CN110285827B - Distance-constrained photogrammetry high-precision target positioning method - Google Patents

Abstract

The invention discloses a distance-constrained photogrammetric high-precision target positioning method. The method comprises the following steps: step 1, camera internal parameter calibration; step 2, shooting a target image containing an indium steel ruler; step 3, target point image coordinates Measurement; step 4, using the least squares method to calculate the three-dimensional coordinates of the target point, as the initial value of the physical coordinates of the target point; step 5, obtaining the initial value of the orientation element outside the image; step 6, calculating the network adjustment of the distance constraint; step 7 , Adjustment iterative convergence judgment, judge the correction number calculated each time, until the tolerance requirement is met; obtain the high-precision coordinates of the target point. The method of the present invention does not need to separately arrange control points, and can obtain millimeter-level target point accuracy. The method is simple and fast, reduces working time, has fast measuring speed, and greatly improves measuring work efficiency.

Description

A distance-constrained high-precision target positioning method for photogrammetry

Technical Field

The invention relates to the technical field of photogrammetry and computer vision geometric positioning, and in particular to a distance-constrained photogrammetry high-precision target positioning method.

Background Art

Photogrammetry target positioning is to use a camera to obtain the image of the measured target, and then obtain the geometric and position information of the photographed object after processing. It has the advantages of non-contact, fast speed, simultaneous acquisition of many observation targets, and high accuracy. It has expanded from the field of aerial photogrammetry to many fields such as archaeology (ancient cultural relics, ancient buildings, etc.), biomedicine, and industrial measurement. However, its main defect is that the measurement target constructed only by images is relative in terms of size, mutual distance and other geometric information, that is, the three-dimensional model constructed by images is similar to the actual object, and there is a scale. In order to obtain the absolute geometric information of the measured target, it is necessary to rely on external control points. The general practice is to use other measurement tools (total station, laser scanner, laser tracker, etc.) to observe the geometric information or coordinate values of a small number of measured targets, use it as an external control, and scale the constructed three-dimensional model to a certain extent, so as to obtain the absolute geometric information of the observed target.

In order to obtain high-precision control information, it is often necessary to establish a three-dimensional control field in front of the target to be measured. For example, when using the total station intersection measurement to establish a three-dimensional control field, first, 2-4 stable forced centering bases should be set up in front of the target to be measured for the placement of the total station. The forced centering base can be welded with steel bars or cast on-site with cement.

After placing the instrument on the forced centering base, firstly perform relative orientation between the two total stations to determine the starting direction of the angle observation. The operation steps of the rotating instrument method are:

a) Fix pins on the two total stations A and B that are to observe each other, strictly level the instruments, and use the coarse aiming device to aim at each other's instruments;

b) Set total station A to face left observation;

c) Sight the pin on instrument B and record the horizontal direction H1, and set the horizontal direction of instrument B to zero;

d) Rotate total station B 180°, aim at the pin on instrument B from total station A and record the horizontal direction H2;

e) Rotate total station B 180° and switch to face-right observation;

f) Repeat steps (c) and (d) to complete a round of measurement, and take the average of the observed values as the starting direction AB;

The same steps can be followed to determine the starting direction of BA.

After determining the starting direction, the standard ruler method is used to measure a known length on the indium steel ruler to reversely calculate the precise distance between measuring stations AB, and then the coordinates of the control points are calculated by angle observation intersection.

This method is cumbersome and labor-intensive, which is seriously inconsistent with the requirement of photogrammetry to quickly obtain a large number of measured targets. Moreover, it is simply impossible to implement it when the measurement accuracy is high, the measurement time is tight, and the measurement space is limited.

Summary of the invention

The technical problem to be solved by the present invention is to provide a distance-constrained photogrammetry high-precision target positioning method in view of the defects in the prior art.

The technical solution adopted by the present invention to solve the technical problem is:

The present invention provides a distance-constrained photogrammetry high-precision target positioning method, the method comprising the following steps:

Step 1: Calibrate the camera parameters.

Step 2, photographing a target image including an indium steel ruler;

Step 3, target point image coordinate measurement: import the captured target image into the measurement software, obtain the coordinates of the target to be measured in the image coordinate system, take one corner of the image as the origin, and simultaneously measure two points with a certain distance on the indium steel ruler as image points;

Step 4: Calculate the three-dimensional coordinates of the target point using the least squares method as the initial value of the physical coordinates of the target point;

Step 5, obtaining the initial values of the exterior orientation elements of the image: after obtaining the three-dimensional coordinates of the target point, make them correspond to the image points, and then calculate the exterior orientation elements of the image, namely, three position parameters and three attitude parameters, according to the single-chip resection method;

Step 6, calculate the network adjustment of distance constraints: the measured image point coordinates are numbered in a certain order so that they correspond to the physical coordinates one by one; then linearized according to the photogrammetry collinearity condition model, and the error equation of adjustment is established; when establishing the error equation, for the two points on the indium steel ruler, the distance constraint is introduced as the adjustment constraint condition;

Step 7, iterative adjustment convergence judgment: After the error equation is established for each image point, the least squares indirect adjustment method with constraints is used to solve the correction number, and the initial value of the physical coordinate is corrected. Repeat steps 6 and 7 to judge the correction number calculated each time until the limit error requirement is met; obtain the high-precision coordinates of the target point.

Furthermore, the method for calibrating the camera internal parameters in step 1 of the present invention is:

For the selected camera, adjust the shooting depth of field in manual mode and then fix it; select a high-precision indoor 3D calibration field to ensure that the camera captures images from the left, center, and right directions of the front of the calibration field, extract the image coordinates of the landmark points on the image, and calculate the camera's internal parameters.

Furthermore, the method for photographing the target image containing the indium steel ruler in step 2 of the present invention is specifically as follows:

Capture images of the target to be measured. Place an indium steel ruler at the specified position of the target to ensure that the captured image contains both the target to be measured and the indium steel ruler. Capture images from five different directions: front, top, bottom, left, and right to ensure that the measured target overlaps in different images.

Furthermore, the method for obtaining the initial value of the physical coordinates of the target point in step 4 of the present invention is:

The initial value of the physical coordinates of the target point is calculated using the motion structure algorithm to restore the three-dimensional geometric shape of the object from multiple images; the basic matrix is solved by least squares using the measured image points of the same name with a constant factor difference, and the essential matrix is obtained by SVD decomposition, the camera motion parameters are calculated by singular value decomposition of the essential matrix, and finally the three-dimensional coordinates of the target point are calculated.

Furthermore, the method for obtaining the initial value of the image exterior orientation element in step 5 of the present invention is:

The exterior orientation element values of the photo are calculated from the three-dimensional coordinates of the target point and the image coordinates corresponding to each image using the single image space resection method based on the collinearity condition; due to the lack of a constant factor when solving the basic matrix, there is a certain ratio between the exterior orientation line elements and the actual values. The ratio value is obtained by the ratio of the actual length on the indium steel ruler to the initial value, and the exterior orientation line elements are corrected.

Further, the method for calculating the distance-constrained network adjustment in step 6 of the present invention is:

When the initial values of the target points and the initial values of the exterior orientation elements are determined, the error equation is established based on the collinearity condition equation of photogrammetry, and finally solved by the least squares adjustment method;

The collinearity condition equation based on photogrammetry is:

The distortion amount is expressed as:

In the formula, (x, y) is the image coordinate; (X, Y, Z) is the object coordinate corresponding to the image point; ( _XS , _YS , _ZS ) is the camera exterior orientation line element; ( _Rij , i, j = 1, 2, 3) is the rotation matrix composed of the camera exterior orientation angle elements; (f, _x0 , _y0 ) is the camera interior orientation element; (k1, _2, 3, p1, ₂ ) is the camera distortion parameter; ^r2 = ^x2 + ^y2 .

The beneficial effects of the present invention are as follows: the distance-constrained photogrammetry high-precision target positioning method of the present invention is based on the principle of photogrammetry, uses a camera to shoot images containing a high-precision indium steel ruler and an observed target, and obtains high-precision coordinates of the target through distance-constrained network adjustment calculation. There is no need to arrange control points separately, and millimeter-level target point accuracy can be obtained. The method is simple and fast, reduces operation time, has a fast measurement speed, and greatly improves measurement operation efficiency.

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:

FIG1 is a flow chart of a method according to an embodiment of the present invention.

DETAILED DESCRIPTION

In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

As shown in FIG1 , the distance-constrained photogrammetry high-precision target positioning method according to an embodiment of the present invention comprises the following steps:

1. Camera internal parameter calibration;

For the selected camera, adjust the depth of field in manual mode and then fix it. The calibration of the intrinsic parameters is calculated by shooting a high-precision three-dimensional control field in the room.

2. Select an indium steel ruler;

According to the actual situation of the shooting scene, select a certain length of indium steel ruler, generally 1 meter, 2 meters or 3 meters long. The selected indium steel ruler must pass the national testing requirements, that is, the average value of the interval length of the indium steel ruler and the error of each decimeter scale have the following requirements: the difference between the average value of the interval length of the indium steel ruler and the nominal value shall not exceed ±0.1mm for one ruler and ±0.05mm for a pair of rulers; the standard deviation of the scale of a row of scales shall not exceed ±13um.

3. Target point image shooting;

Place the indium steel ruler in front of the target point to be measured, and use the selected camera to shoot images containing the indium steel ruler and the target to be measured. It is required to shoot from different positions, and there should be more than 60% overlap between adjacent images. The acquired image should have uniform lighting and clear targets.

4. Target point measurement;

Accurately measuring the target point on the image is the key to calculating its three-dimensional coordinates, which is called feature point extraction in image processing. For multiple images, it is also necessary to match the feature points and automatically establish the correspondence between the image points of the same spatial point in different images. The corresponding points of the same spatial point in different images are called the same-name image points. When measuring, all the same-name points on the image should be measured.

5. Calculate the initial value of physical coordinates;

The initial value of the physical coordinates of the target point is the key to the linearization of the adjustment model. The structure-from-motion algorithm is used to restore the three-dimensional geometric shape of the object from multiple images. The least squares method is used to solve the basic matrix of the measured image points with the same name, and the SVD decomposition is performed to obtain the essential matrix. The singular value decomposition of the essence is performed to estimate the camera motion parameters, and finally the three-dimensional coordinates of the target point are calculated. The specific calculation process is as follows:

量测的同名像点坐标为m＝[u v 1]^T，m′＝[u′ v′ 1]^T，则满足：Assume that the basic matrix is

The measured coordinates of the image points with the same name are m = [uv 1] ^T , m′ = [u′ v′ 1] ^T , which satisfies:

m′Fm＝0 (1)

When more than 8 pairs of points with the same name are measured, the least square method can be used to solve equation (1) to obtain the basic matrix F.

结合基础矩阵F可以计算出本质矩阵E为：The camera internal matrix is formed by the camera calibration parameters

Combined with the basic matrix F, the essential matrix E can be calculated as:

E＝K′FK (2)

Perform singular value decomposition on the essential matrix and get:

E＝UDV′ (3)

The camera motion parameters R and t are calculated using the decomposition results:

R＝UAN′,t＝(0 0 1)′ (4)

in

The projection matrix of cameras at different positions is formed by the camera motion parameters and the camera internal parameter matrix:

If the homogeneous coordinates of the object space points corresponding to the same-name image points m and m′ are (X _w Y _w Z _w 1), then:

The three-dimensional coordinates of the target point can be solved according to formula (6).

6. Calculate the initial value of the image exterior orientation element;

The exterior orientation element values of the photo are calculated by using the single image space resection method based on the collinearity condition from the three-dimensional coordinates of the target point and the image coordinates corresponding to each image. Since the basic matrix is solved under the condition of a constant factor difference, there is a ratio between the exterior orientation line elements and the actual values. The ratio value is obtained by the ratio of the actual length on the indium steel ruler to the initial value, and the exterior orientation line elements are corrected.

7. Free network adjustment calculation;

When the initial values of the target points and the initial values of the exterior orientation elements are determined, the error equation is established based on the collinearity condition equation of photogrammetry, as shown in formula (7), and finally solved by the least squares adjustment method.

The distortion amount is expressed as:

In the formula, (x, y) is the image coordinate; (X, Y, Z) is the object coordinate corresponding to the image point; ( _XS , _YS , _ZS ) is the camera exterior orientation line element; ( _Rij , i, j = 1, 2, 3) is the rotation matrix composed of the camera exterior orientation angle elements; (f, _x0 , _y0 ) is the camera interior orientation element; (k1, _2, 3, p1, ₂ ) is the camera distortion parameter; ^r2 = ^x2 + ^y2 .

Iterative adjustment convergence judgment: After the error equation is established for each image point, the least squares indirect adjustment method with constraints is used to solve the correction number and correct the initial value of the physical coordinates.

Given a correction threshold (according to the accuracy requirements, the threshold range can be set to 0.1mm), the size of the correction value is determined. If the correction value is less than the given threshold, the calculation is stopped to obtain the high-precision coordinates of the target point.

If the correction value is greater than the threshold, the three-dimensional coordinates of the target point are recalculated according to step 6 based on the adjusted camera motion parameters, and step 7 is repeated to perform the adjustment calculation with constraints, and the calculated correction value is judged again until it meets the limit less than the given threshold.

8. Result Verification

In order to verify the correctness and feasibility of the method of the present invention, a digital camera is used to conduct a positioning test on a set of target points set indoors. The camera parameters used in the test are shown in Table 1. The selected indium steel ruler is 2m long and has been verified by the national designated quality inspection office.

Table 1 Camera parameters used in the experiment

An indium steel ruler is placed in front of the measurement target, and the above-mentioned camera is used to shoot images of the target. A total of 9 images are taken from different positions. The image coordinates of each target point on multiple images are obtained using a self-written image measurement program, and the three-dimensional coordinates of the target point are calculated according to the process described in the present invention. The actual three-dimensional spatial coordinates of the target point are obtained by observing the high-precision total station, but because it is placed arbitrarily during the experiment, its three-dimensional coordinate value has changed, but the distance between the target points has not changed. In order to evaluate the accuracy of the method of the present invention in determining the target point, the distance between the points calculated by the present invention is compared with the actual distance, and the results are shown in Table 2.

Table 2 Accuracy evaluation of the method of the present invention

It can be seen from Table 2 that the distance between points determined by the method of the present invention is very close to the actual observation result, and the difference between the two is in the millimeter level. From the statistical results, it can be seen that the maximum distance difference is 1.7mm, and the root mean square of the distance difference is 0.9mm. The results are consistent with the observation results of the traditional high-precision total station. This shows that the target point determination method described in the present invention is correct and feasible, and can simplify the traditional operation method.

It should be understood that those skilled in the art can make improvements or changes based on the above description, and all these improvements and changes should fall within the scope of protection of the appended claims of the present invention.

Claims (6)

1. A distance-constrained photogrammetric high-precision target positioning method, characterized in that the method comprises the following steps: Step 1: Calibrate the camera parameters. Step 2, photographing a target image including an indium steel ruler; Step 3, target point image coordinate measurement: import the captured target image into the measurement software, obtain the coordinates of the target to be measured in the image coordinate system, take one corner of the image as the origin, and simultaneously measure two points with a certain distance on the indium steel ruler as image points; Step 4: Calculate the three-dimensional coordinates of the target point using the least squares method as the initial value of the physical coordinates of the target point; Step 5, obtaining the initial values of the exterior orientation elements of the image: after obtaining the three-dimensional coordinates of the target point, make them correspond to the image points, and then calculate the exterior orientation elements of the image, namely, three position parameters and three attitude parameters, according to the single-chip resection method; Step 6, calculate the network adjustment of distance constraints: the measured image point coordinates are numbered in a certain order so that they correspond to the physical coordinates one by one; then linearized according to the photogrammetry collinearity condition model, and the error equation of adjustment is established; when establishing the error equation, for the two points on the indium steel ruler, the distance constraint is introduced as the adjustment constraint condition; Step 7, iterative adjustment convergence judgment: After the error equation is established for each image point, the least squares indirect adjustment method with constraints is used to solve the correction number, and the initial value of the physical coordinate is corrected. Repeat steps 6 and 7 to judge the correction number calculated each time until the limit error requirement is met; obtain the high-precision coordinates of the target point. 2. The distance-constrained photogrammetry high-precision target positioning method according to claim 1, characterized in that the method for calibrating the camera internal parameters in step 1 is: For the selected camera, adjust the shooting depth of field in manual mode and then fix it; select a high-precision indoor 3D calibration field to ensure that the camera captures images from the left, center, and right directions of the front of the calibration field, extract the image coordinates of the landmark points on the image, and calculate the camera's internal parameters. 3. The distance-constrained photogrammetric high-precision target positioning method according to claim 1, characterized in that the method of capturing the target image containing the indium steel ruler in step 2 is specifically: Capture images of the target to be measured. Place an indium steel ruler at the specified position of the target to ensure that the captured image contains both the target to be measured and the indium steel ruler. Capture images from five different directions: front, top, bottom, left, and right to ensure that the measured target overlaps in different images. 4. The distance-constrained photogrammetric high-precision target positioning method according to claim 1, characterized in that the method for obtaining the initial value of the physical coordinates of the target point in step 4 is: The initial value of the physical coordinates of the target point is calculated using the motion structure algorithm to restore the three-dimensional geometric shape of the object from multiple images; the basic matrix is solved by least squares using the measured image points of the same name with a constant factor difference, and the essential matrix is obtained by SVD decomposition, the camera motion parameters are calculated by singular value decomposition of the essential matrix, and finally the three-dimensional coordinates of the target point are calculated. 5. The distance-constrained photogrammetric high-precision target positioning method according to claim 4, characterized in that the method for obtaining the initial value of the image exterior orientation element in step 5 is: The exterior orientation element values of the photo are calculated from the three-dimensional coordinates of the target point and the image coordinates corresponding to each image using the single image space resection method based on the collinearity condition; due to the lack of a constant factor when solving the basic matrix, there is a certain ratio between the exterior orientation line elements and the actual values. The ratio value is obtained by the ratio of the actual length on the indium steel ruler to the initial value, and the exterior orientation line elements are corrected. 6. The distance-constrained photogrammetric high-precision target positioning method according to claim 1, wherein the method for calculating the distance-constrained network adjustment in step 6 is: When the initial values of the target points and the initial values of the exterior orientation elements are determined, the error equation is established based on the collinearity condition equation of photogrammetry, and finally solved by the least squares adjustment method; The collinearity condition equation based on photogrammetry is:

The distortion amount is expressed as:

In the formula, (x, y) is the image coordinate; (X, Y, Z) is the object coordinate corresponding to the image point; ( _XS , _YS , _ZS ) is the camera exterior orientation line element; ( _Rij , i, j = 1, 2, 3) is the rotation matrix composed of the camera exterior orientation angle elements; (f, _x0 , _y0 ) is the camera interior orientation element; ( _k1,2 , _p1,2 ) is the camera distortion parameter; ^r2 = ^x2 + ^y2 .

Images