CN108986070B - Rock crack propagation experiment monitoring method based on high-speed video measurement - Google Patents

Abstract

The invention relates to an experimental monitoring method for rock fracture expansion based on high-speed video measurement, comprising the following steps: 1) obtaining sequence images of the fracture process of the rock to be measured under uniaxial compression through two high-speed cameras in a binocular vision measurement system 2) In the process of sequence image analysis, the sub-pixel tracking and matching method is used to obtain the coordinates of the image point with the same name in the speckle region of interest, and the time series three-dimensional image of the rock surface to be tested under uniaxial compression is obtained through the photogrammetry analysis algorithm. Point cloud data; 3) For the time-space sequence analysis of the sequence 3D point cloud, obtain the 3D deformation parameters of the rock fracture to be measured, including parameters such as displacement, velocity, acceleration and strain field. Compared with the prior art, the present invention has the advantages of being feasible, effective, flexible and reliable.

Description

An experimental monitoring method of rock fracture propagation based on high-speed video measurement

technical field

The invention relates to a non-contact high-speed video measurement scheme for a rock crack expansion experiment, in particular to a rock crack expansion experiment monitoring method based on high-speed video measurement.

Background technique

In the engineering world, before a material is put into use, its material properties and safety factor often need to be monitored through testing experiments such as tension, compression, and collision. In the measurement of crack propagation in rocks, traditional contact measuring instruments have been gradually replaced by non-contact measuring methods due to many shortcomings such as limited range, few measuring points, increased model quality, and time-consuming and laborious installation. In the research work of optical measurement, digital correlation algorithm has become the mainstream solution method of mechanical calculation. Many scholars have begun to use digital speckle correlation technology to solve displacement field and strain field. However, in most of the experiments, the displacement change of the two-dimensional plane is more measured, and the measurement error caused by the non-parallel between the photosensitive plane and the measured object surface will seriously affect the image measurement results, which is almost uncontrollable. In addition, 3D digital correlation technology has also been widely studied in recent years, but many similar experiments are taken with ordinary cameras or low-performance high-speed cameras, which cannot record the mutation of the measured object in detail, and thus cannot The complete spatial three-dimensional information change of the measurement point is provided, which affects the measurement effect of the experiment.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide an experimental monitoring method for rock crack propagation based on high-speed video measurement in order to overcome the above-mentioned defects of the prior art.

The object of the present invention can be realized through the following technical solutions:

An experimental monitoring method for rock fracture propagation based on high-speed video measurement, comprising the following steps:

1) Obtain and store the sequence images of the fracture process of the rock to be measured under uniaxial compression through two high-speed cameras in the binocular vision measurement system;

2) In the process of sequence image analysis, the sub-pixel-level tracking and matching method is used to calculate the coordinates of the image points with the same name in the speckle region of interest, and the time-series 3D point cloud of the rock surface to be tested under uniaxial compression is obtained through the photogrammetry analysis algorithm. data;

3) For the time-space sequence analysis of the sequence 3D point cloud, obtain the 3D deformation parameters of the rock fracture to be measured, including displacement, velocity, acceleration and strain field.

In the step 1), the high-speed camera in the binocular vision measurement system is an industrial camera, and the synchronization of image acquisition is maintained by a synchronization controller.

The two high-speed cameras described above are placed horizontally and photographed in a cross-direction photography mode, so as to achieve the purpose of increasing the overlapping coverage of the images.

In the described step 1), the speckle area is sprayed on the rock to be tested, specifically:

The surface of the rock to be tested is polished to be flat, and the observation surface is sprayed with white matte paint and air-dried. The observation surface is randomly and evenly sprayed with black matte paint or black ink to form speckles, and the entire speckle area is used as speckle Region of interest, with speckles as target points.

Described step 2) specifically comprises the following steps:

21) Stereo calibration of the high-speed camera: Simultaneously obtain the inner and outer orientation elements of the high-speed camera through the calibration method based on the plane plate;

22) Speckle image preprocessing: select the speckle area of interest in the sequence image and determine the target point in the area of interest;

23) Point matching with the same name: The images collected by two high-speed cameras at the same time are used as matching objects, and the rough point position of the integer pixel level is determined by the normalized correlation coefficient, and the precise point position of the subpixel level is determined by the least squares matching method;

24) Target tracking matching: take the front and rear frame images of the sequence images collected by the high-speed camera as the matching objects, and obtain the time-series two-dimensional image point coordinates of the target points in the sequence images;

25) 3D point cloud reconstruction: According to the image point coordinates of each pair of points with the same name after matching and the calibrated inner and outer orientation elements, the time series 3D points of the target points with the same name in the sequence image are obtained through the forward intersection based on the collinear equation. Bit coordinates.

In the described step 23), the least squares matching method takes the maximum normalized correlation coefficient as the objective function, considers the affine distortion model between the left and right images, and uses the grayscale information and position information in the window to perform adjustment processing, and finally Get exact match points.

In the step 3), the displacement data is obtained by the difference of the time series three-dimensional point coordinates, and the displacement data is subjected to primary differentiation and secondary differentiation in time to obtain velocity data and acceleration data respectively.

Compared with the prior art, the present invention has the following advantages:

By combining the digital speckle matching method with high-speed video measurement, the invention can obtain the three-dimensional shape change of the rock to be measured in a small time interval at all times, and further proposes a complete set of non-contact three-dimensional deformation measurement scheme, which can be used for rock measurement. Quantitative study and analysis of crack propagation provides various deformation parameters.

Description of drawings

FIG. 1 is a technical roadmap of the present invention.

Figure 2 is a schematic diagram of target point sampling.

Figure 3 is a schematic diagram of the same name matching.

FIG. 4 is a schematic diagram of tracking and matching of target points in sequence images.

Figure 5 shows the layout of the high-speed camera network

Figure 6 is a statistical histogram of normalized correlation coefficients.

Figure 7 shows the 3D reconstruction of the rock surface, in which Figure (7a) is the 3D point cloud distribution at the initial moment, and Figure (7b) is the 3D point cloud distribution at the 8.285s.

Figure 8 shows the three-dimensional displacement field of the rock sample at 8.285s, in which Figure (8a) is the displacement in the X direction, Figure (8b) is the displacement in the Y direction, and Figure (8c) is the displacement in the Z direction.

Figure 9 shows the strain field of the rock sample at 8.285s, in which Figure (9a) is the Exx strain, Figure (9b) is the Eyy strain, and Figure (9c) is the Exy strain.

Detailed ways

The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

Example

This patent formulates a detailed high-speed video measurement method for rock fracture propagation experiments. In this embodiment, two high-speed cameras are used for stereoscopic observation, so the hardware device and related measurement theory can be collectively referred to as a binocular vision measurement system. The construction of the system requires hardware such as high-speed cameras, synchronization controllers, and high-speed data acquisition cards. Among them, the high-speed camera is an important part of the system. It belongs to a kind of industrial camera. It has unique advantages such as high stability, high frame rate, high transmission ability and high anti-interference ability, especially suitable for automatic optical inspection, three-dimensional measurement , semiconductor inspection, machine vision and other fields. In addition, the role of the synchronization controller is to maintain the synchronization of the cameras in the joint test. In the subsequent data processing, when solving the three-dimensional space coordinates of a certain target point at a certain moment, it is necessary to obtain an image of the same name shot at that moment. Without synchronous control of the two cameras, post-processing of 3D data cannot be performed, so synchronous control is crucial.

The network construction of the binocular vision measurement system also needs to be designed according to the experimental environment. In order to increase the image overlap coverage, the two high-speed cameras should use cross-direction photography. In front of the rock sample block, the two high-speed cameras should be placed horizontally as much as possible, but the relative posture on the horizontal plane should not be too inclined. This is because if the tilt posture is too large, the relative perspective distortion of the left and right two sets of images will increase accordingly, which will reduce the matching accuracy of sub-pixel images in the later stage. Therefore, before the experiment is carried out, the positions and attitudes of the two cameras can be adjusted according to the field of view of the captured images.

The technical route of the present invention is shown in FIG. 1 .

The present invention specifically includes the following steps:

1. Target matching and tracking

1.1 Image preprocessing

Before image matching, it is necessary to extract the region of interest and determine the target point location. The target point selection process is similar to the sampling process, and the sampling interval can be formulated according to the needs of the experiment. For example, taking the first image (ie, the first frame of image) captured by the left camera as the reference image, the region of interest is selected manually first, and then the target point is selected through a certain length interval (sampling interval). The smaller the sampling interval, the greater the number of target points, as shown in Figure 2.

1.2 Same-name point matching

This experiment requires sub-pixel high-precision matching results, so this patent uses the most conventional matching strategy from coarse to fine. Coarse matching determines the coarse point position of integer pixel matching by calculating the normalized correlation coefficient, and fine matching determines the precise point position of sub-pixel through the least squares matching method. Among them, the least squares matching method aims at the maximum normalized correlation coefficient and regards the left and right image deformation as affine transformation, and uses the grayscale information and position information in the window for adjustment processing, so that it can reach 1/10 or even 1/1/10. Matching accuracy of 100 pixels. In order to speed up the matching speed, the local search area needs to be determined. Since the relative tilt attitude of the two cameras in this experiment is small, the range of the region of interest in the left and right images is not much different. As shown in FIG. 3 , the positional relationship of a certain target point in the left image in the left area of interest can be determined first, and then the approximate range of the point with the same name in the right area of interest can be calculated to determine the image matching search area. In addition, the position of each target point has been determined, so the window size of the target image block and the window size of the search area can be set according to the requirements of the experiment.

1.3 Target tracking matching

The purpose of target tracking and matching is to obtain the image coordinates of each target point sequence, and its sub-pixel matching method is similar to that of the same name point matching. The difference is that the matching object is no longer the left and right images, but the sequence images stored by each camera. Since the matching process of the same name point has provided target image blocks, these image blocks should also be used as target images in the target tracking matching, and the search area of the next frame can be determined by the target position of the previous frame. The schematic diagram of the tracking matching is shown in the figure 4.

Through the above process, sequence image coordinates can be obtained for each target point. Since the captured images have been collected and stored synchronously by the two high-speed cameras, the point with the same name obtained on the first frame of image still maintains the same name in the time series. relation. After 3D reconstruction, the rock surface can form 3D point cloud data at each time scale.

2. 3D reconstruction of binocular vision

2.1 Stereo calibration

This patent uses the calibration method based on the plane plate to obtain the inner and outer orientation elements of the camera synchronously. The inner orientation element not only needs to consider the image principal point (C _x , C _y ) and the image distance (f), but also consider the lens The distortion parameter and the actual size of the pixel (S _x , S _y ). The distortion parameters of the lens mainly include radial distortion (K ₁ , K ₂ , K ₃ ) and tangential distortion (P ₁ , P ₂ ). In addition, the stereo calibration not only needs to determine the internal orientation elements of each camera, but also determines the external orientation elements of the cameras. The external orientation elements (α, β, γ, t _x , _ty , t _z ) are mainly reflected in the conversion relationship between the camera coordinate system and the world coordinate system. Stereo calibration is the most critical step in this experiment, because the accuracy of its calibration orientation affects the final experimental results. Therefore, the detection points can be drawn on the plane calibration plate to further verify the calibration accuracy.

2.2 3D reconstruction

During the experiment, the high-speed camera that has been calibrated cannot move, otherwise it needs to be re-calibrated. In Section 2.1, the stereo calibration has determined the internal and external orientation elements of each camera. Therefore, in the sequence images collected by the two cameras, each time a pair of image point coordinates with the same name is obtained, it can be solved by the forward intersection based on the collinear equation. its three-dimensional point. Through the accumulation in space and time, a large amount of point cloud data is obtained. The formula of the collinear conditional equation for close-range photogrammetry is as follows:

Wherein, R=[a ₁ b ₁ c ₁ ; a ₂ b ₂ c ₂ ; a ₃ b ₃ c ₃ ].

Under the binocular vision measurement system, 4 equations can be listed for a pair of points with the same name, but 3 unknowns need to be solved, so the adjustment solution can be carried out according to the principle of least squares.

3. Deformation parameter solution

In the aforementioned data processing process, 3D point cloud data at any time can already be obtained. From this, various deformation parameters such as displacement, velocity, acceleration, and strain can be formed.

3.1 Displacement, Velocity, Acceleration

Displacement refers to the distance difference between the current position of the target point at a certain moment in the time series and the initial moment of the target point. From this, it can be known that the initial displacement of the tracking point is 0. For example, the displacement formula for 3D point displacement data is as follows:

和

分别表示目标点在X和Y方向在时刻n的位移；X₁，Y₁和Z₁分别表示目标点在X，Y和Z方向于初始时刻的坐标值；X_n，Y_n和Z_n分别表示目标点在X，Y和Z方向于时刻n的坐标值。in,

and

respectively represent the displacement of the target point in the X and Y directions at time n; X ₁ , Y ₁ and Z ₁ respectively represent the coordinate values of the target point in the X, Y and Z directions at the initial moment; X _n , Y _n and Z _n respectively Represents the coordinate value of the target point in the X, Y and Z directions at time n.

Since the acquisition frame rate of the high-speed camera is fixed, the time difference between two adjacent frames can be obtained. The corresponding velocity data and acceleration data can be obtained if the displacement data is differentiated first and second time with respect to time, respectively. Therefore, all the target points on the rock surface are processed above, and the displacement field, velocity field and acceleration field can be formed.

3.2 Strain value

In strain analysis, the strain value of a point can be calculated from the surrounding displacement data, so each target point can be taken as the center, and a displacement window can be selected around it to calculate the strain value. In this displacement window, its displacement distribution can be considered to be linear. The relationship between its displacement and coordinates can be expressed as:

where u(i,j) and v(i,j) are the displacement values of the point (i,j) under the displacement window, a _i=0,1,2 and b _i=0,1,2 are polynomial Undetermined coefficient.

In the case of small deformation, the strain component can be solved by:

The size of the displacement window can be determined according to requirements. If the window size is large, a high-order polynomial can also be used to represent the displacement distribution. But generally speaking, the size of the displacement window should be selected moderately according to the experimental requirements. From the formula (3) and the public announcement (4), the strain components can be solved by knowing the plane coordinates of at least three points. Under uniaxial compression, more attention is paid to the deformation that occurs within the rock surface. Therefore, in the 3D point cloud generated in the early stage, only the strain in the space plane can be considered.

The object of observation in this experiment is a rock sample block used for crack propagation measurement. The sample block is made by mixing medical gypsum and water in a certain proportion, and is processed into a cube with a side length of 70mm after mold processing. In order to meet the measurement requirements, it is necessary to spray speckle to increase the observation points. The specific process is as follows: (1) The observation surface of the sample block needs to be polished to be flat; (2) The observation surface is sprayed with white matte paint, and Air-dry; (3) randomly and evenly spray black matte paint or black ink on the observation surface to form a speckle image. In addition, as shown in Figure 5, two high-speed cameras form a cross-direction photogrammetry method, and a high-power halogen lamp is used to fill the light of the experimental object to ensure the shooting quality of the image.

In the experiment, two high-speed cameras formed binocular vision to measure three-dimensional rock sample blocks. The frame rate was set to 400 frames per second, which can accurately measure dynamic data with a frequency of up to 40 Hz, that is, every 10 times of image data to describe The morphological change of the sample once. The image size of the high-speed camera is 2304×1720 pixels, and the pixel size is 7um. For precise measurement, the high-speed camera is equipped with a 50mm fixed focal length lens.

Through calibration, the inner orientation elements and the relative outer orientation elements of the two high-speed cameras can be obtained. The results are shown in Table 1. Through the inspection of the detection points delineated on the flat panel, the calibration orientation accuracy can reach 0.01mm, and the back projection error is better than 0.2 pixels.

Table 1 Inner orientation elements and outer orientation elements of high-speed cameras

According to the above, image preprocessing needs to be performed before matching. This time, the sampling interval for selecting target points in the area of interest is 5 pixels, resulting in 17423 target observation points, and the rules are carried out according to the number of 131 columns and 133 rows. arrangement. Determine the target image block for all target points, where the size of the target window is set to 30 pixels, and the size of the search window with the same name is set to 50 pixels. After matching the points with the same name, the normalized correlation coefficient statistics chart can be obtained, as shown in Figure 6. It can be seen from the figure that there are 14,084 pairs of points with the same name whose correlation coefficients are in the range of 0.9 to 1.0, and 3,315 pairs of the same name. The correlation coefficient of points is in the range of 0.8 to 0.9. Since the value of the correlation coefficient above 0.8 is high image correlation, almost all target points can be matched to the point with the same name, and the matching accuracy is high. However, the value of some point correlation coefficients is relatively small, which is due to the low matching accuracy caused by less texture information of individual target image blocks. In this case, these bad points can be solved by numerical interpolation.

After all target points are tracked and matched, the coordinates of the image points with the same name in the sequence can be obtained, and then through 3D reconstruction, the 3D point cloud data of the rock surface at any time can be obtained. As shown in Figure 7, the three-dimensional deformation state of the rock surface can also be seen intuitively from the three-dimensional point cloud data.

The three-dimensional displacement field and strain field can be formed by calculating the deformation parameters, as shown in Figure 8 and Figure 9, respectively. During the compression process, cracks are formed at the top of the rock, which in turn damages the rock and causes relatively large deformation. Furthermore, it is evident from the strain field that the fracture trend of the rock will extend downwards.

This patent uses a 400 frame rate/second high-speed camera to measure the rock fracture propagation experiment, introduces the binocular vision 3D reconstruction method in detail, and proposes a speckle image matching strategy and matching method, and then proposes a complete set of 3D deformation measurement plan.

1) This scheme measures the dynamic changes of the displacement field and strain field of the experimental object during the fracture process, and the spatial positioning accuracy of the target point can reach up to 0.01mm, which verifies the feasibility of applying the whole set of measurement schemes in rock fracture experiments. and effectiveness.

2) In the speckle image matching process, the selection interval of the target point and the target image block window can be determined according to the experimental requirements, which increases the flexibility of data analysis.

3) High-speed video measurement technology can measure high-speed moving objects by virtue of its unique high frame rate characteristics, and a high-speed camera with a frame rate of 400 frames per second can accurately measure dynamic data with a frequency of up to 40Hz.

4) This patent also elaborates the solution process of deformation parameters, which provides reliable experimental data for further research and analysis.

Claims (4)

1. A rock crack propagation experiment monitoring method based on high-speed video measurement is characterized by comprising the following steps:

1) acquiring and storing sequence images of the fracture process of the rock to be measured under uniaxial compression by two high-speed cameras in a binocular vision measurement system;

2) in the sequence image analysis process, calculating sequence homonymous image point coordinates in a speckle interest area by a sub-pixel level tracking matching method, and acquiring time sequence three-dimensional point cloud data of the surface of the rock to be detected under uniaxial compression by a photogrammetric analysis algorithm, wherein the method specifically comprises the following steps:

21) three-dimensional calibration of a high-speed camera: simultaneously acquiring an inner orientation element and an outer orientation element of the high-speed camera by a calibration method based on a plane plate;

22) preprocessing a speckle image: selecting a speckle interest area from the sequence image and determining a target point in the interest area;

23) matching points with the same name: specifically, a rough point location of a whole pixel level is determined by calculating a normalized correlation coefficient, then an accurate point location of a sub-pixel level is determined by a least square matching method, wherein the maximum normalized correlation coefficient of the least square matching method is the maximum of a target function, an affine distortion model between a left image and a right image is considered, adjustment processing is carried out by utilizing gray information and position information in a window, and finally an accurate matching point location is obtained;

24) target tracking and matching: taking front and rear frame images of the sequence image collected by the high-speed camera as matching objects, and acquiring time sequence two-dimensional image point coordinates of a target point in the sequence image;

25) reconstructing three-dimensional point cloud: according to the matched image point coordinates of each pair of homonymous points and the calibrated inner orientation element and outer orientation element, acquiring time sequence three-dimensional point position coordinates of homonymous target points in the sequence images through forward intersection based on a collinear equation;

3) and aiming at the time-space sequence analysis of the sequence three-dimensional point cloud, acquiring three-dimensional deformation parameters including displacement, speed, acceleration and a strain field of the rock crack to be detected, acquiring displacement data by the difference of the time-sequence three-dimensional point location coordinates, and performing primary differentiation and secondary differentiation on the displacement data in time to respectively acquire speed data and acceleration data.

2. The method for monitoring the rock fracture propagation experiment based on the high-speed video measurement as claimed in claim 1, wherein in the step 1), the high-speed camera in the binocular vision measurement system is an industrial camera, and the image acquisition is kept synchronous through a synchronous controller.

3. The method for monitoring the rock fracture propagation experiment based on the high-speed video measurement as claimed in claim 2, wherein the two high-speed cameras are horizontally placed and photographed in an intersection manner, so as to achieve the purpose of increasing the image overlapping coverage rate.

4. The method for monitoring the rock crack propagation experiment based on the high-speed video measurement as claimed in claim 1, wherein in the step 1), a speckle region is sprayed on the rock to be measured, specifically:

the method comprises the steps of polishing the surface of a rock to be detected to be flat, spraying white matte paint on the observation surface, air-drying, randomly and uniformly spraying black matte paint or black ink on the observation surface to form speckles, taking the whole speckle area as a speckle interest area, and taking the speckles as target points.

Images