CN118603052A - A terrain survey device for civil engineering - Google Patents

Abstract

The present invention relates to the field of photogrammetry technology, and in particular to a topographic survey device for civil engineering, comprising: a camera, a lifting stepper motor, an upper pan head, a lower pan head, and a rotating stepper motor; a control error C1 is obtained from the measurement result of photogrammetry using the camera, and the upper pan head, the lower pan head, the lifting stepper motor, and the rotating stepper motor are controlled by the control error C1; the camera is used again for photogrammetry to obtain the measurement result t at the next moment and the error C2 of the measurement result t; the difference between the error C2 and the control error C1 is recorded as the photogrammetry error; the measurement result t and the historical measurement result are corrected by the photogrammetry error to obtain the measurement result at the next moment; the corrected measurement result and the measurement result at the next moment are used as the historical measurement result again, and the measurement result at the next moment after the next moment is obtained according to the historical measurement result. The present invention avoids the problem of inaccurate measurement when the survey equipment is fixed or installed unstably.

Description

A topographic survey device for civil engineering

Technical Field

The invention relates to the technical field of photogrammetry, and in particular to a terrain surveying device for civil engineering.

Background Art

In the process of topographic survey for civil engineering, it is often necessary to measure roads, bridges, slopes, etc. to obtain their shapes or settlement and deformation. For example, CN116608829A discloses a road and bridge settlement survey device for civil engineering and a method of using the same, which are used to realize civil engineering survey.

However, existing surveying equipment needs to be fixed on stable ground. However, due to the diversity of surveying environments, in some scenarios, it is necessary to conduct surveys on a ship or using a drone to carry out surveying. Neither the ship nor the drone can provide stable support and fixing conditions for the surveying equipment, which makes it difficult to obtain accurate surveying results.

Summary of the invention

In order to solve the above problems, the present invention provides a terrain surveying device for civil engineering.

An embodiment of the present invention provides a topographic survey device for civil engineering, comprising a camera and a lifting device, wherein the lifting device comprises a screw rod and a lifting stepping motor, and the lifting device further comprises an upper base plate, a lower base plate, and a loading plate, wherein an upper pan head and a lower pan head are respectively installed on the outer surfaces of the upper base plate and the lower base plate, the camera is fixed on a camera base, the camera base is fixed on a rotating shaft lead screw, one end of the rotating shaft lead screw is connected to a rotating stepping motor, and the other end is fixed to a bearing; the rotating stepping motor and the bearing are fixed below the loading plate, wherein the upper pan head, the lower pan head, the lifting stepping motor, and the rotating stepping motor are controlled by a processor installed in a control box;

The processor uses the camera to perform photogrammetry on the target object, obtains the measurement results of all positions on the target object at several moments, records them as historical measurement results, and obtains the measurement results of all positions on the target object at the next moment based on the historical measurement results, including:

The errors of historical measurement results at several moments are recorded as control errors. The control errors at several moments are input into the Kalman filter algorithm to obtain the control error C1 at the next moment after several moments. The processor uses the control error C1 to control the upper pan/tilt, the lower pan/tilt, the lifting stepper motor and the rotating stepper motor. When the control process is completed, the processor uses the camera to perform photogrammetry again and obtains the measurement results t of all positions on the target at the next moment and the error C2 of the measurement result t. The difference between the error C2 and the control error C1 is recorded as the photogrammetry error. The photogrammetry error is used to correct the measurement result t to obtain the measurement results of all positions on the target at the next moment.

The historical measurement results are corrected by using the photogrammetry error; the corrected historical measurement results and the measurement results of all positions on the target object at the next moment are used again as the historical measurement results, and the measurement results of the next moment after the next moment are obtained according to the historical measurement results.

Preferably, the outer support of the surveying device includes an upper fixing pile, a lower fixing pile, and a side fixing pile.

Preferably, the screw rod is hollow, and when the upper gimbal, camera, rotary stepping motor and processor are connected, the connected data line is inserted from the top of the screw rod and connected to the processor through the hollow interior of the screw rod.

Preferably, the measurement results include the settlement amount, deformation amount, distance from the camera and point cloud of each position of the target object, and the target objects include: slopes, bridges, and foundation piles.

Preferably, the specific steps of obtaining the control error are as follows:

Obtain the difference between the measurement result of each position of the target object and the measurement results of all neighboring positions of each position, record it as the difference feature of each position, record the average of the difference between the difference feature of each position at each moment and the difference feature of each position at the two adjacent moments as the target difference feature at each moment, reduce the dimension of the target difference feature of each position obtained at all moments in the historical measurement results to one dimension, record the one-dimensional dimension reduction results at all moments as the error vector of each position, cluster the error vectors of all positions into several categories, obtain the mean of all error vectors in each category, and record a vector composed of the means of all categories as the control error.

Preferably, the method of controlling the upper pan/tilt head, the lower pan/tilt head, the lifting stepping motor and the rotating stepping motor by using the control error C1 includes the following specific steps:

The control error C1 is input into the fully connected neural network, and the fully connected neural network outputs the control parameters of the upper pan-tilt head, the lower pan-tilt head, the lifting stepper motor and the rotating stepper motor. The upper pan-tilt head, the lower pan-tilt head, the lifting stepper motor and the rotating stepper motor rotate according to the control parameters.

Preferably, the method of using the photogrammetry error to correct the measurement result t to obtain the measurement results of all positions on the target object at the next moment includes the following specific steps:

The width of the Gaussian filter kernel is obtained according to the photogrammetry error, and the Gaussian filter kernel of the width is used to perform Gaussian filtering on the measurement results t of all positions on the target object at the next moment to obtain the measurement results of all positions on the target object at the next moment.

Preferably, the dimensionality reduction is performed using a PCA dimensionality reduction algorithm.

Preferably, the clustering uses a K-Means clustering algorithm.

Preferably, the step of obtaining the width of the Gaussian filter kernel according to the photogrammetry error comprises the following specific steps:

All elements in the photogrammetric error are linearly normalized, and the mean of the normalized elements is recorded as the noise level. The noise level is multiplied by the preset value N, and the result is recorded as P. An odd number greater than P and with the smallest difference from P is used as the width of the Gaussian filter kernel.

The beneficial effects of the technical solution of the present invention are:

The surveying device of the present invention can be installed on a ship or a drone or other fixed or unstable installation with a bumpy scene.

The control error C1 is obtained from the measurement results of the photogrammetry using the camera, and the control error C1 is used to control the upper pan/tilt head, the lower pan/tilt head, the lifting stepper motor and the rotating stepper motor. The camera is used again for photogrammetry and the measurement result t at the next moment is obtained. This process avoids interference caused by bumps caused by unstable fixation or installation to a certain extent.

The error C2 of the measurement result t is obtained, and the difference between the error C2 and the control error C1 is recorded as the photogrammetry error. The photogrammetry error is used to correct the measurement result t and the historical measurement results to obtain the measurement result at the next moment; this process reduces part of the photogrammetry error.

The corrected measurement result and the measurement result at the next moment are taken as the historical measurement result again, and the measurement result at the next moment after the next moment is obtained according to the historical measurement result. The present invention avoids the problem of inaccurate measurement when the survey equipment is fixed or installed unstably. This process ensures that the control error C1 and the photogrammetry error can be updated at subsequent moments.

In summary, on the one hand, although the errors of historical measurement results are recorded as control errors to describe the impact of unstable installation or fixation bumps, they must also include the errors in photogrammetry (i.e., photogrammetry errors). Therefore, it is inaccurate to control the upper pan/tilt, lower pan/tilt, lifting stepper motor, and rotating stepper motor based only on the control errors. However, the impact of bumps in the control errors is the main one, and the errors in photogrammetry are relatively minor. Therefore, even if the control errors are used directly for control, some of the impact of bumps can be eliminated. On the other hand, since the control errors can only eliminate some of the impact of bumps, the photogrammetry errors obtained above also include some impact of bumps. Therefore, the accuracy of the measurement results corrected using the photogrammetry errors is still not high. Based on these two aspects, the measurement result of the next moment after the next moment is obtained again on the corrected measurement result. This process involves the recalculation of the control error and the photogrammetry error, so that the control error and the photogrammetry error are updated with each other, making the control process of the upper pan-tilt head, lower pan-tilt head, lifting stepper motor and rotating stepper motor in the surveying device more accurate and reliable, thereby making the measurement result of the next moment after the next moment more accurate.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

FIG1 is a first isometric view of a topographic survey device for civil engineering provided by one embodiment of the present invention;

FIG2 is a second isometric view of a topographic survey device for civil engineering provided by one embodiment of the present invention;

FIG3 is a front view of a terrain survey device for civil engineering provided by one embodiment of the present invention;

FIG. 4 is a left side view of a terrain survey device for civil engineering provided by an embodiment of the present invention.

In the figure: 1. upper gimbal; 2. lower gimbal; 3. screw rod; 4. lifting stepping motor; 5. loading plate; 6. upper base plate; 7. lower base plate; 8. camera base; 9. camera; 10. rotating shaft screw; 11. bearing; 12. rotating stepping motor; 13. control box; 14. processor; 15. lifting device; 16. external bracket; 1601. bracket body; 1602. upper fixing pile; 1603. side fixing pile; 1604. lower fixing pile; 17. mounting frame.

DETAILED DESCRIPTION

In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of a topographic survey device for civil engineering proposed by the present invention, its specific implementation, structure, features and effects, in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

A topographic survey device for civil engineering provided by an embodiment of the present invention is described in detail below with reference to the accompanying drawings. As shown in FIG1, FIG2, FIG3, and FIG4, the device comprises: a camera 9, a lifting device 15, the lifting device 15 comprises a screw rod 3, a lifting stepping motor 4, an upper bottom plate 6, a lower bottom plate 7, and a loading plate 5;

In this embodiment, there are four lifting stepping motors 4 and four screw rods 3 respectively. In other embodiments, more screw rods 3 or lifting stepping motors 4 (for example, six) can be installed to ensure the stability of the lifting device 15; the lower end of the screw rod 3 is connected to the lifting stepping motor 4, and the upper end is fixed to the upper base plate 6 through a bearing; all the lifting stepping motors 4 fixed on the lower base plate 7 rotate synchronously, and when the lifting stepping motor 4 rotates, it drives the loading plate 5 to move up and down;

The upper platform 1 and the lower platform 2 are respectively installed on the outer surfaces of the upper base plate 6 and the lower base plate 7. The upper platform 1 and the lower platform 2 rotate synchronously around the central axis. When the upper platform 1 and the lower platform 2 rotate, the lifting device 15 is driven to rotate.

The camera 9 is fixed on the camera base 8. The image captured by the camera used in this embodiment is a color image, and an image is captured every 0.3 seconds, that is, every 0.3 seconds is a moment. The camera base 8 is fixed on the rotating shaft screw 10, one end of the rotating shaft screw 10 is connected to the rotating stepping motor 12, and the other end is fixed to the bearing 11; the rotating stepping motor 12 and the bearing 11 are fixed under the object carrier 5 through the mounting frame 17. When the rotating stepping motor 12 rotates, the rotating shaft screw 10 will drive the camera base 8 and the camera 9 to rotate, thereby changing the camera viewing angle.

The upper pan-tilt platform 1, the lower pan-tilt platform 2, the lifting stepping motor 4, and the rotating stepping motor 12 are controlled by a processor 14 installed in a control box 13. In addition, the control box 13 also has a lithium battery for power supply.

The screw rod 3 is hollow. When the upper gimbal 1, the camera 9, the rotary stepping motor 12 and the processor 14 are connected, the connected data line is inserted from the top of the screw rod 3, and is connected to the processor 14 through the hollow interior of the screw rod 3, ensuring that the data line does not affect the movement of the field of view of the camera 9.

The working principle is: the survey device is installed on a ship or a drone, and the upper pan/tilt 1, the lower pan/tilt 2, the lifting stepper motor 4, and the rotating stepper motor 12 are remotely controlled to make the camera 9 face the target object. The target object includes bridges, slopes, piles, buildings, etc. This embodiment takes a bridge as an example for description.

Then the processor 14 reads the image captured by the camera, and uses the photogrammetry method to obtain the measurement results of different positions on the target object, wherein the measurement result is the deformation amount at each position, and the measurement results of other embodiments also include: the settlement amount at each position of the target object, and the distance from the camera, wherein the distance between all positions of the target object and the camera describes the shape information of the target object, and the measurement results of other embodiments may also be the point cloud data of each position of the target object, and the camera may be set as a binocular camera to obtain the distance between all positions of the target object and the camera and the point cloud data of each position of the target object.

The photogrammetry method is a commonly used method for topographic surveying, and its specific working principle is not described in detail in this embodiment. For example, the deformation of the pixels with the same name (that is, each position) is obtained by the digital image correlation (DIC) method.

The survey equipment may be installed or fixed unstably and may be jolted. In this embodiment, photogrammetry is performed on the same target object at multiple times in a row. For example, when measuring deformation or settlement, it is necessary to measure at the same position at multiple times in a row. For another example, when measuring the point cloud data of the target, it is necessary to continuously scan the target object at multiple times.

The measurement results of all positions on the target object at several moments are recorded as historical measurement results. For each of the several moments, there will be errors in the measurement results of different positions on the target object at each moment. This error is mainly caused by unstable installation or fixation, such as the bumps of ships or drones. In theory, the error can be greatly reduced by reasonably controlling the upper gimbal 1, the lower gimbal 2, the lifting stepper motor 4, and the rotating stepper motor 12. In other words, this error is caused by the upper gimbal 1, the lower gimbal 2, the lifting stepper motor 4, and the rotating stepper motor 12 not being reasonably controlled or not responding accordingly to the unstable installation or fixation. Therefore, the errors of the historical measurement results at several moments are recorded as control errors.

The control errors at a certain time are input into the Kalman filter algorithm to obtain the control error C1 at the next time after a certain time. The Kalman filter algorithm is a well-known technology and will not be described in detail in this embodiment.

The processor 14 uses the control error C1 at the next moment to control the upper pan/tilt platform 1, the lower pan/tilt platform 2, the lifting stepper motor 4, and the rotating stepper motor 12. This process uses the control errors existing at several moments in the past to perform control at the next moment. Since the control error is obtained from the measurement result, and the measurement result is affected by the unstable installation or fixation, the noise problem caused by the unstable installation or fixation can be eliminated by controlling the upper pan/tilt platform 1, the lower pan/tilt platform 2, the lifting stepper motor 4, and the rotating stepper motor 12 at the control error C1 at the next moment.

After the control process of the upper pan-tilt head 1, the lower pan-tilt head 2, the lifting stepper motor 4 and the rotating stepper motor 12 is completed, the processor 14 uses the camera 9 to perform photogrammetry again and obtains the measurement results 2 of different positions on the target object and the error C2 of the measurement result t. The difference between the error C2 and the control error C1 is recorded as the photogrammetry error. This process takes into account that although a part of the noise error caused by unstable installation or fixation is eliminated after the control is completed, there is still an error in the measurement result t2. The error at this time may be caused by the photogrammetry method. Therefore, the difference between the error C2 and the control error C1 is recorded as the photogrammetry error. Finally, the photogrammetry error is used to correct the measurement result t to obtain the measurement results of all positions on the target object at the next moment.

To summarize the above, the surveying device in this embodiment is used to be installed in scenes where fixed conditions are unstable and there are bumps, such as ships or drones. At the same time, the control process of the upper gimbal 1, the lower gimbal 2, the lifting stepper motor 4 and the rotating stepper motor 12 in the surveying device can be made more accurate and reliable, eliminating the influence of unstable fixed conditions to a certain extent, and obtaining accurate and reliable measurement results.

Furthermore, matching the above application scenario, the surveying device of this embodiment further includes an external bracket 16, and the external bracket 16 includes a bracket body 1601, an upper fixing pile 1602 welded on the bracket body 1601, a lower fixing pile 1604, and a side fixing pile 1603. The function of the external bracket 16 is to enable the surveying device to be fixed on different carriers at different angles, such as a ship or a drone, so that the installation and fixation of the surveying device are more diverse. This embodiment does not limit the specific structure of the external bracket 16, and only needs at least one fixing pile each for the upper and lower parts and the measurement. The function of the fixing pile is to connect and fix with the carrier, such as fixing with bolts and pins or fixing with bearings or even welding. This embodiment does not specifically limit the structure and fixing method of the fixing pile.

In other embodiments, the above-mentioned external bracket 16 may not be used, and the upper gimbal 1 and the lower gimbal 2 may be directly fixed on the drone or the vessel.

As an optional example, the method for obtaining the control error is as follows:

The multiple moments in the historical measurement results of this embodiment refer to the 15 moments closest to the current moment (including the current moment), and other numbers of moments may also be specified in other embodiments.

For each position of the target object, the measurement results of each position at several moments are obtained, and the measurement results of several moments in the historical measurement results are fitted into a polynomial (for example, a cubic polynomial), whose input is the moment and the output is the predicted measurement result. The square of the difference between the measurement result at each moment and the predicted measurement result is obtained, and the mean of the squares of the differences corresponding to all moments is taken as the error size of each position. The error sizes of all positions are clustered into 10 categories using the K-Means clustering algorithm (other embodiments can be clustered into other numbers of categories), the mean of each category is obtained, and the vector composed of the means of all categories is recorded as the control error.

The control error describes the deviation from the original measurement law at different positions when the survey device is installed or fixed unstably (that is, the square of the above difference).

As another optional example, the method for obtaining the control error is as follows:

For the measurement results of all positions at each moment, translate the measurement results of each position so that the difference between all the measurement results corresponding to different positions is minimized. The specific method is:

For the measurement results of all positions at the i-th moment, an undetermined coefficient a(i) is added, and the obtained result is recorded as the translation measurement result of all positions at the i-th moment. For the translation measurement results of the same position at all moments, the variance of these translation measurement results is calculated and recorded as the alignment difference of each position. The alignment differences of all positions are averaged, and the result is recorded as the overall alignment difference. At this time, the overall alignment difference is a data related to the undetermined coefficient a(i). The random gradient descent method (other embodiments can use the simulated annealing algorithm) is used to obtain the undetermined coefficient a(i) that minimizes the overall alignment difference, where i=1, 2, 3, ..., 15; at this time, the alignment difference of each position when the overall alignment difference is minimized is recorded as the error size of each position, and the control error is obtained according to the error size using the above optional example method.

In this optional example, the minimum overall alignment difference means that the measurement results at each moment are translated and aligned. The purpose is to take into account the deformation or settlement of each position at different times. The changes in the measurement results of each position caused by these conditions cannot be regarded as errors. The above-mentioned translation alignment process can avoid the situation where deformation or settlement is mistaken for error to a certain extent.

In other embodiments, in order to further avoid deformation or settlement being mistaken for errors, the above-mentioned "average of alignment differences at all positions" can be replaced by: finding the absolute value of the alignment differences at adjacent moments, and then averaging the absolute values at all adjacent moments.

However, this process requires a large amount of computing power (which will affect the endurance of the survey equipment).

As a preferred example, the method for obtaining the control error is as follows:

Obtain the difference between the measurement result of each position of the target object and the measurement results of all field positions of each position, record it as the difference feature of each position, the average of the difference between the difference feature of each position at each moment and the difference feature of each position at the two adjacent moments is recorded as the target difference feature of each position, reduce the dimension of the target difference feature of each position to one dimension using the PCA algorithm, record the one-dimensional reduction results at several moments in the historical measurement results as the error vector of each position, cluster the error vectors of all positions into 10 categories using the K-means clustering algorithm, obtain the mean of all error vectors in each category, and record a vector formed by splicing the means of all categories end to end as the control error.

It should be noted that when there are no adjacent moments before and after each moment, only the adjacent previous moment and next moment are considered.

In other embodiments, the elements in the error vector at each position may be summed to obtain the error magnitude at each position, and then the control error may be obtained according to the method in the above optional example.

As a preferred example, the control error C1 is used to control the upper pan/tilt platform 1, the lower pan/tilt platform 2, the lifting stepper motor 4 and the rotating stepper motor 12, including the following method:

A fully connected neural network with three hidden layers is constructed, wherein each hidden layer has five neurons, the input layer of the fully connected neural network is the control error, and the output is the control error C1 to control the control parameters of the upper pan/tilt platform 1, the lower pan/tilt platform 2, the lifting stepping motor 4, and the rotating stepping motor 12, such as the rotation angle. In other embodiments, neural networks of other structures may be set, or other control parameters may be introduced, such as the rotation speed. This embodiment is not specifically limited.

Training the neural network includes:

In a laboratory environment or using a simulator to simulate various jolting movements of the surveying device, for example, using a three-axis robotic arm or turntable to control the surveying device to randomly make up, down, left, right, and pitch movements, and record the angles or heights of the up, down, left, right, and pitch movements. When simulating jolting movements of various degrees, perform photogrammetry on the target object, and use the above method of this embodiment to obtain the control error. The control error obtained during the simulation process is used as a sample, and the angle or height of the up, down, left, right, and pitch movements (the height is converted into the rotation angle of the lifting stepper motor) is recorded as the label corresponding to the sample. All samples and labels are used as a data set, and the data set is used to train a fully connected neural network. The loss function used in training is the mean square error loss function, and the parameter update method of the fully connected neural network is the stochastic gradient descent method. The specific training method of the fully connected neural network is well known, and this embodiment will not be described in detail.

It should be noted that the present embodiment uses a simulator to simulate various degrees of bumping actions and simulated targets, and the simulator includes engines such as Unreal Five and Unity. This method does not require too much investment and is convenient and fast.

The control error obtained above is input into the trained fully connected neural network to obtain the control parameters of the upper gimbal 1, the lower gimbal 2, the lifting stepper motor 4 and the rotating stepper motor 12. Then, according to the control parameters, the upper gimbal 1, the lower gimbal 2, the lifting stepper motor 4 and the rotating stepper motor 12 perform control actions, for example, the upper gimbal 1 and the lower gimbal 2 rotate to drive the camera 9 to move left and right, the lifting stepper motor 4 rotates to drive the camera 9 to move up and down, and the rotating stepper motor 12 rotates to drive the camera to pitch.

It should be noted that although the training process of the above-mentioned fully connected neural network takes a large amount of calculation, the training process is completed on a computer outside the surveying equipment. The surveying equipment only needs to run the trained fully connected neural network during specific work, so it will not have a significant impact on the endurance of the surveying equipment. In other embodiments, if it is necessary to further increase the endurance, the hidden layer of the fully connected neural network can be changed to 2 layers, or the number of neurons in each hidden layer can be set to 3.

As an optional example, the control error C1 is used to control the upper pan/tilt platform 1, the lower pan/tilt platform 2, the lifting stepper motor 4, and the rotating stepper motor 12, including the following method:

An acceleration sensor and a gyroscope are installed on the camera base 8, and the motion trajectory of the camera base 8 is solved by the inertial navigation system based on the acceleration sensor and the gyroscope (the motion trajectory includes the position at each moment). The motion trajectory can describe the trajectory of the camera 9 where the survey equipment is located when it is bumpy, but considering that the data collected by the acceleration sensor and the gyroscope contain noise and the inertial navigation system also contains errors, it is necessary to merge the motion trajectory with the above-mentioned control error C1 for control. The specific method is: the motion trajectory of several recent moments (for example, the latest 3 moments) and the control error C1 are spliced into a vector as the input of the fully connected neural network, and then the control parameters are obtained by using the fully connected neural network, and the upper pan-tilt platform 1, the lower pan-tilt platform 2, the lifting stepper motor 4 and the rotating stepper motor 12 are controlled.

The inertial navigation system is a prior art, and its specific principle will not be described in detail in this embodiment.

However, although this method has a more accurate control process, it requires the installation of additional sensor equipment and the setting of more parameters of the fully connected neural network (for example, 7 neurons are required in each hidden layer), which is not conducive to the long-term endurance of the surveying equipment.

When the control process is completed, the camera 9 is used again to perform photogrammetry and obtain the measurement results t of different positions on the target object at the next moment and the error C2 of the measurement result t, wherein the method for obtaining the error C2 is the same as the above-mentioned process of "errors of historical measurement results at several moments" (that is, the same as the process of obtaining the control error). It is only necessary to add the measurement results t of different positions on the target object at the next moment to the historical measurement results. The specific process will not be repeated in this embodiment.

Since the measurement result t is obtained after the upper gimbal 1, the lower gimbal 2, the lifting stepper motor 4 and the rotating stepper motor 12 are controlled, the influence of some bumps caused by unstable installation or fixation is eliminated. At this time, the difference between the error C2 and the control error C1 is recorded as the photogrammetry error. The photogrammetry error mainly describes the error of the photogrammetry algorithm itself, such as the noise caused by image noise.

Furthermore, the measurement results t are corrected by using the photogrammetry error to obtain the measurement results of all positions on the target at the next moment, which can ensure that the measurement results at the next moment can eliminate the influence of the bumpy situation of unstable installation or fixation on the measurement results, and can also eliminate some errors of the photogrammetry algorithm itself, thus ensuring the accuracy of the measurement results.

As a preferred example, the measurement results of all positions on the target object at the next moment are obtained by correcting the measurement result t using the photogrammetry error, including:

All elements in the photogrammetric error are linearly normalized, and the mean of the normalized elements is recorded as the noise level. The noise level is multiplied by the preset value N, and the result is recorded as P. An odd number greater than P and with the smallest difference from P is used as the width of the Gaussian filter kernel. The Gaussian filter kernel is used to perform Gaussian filtering on the measurement results t of all positions on the target at the next moment, and the measurement results of all positions on the target at the next moment are obtained. The above filtering process is the correction process of the measurement result t.

This embodiment is described by taking N=21 as an example. Other embodiments may be set to other values, which will not be described in detail in this embodiment.

Furthermore, the above method is used to correct the historical measurement results with the photogrammetry error to obtain the corrected measurement results. Then the corrected measurement results and the measurement results of all positions on the target at the next moment are used as the historical measurement results again, and the measurement results of the next moment after the next moment are obtained based on the historical measurement results. The process is the same as above, and the general process is:

The errors of historical measurement results at certain moments are recorded as control errors again. The control errors at certain moments are obtained by using the Kalman filter algorithm to obtain the control errors at the next moment after certain moments (that is, the next moment after the above-mentioned next moment), and recorded as C1 again. The control error C1 is used to control the upper pan/tilt platform 1, the lower pan/tilt platform 2, the lifting stepping motor 4, and the rotating stepping motor 12. When the control process is completed, the camera 9 is used again to perform photogrammetry and obtain the measurement results t of different positions on the target object at the next moment after the next moment. The error of the measurement result t is recorded as C2 again. The difference between the error C2 and the control error C1 is recorded as the photogrammetry error. The photogrammetry error is used to correct the measurement result t to obtain the measurement result at the next moment of the next moment. At the same time, the photogrammetry error is used to correct the historical measurement results again to obtain the corrected historical measurement results.

Then, the corrected historical measurement results and the measurement results at the next moment after the next moment are taken as historical measurement results again, and so on, for subsequent photogrammetry.

The above process takes into account the following situation: on the one hand, although the error of the historical measurement result is recorded as the control error to describe the influence of the unstable installation or fixation bumpy situation, it must also contain the error existing in the photogrammetry (i.e., the photogrammetry error). Therefore, it is inaccurate to control the upper pan/tilt platform 1, the lower pan/tilt platform 2, the lifting stepper motor 4, and the rotating stepper motor 12 only according to the control error. However, the influence of the bumpy situation in the control error is the main one, and the error existing in the photogrammetry is relatively minor. Therefore, even if the control error is directly used for control, it is possible to eliminate part of the influence of the bumpy situation. On the other hand, since the control error can only eliminate part of the influence of the bumpy situation, the photogrammetry error obtained above also contains a certain influence of the bumpy situation. Therefore, the accuracy of the measurement result corrected by the photogrammetry error is still not high. Based on these two aspects, this embodiment subsequently obtains the measurement result of the next moment after the next moment based on the corrected measurement result. This process involves recalculation of the control error and the photogrammetry error, so that the control error and the photogrammetry error are updated with each other, making the control process of the upper pan-tilt platform 1, the lower pan-tilt platform 2, the lifting stepping motor 4 and the rotating stepping motor 12 in the surveying device more accurate and reliable, thereby making the measurement result of the next moment after the next moment more accurate.

The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims (10)

1. A topographic survey device for civil engineering, comprising a camera (9) and a lifting device (15), wherein the lifting device (15) comprises a screw rod (3) and a lifting stepping motor (4), characterized in that the lifting device (15) further comprises an upper base plate (6), a lower base plate (7) and a loading plate (5), wherein an upper pan head (1) and a lower pan head (2) are respectively mounted on the outer surfaces of the upper base plate (6) and the lower base plate (7), the camera (9) is fixed on a camera base (8), the camera base (8) is fixed on a rotating shaft screw (10), one end of the rotating shaft screw (10) is connected to a rotating stepping motor (12), and the other end is fixed to a bearing (11); the rotating stepping motor (12) and the bearing (11) are fixed below the loading plate (5), wherein the upper pan head (1), the lower pan head (2), the lifting stepping motor (4) and the rotating stepping motor (12) are controlled by a processor (14) installed in a control box (13); The processor (14) uses the camera (9) to perform photogrammetry on the target object, obtains measurement results of all positions on the target object at a certain time, records them as historical measurement results, and obtains measurement results of all positions on the target object at a next time based on the historical measurement results, including: The errors of historical measurement results at a certain time are recorded as control errors. The control errors at a certain time are input into the Kalman filter algorithm to obtain the control error C1 at the next time after a certain time. The processor (14) uses the control error C1 to control the upper pan/tilt platform (1), the lower pan/tilt platform (2), the lifting stepping motor (4) and the rotating stepping motor (12). When the control process is completed, the processor (14) uses the camera (9) to perform photogrammetry again and obtains the measurement results t of all positions on the target object at the next time and the error C2 of the measurement result t. The difference between the error C2 and the control error C1 is recorded as the photogrammetry error. The photogrammetry error is used to correct the measurement result t to obtain the measurement results of all positions on the target object at the next time. The historical measurement results are corrected by using the photogrammetry error; the corrected historical measurement results and the measurement results of all positions on the target object at the next moment are used again as the historical measurement results, and the measurement results of the next moment after the next moment are obtained according to the historical measurement results. 2. A topographic survey device for civil engineering according to claim 1, characterized in that the outer bracket (16) of the survey device comprises an upper fixed pile (1602), a lower fixed pile (1604), and a side fixed pile (1603). 3. A topographic survey device for civil engineering according to claim 1, characterized in that the screw rod (3) is hollow, and when the upper pan/tilt platform (1), the camera (9), the rotary stepping motor (12) and the processor (14) are connected, the connected data line is inserted from the top of the screw rod (3) and connected to the processor (14) through the hollow interior of the screw rod (3). 4. A topographic survey device for civil engineering according to claim 1, characterized in that the measurement results include the settlement amount, deformation amount, distance from the camera (9) and point cloud of each position of the target object, and the target object includes: slope, bridge, foundation pile. 5. The topographic survey device for civil engineering according to claim 1, characterized in that the specific steps of obtaining the control error are as follows: Obtain the difference between the measurement result of each position of the target object and the measurement results of all neighboring positions of each position, record it as the difference feature of each position, record the average of the difference between the difference feature of each position at each moment and the difference feature of each position at the two adjacent moments as the target difference feature at each moment, reduce the dimension of the target difference feature of each position obtained at all moments in the historical measurement results to one dimension, record the one-dimensional dimension reduction results at all moments as the error vector of each position, cluster the error vectors of all positions into several categories, obtain the mean of all error vectors in each category, and record a vector composed of the means of all categories as the control error. 6. A topographic survey device for civil engineering according to claim 1, characterized in that the use of the control error C1 to control the upper pan/tilt table (1), the lower pan/tilt table (2), the lifting stepping motor (4) and the rotating stepping motor (12) comprises the following specific steps: The control error C1 is input into a fully connected neural network, and the fully connected neural network outputs control parameters of the upper pan/tilt platform (1), the lower pan/tilt platform (2), the lifting stepping motor (4), and the rotating stepping motor (12). The upper pan/tilt platform (1), the lower pan/tilt platform (2), the lifting stepping motor (4), and the rotating stepping motor (12) rotate according to the control parameters. 7. A topographic survey device for civil engineering according to claim 1, characterized in that the method of using the photogrammetry error to correct the measurement result t to obtain the measurement results of all positions on the target object at the next moment comprises the following specific steps: The width of the Gaussian filter kernel is obtained according to the photogrammetry error, and the Gaussian filter kernel of the width is used to perform Gaussian filtering on the measurement results t of all positions on the target object at the next moment to obtain the measurement results of all positions on the target object at the next moment. 8. A terrain surveying device for civil engineering according to claim 5, characterized in that the dimension reduction is performed using a PCA dimension reduction algorithm. 9. A terrain surveying device for civil engineering according to claim 5, characterized in that the clustering is performed using a K-Means clustering algorithm. 10. A topographic survey device for civil engineering according to claim 7, characterized in that the step of obtaining the width of the Gaussian filter kernel according to the photogrammetry error comprises the following specific steps: All elements in the photogrammetric error are linearly normalized, and the mean of the normalized elements is recorded as the noise level. The noise level is multiplied by the preset value N, and the result is recorded as P. An odd number greater than P and with the smallest difference from P is used as the width of the Gaussian filter kernel.