A pavement crack depth measuring device

Abstract

The present application relates to the technical field of road surface survey, in particular to a kind of road surface crack depth measuring device, comprising: mobile platform, data acquisition module, probe array module and data processing module are equipped on mobile platform, data processing module is separately connected with data acquisition module and probe array module and establishes communication connection;Data acquisition module is used to collect the optical data of road surface crack;Data processing module is used to generate measurement path according to optical data;Probe array module is used to collect the bottom data of road surface crack according to measurement path;Data processing module is used to determine the target displacement data of each probe in probe array module by bottom data, then fuses target displacement data and optical data, generates crack depth information, solves the problem that existing road surface crack depth measuring device cannot stably obtain real, representative crack depth data.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of road surface survey, in particular to a road surface crack depth measuring device. BACKGROUND

[0002] Road surface crack depth measurement is a key link for evaluating road surface health condition, determining maintenance scheme and ensuring road safety. Accurate acquisition of crack depth information plays an irreplaceable important role in judging disease severity, predicting service life and optimizing maintenance resource allocation.

[0003] Currently, the road surface crack depth measuring devices on the market mainly adopt probe type measurement method or optical non-contact measurement method. The probe type device directly acquires the depth value by mechanically probing to the bottom of the crack, while the optical device indirectly estimates the crack depth by calculation based on laser ranging or three-dimensional imaging technology. However, whether it is a contact or non-contact measuring device, they all face a common core technical problem, that is, the measuring accuracy is easily disturbed by the complex environment inside the crack (such as dust, debris filler or wet surface). The optical measurement will produce depth estimation error due to the blockage or reflection characteristics change of impurities, while the single probe measurement may lead to the lack of representativeness of the results due to the failure to effectively contact the true bottom or the point instead of the surface, which makes the measurement results of the existing devices lack of reliability and difficult to meet the needs of high-precision road detection. SUMMARY

[0004] In order to solve the technical problem that the existing road surface crack depth measuring device cannot stably obtain true and representative crack depth data, the present application provides a road surface crack depth measuring device.

[0005] The road surface crack depth measuring device provided by the present application adopts the following technical scheme:

[0006] A road surface crack depth measuring device, comprising:

[0007] A mobile platform, a data acquisition module, a probe array module and a data processing module are arranged on the mobile platform, and the data processing module is in communication connection with the data acquisition module and the probe array module;

[0008] The data acquisition module is used for acquiring optical data of the road surface crack;

[0009] The data processing module is used for generating a measurement path according to the optical data;

[0010] The probe array module is used for collecting bottom-touching data of the road surface crack according to the measurement path;

[0011] The data processing module is configured to fuse the target displacement data and the optical data to generate the crack depth information after determining the target displacement data of each probe in the probe array module based on the bottom-touching data.

[0012] Further, the data acquisition module comprises an illumination unit and an imaging unit, and the step of acquiring the optical data of the pavement crack comprises:

[0013] The illumination unit is controlled to project the composite light source onto the pavement crack, and the imaging unit is used to acquire the coherent optical signal scattered and reflected by the composite light source through the pavement crack;

[0014] The coherent optical signal is subjected to phase demodulation operation and speckle statistical operation respectively to obtain a phase map data set and a speckle contrast data set;

[0015] The phase map data set and the speckle contrast data set are fused to generate the optical data.

[0016] Further, the step of generating the measurement path based on the optical data comprises:

[0017] The phase map data set is subjected to structure tensor field calculation to obtain a vector field;

[0018] The vector field is subjected to field line tracking operation to extract the main path and the branch path of the pavement crack from the vector field as an initial path skeleton;

[0019] The speckle contrast data set is subjected to regional credibility scoring to obtain a credibility distribution map of each crack region in the pavement crack;

[0020] The initial path skeleton and the credibility distribution map are fused, and path optimization processing is performed to obtain an optimized path point sequence;

[0021] Motion planning operation is performed based on the optimized path point sequence to generate the measurement path, wherein the measurement path is composed of a plurality of target path points.

[0022] Further, the step of acquiring the bottom-touching data of the pavement crack based on the measurement path comprises:

[0023] Based on the spatial coordinates of each target path point, the deployment interval of the probe array in the probe array module is determined;

[0024] Based on the deployment interval, the central region probes of the probe array are controlled to probe to a preset safe depth of the pavement crack, the pressure feedback data of the central region probes are acquired, and the bottom-touching state of the pavement crack is acquired based on the pressure feedback data;

[0025] After adjusting the probing parameters of each probe in the probe array based on the bottom-touching state, the probes are controlled to perform probing operation on the pavement crack to obtain the bottom-touching data.

[0026] Further, the step of determining the target displacement data of each probe in the probe array module through the bottom-touching data comprises:

[0027] From the bottom-touching data, the pressure change characteristics of each probe are identified;

[0028] According to the pressure change characteristics, the contact state is identified, and the target probe in the stable bottom-touching state is screened out;

[0029] The displacement values of each target probe are respectively compensated for displacement, and a plurality of target displacement data are obtained;

[0030] If it is verified that each target displacement data has continuity and rationality, the plurality of target displacement data are output as target displacement data.

[0031] Further, the step of fusing the target displacement data and the optical data to generate the crack depth information comprises:

[0032] The target displacement data are processed to obtain an initial depth distribution surface, and the optical data are calculated to obtain a surface texture feature distribution;

[0033] According to the surface texture feature distribution, a confidence weight is assigned to each region in the initial depth distribution surface;

[0034] According to the confidence weight of each region, the initial depth distribution surface and the optical data are weighted and averaged to obtain a road surface crack depth distribution map;

[0035] The crack feature parameters are extracted from the road surface crack depth distribution map, and the crack depth information is calculated based on the crack feature parameters.

[0036] Further, the step of processing the target displacement data to obtain an initial depth distribution surface comprises:

[0037] A discrete depth point set is constructed through the plane coordinates of each target probe and the target displacement data of each target probe in the target displacement data;

[0038] The discrete depth point set is processed based on a triangulation algorithm to generate an irregular triangular mesh;

[0039] The irregular triangular mesh is processed based on a preset crack edge constraint condition, and a crack boundary feature line is identified from the irregular triangular mesh;

[0040] According to the crack boundary feature line, the irregular triangular mesh is optimized to generate a triangular mesh graph with a fixed boundary;

[0041] The triangular mesh graph is processed according to a radial basis function interpolation algorithm to obtain an initial depth distribution surface.

[0042] Further, the step of calculating the optical data to obtain the surface texture feature distribution comprises:

[0043] performing a multi-scale gradient field calculation on the phase map dataset to obtain a multi-scale phase gradient amplitude map;

[0044] performing a local binary pattern operator calculation on the speckle contrast dataset to obtain a speckle texture feature map;

[0045] performing feature fusion on the multi-scale phase gradient amplitude map and the speckle texture feature map to obtain a multi-modal texture feature map;

[0046] performing texture consistency analysis on the multi-modal texture feature map to identify a texture difference degree between the road surface crack and the preset complete road surface from the multi-modal texture feature map;

[0047] generating the surface texture feature distribution based on the texture difference degree through a spatial distribution mapping algorithm.

[0048] The beneficial effects achieved are:

[0049] The application provides a road surface crack depth measuring device, comprising a mobile platform, a data acquisition module, a probe array module and a data processing module are arranged on the mobile platform, and the data processing module is in communication connection with the data acquisition module and the probe array module; the data acquisition module is used for acquiring optical data of a road surface crack; the data processing module is used for generating a measurement path according to the optical data; the probe array module is used for acquiring bottom-touching data of the road surface crack according to the measurement path; and the data processing module is used for fusing target displacement data of each probe in the probe array module and the optical data to generate crack depth information after determining the target displacement data through the bottom-touching data.

[0050] That is, in the application, the optical data of the road surface crack is acquired through the data acquisition module, so that the data processing module can generate a measurement path based on the crack macroscopic morphology and surface features reflected by the optical data, the path guides the probe array module to perform bottom-touching data acquisition based on the measurement path, thereby ensuring the representativeness of the measurement points; then, the data processing module filters out a target probe that stably touches the bottom and determines target displacement data of the target probe through the bottom-touching data, the target displacement data is obtained by mechanical bottom touching of the probe and directly reflects the real depth of the crack, effectively eliminating the data distortion caused by internal impurities and other factors in optical measurement; finally, the data processing module fuses the real and reliable target displacement data with the optical data reflecting the full-field morphology of the crack, and the generated result has both the reality of direct bottom touching in mechanical measurement and the representativeness of full-field coverage in optical data, thereby stably obtaining real and representative crack depth information. BRIEF DESCRIPTION OF DRAWINGS

[0051] Figure 1 is a schematic diagram of a module of a road crack depth measuring device of the present application;

[0052] Figure 2 is a schematic diagram of a step flow of collecting optical data of a road crack of the present application;

[0053] Figure 3 is a schematic diagram of a step flow of generating a measurement path according to optical data of the present application;

[0054] Figure 4 is a schematic diagram of a step flow of collecting bottom-touching data of the road crack according to the measurement path of the present application;

[0055] Figure 5 is a schematic diagram of a step flow of generating crack depth information.

[0056] BRIEF DESCRIPTION OF DRAWINGS

[0057] 10, mobile platform; 20, data collection module; 30, data processing module; 40, probe array module. DETAILED DESCRIPTION

[0058] The following will be described in detail in combination with the accompanying Figures 1-5 The present application will be further described in detail.

[0059] In the description of the present application, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application. In addition, the terms "first", "second", "third" are only for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0060] In the description of the present application, it should be noted that unless otherwise explicitly specified and limited, the terms "mounting", "connecting", "connecting" should be understood in a broad sense, for example, it can be fixedly connected, or it can be detachably connected, or integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium; it can be the communication inside two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0061] The embodiment of the present application discloses a road crack depth measuring device.

[0062] Please refer to Figure 1 The embodiment of the present application discloses a road crack depth measuring device, which comprises:

[0063] The mobile platform includes a data acquisition module, a probe array module, and a data processing module. The data processing module establishes communication connections with both the data acquisition module and the probe array module.

[0064] The data acquisition module is used to collect optical data of pavement cracks; the data processing module is used to generate a measurement path based on the optical data; the probe array module is used to collect bottom contact data of pavement cracks based on the measurement path; the data processing module is used to determine the target displacement data of each probe in the probe array module through the bottom contact data, and then fuse the target displacement data and optical data to generate crack depth information.

[0065] To achieve integrated and collaborative operation of various functional modules in the process of measuring pavement crack depth, the data acquisition module, probe array module, and data processing module are integrated and set up on a mobile platform. By establishing communication connections between the data processing module and the data acquisition module and probe array module respectively, a centralized control measurement system is constructed. This makes the mobile platform the physical foundation and motion execution mechanism for carrying and connecting various functional modules. At the same time, it ensures that the data processing module can act as a control center, sending control commands to the data acquisition module and probe array module in real time and receiving their feedback data. This enables coordination and control of the entire measurement process, ensuring that a series of operations from optical data acquisition and path planning to bottom-out measurement and data fusion are carried out in an orderly, automatic, and continuous manner, ultimately effectively improving the overall efficiency, accuracy, and reliability of the measurement.

[0066] In this embodiment, based on the goal of improving the accuracy and reliability of pavement crack depth measurement through multi-source data fusion, an optical data acquisition module collects pavement crack data to obtain the macroscopic morphology and surface characteristics of the pavement crack. A data processing module generates a measurement path based on the optical data to guide the probe array module in selecting measurement points. The probe array module collects bottom-contact data of the pavement crack according to the measurement path to directly obtain true depth information through mechanical contact. Then, the data processing module uses the bottom-contact data to determine the target displacement data of each probe in the probe array module and fuses the target displacement data with the optical data. This combines the full-field coverage of the optical data with the direct authenticity of the target displacement data, thereby achieving the effect of stably generating representative and high-precision crack depth information, effectively solving the measurement error problem caused by the complex internal environment of the crack in traditional devices.

[0067] In one feasible implementation, refer to Figure 2 As shown, the data acquisition module includes an illumination unit and an imaging unit, and the specific execution steps of the data acquisition module include steps S11 to S13:

[0068] Step S11, after the control lighting unit projects the composite light source to the road crack, the imaging unit collects the coherent optical signal scattered and reflected by the composite light source through the road crack.

[0069] The process of controlling the lighting unit to project the composite light source to the road crack is realized by sending a control instruction to the lighting unit through the data processing module, and the control instruction contains specific settings of the wavelength combination, projection angle, modulation frequency and light intensity parameters of the composite light source. After the lighting unit generates the composite light source composed of structured light beams of different wavelengths according to the control instruction, it is projected to the road crack area in a specific space-time coding mode.

[0070] Then, the imaging unit and the lighting unit are strictly synchronized, and at a specific moment of the composite light source projection, the shutter of the imaging unit is triggered, and the image sensor receives and records the light field signal carrying the crack topography and structure information scattered and reflected by the road crack surface and subsurface. The light field signal retains amplitude and phase information because it comes from a coherent light source, forming a coherent optical signal containing interference and speckle characteristics.

[0071] Step S12, the coherent optical signal is respectively subjected to phase demodulation operation and speckle statistical operation to obtain phase map data set and speckle contrast data set.

[0072] The phase demodulation operation of the coherent optical signal is realized by phase shift algorithm or Fourier transform profilometry. In this embodiment, the phase shift algorithm is taken as an example. The embodiment applies to control the lighting unit to project four grating patterns with a phase difference of π / 2 to the road crack in a specific time sequence, and the imaging unit synchronously collects four deformed grating images modulated by the road crack topography. Then, the phase principal value of each pixel point is calculated by using four-step phase shift formula, and the absolute phase value reflecting the three-dimensional profile of the road crack surface is calculated, and finally the phase map data set indexed by pixel coordinates is generated.

[0073] The speckle statistical operation of the coherent optical signal is as follows: after the phase map data set required for phase demodulation is collected, the control lighting unit outputs unmodulated wide-field coherent light and keeps the road crack and the imaging unit relatively stationary, and a single speckle image is collected by the imaging unit. Then, the local speckle contrast of the speckle image is calculated by using a sliding window. Specifically, a pixel window of a certain size is set, and the ratio of the standard deviation to the average value of the gray scale values of all pixels in the pixel window is calculated as the contrast value of the center of the pixel window. After the sliding window traverses the entire speckle image, the speckle contrast data set is generated.

[0074] The two images are obtained from the same lighting unit and imaging unit at different times, i.e., a phase map data set based on a time sequence of multiple images, which accurately represents the micro-geometric morphology of the surface of the road crack, and a speckle contrast data set based on a single image, which can reflect the scattering characteristics of the internal structure of the road crack, both of which constitute a dual-mode feature set, and the collection of the speckle image is independent of the image sequence of the phase demodulation, ensuring clear data sources.

[0075] In step S13, the phase map data set and the speckle contrast data set are fused to generate optical data.

[0076] First, the phase map data set and the speckle contrast data set are pixel-level coordinate aligned to ensure that the phase value and the speckle contrast value at each spatial position correspond to each other, and then the two data sets are reorganized as independent data layers according to a unified coordinate reference system to generate optical data, which completely retains the surface morphology information in the phase map data set and the internal scattering characteristic information in the speckle contrast data set, realizes the generation of optical data with geometric morphology characteristics and internal structure characteristics, maintains the independence and separate processability of the two data sets, and establishes a strict spatial correspondence, providing a data basis for subsequent structure tensor field calculation of the phase map data set and region reliability score calculation of the speckle contrast data set.

[0077] In a feasible implementation, referring to FIG. 6, the specific execution steps of the data processing module include steps S21-S25. Figure 3

[0078] In step S21, the structure tensor field of the phase map data set is calculated to obtain a vector field.

[0079] The gradient values in the horizontal direction and the vertical direction of each pixel point in the phase map data set are calculated to obtain a gradient field, then a local window is taken with each pixel point as the center and the weighted sum of the outer products of the gradient vectors of all pixel points in the local window is calculated to construct a 2x2 structure tensor matrix, and the principal eigenvector direction of the structure tensor matrix after eigenvalue decomposition represents the direction in which the gray scale changes most slowly in the local area of the point, i.e., the tangent direction of the crack edge, and the field composed of the principal eigenvectors of all pixel points is the vector field, which quantifies the geometric trend of the road crack.

[0080] In step S22, a field line tracking operation is performed on the vector field to extract the main stem path and the branch path of the road crack as an initial path skeleton from the vector field.

[0081] ​The point with the maximum modulus value in the vector field is searched as the main stem seed point, and then the second-order Runge-Kutta numerical integration is performed from the main stem seed point along the vector direction and the reverse direction, and the integration step is adaptively adjusted according to the local vector field curvature, and the tracking is stopped when the integration path encounters a region with a sudden change in the vector direction or a too small modulus value, and the main stem path extraction is completed.

[0082] Then, whether there is a point satisfying the branch condition on both sides of the tracked path is detected, and when the angle between the vector direction of the point and the main stem path is greater than a set first threshold value and the modulus value reaches a certain level, the point is taken as a branch seed point, and then the second-order Runge-Kutta numerical integration is performed from the branch seed point along the vector direction and the reverse direction, and the integration step is adaptively adjusted according to the local vector field curvature, and the tracking is stopped when the integration path encounters a region with a sudden change in the vector direction or a too small modulus value, and all branch paths are tracked.

[0083] The main stem path and the branch path obtained by tracking are first subjected to connectivity analysis, the Euclidean distance between the end points of the paths is calculated, and when the distance is less than a set second threshold value, it is determined that these path segments should be connected to each other, and the path topological structure is completed by inserting a new connecting line segment to ensure the continuity of the skeleton. Subsequently, smoothing processing is performed, a smoothing algorithm based on a cubic B-spline curve is used to interpolate and fit the path point sequence after connection, the path curvature is uniformly changed by adjusting the position of the control point, local jitter and sharp corners caused by tracking errors are eliminated, and the vector field representing the local trend is converted into a path skeleton fully reflecting the global topological structure of the road surface crack, thereby providing an accurate and reliable geometric basis for subsequent path optimization.

[0084] The first threshold value is an angle threshold value in branch detection, and is used to determine whether the angle between the point and the main stem path is large enough to be considered as the starting point of the branch; and the second threshold value is a distance threshold value in connectivity analysis, and is used to determine whether the distance between the end points of the paths is small enough to be connected.

[0085] In step S23, the region credibility of the speckle contrast data set is scored to obtain a credibility distribution map of each crack region in the road surface crack.

[0086] The speckle contrast data set is preprocessed by Gaussian filtering to suppress noise, and then the speckle contrast variance of the speckle contrast in each local window is calculated by using a sliding window. When the speckle contrast variance of the local area is low, it indicates that the scattering characteristics of the local area are uniform, and the surface of the pavement crack is clean or the internal filler is single, which corresponds to reliable measurement conditions. When the speckle contrast variance of the local area is high, it indicates that there is obvious scattering anomaly in the local area, which corresponds to unreliable measurement conditions caused by environmental interference factors such as surface pollution, internal debris or moisture of the pavement crack. Finally, the speckle contrast variance is mapped to a reliability score between 0 and 1, thereby generating a reliability distribution map corresponding to the spatial coordinates of the speckle contrast data set, and realizing quantitative identification of the measurement reliability difference of each local area on the surface of the pavement crack, providing a key basis for subsequent path optimization, so that the measurement path can actively avoid low-reliability areas caused by environmental interference, thereby improving the accuracy and robustness of the touch data collection from the source.

[0087] Step S24, fuse the initial path skeleton and the reliability distribution map, and perform path optimization processing to obtain an optimized path point sequence.

[0088] The initial path skeleton is discretized into a path point sequence, each path point in the path point sequence is mapped to a reliability score in the reliability distribution map, and then the total length and average reliability of the path point sequence are taken as the optimization target, and iterative optimization is performed. In each iteration, the gradient direction of each path point movement is calculated, and the gradient is determined by the geometric distance change between the path point and the adjacent path point and the reliability score of the path point. By adjusting the position of the path point, the path gradually shifts to the high-reliability area while maintaining the continuity of the pavement crack shape. When the average reliability of the path point sequence is greater than the preset average reliability and the total length change tends to be stable, the iteration is stopped, and finally the optimized path point sequence is output, realizing a measurement path that accurately traces the geometric contour of the pavement crack and actively avoids low-reliability areas.

[0089] Wherein, the preset average reliability is the lowest acceptable reliability threshold set for the path point sequence as a whole. When the average reliability of the optimized path reaches or exceeds this threshold, it is considered that the reliability of the path meets the measurement requirements.

[0090] Step S25, perform motion planning operation on the basis of the optimized path point sequence to generate a measurement path, wherein the measurement path is composed of a plurality of target path points.

[0091] The optimized path point sequence is input as key points, a cubic B-spline curve interpolation method is used to smoothly connect the path points in the optimized path point sequence to generate a continuous and derivable smooth trajectory, and the smooth trajectory is time parameterized according to the dynamic constraints (such as maximum speed, acceleration, etc.) of the mobile platform to assign time stamps and motion parameters to each path point, so as to ensure smooth motion of the mobile platform when moving along the path without impact. Then, coordinates of the path points are extracted along the smooth trajectory at a fixed step size to generate a plurality of target path points, and the target path points constitute a measurement path, so as to realize conversion of the optimized path point sequence into action instructions that can be accurately executed by the mobile platform, and ensure that the probe array module can automatically measure along the pavement crack profile in a smooth and efficient manner, thereby ensuring the accuracy and consistency of the bottom touch data collection.

[0092] In one possible implementation, referring to FIG. 6, the specific execution steps of the probe array module include steps S31-S33. Figure 4

[0093] In step S31, the deployment interval of the probe array in the probe array module is determined based on the spatial coordinates of each target path point.

[0094] According to the spatial coordinates of the target path point and its adjacent path points, the local tangent direction of the position crack corresponding to the target path point is calculated, and then the reference axis of the probe array is aligned to the tangent direction to ensure that the probe arrangement direction is perpendicular to the pavement crack direction, and the deployment interval of each probe is determined according to the preset probe coverage width parameter at the target path point, so that the probe array forms a spatial configuration matching the local geometric features of the pavement crack before contacting the pavement crack, and the preliminary matching between the probe array and the pavement crack morphology is completed before the measurement starts, thereby improving the efficiency and success rate of the first probing operation.

[0095] In step S32, the center area probes of the probe array are controlled to probe to a preset safe depth of the pavement crack according to the deployment interval, the pressure feedback data of the center area probes are acquired, and the bottom touch state of the pavement crack is acquired according to the pressure feedback data.

[0096] It should be noted that the preset safe depth refers to a maximum probing depth limit value that is preset to prevent the probe from excessively probing and colliding with the bottom of the pavement crack.

[0097] A position instruction is generated according to the preset safe depth value to drive the micro-drive mechanism of the center area probe to move downward at a constant speed, and when the displacement sensor detects that the probe tip of the center area probe reaches the preset safe depth, the probing is immediately stopped.

[0098] ​Then, the micro pressure sensor integrated at each probe root of the central region probe is used to collect pressure signals in real time, and the pressure signals are subjected to sliding average filtering to eliminate random noise interference, so as to output stable pressure feedback data.

[0099] When the pressure feedback data continuously exceeds the preset bottom threshold and remains stable, it is determined that the reliable bottom state is reached; when the pressure data fluctuates sharply or is always below the threshold, it is determined that the bottom is not reached, so that the central region probe is used for preliminary detection, and preliminary information of the bottom of the road crack is quickly obtained, so as to provide a basis for optimization of the measurement parameters of the subsequent full-array probe, thereby significantly improving the overall measurement efficiency under the premise of ensuring measurement safety.

[0100] The preset bottom threshold is a minimum pressure threshold value set for determining whether the probe is in a reliable bottom state.

[0101] In step S33, after adjusting the probe parameters in the probe array according to the bottom state, the control unit controls each probe to perform a probing operation on the road crack to obtain bottom data.

[0102] When the central region probe feedbacks the reliable bottom state, the displacement value thereof is recorded as a reference depth, and the preset safety depth value of the peripheral region probe is reduced by a preset proportion according to the reference depth, for example, by multiplying the displacement value of the central region probe by a preset proportion coefficient less than 1, and directly using the calculation result as the preset safety depth value of the peripheral region probe, and adjusting the probing priority of each probe according to the trend of the road crack to make the probe close to the bottomed region have a higher probing priority.

[0103] The probing operation of each probe is triggered in steps, the high-priority probe is first driven to perform fine probing at a low speed, the data is locked immediately after the probe touches the bottom, and then the probing operation of the adjacent low-priority probe is triggered, and the pressure data and displacement value, i.e. the bottom data, are continuously collected during the probing process of all the probes, so that the preset safety depth value of the peripheral region probe is optimized by using the displacement value of the reliable bottom state, the mutual interference between the probes is avoided by the step-by-step triggering mechanism, and finally the bottom data fully reflecting the cross-sectional shape of the crack is obtained.

[0104] The preset proportion coefficient is an empirical value obtained from a large amount of experimental data, and is used to ensure that the new preset safety depth of the peripheral region probe can fully utilize the displacement value of the central region probe and reserve a safety margin for a deeper crack region possibly existing below the peripheral region probe.

[0105] In a feasible implementation, referring to FIG. 4, the specific execution steps of the data processing module include steps S41-S48. Figure 5

[0106] ​Step S41, identify the pressure change characteristics of each probe from the bottoming data.

[0107] The pressure signal in the bottoming data is pre-processed by noise reduction, and then the first and second derivatives of the pressure signal are calculated to represent the pressure change rate and acceleration. Then the key feature points on the pressure-time curve are identified, including the pressure initial rising point, the maximum pressure change rate point, the pressure stable point and the possible pressure sudden drop point. The pressure change characteristics of each probe are quantified by the combination of the positions and values of these feature points.

[0108] Step S42, according to the pressure change characteristics, the contact state is identified, and the target probe in stable bottoming state is screened out.

[0109] First, set the classification rule: if the pressure-time curve of a probe shows the feature combination of clear pressure initial rising point, maximum pressure change rate exceeding the threshold value, and fluctuation amplitude of pressure stable point less than the tolerance value, it is determined as stable bottoming state; if the pressure curve has no significant rise or the pressure stable point fluctuates sharply, it is determined as not bottoming state; if the pressure curve appears a short peak and then falls rapidly, it is determined as transient contact state. Then according to the determination result, the probe that meets the condition of stable bottoming state is marked as target probe.

[0110] Step S43, displacement compensation is performed on the displacement values of each target probe respectively, and a plurality of target displacement data is obtained.

[0111] A comprehensive compensation algorithm combining temperature drift compensation and probe bending deformation compensation is adopted. Specifically, two-step subtraction calculation of "original displacement value minus temperature drift amount" and "then minus probe bending deformation amount" is performed to directly correct the displacement sensor reading value, so as to obtain the target displacement data. The temperature drift compensation is to linearly correct the displacement value read by the displacement sensor by real-time monitoring of the environmental temperature and using the pre-calibrated temperature-displacement drift coefficient lookup table. The probe bending deformation compensation is to establish a deformation model of the probe under different pressures by finite element analysis, calculate the measurement error caused by lateral bending of the probe according to its pressure value and perform vector correction, so as to effectively eliminate the system error introduced by environmental temperature change and probe mechanical deformation and perform vector correction.

[0112] Step S44, if it is verified that each accurate displacement value has continuity and reasonableness, the plurality of target displacement data is output as target displacement data.

[0113] The absolute value of the difference between the accurate displacement values of adjacent target probes is calculated, and if the absolute value is less than a continuity threshold value set based on the expected width of the road surface crack and the measurement accuracy, it is determined that the displacement values are continuous in space; at the same time, it is checked whether each accurate displacement value is within a reasonable range determined by prior knowledge of the depth of the road surface crack, and if all displacement values are between the minimum road surface crack depth and the maximum road surface crack depth, it is determined that the displacement values are reasonable.

[0114] When the spatial continuity condition and the reasonableness condition are both satisfied, the multiple target displacement data are marked as valid and output as target displacement data, and by identifying and excluding outlier data points caused by measurement abnormalities or external interference, it is ensured that the target displacement data are physically credible and consistent in spatial distribution.

[0115] In step S45, the target displacement data are processed to obtain an initial depth distribution surface, and the optical data are calculated to obtain a surface texture feature distribution.

[0116] The initial depth distribution surface describing the depth characteristics of the crack in the vertical direction and the surface texture feature distribution describing the surface characteristics in the horizontal direction are generated in parallel, and the spatial correspondence between the depth and the surface characteristics of the road surface crack is constructed.

[0117] In step S46, the confidence weight is assigned to each region in the initial depth distribution surface according to the surface texture feature distribution.

[0118] First, the texture gradient amplitude of each pixel point in the surface texture feature distribution is extracted as a feature value, and then a pre-calibrated S-shaped function is used to map the feature value to a confidence weight in the range of 0-1, and the weight calculation formula is .

[0119] wherein, is the confidence weight; is the texture gradient amplitude; is a pre-set threshold value for distinguishing the strength of the feature value, and when the feature value is higher than the pre-set threshold value, the output confidence weight tends to 1; is a gain coefficient.

[0120] The regions with high texture gradient amplitude (representing clear and reliable surface) are assigned with high confidence weight, and the regions with low texture gradient amplitude (representing fuzzy and unreliable surface) are assigned with low confidence weight, so as to dynamically adjust the contribution of each region in data fusion according to the reliability of the surface texture feature distribution, so that the final road surface crack depth estimation is more inclined to trust the reliable regions with clear texture, thereby effectively reducing the influence of noise and interference and improving the accuracy and robustness of the road surface crack depth information.

[0121] Step S47, according to the confidence weight of each region, the initial depth distribution surface and the optical data are weighted and averaged to obtain the pavement crack depth distribution map.

[0122] The optical depth estimate value obtained by inverting the optical data through optical measurement technology (such as three-dimensional reconstruction of phase information) is pixel-level registered with the initial depth distribution surface to ensure spatial alignment. Then for each region, the corresponding confidence weight value is taken as the reliability coefficient of the initial depth distribution surface at that position depth, and a weighted average formula is used for calculation: depth value = confidence weight value × initial depth distribution surface depth value + (1-confidence weight value) × optical depth estimate value. Subsequently, the depth values calculated by all regions are arranged according to their spatial coordinates to form a complete depth value matrix corresponding to the pavement crack region one by one, i.e. the pavement crack depth distribution map.

[0123] Step S48, crack feature parameters are extracted from the pavement crack depth distribution map, and crack depth information is calculated based on the crack feature parameters.

[0124] A depth threshold is set, and all pixel points in the pavement crack depth distribution map with a depth value greater than the depth threshold are preliminarily determined as candidate points belonging to the crack region. Then, connected region analysis is performed on these candidate points to filter out continuous regions that meet the length-width ratio and area characteristics of the pavement crack, and small cavities in the region are filled to accurately identify the complete crack region.

[0125] The maximum depth value is extracted by traversing the depth value matrix as the maximum crack depth parameter, the average value of the depth values in all crack regions is calculated as the average depth parameter, and the pixel distribution frequency in different depth intervals is counted as the depth distribution feature parameter. Then, based on these crack feature parameters, a weighted comprehensive calculation method is used to obtain the pavement crack depth information, which is specifically: the maximum depth parameter, the average depth parameter and the depth distribution feature parameter are multiplied by the empirical weight coefficient respectively and then added to generate a crack depth information that comprehensively represents the crack depth, which realizes the condensation of the two-dimensional depth distribution map into quantitative depth information with clear physical meaning, providing intuitive and reliable depth data support for pavement condition assessment and maintenance decision-making.

[0126] The empirical weight coefficient is a value pre-calibrated based on a large amount of historical measurement data and expert knowledge, which is used to quantify the importance of the maximum crack depth, the average depth and the depth distribution feature in the comprehensive evaluation of crack severity.

[0127] Further, step S45 can further include steps S451-S4510:

[0128] ​​Step S451, constructing a discrete depth point set by the planar coordinates of each target probe and the target displacement data of each target probe represented in the target displacement data.

[0129] The planar coordinates and target displacement data corresponding to each target probe are read from the target displacement data one by one, and then the planar coordinates and target displacement data corresponding to each target probe are combined into a three-dimensional coordinate point, which is combined into an ordered point set according to its spatial position relationship, i.e. a discrete depth point set. In this way, the discretely distributed probe measurement points are converted into a structured spatial data set.

[0130] Step S452, processing the discrete depth point set based on a triangulation algorithm to generate an irregular triangular mesh.

[0131] Each three-dimensional coordinate point in the discrete depth point set is regarded as a planar scattered point, and then a triangular mesh is iteratively constructed by the Bowyer-Watson algorithm. The core of this algorithm is to satisfy the empty circle characteristic (i.e. no other points are contained in the circumcircle of each triangle) and the principle of maximizing the minimum angle (to avoid the appearance of narrow triangles). In the calculation process, the global optimality of the mesh quality is ensured by continuously inserting points and locally optimizing the triangular mesh, so as to obtain an irregular triangular mesh, which realizes the connection of discrete depth measurement points into a continuous triangular mesh surface. The mesh can reflect the spatial topological relationship between the depth measurement points.

[0132] Step S453, processing the irregular triangular mesh based on a preset crack edge constraint condition to identify the crack boundary feature line from the irregular triangular mesh.

[0133] It should be noted that the preset crack edge constraint condition includes an edge length threshold and a two-side triangle height difference threshold, and the edge length threshold is used to determine whether the length of the triangular edge is sufficient to become the critical length value of the crack boundary. The two-side triangle height difference threshold is used to determine whether there is a large enough topographic mutation at the edge by the included angle of the normal vectors of the two adjacent triangles.

[0134] By traversing all the triangular edges in the irregular triangular mesh, the length of each edge and the included angle cosine value of the normal vectors of the two adjacent triangles are calculated. When the length of a certain edge exceeds the edge length threshold, and the included angle cosine value of the normal vectors of the two adjacent triangles is less than the two-side triangle height difference threshold, the edge is marked as a candidate boundary edge. Finally, the candidate boundary edges that are continuous in space and consistent in direction are connected to form a complete crack boundary feature line, so as to identify the geometric boundary between the pavement crack and the complete pavement area.

[0135] Step S454, optimizing the irregular triangular mesh based on the crack boundary feature line to generate a triangular mesh graph with fixed boundaries.

[0136] The identified crack boundary feature line is taken as an impassable rigid constraint edge, and then the original irregular triangle mesh is reconstructed, all triangles intersecting with the constraint edge are deleted, and Delaunay triangulation conforming to the empty circle characteristic is performed inside and outside the boundary defined by the constraint edge, so that the crack boundary feature line becomes a permanent triangle edge in the triangle mesh, thereby generating a triangle mesh map strictly following the real geometric boundary of the crack, effectively avoiding the problem of boundary ambiguity or triangle crossing the crack in the irregular triangle mesh.

[0137] In step S455, the triangle mesh map is processed according to the radial basis function interpolation algorithm to obtain an initial depth distribution surface.

[0138] All vertices of the triangle mesh map are input as known data points, each vertex including its two-dimensional plane coordinates and target displacement data representation, a Gaussian kernel function is selected as the radial basis function, and the form is wherein is the Euclidean distance between the vertices, is a shape parameter.

[0139] All the above vertices are regarded as known interpolation nodes, the coordinates of the ith interpolation node are , and the known depth value is . A Gaussian radial basis function is selected, and an interpolation function is constructed:

[0140]

[0141] wherein, represents the interpolation function; represents the independent variable of the interpolation function, representing any point on the two-dimensional plane; j represents the jth known interpolation node; represents the weight coefficient of the jth interpolation node. It strictly satisfies the interpolation condition on all interpolation nodes, that is, for each interpolation node i, there is .

[0142] The coordinates of each interpolation node i are substituted into the interpolation function, and a linear equation with the weight coefficient as an unknown quantity is obtained:

[0143]

[0144] All the equations corresponding to the n interpolation nodes are combined to form an n-linear equation group, which can be simply expressed in the form of a matrix Aw=z. Wherein, A is an n*n symmetric matrix, called an interpolation matrix, and the elements are , where Rij is the radial basis function value between the ith interpolation node and the jth interpolation node; w is a column vector composed of the weight coefficients w1, w2, w3, …, wn to be solved; z is a column vector composed of the known depth values z1, z2, z3, …, zn.

[0145] By solving the linear equation set, the weight coefficients that make the interpolation surface pass through all the known data points accurately can be obtained by using a numerical method such as LU decomposition, so that the interpolation surface passes through each vertex accurately.

[0146] In the range of the crack area defined by the established triangular mesh, uniformly distributed dense sampling point coordinates are generated according to a set resolution (for example, 10,000 points per square meter), and then for each sampling point, the following formula is used for calculation:

[0147]

[0148] The depth estimation value of the sampling point is calculated by weighted summation .

[0149] Finally, the spatial coordinates (x, y) of all sampling points and the corresponding depth values f(x, y) are combined into a complete depth data set, and a continuous initial depth distribution surface covering the entire crack area is generated by a surface fitting algorithm.

[0150] Step S456, multi-scale gradient field calculation is performed on the phase map data set to obtain a multi-scale phase gradient amplitude map.

[0151] The phase map data set is decomposed by a Gaussian pyramid, a series of phase images of different scales are generated by applying Gaussian blur and downsampling multiple times, then the Sobel operator is used to calculate the gradient components in the x direction and the y direction at each scale, and then the gradient amplitude of each pixel point is calculated according to the gradient components, where the amplitudes , and represent the gradient amplitudes in the x direction and the y direction, respectively, and finally the gradient amplitude maps at all scales are integrated to form a multi-scale phase gradient amplitude map.

[0152] Step S457, local binary pattern operator calculation is performed on the speckle contrast data set to obtain a speckle texture feature map.

[0153] A 3x3 neighborhood window centered on each pixel is set, and the gray value of the center pixel and its 8 neighborhood pixels are compared. If the neighborhood pixel value is greater than or equal to the center pixel value, it is marked as 1, otherwise it is marked as 0. The 8 binary bits are combined in a clockwise direction to form an 8-bit binary number, which is converted to a decimal number. The decimal number is the local binary pattern feature value of the center pixel. After processing each pixel in the speckle contrast dataset, the corresponding speckle texture feature map is generated.

[0154] In step S458, the multi-scale phase gradient amplitude map and the speckle texture feature map are fused to obtain a multi-modal texture feature map.

[0155] The pixel-level coordinate registration of the multi-scale phase gradient amplitude map and the speckle texture feature map ensures spatial alignment. Then, for each pixel position, its phase gradient amplitude feature sequence and local binary pattern texture feature value at different scales are extracted. These two types of feature vectors are concatenated to form a high-dimensional fusion feature vector. Finally, the high-dimensional fusion feature vector is processed by principal component analysis algorithm to retain the main feature components and generate a low-dimensional multi-modal feature descriptor for each pixel. The multi-modal feature descriptors of all pixels are arranged according to the spatial position to form a multi-modal texture feature map.

[0156] In step S459, texture consistency analysis is performed on the multi-modal texture feature map to identify the texture difference between the pavement crack and the pre-set complete pavement.

[0157] In the experimental stage, a large number of multi-modal texture feature maps of complete pavement samples are collected. The reference feature cluster center of the pre-set complete pavement is constructed in the feature space by K-Means clustering algorithm. Then, for each pixel in the multi-modal texture feature map of the pavement to be tested, the distance value between its multi-modal feature vector and the reference feature cluster center of the pre-set complete pavement is calculated. The distance value is the texture difference between the pixel and the pre-set complete pavement.

[0158] Then, the difference values of all pixels in the whole image are normalized to make the difference value range between 0 and 1. A difference value close to 0 indicates a high degree of consistency with the texture characteristics of the complete pavement, while a difference value close to 1 indicates a significant difference. This converts the abstract multi-modal texture feature into a quantitative and interpretable difference index, accurately identifying the degree of deviation of the crack region from the pre-set complete pavement in terms of texture characteristics.

[0159] In step S4510, based on the texture difference, a surface texture feature distribution is generated by a spatial distribution mapping algorithm.

[0160] The texture difference value corresponding to each pixel point is taken as the initial feature intensity of the point, then a bilinear interpolation algorithm is used to perform spatial interpolation calculation on the discrete pixel point feature values, a continuous feature intensity distribution surface is generated, and finally an extreme value normalization algorithm is used to map the feature intensity value into a standard interval of 0-1, forming a final surface texture feature distribution map, so as to convert the discrete pixel-level texture difference into a continuous spatial distribution model, and intuitively present the texture feature gradient rule of the crack area and the intact pavement.

[0161] The above are preferred embodiments of the present application, and do not limit the protection scope of the present application, therefore: any equivalent changes made on the structure, shape, principle of the present application shall be covered within the protection scope of the present application.

Claims

1. A device for measuring the depth of road surface cracks, characterized in that, include: A mobile platform is provided with a data acquisition module, a probe array module and a data processing module, wherein the data processing module establishes communication connections with the data acquisition module and the probe array module respectively; The data acquisition module is used to acquire optical data of road surface cracks; The data processing module is used to generate a measurement path based on the optical data; The probe array module is used to collect the bottom contact data of the road surface crack according to the measurement path; The data processing module is used to determine the target displacement data of each probe in the probe array module through the bottoming data, and then fuse the target displacement data and the optical data to generate crack depth information. The step of determining the target displacement data of each probe in the probe array module using the bottom-touching data includes: From the bottoming data, the pressure change characteristics of each probe can be identified; Based on the pressure change characteristics, the contact state is identified, and target probes in a stable bottoming state are selected. Displacement compensation is performed on the displacement values ​​of each target probe to obtain multiple target displacement data; If the continuity and rationality of the target displacement data are verified, then the multiple target displacement data are output as the target displacement data.

2. The pavement crack depth measuring device according to claim 1, characterized in that, The data acquisition module includes an illumination unit and an imaging unit, and the step of acquiring optical data of road surface cracks includes: After the lighting unit projects a composite light source onto the road surface crack, the imaging unit collects the coherent optical signals of the composite light source scattered and reflected by the road surface crack. Phase demodulation and speckle statistics operations are performed on the coherent optical signals to obtain phase map datasets and speckle contrast datasets. The optical data is generated by fusing the phase map dataset and the speckle contrast dataset.

3. The pavement crack depth measuring device according to claim 2, characterized in that, The step of generating a measurement path based on the optical data includes: The structure tensor field is calculated on the phase map dataset to obtain the vector field; A field line tracing operation is performed on the vector field to extract the main path and branch path of the road surface crack as the initial path skeleton. The region confidence score is performed on the speckle contrast dataset to obtain the confidence distribution map of each crack region in the pavement crack; The initial path skeleton and the confidence distribution map are fused together, and path optimization is performed to obtain an optimized path point sequence; Motion planning is performed based on the optimized path point sequence to generate the measurement path, wherein the measurement path consists of multiple target path points.

4. The pavement crack depth measuring device according to claim 3, characterized in that, The step of collecting the bottom contact data of the road surface cracks according to the measurement path includes: Based on the spatial coordinates of each target path point, the deployment spacing of the probe array in the probe array module is determined; Based on the aforementioned deployment spacing, the probes in the central region of the probe array are controlled to probe down to a preset safe depth of the road surface crack, pressure feedback data of the probes in the central region is obtained, and the bottoming state of the road surface crack is obtained based on the pressure feedback data. Based on the bottoming-out state, after adjusting the probing parameters of each probe in the probe array, the probes are controlled to perform a probing operation on the road surface crack to obtain the bottoming-out data.

5. The pavement crack depth measuring device according to claim 4, characterized in that, The step of fusing the target displacement data and the optical data to generate crack depth information includes: The target displacement data is processed to obtain an initial depth distribution surface, and the optical data is calculated to obtain a surface texture feature distribution; Based on the surface texture feature distribution, confidence weights are assigned to each region in the initial depth distribution surface. Based on the confidence weight of each region, a weighted average calculation is performed on the initial depth distribution surface and the optical data to obtain the road surface crack depth distribution map; Crack feature parameters are extracted from the road surface crack depth distribution map, and the crack depth information is calculated based on the crack feature parameters.

6. The pavement crack depth measuring device according to claim 5, characterized in that, The step of processing the target displacement data to obtain the initial depth distribution surface includes: A discrete depth point set is constructed using the planar coordinates of each target probe and the target displacement data of each target probe as represented in the target displacement data. The discrete depth point set is processed using a triangulation algorithm to generate an irregular triangular mesh. The irregular triangular mesh is processed based on the preset crack edge constraint conditions, and crack boundary feature lines are identified from the irregular triangular mesh. Based on the crack boundary feature lines, the irregular triangular mesh is optimized to generate a triangular mesh diagram with fixed boundaries; The triangular mesh is processed using a radial basis function interpolation algorithm to obtain the initial depth distribution surface.

7. The pavement crack depth measuring device according to claim 6, characterized in that, The step of calculating the surface texture feature distribution from the optical data includes: Multi-scale gradient field calculations are performed on the phase map dataset to obtain a multi-scale phase gradient magnitude map; The speckle contrast dataset is subjected to local binary mode operator calculation to obtain speckle texture feature map; The multi-scale phase gradient magnitude map and the speckle texture feature map are fused to obtain a multimodal texture feature map; A texture consistency analysis is performed on the multimodal texture feature map to identify the texture difference between the road surface crack and the preset complete road surface. Based on the texture difference, the surface texture feature distribution is generated using a spatial distribution mapping algorithm.

Images