CN117218089B - A method for detecting the structural depth of asphalt pavement - Google Patents

Abstract

The present application is applicable to the technical field of road surface detection, and provides a method for detecting the structural depth of an asphalt road surface, including: obtaining images of an asphalt road surface from multiple perspectives; the images of the multiple perspectives include a reference image and n source images taken around the asphalt road surface, and the optical axis of the camera taking the reference image is perpendicular to the asphalt road surface; inputting the images of the multiple perspectives into a depth map reconstruction model to reconstruct the depth map, and obtaining a depth map of the reference image; obtaining the absolute depth value of the depth map, and cropping the depth map to obtain a depth map with an absolute depth value; fitting the depth map with the absolute depth value to obtain a fitting plane; correcting the tilt error based on the fitting plane to obtain an accurate depth map of the asphalt road surface; and obtaining the structural depth of the asphalt road surface according to the accurate depth map. The present application can improve the efficiency and accuracy of detecting the structural depth of an asphalt road surface.

Description

A method for detecting the structural depth of asphalt pavement

Technical Field

The present application belongs to the technical field of road surface detection, and in particular to a method for detecting the structural depth of an asphalt road surface.

Background technique

The surface structure and friction of asphalt pavement not only affect driving comfort, but are also important factors affecting traffic safety when vehicles accelerate or decelerate. In the standards of various countries, the construction depth of the pavement is an important technical indicator of the skid resistance of asphalt pavement. The greater the pavement construction depth, the better the skid resistance. The construction depth of newly built pavement is the largest. As the service life increases, the road surface will gradually wear out, resulting in a decrease in friction between the wheels and the pavement. Traffic accidents caused by insufficient road friction are common. Therefore, construction depth is an important detection indicator in the acceptance of newly built pavements and the detection of quality decay of operating pavements.

Commonly used methods for testing the skid resistance of asphalt pavement include the braking distance method, the pendulum instrument method, the sand spreading method, the laser method, and the lateral force coefficient method. Among them, the sand spreading method is the main method for testing the texture depth of the pavement. According to the specifications such as ASTM E 965 and T0961-95, it is divided into the electric sand spreading method and the manual sand spreading method. The sand spreading method spreads fine sand on the pavement in the shape of a round cake, records the geometric dimensions of the circle, and calculates the mean texture depth (MTD). The specification EN 13036-1 replaces fine sand with glass balls, which have a smaller particle size range and are more easily flattened during operation. The sand spreading method has some disadvantages, such as the operating method has a great influence on the results and is time-consuming and labor-intensive. Fine sand is difficult to recycle and is not environmentally friendly. It cannot be used on slopes with large angles. It is greatly affected by the climate and cannot be measured on rainy or windy days. Despite this, the sand spreading method is still used as the accuracy basis for many new methods for measuring MTD.

The laser method uses a line laser device to measure the distance of the road surface, and uses the distance value to draw a cross-section elevation scatter diagram to reflect the concave and convex texture characteristics of the road surface. The laser cross-section analysis method obtains the mean profile depth (MPD, Mean Profile Depth) and the MTD, which requires the calculation of the correlation relationship. The three-dimensional analysis method generally obtains the MTD through relative elevation or volume calculation.

Road inspection vehicles equipped with line laser scanning equipment can record the micro-texture fluctuations of the road surface while driving. Some portable laser equipment can help indoor and small-scale detection. Cigada et al. installed two industrial laser triangulation displacement sensors on a road inspection vehicle to estimate the road surface texture morphology based on the driving speed estimated by the sensor. Slower speeds such as 10km/h can measure wavelengths less than 0.5mm. White et al. developed a handheld laser MTD device, and the ETD data obtained in the sand-paving MTD range of 0.5mm to 1.5mm was relatively accurate. The ring texture tester (CTMeter) is also a movable MTD detection device based on laser measurement. Abe et al. and Hanson et al. compared the MPD data of the measuring points obtained by CTMeter with MTD and obtained good correlation and small error results. Laser measurement is very accurate when used for micro-texture analysis, but invalid readings are easily generated under unfavorable conditions of light or light shadows, and triangulation measurement is affected by the external environment and the surface reflection characteristics of the road material itself. There are problems with resampling and filtering effects.

In addition, Zheng Mulian and others used the SAFEGATE friction factor test vehicle to measure the anti-skid performance of the pavement. Wasilewska and others used an anti-skid resistance tester and a friction force observer to test the anti-skid force, and then evaluated the anti-skid performance of the pavement. Dou Guangwu used a high-precision laser ranging sensor to measure the pavement structure depth to evaluate the anti-skid performance of the pavement. Zhou Xinglin and others measured the structure depth of the asphalt pavement based on the laser vision method, and estimated the pavement section depth (MPD) through image processing methods to evaluate the anti-skid performance of the pavement. Qian Zhendong and others used digital image processing technology to reconstruct the three-dimensional texture model of the rutting plate surface, calculated the fractal dimension of the three-dimensional texture model through the differential box dimension method, and studied the relationship between the fractal dimension and the anti-skid performance. Ueckermann and others measured the pavement texture based on the non-contact anti-skid resistance of optical texture measurement to evaluate the anti-skid performance of the pavement. Nejad and others used an image-based automatic system to capture the pavement texture image through the automatic image acquisition system (IAS) to evaluate the anti-skid performance of the pavement. Liang et al. generated a 3D virtual model of the road surface texture based on the point cloud data in the 3D detection system to calculate the mean texture depth (MTD) and evaluate the anti-skid performance of the road surface. Cui et al. measured the average texture depth of asphalt pavement based on multi-line laser and binocular vision technology, introduced multi-line laser pairing and epipolar constraint technology to achieve image matching between multi-line laser and binocular vision, and calculated the average profile depth of the asphalt pavement in turn.

In view of this, with the development of photographic equipment and 3D reconstruction technology, this technology has gradually been applied in the analysis of surface texture and surface crack defects of asphalt concrete. Chen et al. used a trinocular camera to collect pavement texture images and perform 3D reconstruction, extracting pavement elevation data to calculate MTD. et al. developed a scanning device consisting of a line laser and a dual camera, using the camera image to perform horizontal and vertical corrections on the scanned object. In fact, both the laser and structured light methods are active vision methods that emit active light sources, while the monocular, binocular and multi-eye vision methods captured by cameras are passive vision methods that use external light sources. However, the efficiency and accuracy of these methods for detecting the structural depth of asphalt pavements are currently poor.

Summary of the invention

The embodiment of the present application provides a method for detecting the structural depth of an asphalt pavement, which can solve the problem of poor efficiency and accuracy in detecting the structural depth of an asphalt pavement.

The present application provides a method for detecting the depth of an asphalt pavement structure, comprising:

Acquire images of an asphalt road surface from multiple perspectives; the images from multiple perspectives include a reference image and n source images shot around the asphalt road surface, and the optical axis of a camera shooting the reference image is perpendicular to the asphalt road surface;

Input the images from multiple perspectives into the depth map reconstruction model to reconstruct the depth map, and obtain the depth map of the reference image;

Obtaining an absolute depth value of the depth map, and cropping the depth map to obtain a depth map with an absolute depth value;

Fitting the depth map with absolute depth values to obtain a fitting plane;

The accurate depth map of the asphalt pavement is obtained by correcting the tilt error based on the fitting plane;

Obtain the construction depth of asphalt pavement based on accurate depth map.

Optionally, get the absolute depth value of the depth map, including:

Obtain four pairs of calibration points at the four corners inside the calibration plate of known thickness; the four corners correspond to the four pairs of calibration points one by one;

Obtain the absolute depth value of the depth map through the depth value calculation formula;

The depth value calculation formula is:

Among them, _Zabs represents the absolute depth value of the depth map, _Zrel represents the relative depth value of the depth map, scale represents the scale coefficient of the calibration plate, x represents the thickness of the calibration plate, _{Zblue_i} represents the relative depth value of one calibration point in the i-th pair of calibration points, and _{Zred_i} represents the relative depth value of the other calibration point in the i-th pair of calibration points.

Optionally, the depth map is cropped to obtain a depth map with absolute depth values, including:

The depth map is cropped using the calibration plate to obtain a depth map with absolute depth values; the size of the cropped depth map is the same as the size of the calibration plate.

Optionally, fitting the depth map with absolute depth values to obtain a fitting plane includes:

The RANSAC algorithm is used to fit the depth map with absolute depth values to obtain a fitting plane.

Optionally, the accurate depth map of the asphalt pavement is obtained by correcting the tilt error based on the fitted plane, including:

Get the normal vector n of the fitting plane;

Obtain accurate depth map of asphalt pavement through slope correction formula;

The tilt correction formula is:

Among them, Z′ represents the accurate depth map of the asphalt road surface, T represents the coordinate transformation matrix, X, Y represent the pixel coordinates in the depth map, Z represents the absolute depth value of the depth map, and R represents the rotation matrix. θ represents the angle between the normal vector n and the Z axis,/> t represents the three-dimensional translation vector, /> Represents the transpose of a 3×1 zero vector.

Optionally, obtain the structural depth of the asphalt pavement based on an accurate depth map, including:

By formula The structural depth of the asphalt pavement is calculated;

Where MTD _p represents the structural depth of the asphalt pavement, M and N represent the number of pixels in the length and width directions of the accurate depth map, respectively, Z _mn represents the absolute depth value of the pixel in the mth row and nth column of the accurate depth map, and Z _p represents the absolute depth value of the selected texture reference surface. Y represents the area inside the calibration plate.

Optionally, obtain the structural depth of the asphalt pavement based on an accurate depth map, including:

By formula The structural depth of the asphalt pavement is calculated;

Among them, MPD _p represents the structural depth of the asphalt pavement, N is the total number of pixel rows in the accurate depth map, MSD represents the mean cross-sectional depth, represents the depth value of the average elevation of the pixels in row j in the accurate depth map,/> The depth value representing the peak elevation of the pixel in row j in the accurate depth map.

Optionally, the resolutions of the images from multiple perspectives are all 3024×3024, and the value of n is 6.

The above solution of the present application has the following beneficial effects:

In an embodiment of the present application, a reference image of an asphalt pavement and n source images taken around the asphalt pavement are input into a depth map reconstruction model for depth map reconstruction to obtain a depth map of the reference image, and the depth map is cropped by obtaining the absolute depth value of the depth map to obtain the effective area of the depth map, and then the cropped depth map is fitted, and the tilt error is corrected based on the fitting plane to obtain an accurate depth map of the asphalt pavement, so as to obtain the structural depth of the asphalt pavement based on the accurate depth map. Among them, since the depth map of the present application is obtained by combining computer vision technology and deep learning technology, the efficiency and accuracy of depth map reconstruction can be greatly enhanced, so as to improve the efficiency and accuracy of detecting the structural depth of the asphalt pavement. At the same time, since the detection of the structural depth is based on the effective area of the depth map and is completed based on the depth map after the tilt error correction, the efficiency and accuracy of detecting the structural depth of the asphalt pavement can be greatly improved.

In addition, since the detection method of the present application integrates automated image acquisition and analysis, the detection method of the present application is a truly complete automated detection method. At the same time, it is applicable to images of various types of asphalt pavements and has extremely strong practical significance.

Other beneficial effects of the present application will be described in detail in the subsequent specific implementation section.

BRIEF DESCRIPTION OF THE DRAWINGS

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

FIG1 is a flow chart of a method for detecting the depth of an asphalt pavement structure provided by an embodiment of the present application;

FIG2 is a schematic diagram of the structure of a calibration plate in one embodiment of the present application;

FIG3 is a diagram showing a change in depth value of a certain cross section before and after tilt error processing in an embodiment of the present application;

FIG4 is a schematic diagram of MAPE statistical indicators corresponding to different texture reference surfaces when calculating structural depths in one embodiment of the present application;

FIG. 5 is a schematic diagram of a calculation process of MPD _p in an embodiment of the present application.

Detailed ways

In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and/or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and/or combinations thereof.

It should also be understood that the term “and/or” used in the specification and appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

As used in the specification and appended claims of this application, the term "if" can be interpreted as "when" or "uponce" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "uponce it is determined" or "in response to determining" or "uponce [described condition or event] is detected" or "in response to detecting [described condition or event]", depending on the context.

In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

In response to the current problem of low efficiency and accuracy in detecting the structural depth of asphalt pavements, an embodiment of the present application provides a method for detecting the structural depth of an asphalt pavement. The method reconstructs the depth map by inputting a reference image of the asphalt pavement and n source images shot around the asphalt pavement into a depth map reconstruction model to obtain a depth map of the reference image, and crops the depth map by obtaining the absolute depth value of the depth map to obtain a valid area of the depth map. The cropped depth map is then fitted, and the tilt error is corrected based on the fitting plane to obtain an accurate depth map of the asphalt pavement, so as to obtain the structural depth of the asphalt pavement based on the accurate depth map.

Among them, since the depth map of the present application is obtained by combining computer vision technology and deep learning technology, the efficiency and accuracy of depth map reconstruction can be greatly enhanced, so as to improve the efficiency and accuracy of detecting the structural depth of asphalt pavement. At the same time, since the detection of the structural depth is based on the effective area of the depth map and is based on the depth map after tilt error correction, the efficiency and accuracy of detecting the structural depth of the asphalt pavement can be greatly improved.

The asphalt pavement structure depth detection method provided in the embodiment of the present application is exemplarily described below in conjunction with specific embodiments.

As shown in FIG1 , the asphalt pavement structure depth detection method provided in the embodiment of the present application includes the following steps:

Step 11, acquiring images of the asphalt road surface from multiple perspectives.

The images from multiple perspectives include a reference image and n source images shot around the asphalt road surface. The optical axis of the camera shooting the reference image is perpendicular to the asphalt road surface, and the images from multiple perspectives are all RGB images, where RGB represents the colors of the three channels of red, green and blue.

The resolution and number of the images of the above multiple perspectives can be appropriately adjusted according to the performance of the camera. As a preferred example, the resolution of the images of the above multiple perspectives can be 3024×3024, the value of n can be 6, 7, 8, 9, 10, etc., and the camera can be a digital camera, such as a camera of a smart phone.

Step 12: Input the images from multiple perspectives into a depth map reconstruction model to reconstruct the depth map, and obtain a depth map of the reference image.

The above n source images are mainly used to assist in reconstructing the depth map of the reference image.

In some embodiments of the present application, the above-mentioned depth map reconstruction model may be a commonly used depth map reconstruction model. As a preferred example, PatchmatchNet (PatchmatchNet is a deep learning model for image processing) may be used.

It should be noted that the depth map reconstruction model in the above step 12 is a trained model. Here, PatchmatchNet is taken as an example to illustrate the training process of the model.

Firstly, the depth camera Intel RealSense D405 is used to collect image data to obtain the RGB-D (RGB-D = ordinary RGB three-channel color image + depth image) training set (covering common asphalt mixture pavements such as dense-graded asphalt concrete mixture (AC), mastic asphalt macadamia mixture (SMA), upgraded drainable wearing course mixture (OGFC)), and other preprocessing operations such as completion and cropping are performed on the RGB-D training set; then, a large-scale public dataset (such as the DTU dataset) is used to pre-train the model, and the weight of the pre-trained model is used as the initialization parameter, and then the pre-processed RGB-D training set is used to fine-tune the model.

It is worth mentioning that by training the model through the above transfer learning training method, the features and parameter initialization values learned by the pre-trained model can be used to accelerate the training of the model and improve the performance, reducing the data volume and training time. The model will further learn the road texture features from the road texture RGB-D dataset based on the pre-trained weights, and the model will eventually achieve good reconstruction performance.

After the model training is completed, the IoU indicator (IoU indicator is a standard for measuring the accuracy of detecting corresponding objects in a specific data set) can be used to evaluate the accuracy of the neural network training and compare the point cloud reconstructed by the model with the true point cloud.

Specifically, the accuracy of the model can be evaluated on the RGB validation set (which can be obtained by taking a digital camera). It is defined as Among them, V _GT∩Pre represents the intersection of the true point cloud and the point cloud reconstructed by the model, and V _GT∪Pre represents the union of the true point cloud and the point cloud reconstructed by the model. The true point cloud can be reconstructed by inputting the RGB verification set image into the PhotoScan software.

The IoU indicator ranges from 0 to 1. When the IoU value is closer to 1, it means that the spatial forms of the two 3D models are closer, that is, the accuracy of PatchmatchNet's predicted point cloud is better, and it can be inferred that the model can reconstruct depth maps with good quality.

In some embodiments of the present application, in order to improve the accuracy of the depth map reconstructed by PatchmatchNet, after completing the model training and verification, the impact of different image resolutions and numbers of viewing angles and road material types on the reconstruction quality can also be analyzed.

To ensure that the dataset contains diverse scenes and conditions and obtain comprehensive analysis results, the analysis can be performed on the RGB validation set containing different image resolutions, number of viewing angles, and road material types. In the specific analysis process, different image resolutions (2048, 3024, 3400) and number of viewing angles (5, 7, 10) are used for combination and experimentation. Then the image resolution (3024) and number of viewing angles (7) are fixed, and experiments are performed on different road material types, and the corresponding reconstruction results and evaluation indicators (IoU indicators can be used) are recorded.

By statistically analyzing the experimental results and comparing the effects of different factors on the reconstruction quality, it is concluded that the reconstruction quality is best when the image resolution is 3024×3024 and the viewing angle is 7. The reconstruction quality on different road material types is not much different and the effect is stable, so good results can be achieved in the reconstruction tasks in practical applications.

In some embodiments of the present application, in order to prove the accuracy of PatchmatchNet in the present application, an RGB validation set is selected as the test object (specifically screening images with an image resolution of 3024×3024, 7 viewing angles and different pavement material types); a traditional 3D reconstruction model colmap is selected as a benchmark model, and the model is used to perform 3D reconstruction on the images in the data set, and the corresponding reconstruction results are obtained; the IoU indicator is used to measure the point cloud reconstruction quality under different models.

For the selected RGB validation set, PatchmatchNet and the traditional 3D reconstruction model are used for reconstruction. The reconstruction results obtained by the two methods are compared, including visual comparison of the reconstructed depth map and quantitative comparison of the IoU indicator.

The two reconstruction results are quantitatively evaluated using the selected evaluation indicators. PatchmatchNet has a reconstruction quality of IoU = 0.77 for point clouds, which is better than the reconstruction quality of the traditional 3D reconstruction model of IoU = 0.64, and is far superior to the traditional 3D reconstruction model in terms of reconstruction efficiency.

Step 13: Obtain an absolute depth value of the depth map, and crop the depth map to obtain a depth map with an absolute depth value.

In some embodiments of the present application, a calibration plate with a known thickness may be used to determine the absolute depth value of the depth map, and the calibration plate may be cropped to obtain a depth map with the absolute depth value.

Step 14: Fit the depth map with absolute depth values to obtain a fitting plane.

In some embodiments of the present application, a random sampling consensus (RANSAC) algorithm may be used to fit a depth map with absolute depth values to obtain a fitting plane.

Step 15: Correct the tilt error based on the fitting plane to obtain an accurate depth map of the asphalt pavement.

In some embodiments of the present application, an accurate depth map of the asphalt pavement can be obtained by obtaining the normal vector n of the fitting plane and then projecting the points in the depth map into the rotated coordinate system through the coordinate transformation matrix T.

Step 16, obtaining the structural depth of the asphalt pavement according to the accurate depth map.

In some embodiments of the present application, the construction depth of the asphalt pavement may be obtained based on a three-dimensional index of volume or a two-dimensional index based on a cross-section.

It is worth mentioning that since the depth map of the present application is obtained by combining computer vision technology and deep learning technology, the efficiency and accuracy of depth map reconstruction can be greatly enhanced, so as to improve the efficiency and accuracy of detecting the structural depth of asphalt pavement. At the same time, since the detection of the structural depth is based on the effective area of the depth map and is based on the depth map after tilt error correction, the efficiency and accuracy of detecting the structural depth of the asphalt pavement can be greatly improved.

The following is an exemplary description of a specific implementation method of the above step 13, obtaining the absolute depth value of the depth map, and cropping the depth map to obtain the depth map with the absolute depth value.

In some embodiments of the present application, four pairs of calibration points can be first obtained at the four corners of a calibration plate of known thickness, and then the absolute depth value of the depth map can be obtained through a depth value calculation formula, and the depth map can be cropped using the calibration plate to obtain a depth map with an absolute depth value.

Among them, the above four corners correspond to four pairs of calibration points one by one, that is, each corner has a pair of calibration points, and the calibration points of the four corners are set in the same way. To facilitate understanding of the calibration points, the calibration points are exemplified by the calibration plate shown in Figure 2. Figure 2 only illustrates a pair of calibration points of one of the four corners, and the pair of calibration points includes a circular calibration point on the upper side of the corner and a triangular calibration point on the lower side.

The above depth value calculation formula is:

Among them, _Zabs represents the absolute depth value of the depth map; _Zrel represents the relative depth value of the depth map; scale represents the scale factor of the calibration plate, x represents the thickness of the calibration plate, which can be set according to the actual value, such as 3 mm; Z _{blue_i} represents the relative depth value of one calibration point in the i-th pair of calibration points (such as the circular calibration point in Figure 2); Z _{red_i} represents the relative depth value of another calibration point in the i-th pair of calibration points (such as the triangular calibration point in Figure 2).

It should be noted that the depth map reconstructed by PatchmatchNet can reflect the relative depth value. Z _{blue_i} and Z _{red_i} refer to the relative depth values in the depth map reconstructed by PatchmatchNet on the calibration plate.

In some embodiments of the present application, the depth map is cropped using a calibration plate in order to filter out a valid area of the depth map (also referred to as a range of interest). It should be noted that the size of the cropped depth map is the same as the size of the calibration plate. As a preferred example, the size of the cropped depth map can be 100 mm × 100 mm.

The specific implementation method of the above step 15, which corrects the tilt error based on the fitting plane to obtain the accurate depth map of the asphalt pavement, is exemplarily described below in conjunction with a specific embodiment.

In some embodiments of the present application, the normal vector n of the fitting plane can be obtained first, and then the accurate depth map of the asphalt pavement can be obtained by the inclination correction formula.

The above tilt correction formula is:

Where Z′ represents the accurate depth map of the asphalt pavement; T represents the coordinate transformation matrix, which is a matrix form used to describe the transformation relationship from one coordinate system to another, and is set according to the selected coordinate system and transformation rules. X, Y represent the pixel coordinates in the depth map; Z represents the absolute depth value of the depth map; R represents the rotation matrix, /> θ represents the angle between the normal vector n and the Z axis, where the Z axis is the Z axis of the camera coordinate system (specifically, it can be a camera that obtains images from multiple perspectives). /> t represents a three-dimensional translation vector, which can be used to describe the translation transformation from one coordinate system to another. When a vector is added to the translation vector, they will be translated accordingly on each coordinate axis; /> Represents the transpose of a 3×1 zero vector.

It is worth mentioning that the correction of the tilt error can effectively correct the tilt error of the road surface and improve the accuracy of the structural depth. As shown in curve A (curve A is the depth value change curve after the tilt error correction) and curve B (curve B is the depth value change curve before the tilt error correction) in Figure 3, the absolute difference of the depth value after the tilt error correction becomes smaller, and the overall slope of the profile depth value also becomes smaller, indicating that the tilt error of the road surface has been effectively corrected. Among them, the horizontal axis in Figure 3 represents the pixel value of a row of the depth map, and the vertical axis represents the depth value relative to the texture reference surface.

The specific implementation method of the above step 16, that is, obtaining the structural depth of the asphalt pavement according to the accurate depth map, is exemplarily described below in conjunction with a specific embodiment.

The process of obtaining the structural depth of the asphalt pavement based on the volumetric three-dimensional index is as follows:

By formula The structural depth of the asphalt pavement is calculated.

Where MTD _p represents the structural depth of the asphalt pavement, M and N represent the number of pixels in the length and width directions of the accurate depth map, respectively, Z _mn represents the absolute depth value of the pixel in the mth row and nth column of the accurate depth map, and Z _p represents the absolute depth value of the selected texture reference surface. Y represents the area in the calibration plate. It should be noted that the plane where the p% minimum depth quantile value in the accurate depth map is located can be selected as the texture reference surface.

In some embodiments of the present application, the plane where the depth percentile value at 5% in the accurate depth map is located can be selected as the texture reference surface, at which time the prediction error of each type of road surface is close to the minimum. It can be understood that in the embodiments of the present application, the texture reference surface is not limited to which plane of the depth map.

In order to make it easier to understand the influence of texture reference surface on structural depth, different texture reference surfaces are selected to calculate the structural depth, and the MAPE statistical index is used to measure the average relative error percentage between the structural depth and the measured value.

Wherein, MTD _e represents the measured value.

As shown in Figure 4, four different types of asphalt pavements, AC-13, AC-16, SMA-13, and OGFC-16, were tested to obtain the corresponding MAPE statistical index when the plane at different depth quantile values was used as the texture reference surface to calculate the structural depth. In Figure 4, the horizontal axis represents the value of p, and the vertical axis represents the MAPE statistical index value.

The process of obtaining the structural depth of the asphalt pavement based on the two-dimensional index of the profile is as follows:

By formula The structural depth of the asphalt pavement is calculated.

Among them, MPD _p represents the structural depth of the asphalt pavement, N is the total number of pixel rows in the accurate depth map, MSD represents the mean cross-sectional depth, represents the depth value of the average elevation of the pixels in row j in the accurate depth map,/> The depth value representing the peak elevation of the pixel in row j in the accurate depth map.

It should be noted that the depth value of the average elevation and the depth value of the peak elevation can be selected from the schematic diagram of the calculation process of MPD _p shown in Figure 5. Among them, the abscissa of Figure 5 represents the pixel value of a row of the depth map, the ordinate represents the depth value relative to the texture reference surface, the line C represents the peak elevation surface, and the line D represents the average elevation plane.

It should be further explained that, in actual application, the structural depth of the asphalt pavement can be obtained by selecting a volume-based three-dimensional index or a profile-based two-dimensional index according to actual conditions.

The following is an exemplary description of the relevant equipment for implementing the detection method provided in this embodiment.

The core components of the equipment include a wide-angle macro lens, a SLR camera, a rotary servo motor, a horizontal sliding module and a fixed bracket. The servo motor and the sliding module control the camera's shooting angle, and the number of measurement point shots is 1 (reference image shot perpendicular to the road surface) + 8 (multi-view images), a total of 9. The camera is directly connected to the computing server, and photos are exported in real time for road structure depth prediction. It is necessary to design a sunshade, and replace external light sources such as natural light with internal light sources to avoid the impact of uneven or severely tilted ambient light on the imaging results. The equipment should be portable and lightweight to avoid the center of gravity shift of the equipment.

In summary, the asphalt pavement structure depth detection method provided in the embodiment of the present application has the following advantages:

First, non-contact: It is a non-contact measurement method that does not require physical contact with the object being measured like the sand laying method, which is more convenient and safer;

Second, rich information: it can provide rich additional information. In addition to the structural depth value, the depth image can also provide other information about the structural surface, such as texture, color, etc. This information can be used for further analysis and applications, such as road surface quality assessment, defect detection, etc.

Third, visualization: Visual depth images can be generated to intuitively display the distribution and changes of structural depth values. This helps researchers and decision makers understand and analyze structural depth and supports subsequent decision-making and processing.

The above is a preferred embodiment of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles described in the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims (5)

1. A method for detecting the depth of an asphalt pavement structure, comprising the steps of:

Acquiring images of multiple visual angles of an asphalt pavement; the images of the multiple visual angles comprise a reference image and n source images shot around the asphalt pavement, and the optical axis of a camera shooting the reference image is perpendicular to the asphalt pavement;

Inputting the images of the multiple view angles into a depth map reconstruction model to reconstruct a depth map, and obtaining a depth map of the reference image;

Acquiring an absolute depth value of the depth map, and cutting the depth map to obtain a depth map with the absolute depth value;

fitting the depth map with the absolute depth value to obtain a fitting plane;

Correcting the inclination error based on the fitting plane to obtain an accurate depth map of the asphalt pavement;

acquiring the construction depth of the asphalt pavement according to the accurate depth map;

Wherein the obtaining the absolute depth value of the depth map includes:

Four pairs of calibration points are obtained at four corners inside a calibration plate with known thickness; the four corners are in one-to-one correspondence with the four pairs of calibration points;

Obtaining an absolute depth value of the depth map through a depth value calculation formula;

The depth value calculation formula is as follows:

Wherein Z _abs represents the absolute depth value of the depth map, Z _rel represents the relative depth value of the depth map, scale represents the scale factor of the calibration plate, x represents the thickness of the calibration plate, Z _{blue_i} represents the relative depth value of one of the ith pair of calibration points, and Z _{red_i} represents the relative depth value of the other of the ith pair of calibration points;

The obtaining the construction depth of the asphalt pavement according to the accurate depth map comprises the following steps:

By the formula Calculating to obtain the construction depth of the asphalt pavement; wherein MTD _p represents the construction depth of the asphalt pavement, M and N represent the pixel numbers of the accurate depth map in the length and width directions respectively, Z _mn represents the absolute depth value of the nth pixel of the M-th row in the accurate depth map, Z _p represents the absolute depth value of the selected texture reference surface,/>Y represents the area in the calibration plate; or alternatively

By the formulaCalculating to obtain the construction depth of the asphalt pavement; wherein MPD _p represents the construction depth of the asphalt pavement, N is the total number of pixels of the accurate depth map, MSD represents the average section depth,/>, andDepth value representing average elevation of jth row of pixels in the accurate depth map,/>And a depth value representing the peak elevation of the j-th row of pixels in the accurate depth map.

2. The method of claim 1, wherein cropping the depth map to obtain a depth map having the absolute depth value comprises:

Cutting the depth map by using the calibration plate to obtain a depth map with the absolute depth value; the size of the cut depth map is the same as that of the calibration plate.

3. The method of claim 1, wherein fitting the depth map having the absolute depth values to obtain a fit plane comprises:

and fitting the depth map with the absolute depth value by using a RANSAC algorithm to obtain a fitting plane.

4. The method of claim 1, wherein the correcting for tilt errors based on the fitted plane obtains an accurate depth map of the asphalt pavement, comprising:

obtaining a normal vector n of the fitting plane;

obtaining an accurate depth map of the asphalt pavement through an inclination correction formula;

the inclination correction formula is as follows:

Wherein Z' represents an accurate depth map of the asphalt pavement, T represents a coordinate transformation matrix, X, Y represents pixel coordinates in the depth map, Z represents an absolute depth value of the depth map, R represents a rotation matrix, Θ represents the angle between normal vector n and Z axis,/> T represents a three-dimensional translation vector,/>Representing a 3 x1 transpose of the zero vector.

5. The method of claim 1, wherein the resolution of the images at the plurality of view angles is 3024 x 3024 and n has a value of 6.