: a loading jack 1 , a pressure sensor 2 , a dial indicator 3 , a light 4 , and a CCD camera 5 .

A double-precision displacement measurement method based on adaptive search, comprising the following steps:

Image acquisition step: acquire the image of the object under test and generate image information;

Storage step: store calculation rules;

Calculation steps: calculate the image information according to the calculation rules to obtain the displacement;

Among them: calculation rules include parameter algorithm, coefficient threshold algorithm, size algorithm, local exhaustive search method, gradient method and weight calculation method;

Calculation steps include:

Parameter calculation step: calculate the two image information of the measured object before and after a period of time according to the parameter algorithm to obtain the parameter value;

The coefficient threshold calculation step: calculate the parameter value according to the coefficient threshold algorithm to obtain the coefficient threshold;

Subset selection step: Calculate the size of the subset satisfying the coefficient threshold according to the size algorithm;

Integer pixel displacement search step: used to obtain the initial value of the integer pixel according to the local exhaustive search of the initial calculation subset;

The sub-pixel displacement initial value search step: used to determine the sub-pixel area subset according to the initial value of the integer pixel and the subset selection method, and calculate the sub-pixel area subsets according to the gradient method for the integer pixel initial value points at both ends of the sub-pixel area subset sub-pixel displacement within;

Calculation steps of accurate displacement value: perform weight calculation on sub-pixel displacement to obtain accurate displacement value;

Alarming step: the calculation step also calculates the obtained accurate displacement value and displacement threshold, and sends an alarm message when the calculated accurate displacement value is greater than or equal to the displacement threshold.

Specifically, it also includes a timing step: sending a start signal according to the stored timing information; a control step: after receiving the start signal, controlling the start of the image acquisition step;

Input step: input the displacement threshold value, store the displacement threshold value; the displacement threshold value includes multiple sets of matching painted beam types and critical threshold values, and also includes an input step: input the painted beam type; matching step: cooperate according to the input painted beam type as the current The critical threshold of the displacement threshold of the painted beam; the processing step: remove the edge processing on the image information, and calculate the image information after the edge removal processing in the calculation step, specifically, include the input step: input the edge processing size, according to the processing step The edge processing size performs edge removal processing on the image information.

Based on the above measurement method, as shown in FIG. 1 , this embodiment also discloses a measurement system, including an image acquisition module, which is used to acquire an image of a measured object and generate image information;

A storage module for storing calculation rules and displacement thresholds;

The input module is used to input timing information and edge processing size, and the storage module stores the timing information;

The processing module is used to perform edge removal processing on the image information; specifically, the processing module removes the edge processing on the left and right edges of the image information according to the edge processing size;

A calculation module, configured to calculate the displacement of the image information after edge removal processing according to calculation rules;

Specifically, the calculation rules include parameter algorithm, coefficient threshold algorithm, size algorithm, local exhaustive search method, gradient method and weight calculation method;

A timing module, the storage module pre-stores timing information, and the timing module is used to send a start signal according to the timing information;

The control module is used to control the image acquisition module to acquire the image of the measured object and generate image information after receiving the start signal;

Computing modules include:

Parameter calculation unit: used to calculate the two image information of the measured object before and after a period of time according to the parameter algorithm to obtain the parameter value; if the timing information is T, the two image information before and after are the image information at time t and the time Image information at t+T;

Coefficient threshold calculation unit: used to calculate the parameter value according to the coefficient threshold algorithm to obtain the coefficient threshold;

Subset selection unit: used to calculate the size of the subset satisfying the coefficient threshold according to the size algorithm;

Integer pixel displacement search unit: used to obtain the initial value of the integer pixel according to the local exhaustive search of the initial calculation subset;

Sub-pixel displacement initial value search unit: used to determine the sub-pixel area subset according to the initial value of the integer pixel and the subset selection method, and respectively calculate the sub-pixel area subsets for the integer pixel initial value points at both ends of the sub-pixel area subset according to the gradient method sub-pixel displacement within;

Accurate displacement value calculation unit: perform weight calculation on sub-pixel displacement to obtain accurate displacement value;

The calculation module is also used to calculate the obtained accurate displacement value and displacement threshold;

The alarm module is configured to send an alarm message when the calculated precise displacement value is greater than or equal to the displacement threshold.

The above size algorithm is as follows:

(1)

(2)

function[TH]=threshold(imgT1,imgT2,imgR,m,n,Srr,Scr,sx,sy,SD)

%imgT1: pre-processing image 1

%imgT2: pre-processing image 2

%imgR: reference image

%m: The y coordinate of the center point of the reference image to calculate the subset

%n: x-coordinate of the reference image to calculate the center point of the subset

%SD: set the standard deviation of displacement identification error

%Srr: Integer pixel search reference image calculation subset y-direction radius

%Scr: Integer pixel search reference image calculation subset x-direction radius

%sx: reference image calculation subset internal x-direction calculation interval

%sy: calculation interval in the y direction within the reference image calculation subset

[H]=crossCorrelation(imgR,m,n,Srr,Scr,sx,sy);

[Dn]=varianceNoise(imgT1, imgT2);

TH＝H*Dn/(SD^2)

(3)

clc, clear;

imgT1=double(imread('T1.tif'));% pre-processing picture 1

imgT2=double(imread('T2.tif'));% pre-processing picture 2

imgR=double(imread('P_1.tif'));% reference image to be calculated

SD=0.01; % displacement recognition error (standard deviation, unit: pixel)

m=1024;

n=1024;

[Ssrr,Sscr]=subpixelsubsetsize(imgT1,imgT2,imgR,m,n,SD)

(4)

For the convenience of description, the above calculation schemes are collectively referred to as the adaptive search double-precision gradient DIC method.

In the above process, the calculation of the deformation displacement is based on the DIC data image processing method. In the theoretical system of the DIC digital image processing method, it is generally believed that the gray information on the surface of the material will be displaced and deformed synchronously with the material. It is also based on this assumption. The mathematical relationship between the measured area before deformation and the deformed image is established. The DIC digital image correlation method obtains the digital image of the surface of the measured plane object before and after deformation through the camera, and then obtains the displacement of each point on the surface of the measured object by matching the corresponding image subsets in the digital image before and after deformation, that is, obtains the displacement of each point on the surface of the measured object through a camera or a video camera. The digital image of the measured object, and then the displacement of the numerical image movement is obtained through the solution method. The solution method firstly selects the size of the subset manually through the selection of the correlation function and the shape function, and then continues to adjust the pixel displacement, sub-pixel Displacement search, and finally through the calibration of pixel displacement to complete the deformation measurement.

In this scheme, the displacement initial value vector in the traditional self-adaptive search method is updated through an exhaustive method, and an accurate displacement initial value vector is obtained. The search method consists of two processes. First, in the same image, a local exhaustive search is performed on the first search point according to a given search area to find out the precise displacement; then, the accurate displacement information of the first point is used as The initial value is to perform recursive adaptive path search on other adjacent points, and calculate the displacement information of other measurement points in the image. The specific solution process is as follows:

Step 1: Based on the given search area, perform a local exhaustive search for the displacement of the first calculation point in the image to obtain the initial displacement vector (u, v).

The second step: take the max(u,v) calculated in the first step as the calculation step of the adjacent point to be measured, and respectively calculate the four prediction points up, down, left and right with the step of max(u,v) away from the point and the first The displacement vector points obtained by one-step calculation, a total of five points (only four points if coincident) are matched and calculated to obtain the best matching position.

Part 3: Search in the small diamond search mode around the best matching position obtained in the second step. If the best matching position is the center of the small diamond, end the search, and this point will be the final displacement search point; otherwise, move the center of the small diamond To the new best matching point, repeat the small diamond search mode search until the best matching point is the center of the small diamond, and obtain the displacement value of this point.

Step 4: Repeat steps 2 to 3 for the next adjacent point until all points to be measured are searched.

The schematic diagram of the search process is shown in Figure 8.

In this embodiment, the adaptive size subset using the local exhaustive search method is obviously smaller than the current square subset, and the calculation speed is improved without reducing the accuracy. At the same time, the example compared with the square Proved, the results are shown in Table 1.

Appendix Table 1. Comparison of Adaptive Size Subset and Square Subset

Compared with the medium-coarse-fine search method, three-step search method and diamond search method, the comparison results are shown in Figure 13.

The process of using the gradient method to calculate the sub-pixel displacement in the subset of the sub-pixel area is as follows:

处的权值系数ɑ近似为亚像素位移精确值位置x处的权值系数，x由公式4计算得到，选用该点进行近似代替是由于越靠近整像素初值点的亚像素位移计算精度越高，越能够近似代替位移精确值进行权值系数的计算。根据图10中计算得到的亚像素位移数据，由公式3反算出在不同噪声情况下0～0.1pixel间隔为0.1pixel的各个亚像素位移控制点处权值系数ɑ_i,i＝0.1,0.2,0.3…1；其中本文假定亚像素位移为0pixel时，在相应整像素初值点计算得到的亚像素位移即为精确值。故由理论分析可，ɑ₀＝0,ɑ₁＝1，又由于亚像素位移为0.5pixel时从两个方向计算该点位移的均值误差互为相反数，因此有ɑ_0.5＝0.5；然后采用指数函数对剩余八个亚像素控制点权值系数ɑ_i与图像噪声标准差SD采用公式5进行拟合，其中c₁,c₂,c₃为拟合系数，结果如图11所示。As shown in Fig. 9, the position of the exact value of the sub-pixel displacement is recorded as E, and its coordinate is x, and the sub-pixel displacement calculated from the initial value point obtained by the integer pixel search is recorded as A, and the coordinate is x ₁ , and the adjacent integer The sub-pixel displacement calculated by the pixel initial value point is recorded as B, and the coordinate is x ₂ ; the coordinate x of the sub-pixel displacement exact value E is calculated from x ₁ and x ₂ , as shown in formula 3, where ɑ is the corresponding sub-pixel displacement The weight coefficient at the exact value position x, but since the exact position cannot be known, this paper uses

The weight coefficient at the position ɑ is approximately the weight coefficient at the position x of the exact value of the sub-pixel displacement. The higher the value, the more it can approximate the exact value of the displacement to calculate the weight coefficient. According to the sub-pixel displacement data calculated in Figure 10, the weight coefficients at each sub-pixel displacement control point ɑ _i , i=0.1,0.2, 0.3…1; where this article assumes that the sub-pixel displacement is 0pixel, the sub-pixel displacement calculated at the corresponding integer pixel initial value point is the exact value. Therefore, according to theoretical analysis, ɑ ₀ = 0, ɑ ₁ = 1, and because the average error calculated from the two directions of the point displacement when the sub-pixel displacement is 0.5pixel is opposite to each other, so there is ɑ _0.5 = 0.5; then use The exponential function fits the remaining eight sub-pixel control point weight coefficients ɑ _i and the image noise standard deviation SD using formula 5, where c ₁ , c ₂ , and c ₃ are fitting coefficients, and the results are shown in Figure 11.

After obtaining 11 sub-pixel displacement weight coefficient control points with an interval of 0.1 pixel from 0 to 0.1 pixel under different noise conditions, perform cubic spline interpolation on the weight coefficients at these 11 control points according to formula 6 respectively. The weight coefficient ɑ of any sub-pixel displacement point under the condition of arbitrary noise standard deviation is obtained, an example is shown in Figure 12.

x＝(1-a) x ₁ +a x ₂ (3)

Among them, SD is the standard deviation; c ₁ , c ₂ , c ₃ are fitting coefficients.

The above is the analysis result for the x direction. The calculation method for the sub-pixel displacement in the y direction is the same.

For the convenience of description, the above sub-pixel displacement calculation methods are collectively referred to as double-precision algorithms.

The above calculation method not only maintains the calculation stability of the gradient method in a noisy environment, but also effectively overcomes the shortcomings of the traditional gradient method's weak calculation accuracy and anti-noise ability. At the same time, the accuracy and stability of the N-R method and the gradient method were compared, and the results are shown in Figure 14.

The specific implementation process is as follows: In this embodiment, Hexi painted beams are taken as an example. As shown in Figure 2, the painted wooden beam component size is 1400mm×1400mm×50mm, and the painted image size on the beam surface is 1300mm×100mm. The influence of the pattern junction in the edge area on the displacement recognition results is removed by removing the 50mm close to the left and right edges, so the size of the colored drawing in the loading area is selected to be 1200mm×100mm. The loading method of the simply supported painted beam is selected as two-point loading of the distribution beam, and the loading interval is 400mm. The support spacing is 1200mm, and the loading control method is mid-span deflection control. In order to better compare the recognition effect of DIC, in this example, three dial gauges are placed on both sides of the wooden beam mid-span and 300mm away from the mid-span to measure the deflection of the painted beam, and the values measured by the dial gauges are taken as the real displacement. At the same time, a pressure sensor is placed under the loader to measure the real-time pressure. The experimental device is shown in Figure 2. In the actual measurement environment, the light intensity has a great influence on the calculation accuracy of the DIC displacement measurement method. Therefore, in this example, different light intensities are set under the same displacement condition to simulate the measurement situation under complex light intensities in the actual measurement process. In this experiment, the light intensity The conditions are divided into two levels 1 and 2, corresponding to weak and strong light intensity respectively.

First of all, preload the Hexi painted beams, and then formally load them after confirming that everything is normal.

After the experimental environment is set up, each beam is taken with a CCD camera to take two identical images before formal loading to perform adaptive subset selection for DIC calculation. Six working conditions (including the initial working condition) were set for each beam according to the maximum loading displacement, and each working condition was divided into 3 control groups according to the light intensity for image shooting. According to GB 50005-2017 "Standards for Design of Wooden Structures", the maximum deflection of the roof beam in the wooden structure should be less than l/250 in the flexural span, and l is used to calculate the span. In this experiment, l=1200mm. The maximum deflection condition of the painted beam set in this test is 12mm, which is 1/100 of the span of simply supported wooden beams and 2.5 times of the specification limit. Based on the maximum mid-span deflection of 12mm, the working condition numbers and working condition displacements are shown in Attached Table 2.

Attached Table 2 Working Condition Number and Working Condition Displacement

The experimental loading device and measuring device are arranged as shown in Figure 2. There are gaskets between the simply supported support and the loading point of the two distribution beams and the painted beam to prevent local crushing of the wooden beam. Three The dial gauge measures the deflection at different positions, and the upper part is the distribution beam, the loading jack and the pressure sensor (model: CFBLZ S-shaped tension and pressure sensor, acquisition instrument model: SDY2202 static strain gauge). CCD camera (model GZL-CL-41C6M-C, image resolution 2048×2048 pixels) is placed on a tripod about 4 meters in front of the painted beam, and its optical axis is aligned with the centroid of the beam and aligned with the painted pattern The surface is vertical, connect the camera data line to the computer, and use the software for image acquisition. During the experiment, two groups of white light sources with adjustable brightness were used for illumination.

The loading and measurement steps are as follows:

1) After the components, jacks, and pressure sensors are placed in place, pre-press the pressure. The pressure is gradually applied from small to large, so that the contact between each part is tight, and at the same time, the readings of the pressure sensor collector are reset to zero and recorded. The light condition is set to 2. Define the The status is 0 working condition. At this time, a CCD camera is used to capture two initial test images for image noise variance analysis and adaptive subset size selection. Then take pictures under this lighting condition and 1 lighting condition, and record the initial values of the three dial gauges.

2) Corresponding to the mid-span displacement control value of each working condition in Attached Table 2, use the loading device to carry out displacement loading under each working condition, record the readings of the pressure sensor and the readings of the three dial indicators after each level of loading is stable, and use the CCD camera to take two pictures at the same time. A deformed image of a painted beam under the set of lighting conditions.

3) After the loading is completed, check and organize the collected data, and slowly unload the loader after checking.

The initial condition of the wooden beam and the field test conditions of the final loading condition are shown in Figure 3(a) and Figure 3(b), and the loading data records of the wooden beam are shown in Table 3.

Attached Table 3. Loading Data of Hexi Painted Beam

First, preload the test beam, and then proceed to formal loading after confirming that everything is normal.

After the experimental environment is set up, each beam is taken with a CCD camera to take two identical images before formal loading to perform adaptive subset selection for DIC calculation. Six working conditions (including the initial working condition) were set for each beam according to the maximum loading displacement, and each working condition was divided into 3 control groups according to the light intensity for image shooting. According to GB 50005-2017 "Standards for Design of Wooden Structures", the maximum deflection of the roof beam in the wooden structure should be less than l/250 in the flexural span, and l is used to calculate the span. In this experiment, l=1200mm. The maximum deflection condition of the painted beam set in this test is 12mm, which is 1/100 of the span of simply supported wooden beams and 2.5 times of the specification limit. Based on the maximum mid-span deflection of 12mm, the working condition numbers and working condition displacements are shown in Attached Table 2.

The experimental loading device and measuring device are shown in Figure 2. There are gaskets between the simply supported support and the loading points of the two distribution beams and the painted beams to prevent local crushing of the wooden beams. The meter measures the deflection at different positions, the upper part is the distribution beam, the loading jack and the pressure sensor (model: CFBLZ S-shaped tension and pressure sensor, the acquisition instrument model: SDY2202 static strain gauge). CCD camera (model GZL-CL-41C6M-C, image resolution 2048×2048 pixels) is placed on a tripod about 4 meters in front of the painted beam, and its optical axis is aligned with the centroid of the beam and aligned with the painted pattern The surface is vertical, connect the camera data line to the computer, and use the software for image acquisition. During the experiment, two groups of white light sources with adjustable brightness were used for illumination.

At the same time, the DIC method (Coarse-Fine Gradient DIC, CG-DIC) based on manual selection of subsets, coarse-fine pixel search method and gradient sub-pixel displacement search method and the adaptive search double-precision gradient DIC method ( Adaptive-Search Double-Precision-Gradient DIC, AD-DIC) for comparison.

Through the analysis of the image data of Hexi colored painted beams, it can be seen that the corresponding length pixels in the image corresponding to the actual 100mm beam height are shown in Figure 4, and then the relationship between the pixel displacement and the real displacement in each group of experiments can be calculated by formula 7 quantitative relationship.

For the analysis of calculation accuracy, the DIC method is used to identify the deflection data within the width range of the painted beam pattern corresponding to the position of the dial gauge, and compare it with the data of the dial gauge. Calculate the standard deviation (StandardDeviation, SD) of the deflection data identified by the DIC method and the average error (Averaging Error, AE) after subtracting the corresponding dial gauge measurement data, which are used as the calculation stability and accuracy evaluation standards, respectively. Let the deflection value identified by DIC method at each pixel point in the measurement range be d _i , i∈N, where N is the coordinate set of all pixel points in the measurement area, and d is the deflection of the wooden beam measured by the dial gauge in the corresponding area, Therefore, the mean error and standard deviation can be represented by Equation 8 and Equation 9, respectively.

Among them, the actual size of the reference object is LR, and the pixel size in _{the R} image is _LP ; the actual displacement is DR, and the pixel displacement in the _R image is _DP ; the recognized deflection value is d _i ; AE is the mean error; SD is the standard deviation, and N is the calculation points.

The average error and standard deviation of the identification results of the two DIC methods under each working condition at the three dial gauges of the Hexi painted beam bending test are shown in Attached Table 4 and Attached Table 5.

Attached table 4. Accuracy analysis of deflection identification of Hexi colored painted beams by DIC method

Attached Table 5. Stability Analysis of Deflection Identification of Hexi Painted Beam DIC Method

It can be seen from the attached table 4 that during the loading process of the Hexi painted beam, as the bending degree of the Hexi painted beam gradually increases, the non-contact deflection identification results of the two DIC methods at the mid-span measurement point are compared with the dial gauge contact measurement The mean error of the results has no obvious trend, but remains relatively stable, indicating that the bending degree of Hexi painted beam has little effect on the mid-span DIC deflection measurement. This is because the mid-span pattern of the Hexi painted beam is basically a translational displacement when it is bent, without obvious rotation and uneven stretching. Therefore, within a certain range, the deflection of the Hexi painted beam has a great influence on the identification of the mid-span deflection by the DIC method. Small. For the measurement positions on both sides of the mid-span, as the bending degree of the Hexi painted beam gradually increases, the mean errors of the identification results of the two DIC methods generally increase. This is because the deflection of the Hexi painted beam increases increases, the painted patterns of the deflection measurement positions on both sides of the mid-span not only increase in translational displacement, but also gradually increase the rotational displacement in the direction of the mid-span, so the interference to the image recognition method also gradually increases, resulting in errors in the recognition results increase. Comparing the AD-DIC method and the CG-DIC method proposed in this application for the deflection identification results of Hexi painted beams, it can be seen that the former is generally better than the latter, and the average error of the AD-DIC method deflection identification results is generally 0.1 mm or less, the maximum absolute value is 0.198mm, while the mean error of the deflection identification results by the CG-DIC method is generally above 0.1mm, and the maximum absolute value is 0.908mm.

It can be seen from the attached table 5 that with the increase of the deflection of Hexi painted beam, the standard deviation of the deflection identification results of the AD-DIC method at the three measuring points gradually increases. First, for the mid-span measurement position, although there is no rotational displacement within the beam height range, as the bending degree of the Hexi color painting beam increases, the difference in the bending tensile length of the upper and lower sides of the calculated pixel point subset within the beam height range gradually increases. increases, the small bending deformation of the subset gradually increases, and the interference increases when the front and rear image subsets are matched, so the discreteness of the calculated data will gradually increase, but since this error is random, the deflection within the beam height range The identified mean value error has little influence. For the measurement area on both sides of the mid-span, not only the bending deformation of the subset, but also the rotation angle of the subset exist. The resulting standard deviation will also increase. At the same time, it can be seen that the standard deviation of the deflection identification results of the CG-DIC method changes irregularly with the increase of the deflection of Hexi painted beams, and the discrete type is relatively large. Due to the inaccurate positioning of the whole pixel of the pixel. From the horizontal comparison of the two algorithms, it can be seen that the AD-DIC method is much more stable in deflection identification than the CG-DIC method, and the standard deviation of the deflection identification results is always kept below 0.06mm.

Based on the measurement data of the dial gauge, the difference (recognition error) between the recognition results of the two DIC algorithms in the three measurement areas in working condition 5 and the corresponding dial gauge measured data (recognition error) was analyzed, as shown in Figure 5.

It can be seen from Figure 5 that the CG-DIC method will produce a large positioning error of the whole pixel displacement (large peak) at some positions when the deflection is identified in the three measurement areas, while the AD-DIC method will There is always a high calculation accuracy, because the gray feature of the painted pattern is weaker than that of the speckle pattern, coupled with the influence of image noise, the coarse-fine search method adopted by the former is easy to fall into the wrong local optimal solution, while the latter The local exhaustive-adaptive search algorithm based on the continuous deformation property of the image overcomes this situation well. When the mid-span measurement position is close to the bottom of the beam, the deflection identified by the AD-DIC method produces a small error. This is because the image recognition will be affected by the background image when the boundary of the calculated pixel point subset touches the bottom edge of the wooden beam or exceeds the bottom edge of the wooden beam. Therefore, the recognition results may fluctuate. Therefore, when the DIC method is actually used to measure the displacement, a certain distance should be left between the measurement area and the boundary of the component.

Embodiment two

Compared with Embodiment 1, in this embodiment, the displacement threshold includes multiple sets of matching painted beam types and critical thresholds, and also includes an input step, inputting the painted beam type, and a matching step, matching the inputted painted beam type as The critical threshold for the current painted beam displacement threshold.

Considering that for different painted beams, the critical point of deformation is different, that is to say, the displacement threshold is different, so the displacement threshold in this scheme includes the type of painted beam and the matching critical threshold, and an input module is also set up to facilitate the staff to input painted Beam type, the matching module automatically matches the critical threshold of the current painted beam for calculation, thus ensuring the accuracy of the reminder.

What is described above is only an embodiment of the present invention, and the common knowledge such as the specific structure and characteristics known in the scheme is not described too much here, and those of ordinary skill in the art know all the common knowledge in the technical field to which the invention belongs before the filing date or the priority date Technical knowledge, being able to know all the existing technologies in this field, and having the ability to apply conventional experimental methods before this date, those of ordinary skill in the art can improve and implement this plan based on their own abilities under the inspiration given by this application, Some typical known structures or known methods should not be obstacles for those of ordinary skill in the art to implement the present application. It should be pointed out that for those skilled in the art, under the premise of not departing from the structure of the present invention, several modifications and improvements can also be made, and these should also be regarded as the protection scope of the present invention, and these will not affect the implementation of the present invention. Effects and utility of patents. The scope of protection required by this application shall be based on the content of the claims, and the specific implementation methods and other records in the specification may be used to interpret the content of the claims.

Claims (10)

1. The double-precision displacement measurement method based on the self-adaptive search comprises the following steps:

an image acquisition step: acquiring an image of a measured object and generating image information;

a storage step: storing the calculation rule;

a calculation step: calculating the image information according to a calculation rule to obtain a displacement;

the method is characterized in that: the calculation rules comprise a parameter algorithm, a coefficient threshold algorithm, a size algorithm, a local exhaustive search method, a gradient method and a weight operation method;

the calculating step includes:

and (3) parameter calculation: calculating two image information of a measured object before and after a period of time according to a parameter algorithm to obtain a parameter value;

coefficient threshold calculation step: calculating the parameter value according to a coefficient threshold algorithm to obtain a coefficient threshold;

a subset selection step: calculating the size of the subset meeting the coefficient threshold according to a size algorithm;

and searching for the integral pixel displacement: the method comprises the steps of obtaining an initial value of a whole pixel according to the local exhaustive search of an initial calculation subset;

searching the initial value of the sub-pixel displacement: the method comprises the steps of determining a sub-pixel region subset according to an integer pixel initial value and a subset selection method, and respectively calculating sub-pixel displacement in the sub-pixel region subset for integer pixel initial value points at two ends of the sub-pixel region subset according to a gradient method;

and calculating a displacement accurate value: and performing weight calculation on the sub-pixel displacement to obtain a displacement accurate value.

2. The adaptive search based dual-precision displacement measurement method according to claim 1, wherein: also comprises a timing step: sending a starting signal according to the stored timing information;

the control steps are as follows: and after receiving a starting signal, controlling the image acquisition step to start.

3. The adaptive search based dual-precision displacement measurement method according to claim 1, wherein: the method also comprises the following alarming steps: and the calculation step is also used for calculating the obtained displacement accurate value and the displacement threshold value, and sending alarm information when the calculated displacement accurate value is greater than or equal to the displacement threshold value.

4. The dual-precision displacement measurement method based on adaptive search according to claim 3, wherein: also comprises the input step: inputting a displacement threshold value and storing the displacement threshold value.

5. The dual-precision displacement measurement method based on adaptive search according to claim 3, wherein: the displacement threshold comprises a plurality of groups of mutually matched colored drawing beam types and critical thresholds, and further comprises the following input steps: inputting the type of the colored drawing beam; matching: and matching a critical threshold value serving as a current displacement threshold value of the colored drawing beam according to the type of the input colored drawing beam.

6. The dual-precision displacement measurement method based on adaptive search according to claim 1, wherein: and a CCD camera is adopted in the image acquisition step.

7. The adaptive search based dual-precision displacement measurement method according to claim 6, wherein: in the image acquisition step, the optical axis of the CCD camera is opposite to the centroid of the measured object and is perpendicular to the pattern surface of the measured object.

8. The dual-precision displacement measurement method based on adaptive search according to claim 1, wherein: also comprises the following processing steps: and performing edge removal processing on the image information, wherein the image information subjected to the edge removal processing is calculated in the calculating step.

9. The dual-precision displacement measurement method based on adaptive search according to claim 8, wherein: in the processing step, edge removing processing is performed on the left edge and the right edge of the image information.

10. The dual-precision displacement measurement method based on adaptive search according to claim 9, wherein: the method also comprises the following input steps: inputting an edge processing size, and performing edge removing processing on the image information according to the edge processing size in the processing step.

Images