A linear dimension measurement method and device for the total length of highway guardrail columns

Abstract

The present invention provides a linear dimension measurement method and device for the total length of highway guardrail columns, which relates to the field of linear dimension measurement. The device is deployed on a detection vehicle and includes a control module, a ground penetrating radar measurement module, a 3D camera measurement module, a sensor module, a lower computer and an upper computer. Among them, the present invention realizes accurate data acquisition through multi-module collaborative control, and the lower computer combines AI image recognition technology to achieve accurate acquisition of contour data. The lower computer transmits the radar raw data (reflected wave + diffracted wave) and the 3D contour data (spatial coordinates of the column cap) to the upper computer, and the upper computer realizes data fusion through a layered velocity model (time-depth conversion) and a coordinate system mapping algorithm (3D → road coordinate system). The total length measurement error has been greatly improved in accuracy compared with single-module detection. The present invention solves the problems of low efficiency, large destructiveness and poor accuracy of traditional manual detection, and combines method innovation with engineering implementation.

Description

Technical Field

[0001] The present invention relates to the field of linear dimension measurement, and particularly to a method and device for linear dimension measurement of the total length of highway guardrail columns. Background Art

[0002] Highway guardrails are important facilities for ensuring traffic safety and are the last line of defense for protecting lives on the road. Their anti-collision performance is mainly reflected in two indicators: the underground burial depth of the columns and the exposed height on the ground surface; among them, the columns, as the main body that bears the impact force of vehicles running off the road, whether they have sufficient burial depth and exposed height according to the design requirements directly affects their protection ability for vehicles; therefore, it is very necessary to measure the linear dimension of the total length of highway guardrail columns (underground burial depth + exposed height). At present, the mainstream method for measuring the total length of highway guardrail columns is manual operation.

[0003] The method of manually detecting the total length of highway guardrail columns is to pull out the columns on-site, that is, after pulling out the columns, measure their linear dimensions with a ruler. This method has the highest accuracy for measuring the total length of the columns, but for the measurement of the burial depth, due to the objective reason that the boundary between dry and wet parts of the column buried in the stratum is not obvious, the measurement accuracy of the burial depth is greatly affected by subjective factors of different manual detections; at the same time, column pulling detection is time-consuming and laborious, easily damages the integrity of the slope and roadbed and is not easy to restore the construction site, with great destructiveness and traffic impact, and cannot be used as a daily inspection means.

[0004] Therefore, it is necessary to provide a method and device for linear dimension measurement of the total length of highway guardrail columns to solve the technical problems existing in the prior art, such as being unable to accurately judge the underground burial depth and exposed height, occupying the road and closing the road during detection will affect the normal traffic of the highway, and there are also potential safety hazards and low efficiency. Summary of the Invention

[0005] The purpose of the present invention is to provide a method and device for linear dimension measurement of the total length of highway guardrail columns, aiming to achieve accurate measurement of the underground burial depth and exposed height, avoid occupying the road and closing the road to affect the normal traffic of the highway, and improve the safety and measurement efficiency during the implementation process.

[0006] To solve the above technical problems, the present invention adopts the following solutions:

[0007] A device for linear dimension measurement of the total length of highway guardrail columns is deployed on a detection vehicle and close to one side of the highway guardrail column, and includes a ground penetrating radar measurement module, a 3D camera measurement module, a sensor module, a control module, a lower computer and an upper computer; wherein,

[0008] The control module is electrically connected to the ground penetrating radar measurement module, the 3D camera measurement module, the sensor module, and the lower computer respectively, and obtains linear dimension measurement data by executing linear dimension measurement control;

[0009] The sensor module is used to obtain the sensor data of the inspection vehicle in real time, and the sensor data is used to trigger linear dimension measurement control;

[0010] The 3D camera measurement module is used to perform the measurement of the exposed height of the column on the ground, including:

[0011] A 3D camera, which is used to collect the contour data of the exposed part of the column on the ground;

[0012] An observation camera, which is used to collect the image data of the exposed part of the column on the ground;

[0013] The ground penetrating radar measurement module is used to perform the measurement of the underground embedding depth of the column, including:

[0014] A ground penetrating radar, which is used to collect the original radar data under the road surface where the inspection vehicle is located;

[0015] The lower computer is used to collect linear dimension measurement data and perform local processing on the image data, including a ground penetrating radar acquisition module, a 3D camera acquisition module, an observation camera acquisition module, and an observation camera processing module;

[0016] The upper computer is electrically connected to the lower computer, including:

[0017] A ground penetrating radar data processing module, which processes the original radar data to obtain the underground embedding depth of the column;

[0018] A 3D camera data processing module, which processes the contour data to obtain the exposed height of the column on the ground.

[0019] Further, the sensor module includes:

[0020] A distance sensor, which is used to obtain the distance data in the driving direction of the inspection vehicle;

[0021] A trigger sensor: which is used to generate a corresponding trigger signal according to the highway guardrail columns passed by the inspection vehicle.

[0022] Further, the ground penetrating radar measurement module is installed in the middle and lower part of the inspection vehicle through a radar mounting bracket, and the 3D camera measurement module is installed in the middle and upper part of the inspection vehicle through a camera mounting bracket; wherein, the ground penetrating radar measurement module and the 3D camera measurement module are installed on the same vertical line.

[0023] On the other hand, a linear dimension measurement method for the total length of highway guardrail columns is provided, which is applied to the linear dimension measurement device as described above. The total length of highway guardrail columns is measured linearly through the following steps:

[0024] Step 1: The detection vehicle travels along the road surface, and the sensor module continuously and real-time acquires the sensor data of the detection vehicle and sends it to the control module;

[0025] Step 2: When the control module triggers the linear dimension measurement control, it respectively controls the ground penetrating radar measurement module and the 3D camera measurement module to execute the linear dimension measurement control;

[0026] Step 3: The control module completes the linear dimension measurement control, obtains the linear dimension measurement data and controls the corresponding module to send it to the lower computer; among them, the linear dimension measurement data includes contour data, image data and radar raw data;

[0027] Step 4: The lower computer collects the linear dimension measurement data and locally processes the image data;

[0028] Step 5: The upper computer receives the contour data and radar raw data sent by the lower computer, and respectively processes the data to obtain the underground embedding depth of the column and the exposed height of the column on the ground;

[0029] Step 6: Repeat steps 1 to 5 to obtain the total length of each highway guardrail column passed by the detection vehicle until the linear dimension measurement ends.

[0030] Furthermore, the control module controls the ground penetrating radar measurement module for linear dimension measurement through the following steps:

[0031] Control the ground penetrating radar to start with a preset center frequency, wavelength and scanning rate;

[0032] The ground penetrating radar emits radar waves to the road surface under the detection vehicle; among them, the radar wave is a radiation wave;

[0033] The ground penetrating radar collects the reflected waves and diffracted waves under the road surface; among them, the reflected waves are obtained by formation reflection, and the diffracted waves are generated by the corresponding diffraction points of the highway guardrail columns;

[0034] The control module controls the ground penetrating radar to pack the reflected waves and diffracted waves into corresponding radar raw data and send them to the lower computer, and the lower computer receives the radar raw data through the ground penetrating radar acquisition module.

[0035] Furthermore, the control module controls the 3D camera measurement module for linear dimension measurement through the following steps:

[0036] Adjust the detection vehicle to park at the detection position and control the observation camera to start;

[0037] Send the image data to the lower computer, and the lower computer receives the image data through the observation camera acquisition module;

[0038] The lower computer processes the image data through the observation camera processing module, and infers and identifies whether there is a column cap in the preset area of the image;

[0039] If so, start the 3D camera to obtain contour data;

[0040] Otherwise, adjust the shooting angle until the column cap is located in the preset area of the image;

[0041] The control module controls the 3D camera to send the contour data to the lower computer, and the lower computer receives the contour data through the 3D camera acquisition module.

[0042] Furthermore, the observation camera processing module identifies the column cap through a target detection model; among them, the target detection model is trained through the YOLOV8 network and deployed using the 16-bit inference engine of TensorRT.

[0043] Furthermore, the ground penetrating radar data processing module processes the original radar data through the following steps:

[0044] Preprocess the original radar data to obtain ground penetrating radar data;

[0045] Conduct formation interface analysis through the reflected wave to obtain the number of formation layers;

[0046] Perform velocity fitting on the radar waves of each formation to obtain the electromagnetic wave velocity of each formation;

[0047] Build a velocity model and perform time-depth conversion to convert the time domain of the velocity model to the depth domain;

[0048] Extract specific parameter values based on the ground penetrating radar data, and substitute them into the velocity model to solve for the underground burial depth of the column.

[0049] Furthermore, the 3D camera data processing module processes the contour data through the following steps:

[0050] Preprocess the contour data and convert the contour data into contour image data;

[0051] Preprocess the contour image data, and successively go through image filtering, binary threshold extraction, and removal of discrete points to fit a straight line to obtain the image to be analyzed;

[0052] Extract the contour lines that meet the requirements in the image to be analyzed as the road reference line and calculate the mapping relationship, and use the mapping relationship to map the image to be analyzed to the road coordinate system;

[0053] According to the characteristics of the column cap, the contour of the column cap is extracted from the image to be analyzed, and the corresponding reference line of the column cap is obtained by combining the spatial coordinate information of the contour data.

[0054] The exposed height of the column on the ground is calculated through the spatial position information corresponding to the road reference line and the reference line of the column cap.

[0055] Furthermore, the mapping relationship between the 3D camera coordinate system and the road coordinate system is obtained through prior knowledge; wherein, the prior knowledge is:

[0056] The z-axis of the road coordinate system is the horizontal axis, perpendicular to the road direction; the x-axis of the road coordinate system is the vertical axis, perpendicular to the road surface; the y-axis of the road coordinate system is along the road direction and parallel to the road surface.

[0057] The z-axis of the 3D camera coordinate system is the extension line of the receiver optical axis of the 3D camera; the x-axis of the 3D camera coordinate system is perpendicular to the z-axis and points to the ground; the y-axis of the 3D camera coordinate system is parallel to the driving direction.

[0058] The linear dimension measurement method and device for the total length of highway guardrail columns provided by the present invention have the following beneficial effects:

[0059] 1. The present invention realizes precise data synchronous acquisition through multi-module collaborative control. Through the core scheduling function of the control module, the operation timings of the ground penetrating radar and the 3D camera module are dynamically coordinated. Based on the real-time data of the sensor module, the relative position between the detection vehicle and the column is accurately determined, and the ground penetrating radar is triggered to emit radar waves and the 3D camera is triggered to collect contour data synchronously, ensuring the spatio-temporal consistency of the underground burial depth and the exposed height.

[0060] 2. The lower computer of the present invention combines AI image recognition technology to achieve accurate acquisition of contour data; the lower computer transmits the original radar data (reflected wave + diffracted wave) and 3D contour data (spatial coordinates of the column cap) to the upper computer, and the upper computer realizes data fusion through a layered velocity model (time-depth conversion) and a coordinate system mapping algorithm (3D → road coordinate system), and the measurement error of the total length is greatly improved in accuracy compared with single-module detection.

[0061] 3. The present invention improves the measurement reliability through a hierarchical control logic; ground penetrating radar control chain: preset radar parameters → emit radiation wave → capture diffracted wave → data preprocessing → formation interface analysis → velocity model fitting, to solve the burial depth calculation deviation caused by the electromagnetic wave velocity difference in multiple strata; 3D camera control chain: observe camera target detection → trigger 3D camera → collect contour data → extract road reference line → coordinate system mapping → calculate exposed height, to achieve millimeter-level positioning accuracy of the column cap under complex lighting conditions.

[0062] 4. The present invention deeply couples hardware deployment with logical control. The ground penetrating radar and the 3D camera are vertically aligned and installed through a bracket. The control module calibrates the relative pose of the two in real time to ensure the spatial matching between the radar wave emission direction and the optical axis of the 3D camera, and eliminates the collaborative error caused by the perspective deviation. Description of the Drawings

[0063] Figure 1 Schematic structural diagram of the linear dimension measurement device provided by the present invention;

[0064] Figure 2 Schematic installation diagram of the linear dimension measurement device provided by the present invention;

[0065] Figure 3 Schematic diagram of the composition of the lower computer module provided by the present invention;

[0066] Figure 4 Schematic diagram of the processing flow of the ground penetrating radar data processing module provided by the present invention;

[0067] Figure 5 Schematic diagram of a single-stratum model provided by the present invention;

[0068] Figure 6 Schematic diagram of the calculation scenario of a multi-stratum model provided by the present invention;

[0069] Figure 7 Schematic diagram of the calculation process of the exposed height of the column provided by the present invention;

[0070] Figure 8 Schematic diagram of the calculation scenario of the exposed height of the column provided by the present invention. Detailed Embodiments

[0071] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. The following description of at least one exemplary embodiment is actually only illustrative and in no way limits the present invention and its application or use. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0072] Unless otherwise specifically stated, the relative arrangements of components and steps, numerical expressions and values set forth in these embodiments do not limit the scope of the present invention.

[0073] At the same time, it should be understood that, for the sake of convenience of description, the dimensions of each part shown in the drawings are not drawn in actual proportional relationship.

[0074] In addition, for the sake of clarity and conciseness, descriptions of well-known structures, functions, and configurations may be omitted. Those of ordinary skill in the art will recognize that various changes and modifications can be made to the examples described herein without departing from the spirit and scope of the present disclosure.

[0075] Technologies, methods, and devices known to those of ordinary skill in the relevant art may not be discussed in detail, but where appropriate, such technologies, methods, and devices should be regarded as part of the authorization specification.

[0076] In all the examples shown and discussed here, any specific values should be construed as merely exemplary and not as limitations. Therefore, other examples of the exemplary embodiments may have different values.

[0077] The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments:

[0078] Embodiment 1

[0079] Please refer to Figure 1 , this embodiment provides a linear dimension measurement device for the total length of highway guardrail columns. The device mainly consists of a ground-penetrating radar measurement module, a 3D camera measurement module, a sensor module, a control module, a lower computer, and an upper computer, and is deployed on a detection vehicle and close to one side of the highway guardrail column through the installation method as Figure 2 shown.

[0080] In order to accurately judge the underground burial depth and the exposed height in this application, the control module needs to cooperate with the ground-penetrating radar measurement module and the 3D camera measurement module to collect corresponding data information respectively; among them, the control module consists of a single-chip microcomputer and necessary peripheral circuits, and is mainly used to execute linear dimension measurement control, so as to obtain linear dimension measurement data, and control the corresponding module to send it to the lower computer for further processing.

[0081] The linear dimension measurement includes a trigger process and a control process. Among them, in the trigger process: the sensor module continuously obtains the sensor data of the detection vehicle, and the control module judges whether to trigger the linear dimension measurement control according to the sensor data; when executing the control process, the control module needs to cooperate with the ground-penetrating radar measurement module, the 3D camera measurement module, and the lower computer for processing.

[0082] The sensor module mainly consists of a distance sensor and a trigger sensor; among them, for the distance sensor, we can set an optoelectronic encoder as the core and install it on the wheel hub of the detection vehicle, which can accurately obtain the driving mileage of the detection vehicle, and perform a calibration operation before use, that is, correspond the actual distance traveled by the wheel and the number of pulses sent by the distance sensor.

[0083] Calibration operation: First, determine the distance to be calibrated. Mark the starting point and the ending point on a straight road. When the detection vehicle passes the starting point, start counting the pulses of the distance sensor. When the detection vehicle passes the ending point mark, stop counting. Repeat the measurement multiple times and record the average value of the calibrated distance and the pulse count. The distance sensor sends a pulse signal to the control module every time the wheel travels a certain distance.

[0084] The trigger sensor uses a reflective laser sensor, which can set the output of the trigger signal when the object to be measured is within a certain distance range, avoiding false triggering of objects that are too close or too far away; the trigger sensor can measure the distance from the column to the vehicle body and can be used to correct the horizontal distance of the ground penetrating radar reaching the column surface, reducing errors; the trigger sensor is installed at the lower part of the vehicle body, and the specific height can be adjusted according to the actual road conditions.

[0085] The ground penetrating radar measurement module is installed in the middle and lower part of the detection vehicle through a radar mounting bracket. The radar mounting bracket can adjust the height of the ground penetrating radar from the ground, the horizontal distance from the guardrail column, and the inclination of the radar with respect to the ground. The 3D camera measurement module is installed in the middle and upper part of the detection vehicle through a camera mounting bracket and is on the same vertical line as the ground penetrating radar measurement module. According to specific detection requirements, the two detection modules can be installed on either one side or both sides of the detection vehicle.

[0086] In specific implementation, the detection vehicle can be a modified ordinary medium or large vehicle. The control module, the lower computer, and the upper computer are installed inside the detection vehicle. The detection vehicle can be driven to the corresponding road position according to the detection requirements, making the device have the advantages of flexible deployment and convenient transportation. During detection, only the detection vehicle needs to be parked at the corresponding position, without occupying the road or closing the road, and it will not affect the normal traffic of the highway. There is no need to get out of the vehicle, thus improving the safety of the execution process.

[0087] The ground penetrating radar measurement module consists of a ground penetrating radar and its related supporting devices. The ground penetrating radar is fixed to the middle and rear part of the vehicle body through a connecting device to collect the original radar data under the road surface where the detection vehicle is located. During the collection process, the ground penetrating radar slides along the ground, ensuring that the horizontal distance from the column is basically the same. The ground penetrating radar and the lower computer use a wireless communication protocol to transmit data, expanding the convenience and operability of the ground penetrating radar. The transmission is stable and reliable, and at the same time, it avoids the frequent plugging and unplugging of wired connections. The lower computer can display the radar data in real time and store it.

[0088] The 3D camera measurement module includes a 3D camera and an observation camera; among them, the observation camera is an ordinary monocular camera, and the 3D camera is set based on a laser emitter and receiver; when collecting data, the observation camera detects whether there are columns in the image through the lower computer observation camera processing module, and can provide a basis for manual review at the same time; the 3D camera projects the laser emitted by the laser on the surface of the object to be measured, and captures the position of the reflected laser through the receiver to achieve contour data collection.

[0089] As Figure 3 shown, the lower computer integrates a ground penetrating radar acquisition module, a 3D camera acquisition module, an observation camera acquisition module, and an observation camera processing module; data is transmitted between the lower computer and the upper computer in a wireless or wired manner; among them, the observation camera processing module is specifically a deep learning target recognition program; the upper computer is installed with a ground penetrating radar data processing module and a 3D camera data processing module, which are used to analyze and process the data obtained from the lower computer and obtain results; of course, if the computer performance is sufficient, the upper computer and the lower computer can be implemented on one computer.

[0090] Embodiment 2

[0091] Based on the above Embodiment 1, this embodiment measures the total length of the highway guardrail columns as a linear dimension through the following steps:

[0092] Step 1: The detection vehicle drives along the road surface, and the sensor module continuously and real-time obtains the sensor data of the detection vehicle and sends it to the control module;

[0093] Step 2: When triggering the linear dimension measurement control, the control module respectively controls the ground penetrating radar measurement module and the 3D camera measurement module to execute the linear dimension measurement control;

[0094] Step 3: The control module completes the linear dimension measurement control, obtains the linear dimension measurement data and controls the corresponding module to send it to the lower computer; among them, the linear dimension measurement data includes contour data, image data, and radar raw data;

[0095] Step 4: The lower computer collects the linear dimension measurement data and locally processes the image data;

[0096] Step 5: The upper computer receives the contour data and radar raw data sent by the lower computer, and respectively processes the data to obtain the underground burial depth of the column and the exposed height of the column on the ground;

[0097] Step 6: Repeat Steps 1 to 5 to obtain the total length of each highway guardrail column passed by the detection vehicle until the linear dimension measurement ends.

[0098] It should be noted that: in order to accurately determine the underground burial depth and the exposed height, in this embodiment, the control logic for each link is set in detail, aiming to cooperate with the corresponding module to accurately obtain the corresponding data according to the actual situation, and control the upper computer and the lower computer to process the collected data in an orderly manner, so as to obtain the total length of the highway guardrail column. The following will elaborate on the control and processing steps for each link in detail.

[0099] The control module performs linear dimension measurement control on the ground penetrating radar measurement module through the following steps:

[0100] Control the ground penetrating radar to start with a preset center frequency, wavelength, and scanning rate;

[0101] The ground penetrating radar emits radar waves into the road surface under the detection vehicle; among them, the radar waves are radiation waves;

[0102] The ground penetrating radar collects the reflected waves and diffracted waves under the road surface; among them, the reflected waves are obtained by stratigraphic reflection, and the diffracted waves are generated by the corresponding diffraction points of the highway guardrail column;

[0103] The control module controls the ground penetrating radar to package the reflected waves and diffracted waves into corresponding raw radar data and send them to the lower computer, and the lower computer receives the raw radar data through the ground penetrating radar acquisition module.

[0104] In a specific embodiment, the center frequency of the ground penetrating radar used is 200MHZ, the wavelength is about 1.5 meters, the detection depth can reach up to 15 meters at most, and the scanning rate can reach 1000 scans / second; during on-site detection, the ground penetrating radar is placed on a bracket connected to the detection vehicle, close to the ground, and moved along the column arrangement direction on the road surface beside the column, that is, the road driving direction. The control module triggers the acquisition of data through a distance sensor to obtain the radar data under the road surface.

[0105] The radiation beam of the ground penetrating radar antenna has a certain width. Therefore, the ground penetrating radar can not only detect the target directly below the antenna, but also detect the targets deviating from directly below within a certain range; using the beam width of the ground penetrating radar for side target detection can not only detect the surface layer structure characteristics directly below the instrument, but also detect the signal at the bottom of the side column. Among them, the diffracted wave generated when the radar wave encounters the bottom of the column is the key signal for detecting the column burial depth. The present invention simultaneously uses the reflected waves from each layer interface and the diffracted waves from each diffraction point to distinguish the strata and calculate the column burial depth.

[0106] Among them, when radar waves encounter a target, electromagnetic scattering occurs. When the target is much larger than the wavelength, the effect of the wave is called reflection, and reflection is a special case of scattering; when the target is less than or equal to the wavelength, diffraction occurs; radar waves are reflected when encountering an interface; diffraction occurs when encountering a diffraction point; this solution uses the diffraction signal generated when the radar wave reaches the bottom of the column for target detection, and then obtains the accurate position of the bottom of the column. The original radar data contains both the reflection signals of each interface and the diffraction signals of each diffraction point.

[0107] As Figure 4 shown, the ground penetrating radar data processing module further processes the processed radar raw data; among them, it includes:

[0108] Preprocess the original radar data to obtain ground penetrating radar data; among them, data preprocessing includes: removing the direct wave, removing the DC component, gain, filtering, removing the background, etc.

[0109] Remove the direct wave. The transmitting antenna is very close to the receiving antenna. The energy of the transmitting antenna reaches the receiving antenna directly without being reflected by the underground target. This received signal is called the direct wave. Generally, there is a strong direct wave in the ground penetrating radar data. The direct wave is one of the main sources of interference in the ground penetrating radar. The intensity of the direct wave may be greater than the amplitude of the target echo. Therefore, it is necessary to first remove the strong direct wave. It is relatively simple to use the mean method. Calculate the average value of all traces, and then subtract this average value from each trace.

[0110] Remove the DC component. A normal trace of data fluctuates near zero, but due to factors such as unstable electronic devices during the detection process, the data deviates from zero, and it is necessary to correct the waveform to ensure that the average value of each item of data is close to zero. The method is to sum up each trace of data, then divide by the number of sampling points of each trace to obtain the mean value, and then subtract the mean value from each data point of this trace to obtain the data after removing the DC component.

[0111] The calculation formula is:

[0112] is the data after correction, is the data before correction, is the number of traces, is the time window, is the number of sampling points of each trace, is the summation iteration variable.

[0113] Gain. Electromagnetic waves attenuate very quickly underground, resulting in a decrease in the amplitude of the reflected echo. This attenuation may lead to inaccurate target identification. The method of gain is used to restore the waveform amplitude, amplify the amplitude of the attenuated part of the signal, observe the waveform more clearly, and optimize the signal visibility.

[0114] Filtering: Affected by complex underground media and external signal interference, ground penetrating radar data often contains noise. Therefore, it is necessary to effectively suppress noise, and mean filtering or median filtering is commonly used. FIRBPF (Finite Impulse Response Bandpass Filter) is used to selectively retain or remove signal components within a specific frequency range during signal processing. It allows signals within a specific frequency range to pass through while blocking signals below or above that range.

[0115] Analyze the formation interface through reflected waves to obtain the number of formation layers. When radar waves encounter a layer interface, reflection occurs, and its characteristic in the radar image is a straight line with a distinct light and dark boundary. Based on this, it is easy to obtain the number of formation layers.

[0116] Perform velocity fitting on the radar waves of each formation to obtain the electromagnetic wave velocity of each formation. The velocities of electromagnetic waves in different formations are different, and the hyperbolic characteristics formed by diffraction when encountering underground targets are also different. The image characteristic formed by the diffraction of electromagnetic waves encountering an underground column is a hyperbola, and the characteristic curve in the radar profile can be represented by the standard equation of the center ( ) of the hyperbola at a certain point in the image plane:

[0117]

[0118] where a is the semi-major axis length and b is the semi-minor axis length.

[0119] Adopt the hyperbola fitting method to fit the electromagnetic wave velocities of different formations. This is a fast method. Use the Gauss-Newton fitting method to fit the hyperbola and obtain the characteristic parameters of the hyperbola. Since the electromagnetic wave velocities of different formations are only related to the eccentricity of the hyperbola, the characteristic parameters of the hyperbola can be obtained through fitting, and the electromagnetic wave velocity of the current formation can be calculated.

[0120] Taking Figure 5 the single-formation model shown as an example, there is an equation:

[0121] Converted into the general hyperbola form:

[0122] , its foci are in the vertical direction,

[0123] where is the distance from the radar to the bottom of the column, is the electromagnetic wave velocity of the current formation, is the shortest time of the diffracted wave at the bottom of the column, is the horizontal distance between the radar and the column, is the buried depth of the guardrail column.

[0124] It can be seen that , , The eccentricity of the hyperbola can be calculated as , where is the semi-major axis length, is the semi-minor axis length.

[0125] Build a velocity model and perform time-depth conversion to convert the time domain of the velocity model to the depth domain; for further explanation and without loss of generality, combined with the calculation scenario of a multi-layer model as shown in Figure 6 , the underground of the road is a conventional three-layer structure. In addition to the direct wave, the radar also receives diffracted waves at the intersection of the column and the layer interface, diffracted waves at the bottom of the column, and reflected waves of the interface. The isochrones composed of the direct wave and the formation reflected waves are linearly distributed, and each diffracted wave is hyperbolically distributed; the relationship of each parameter can be expressed as:

[0126]

[0127] where

[0128]

[0129]

[0130] —— Thickness of formation 1, —— Thickness of formation 2, —— Embedded depth of the column, —— Shortest time of the diffracted wave at the bottom of the column, —— Arrival time of the reflected wave of the first interface, —— Arrival time of the reflected wave of the second interface, —— Layer velocity of formation 1, —— Layer velocity of formation 2, —— Layer velocity of formation 3, —— Horizontal distance between the radar and the column, —— Incident angle of the radar wave in formation 1, —— Refraction angle / incident angle of the radar wave in formation 2, —— Refraction angle of the radar wave in formation 3.

[0131] where the velocity , , is obtained through velocity fitting, and the arrival times , , are obtained through the radar image, D is obtained through measurement, and is calculated by the formula , , , , After that, then through Obtain the buried depth of the column.

[0132] Tilt correction: In some complex road conditions, detection is difficult. The ground penetrating radar can be installed obliquely to increase the radiation energy of the antenna towards the bottom of the column. In this case, the final buried depth result needs to be corrected for tilt.

[0133] The correction equation is:

[0134]

[0135] Where, is the actual buried depth of the column, is the measured buried depth of the column, is the horizontal distance between the ground penetrating radar and the column.

[0136] Extract the specific parameter values based on the ground penetrating radar data and substitute them into the velocity model for solution. Thus, the measurement of the underground buried depth of the column is completed.

[0137] On the other hand, the control module controls the linear dimension measurement of the 3D camera measurement module through the following steps:

[0138] Adjust the detection vehicle to park at the detection position and control the observation camera to start;

[0139] Send the image data to the lower computer, and the lower computer receives the image data through the observation camera acquisition module;

[0140] The lower computer processes the image data through the observation camera processing module and infers whether there is a column cap in the preset area of the image;

[0141] If so, start the 3D camera to obtain contour data;

[0142] Otherwise, adjust the shooting angle until the column cap is located in the preset area of the image;

[0143] The control module controls the 3D camera to send the contour data to the lower computer, and the lower computer receives the contour data through the 3D camera acquisition module.

[0144] During specific execution, the detection vehicle travels on the road to be detected, and the observation camera and distance sensor start working first. The observation camera continuously captures images, and the lower computer's observation camera processing module, namely the deep learning object detection algorithm, identifies whether there is a column in the preset area of the current image.

[0145] When the trigger sensor detects the presence of a column within a preset distance, it sends a signal to the control module, which then communicates with the lower computer to report that the 3D camera can be triggered for shooting at this time. The observation camera processing module is a deep learning object detection program using YOLOV8, and is deployed for inference using the 16-bit inference engine of TensorRT to quickly identify whether there is a column cap in the image.

[0146] The process of using the YOLOV8 network to identify column caps is as follows:

[0147] First, obtain a large number of various column cap images collected by the observation camera. After operations such as data augmentation, data balancing, and data cleaning, label them according to a unified specification to ensure the consistency, integrity, and accuracy of the labels. Use the YOLOV8 network to train a target detection model, and deploy the target detection model using the 16-bit inference engine of TensorRT.

[0148] During actual detection, input the image collected by the observation camera into the YOLOV8 object detection network, use the pre-trained target detection model, and then combine parameters such as the preset detection target label category and confidence level to screen the target, obtain the anchor box of the column cap, calculate and extract the horizontal and vertical coordinates and area of the region where the anchor box is located, and perform target detection in combination with the identification scheme for columns; among them,

[0149] A feasible scheme is provided for the identification of columns:

[0150] The spacing between columns on the road is generally fixed at 1 meter, 2 meters, or 4 meters. In addition to identifying column caps, further compare the driving distance of the detection vehicle passing between two adjacent columns with the fixed spacing between the columns to more accurately determine the position of the columns. For example, the spacing between adjacent columns on the detected road is 4 meters. After the first column triggers the detection, record the driving distance of the detection vehicle from the current column to the next column, and compare this driving distance with the fixed spacing between the columns. If it is within the error range, the comparison is successful. Then, combine the column cap image recognition to accurately judge whether there is a column. Compare whether the anchor box of the column cap is within the ROI of the preset observation camera. If the anchor box is located within the preset ROI and at the same time the driving distance is successfully compared with the fixed spacing between the columns, it is determined that the 3D camera can be triggered for shooting at this moment. If not, the detection of the next image is performed.

[0151] Another feasible scheme is provided for the identification of columns:

[0152] The column is installed behind the guardrail. Deep learning is performed on the special cross-sectional contour shapes of the column and the guardrail to determine whether there is a column. Specifically, guardrails generally have double-wave and triple-wave types. The contours formed by the same type of guardrail and the exposed part of the column are consistent and universal. The cross-sectional contour shape features formed by different types of guardrails and columns are different. The 3D camera captures the complete contour information of the column and the guardrail. After a series of processes, data screening and data cleaning are carried out, and a target model is trained using the YOLO8 network. The target model is used to identify whether there is a column in the preset area of the current observation camera.

[0153] The lower computer synthesizes the detection results of the observation camera images and the status of the trigger sensor. When both conditions are met, after a specific distance delay of the distance between the distance sensor and the 3D module, the control module controls the 3D camera to take continuous shots and collect multiple groups of data to ensure that the entire column cap is completely collected. The horizontal physical interval between the observation camera and the 3D camera is very small. Even at a relatively high vehicle speed, the trigger delay is very small and can be regarded as real-time triggering.

[0154] After receiving the current column 3D contour data, the lower computer saves the data and transmits it to the upper computer for analysis and processing. There is a data acquisition module in the lower computer, which can collect and display the data uploaded by the ground penetrating radar, 3D camera, and observation camera in real time. The analysis and processing of the data can be carried out offline or online in real time. The data saved by the lower computer can be read in an offline manner for analysis and processing to obtain results, or the data forwarded by the lower computer can be obtained in real time for online analysis and processing to give processing results.

[0155] The 3D camera processing module in the upper computer is used to load raw data, process data, export data, and modify detection parameters. Please refer to Figure 7 , and the specific process of data analysis and processing is

[0156] Preprocess the contour data and convert the contour data into contour image data;

[0157] Preprocess the contour image data, which successively undergoes image filtering, binary threshold extraction, and removal of discrete points to fit a straight line to obtain the image to be analyzed;

[0158] Extract the road reference line according to the road surface characteristics. Specifically: extract the contour line that meets the requirements in the image to be analyzed as the road reference line and calculate the mapping relationship, and then use the mapping relationship to map the image to be analyzed into the road coordinate system;

[0159] Extract the column cap contour from the image to be analyzed according to the characteristics of the column cap, and combine the spatial coordinate information of the contour data to obtain the corresponding column cap reference line;

[0160] The exposed height of the column above the ground is calculated based on the spatial position information corresponding to the road reference line and the column cap reference line.

[0161] In the image to be analyzed, first, extract contour lines that are parallel to each other and satisfy a certain distance interval. Then, based on the differences in characteristics such as the lengths and spatial distributions of the road reference line and the column cap reference line, extract the road reference line and the column cap reference line respectively.

[0162] The basis for extracting the contour lines that meet the requirements is that the road reference line and the column cap reference line are straight line segments that are parallel to each other and have a fixed length, and the road reference line has a longer length than the column cap reference line. Obtain the geometric parameters of the road reference line, and through prior knowledge, obtain the mapping relationship between the 3D camera coordinate system and the road coordinate system. Use the mapping relationship to map the image to be analyzed into the road coordinate system.

[0163] The prior knowledge is as follows:

[0164] The z-axis of the road coordinate system is the horizontal axis, perpendicular to the road direction; the x-axis of the road coordinate system is the vertical axis, perpendicular to the road surface; the y-axis of the road coordinate system is along the road direction, parallel to the road surface.

[0165] The z-axis of the 3D camera coordinate system is the extension line of the receiver optical axis of the 3D camera; the x-axis of the 3D camera coordinate system is perpendicular to the z-axis, pointing to the ground; the y-axis of the 3D camera coordinate system is parallel to the driving direction.

[0166] The laser contour data contains the spatial coordinate information of each data point. Using the spatial position information corresponding to the extracted road reference line and column cap reference line, calculate the height of the column cap above the road surface.

[0167] As Figure 8 shown, the distance between any point A on the column cap and any point B on the road surface is the exposed height of the column:

[0168]

[0169] Among them, the coordinates of point A are , and the coordinates of point B are .

[0170] For a single column, multiple sets of contour data are collected, and an exposed height result can be calculated for each set of contour data. Calculate the average value of the heights of multiple sets of contour data as the final exposed height data.

[0171] Every time the inspection vehicle passes by a column, it will collect the data underground and on the ground of the current column simultaneously. Through the above steps in this article, the buried depth and exposed height of the column are obtained respectively, and then the overall length of the current column is obtained by adding them up by the upper computer software.

[0172] In summary, in addition, the present application can not only measure the highway guardrail columns, but also measure the linear dimensions of other "semi-hidden" or "partially hidden" objects.

[0173] In summary, the present invention constructs a dual-modal data fusion measurement system for underground burial depth and ground-exposed height through the collaborative scheduling of a ground-penetrating radar and a 3D camera module by a control module:

[0174] Precision control logic: Based on the real-time feedback of sensors, dynamically trigger the synchronous acquisition of multiple modules, and combine the layered velocity model and the spatial coordinate system mapping algorithm to greatly reduce the total length measurement error and break through the bottleneck of the traditional column-pulling method restricted by subjective factors.

[0175] Deep coupling of hardware and algorithm: The vertical installation structure and the dynamic parameter calibration mechanism ensure the spatial consistency of multiple modules. Combined with YOLOv8 object detection and TensorRT accelerated inference, it realizes the millisecond-level positioning and millimeter-level accuracy of the column cap under complex road conditions.

[0176] Engineering practical value: The non-contact detection scheme greatly compresses the time-consuming of single-column detection and provides efficient and reliable technical support for the safety inspection of highway guardrails.

[0177] The present invention combines method innovation and engineering implementation, and provides a standardized solution for the field of total length measurement of highway guardrail columns.

[0178] The above is only a preferred embodiment of the present invention, and does not impose any form of limitation on the present invention. According to the technical essence of the present invention, within the spirit and principle of the present invention, any simple modification, equivalent replacement, and improvement made to the above embodiments still fall within the protection scope of the technical solution of the present invention.

Claims

1. A linear dimension measuring device for the total length of highway guardrail columns, which is deployed on a detection vehicle and close to one side of the highway guardrail columns, and is characterized in that, It includes a ground penetrating radar measurement module, a 3D camera measurement module, a sensor module, a control module, a lower computer and an upper computer; among them, The control module is electrically connected to the ground penetrating radar measurement module, the 3D camera measurement module, the sensor module and the lower computer respectively, and obtains linear dimension measurement data by executing linear dimension measurement control; The sensor module is used to obtain the sensor data of the inspection vehicle in real time, and the sensor data is used to trigger linear dimension measurement control; The 3D camera measurement module is used to perform the measurement of the exposed height of the column on the ground, including: A 3D camera, which is used to collect the contour data of the exposed part of the column on the ground; An observation camera, which is used to collect the image data of the exposed part of the column on the ground; The ground penetrating radar measurement module is used to perform the measurement of the buried depth of the column underground, including: A ground penetrating radar, which is used to collect the original radar data under the road surface where the inspection vehicle is located; The lower computer is used to collect linear dimension measurement data and perform local processing on the image data, including a ground penetrating radar acquisition module, a 3D camera acquisition module, an observation camera acquisition module and an observation camera processing module; The upper computer is electrically connected to the lower computer, including: A ground penetrating radar data processing module, which processes the original radar data to obtain the buried depth of the column underground; A 3D camera data processing module, which processes the contour data to obtain the exposed height of the column on the ground; The ground penetrating radar data processing module processes the original radar data through the following steps: Preprocess the original radar data to obtain ground penetrating radar data; Analyze the formation interface through the reflected wave to obtain the number of formation layers; Perform velocity fitting on the radar waves of each formation to obtain the electromagnetic wave velocity of each formation; Build a velocity model and perform time-depth conversion to convert the time domain of the velocity model into the depth domain; Extract specific parameter values based on the ground penetrating radar data and substitute them into the velocity model to solve for the buried depth of the column underground; The velocity model is set by the following formula: Among them, is the thickness of formation 1, is the thickness of formation 2, is the embedding depth of the column, is the shortest diffraction wave time at the bottom of the column, is the arrival time of the reflection wave at the first interface, is the arrival time of the reflection wave at the second interface, is the layer velocity of formation 1, is the layer velocity of formation 2, is the layer velocity of formation 3, is the horizontal distance between the radar and the column, is the incident angle of the radar wave in formation 1, is the refraction angle / incident angle of the radar wave in formation 2, is the refraction angle of the radar wave in formation 3.

2. The linear dimension measuring device for the total length of highway guardrail columns according to claim 1, wherein, The sensor module includes: A distance sensor, which is used to obtain the distance data in the driving direction of the inspection vehicle; A trigger sensor: which is used to generate a corresponding trigger signal according to the highway guardrail column passed by the inspection vehicle.

3. The linear dimension measuring device for the total length of highway guardrail columns according to claim 1, characterized in that, The ground penetrating radar measurement module is installed in the middle and lower part of the inspection vehicle through a radar mounting bracket, and the 3D camera measurement module is installed in the middle and upper part of the inspection vehicle through a camera mounting bracket; among them, the ground penetrating radar measurement module and the 3D camera measurement module are installed on the same vertical line.

4. A linear dimension measurement method for the total length of highway guardrail columns, which is applied to the linear dimension measurement device according to any one of claims 1 to 3, characterized in that, Perform linear dimension measurement on the total length of the highway guardrail column through the following steps: Step 1: The inspection vehicle travels along the road surface where it is located, and the sensor module continuously obtains the sensor data of the inspection vehicle in real time and sends it to the control module; Step 2: When the control module triggers linear dimension measurement control, it respectively controls the ground penetrating radar measurement module and the 3D camera measurement module to execute linear dimension measurement control; Step 3: The control module completes the linear dimension measurement control, obtains the linear dimension measurement data and controls the corresponding module to send it to the lower computer; among them, the linear dimension measurement data includes contour data, image data and original radar data; Step 4: The lower computer collects the linear dimension measurement data and performs local processing on the image data; Step 5: The host computer receives the contour data and the original radar data sent by the slave computer, and processes the data respectively to obtain the underground burial depth of the column and the exposed height of the column above the ground; Step 6: Repeat Steps 1 to 5 to obtain the total length of each highway guardrail column passed by the inspection vehicle until the linear dimension measurement ends; Among them, the control module controls the linear dimension measurement of the 3D camera measurement module through the following steps: Adjust the inspection vehicle to park at the inspection position and control the observation camera to start; Send the image data to the slave computer, and the slave computer receives the image data through the observation camera acquisition module; The slave computer processes the image data through the observation camera processing module to infer and identify whether there is a column cap in the preset area in the image; If so, start the 3D camera to obtain contour data; Otherwise, adjust the shooting angle until the column cap is located in the preset area in the image; The control module controls the 3D camera to send the contour data to the slave computer, and the slave computer receives the contour data through the 3D camera acquisition module; The 3D camera data processing module processes the contour data through the following steps: Preprocess the contour data and convert the contour data into contour image data; Preprocess the contour image data, and successively perform image filtering, binary threshold extraction, and removal of discrete points to fit a straight line to obtain an image to be analyzed; Extract the contour line that meets the requirements in the image to be analyzed as the road surface reference line and calculate the mapping relationship, and use the mapping relationship to map the image to be analyzed into the road coordinate system; According to the characteristics of the column cap, extract the column cap contour from the image to be analyzed, and combine the spatial coordinate information of the contour data to obtain the corresponding column cap reference line; Calculate the exposed height of the column above the ground through the spatial position information corresponding to the road surface reference line and the column cap reference line; Among them, the requirements for the contour line are: the road surface reference line and the column cap reference line are straight line segments that are parallel to each other and have a fixed length, and the road surface reference line has a longer length than the column cap reference line; The distance between any point A on the column cap and any point B on the road surface is the exposed height of the column: Among them, the coordinates of point A are , and the coordinates of point B are ; The ground penetrating radar data processing module processes the original radar data through the following steps: Preprocess the original radar data to obtain ground penetrating radar data; Analyze the formation interface through the reflected wave to obtain the number of formation layers; Fit the radar waves of each formation to obtain the electromagnetic wave velocity of each formation; Build a velocity model and perform time-depth conversion to convert the time domain of the velocity model into the depth domain; Extract specific parameter values based on the ground penetrating radar data and substitute them into the velocity model to solve for the underground burial depth of the column; The velocity model is set by the following formula: Among them, is the thickness of formation 1, is the thickness of formation 2, is the embedding depth of the column, is the shortest diffraction wave time at the bottom of the column, is the arrival time of the reflection wave at the first interface, is the arrival time of the reflection wave at the second interface, is the layer velocity of formation 1, is the layer velocity of formation 2, is the layer velocity of formation 3, is the horizontal distance between the radar and the column, is the incident angle of the radar wave in formation 1, is the refraction angle / incident angle of the radar wave in formation 2, is the refraction angle of the radar wave in formation 3.

5. A linear dimension measurement method for the total length of highway guardrail columns according to claim 4, characterized in that, The control module controls the linear dimension measurement of the ground penetrating radar measurement module through the following steps: Control the ground penetrating radar to start with a preset center frequency, wavelength, and scanning rate; The ground penetrating radar emits radar waves under the road surface where the inspection vehicle is located; among them, the radar wave is a radiation wave; The ground penetrating radar collects the reflected wave and diffracted wave under the road surface where it is located; among them, the reflected wave is obtained by formation reflection, and the diffracted wave is generated by the diffraction point corresponding to the highway guardrail column; The control module controls the ground penetrating radar to pack the reflected waves and diffracted waves into corresponding raw radar data and send it to the lower computer, and the lower computer receives the raw radar data through the ground penetrating radar acquisition module.

6. A linear dimension measurement method for the total length of highway guardrail columns according to claim 4, characterized in that The observation camera processing module identifies the column cap of the column through the target detection model; among them, the target detection model is trained through the YOLOV8 network and deployed using the 16-bit inference engine of TensorRT.

7. A linear dimension measurement method for the total length of highway guardrail columns according to claim 4, characterized in that The mapping relationship between the 3D camera coordinate system and the road coordinate system is obtained through prior knowledge; among them, the prior knowledge is: The z-axis of the road coordinate system is the horizontal axis, perpendicular to the road direction; the x-axis of the road coordinate system is the vertical axis, perpendicular to the road surface; the y-axis of the road coordinate system is parallel to the road surface along the road direction. The z-axis of the 3D camera coordinate system is the extension line of the receiver optical axis of the 3D camera; the x-axis of the 3D camera coordinate system is perpendicular to the z-axis and points to the ground; the y-axis of the 3D camera coordinate system is parallel to the driving direction.

8. A linear dimension measurement method for the total length of highway guardrail columns according to claim 4, characterized in that, By + + After obtaining the measured underground embedding depth of the column, tilt correction is also carried out through a correction equation, and the correction equation is: Among them, is the actual underground embedding depth of the column, is the measured underground embedding depth of the column, is the horizontal distance between the ground penetrating radar and the column.

Images