Pavement flatness and compactness visual detection device carried on road roller platform

Abstract

The invention relates to a road surface flatness and compactness visual detection device carried on a road roller platform, which comprises a binocular camera and a mounting table for mounting the binocular camera, and the mounting table is fixed on the road roller platform; the controller is connected with the binocular camera and is configured to execute the following steps: S1, acquiring a binocular image of a road surface acquired by the binocular camera, and correcting based on a calibration parameter to obtain a corrected binocular image; s2, obtaining a depth map based on the corrected binocular image, and obtaining the flatness of the road surface based on the depth map; s3, judging whether the flatness of the road surface is within a pre-configured first flatness range or not, and if yes, executing the step S4; s4, the particle size is calculated based on the plane graph of the vertical view angle of the road surface, and second particle grading information is obtained; and S5, calculating the difference between the second grain composition information and the original grain composition information, and determining the compactness in combination with a measurement proportion obtained based on the aggregate information. Compared with the prior art, the method has the advantages that under the condition that high computing power and training samples are not needed, the flatness and the compactness can be synchronously measured while compaction operation is carried out in the pavement compaction operation process.

Description

Technical Field

[0001] The invention relates to the field of roadside flatness and compactness measurement, and in particular to a road surface flatness and compactness visual detection device mounted on a roller platform. Background Art

[0002] Binocular vision is a 3D imaging technology that mimics the principle of human binocular vision. It uses two cameras to obtain the left and right views of the scene and uses the parallax in the image to calculate the distance between the object and the camera. This method can generate depth information of the scene in real time and is widely used in fields that require the perception of spatial structure, such as autonomous driving, robot navigation, and 3D reconstruction.

[0003] During the operation of a road roller, it is often necessary to manually judge the flatness and density of the road surface that has been operated. The current flatness detection methods include: a three-meter ruler with a feeler gauge, an eight-wheel meter, a laser flatness meter, a bump accumulation meter, etc. The three-meter ruler is a classic detection technology. The eight-wheel meter and the laser flatness meter both use the detection principle of a three-meter ruler. The current density detection methods include: core sampling method, acoustic wave sensor method, and vibration density meter. The current detection method relies on manual professionalism, the equipment operation is complicated, and the detection efficiency is low. The present invention aims to propose an efficient and fast detection method to detect the flatness and density of the road surface after the roller operation in real time, enabling intelligent construction.

[0004] With the introduction of intelligent construction, the road surface detection technology of rollers is also constantly innovating and developing. Traditional methods are time-consuming and have a certain dependence on the environment. In order to improve the detection efficiency and automation level, the innovative binocular vision detection technology installs a binocular camera at the rear of the roller, and dynamically analyzes the flatness and compaction effect of the road surface by obtaining the three-dimensional depth information of the road surface after compaction. Combined with image processing and intelligent algorithms, this technology can monitor and evaluate the quality of the road surface in real time, significantly reduce human intervention, and improve the accuracy and efficiency of detection. At the same time, the binocular vision system has the advantage of non-contact and can adapt to complex construction environments, which has a positive role in promoting intelligent construction.

[0005] For example, Chinese patent CN118375034A discloses a large-scale road surface smoothness three-dimensional detector, which involves the field of road surface smoothness detection, including a binocular line structured light measurement module and a display control component, wherein the binocular line structured light measurement module is installed on the rear side of the roller, and the display control component is installed in the cab of the roller; the binocular line structured light measurement module is used to reconstruct the three-dimensional point cloud of the road surface after rolling in real time, and calculate the international smoothness index based on the three-dimensional point cloud; the binocular line structured light measurement module includes a line laser, a binocular industrial camera, a posture sensor, a speed encoder and a data processor; the display control component is used to display the three-dimensional information of the road surface and the international smoothness index. However, the above detector can only detect smoothness, not density.

[0006] However, if the flatness and density are detected in stages, the detection process will remain unchanged. In this regard, technical personnel in this field may choose to combine some density detection schemes that are also based on vision. For example, Chinese patent CN114821175A discloses a filler density detection method, system, electronic equipment and storage medium, including: imaging analysis of compacted road sections to obtain multiple shooting blocks and their numbers corresponding to the compacted road sections; acquiring images at multiple shooting blocks to obtain a road surface image set; using a pre-built road surface image recognition model to identify the road surface image set, and extracting abnormal images from the road surface image set according to the recognition results; and presenting the abnormal category corresponding to the abnormal image and the number of the corresponding shooting block. Alternatively, Chinese patent CN117576563A discloses an intelligent detection method and system for the density of stone-filled roadbed based on image processing, including: obtaining the corresponding relationship between different gradations, dosages and crushed stone aggregate density through experiments; extracting image indicators that match the density of the stone-filled roadbed; converting the parameters corresponding to the gradation, dosage and image indicators into vectors, and using a fully connected neural network to assign different weights to each parameter in the vector to form an n×n matrix; using a convolutional neural network to process the matrix to obtain a stone-filled roadbed density detection model; obtaining the gradation, dosage information and surface image information of the stone-filled roadbed to be tested, and using the stone-filled roadbed density detection model to obtain the stone-filled roadbed density detection result.

[0007] However, no matter which of the above methods is used, they are overly dependent on the number of sample sets and extremely high computing power. At present, there is not enough financial support for the detection of road compaction operations to meet the sample data and computing power requirements. Therefore, how to achieve the simultaneous measurement of flatness and density during the road compaction operation without the need for higher computing power and training samples is a technical problem that technical personnel in this field are eager to solve. Summary of the invention

[0008] The purpose of the present invention is to provide a visual detection device for road surface flatness and density mounted on a roller platform.

[0009] The purpose of the present invention can be achieved by the following technical solutions:

[0010] A road surface flatness and compactness visual detection device mounted on a roller platform, comprising:

[0011] A binocular camera and a mounting platform for mounting the binocular camera, wherein the mounting platform is fixed on the roller platform;

[0012] The controller, connected to the binocular camera, is configured to perform the following steps:

[0013] Step S1: obtaining a binocular image of a road surface captured by a binocular camera, and performing correction based on calibration parameters to obtain a corrected binocular image;

[0014] Step S2: obtaining a depth map based on the corrected binocular image, and obtaining the flatness of the road surface based on the depth map;

[0015] Step S3: determining whether the flatness of the road surface is within a pre-configured first flatness range, if yes, executing step S4;

[0016] Step S4: calculating the particle size based on the plane view of the road surface at a vertical viewing angle, and obtaining second particle grading information;

[0017] Step S5: Calculate the difference between the second particle gradation information and the original particle gradation information, and determine the density in combination with the metric ratio obtained based on the aggregate information.

[0018] The step S2 comprises:

[0019] Step S2-1: Based on the corrected binocular image, key points are matched by key point matching, and based on the key positions in the corrected binocular image, a disparity map is calculated;

[0020] Step S2-2: Filling holes in the obtained disparity map using a neighborhood interpolation method, and performing depth calculation based on the disparity principle to obtain a depth map;

[0021] Step S2-3: Generate three-dimensional point cloud data according to the depth map, and perform a road surface flatness analysis on the three-dimensional point cloud data to obtain the road surface flatness.

[0022] The plan view of the road surface at a vertical viewing angle is obtained through the corrected binocular image.

[0023] The step S4 comprises:

[0024] Step S4-1: using a histogram equalization method to process the vertical viewing angle plan view of the road surface to obtain a first intermediate image to increase contrast;

[0025] Step S4-2: After filtering the first intermediate image, a threshold segmentation method is used to segment the image to obtain particles in the image;

[0026] Step S4-3: After performing morphological processing on the particles in the image to remove burrs and isolated spots on the particle surface, the particle size of each particle is calculated;

[0027] Step S4-4: obtaining second particle grading information based on the proportions of particles with different particle sizes.

[0028] The step S4-4 specifically includes:

[0029] Step S4-4-1: sort all particles from small to large in size and divide them into 10 equal parts;

[0030] Step S4-4-2: The minimum value of the particle size, the particle size of each segmentation point, and the maximum value of the particle size are sequenced as the second particle grading information.

[0031] The original particle gradation information is obtained based on the original gradation of the pavement filler, and the acquisition process includes:

[0032] Draw a distribution histogram based on the original gradation of the pavement filler and divide it into 10 equal parts;

[0033] The sequence of the minimum particle size, the particle size of each segmentation point, and the maximum particle size is taken as the original particle grading information.

[0034] The mounting platform comprises:

[0035] A first support frame is fixed to the roller platform through a plurality of pillars;

[0036] The second steering gear is fixed on the first support frame, and the output shaft is vertically arranged;

[0037] A base plate and a first steering gear arranged on the base plate, wherein the base plate is arranged horizontally, located above the first support frame and the bottom of the base plate is connected to the output shaft of the second steering gear, wherein the output shaft of the first steering gear and the output shaft of the second steering gear are arranged vertically;

[0038] A camera bracket and an acceleration sensor arranged on the camera bracket, wherein the acceleration sensor and the binocular camera are both fixed on the camera bracket and move synchronously with the movement of the camera bracket, and the camera bracket is connected to the output shaft of the first servo.

[0039] The first support frame comprises:

[0040] a first mounting plate, wherein the second steering gear is mounted on the first mounting plate;

[0041] The second supporting plate has a through hole at the center thereof for the output shaft of the second steering gear to pass through.

[0042] The mounting platform further comprises a mounting seat, the mounting seat is arranged on the base plate, and the first steering gear is arranged on the base plate through the mounting seat.

[0043] The controller is connected to the second steering engine, the first steering engine and the acceleration sensor, and is configured to further perform the following steps:

[0044] Obtain vibration data collected by the acceleration sensor and analyze it to obtain the frequency, amplitude and direction of the vibration;

[0045] The compensation angle is calculated according to the frequency, amplitude and direction of the vibration, and the compensation angle is decomposed into the heading angle and the azimuth angle, and the actions of the second servo and the first servo are controlled respectively.

[0046] Compared with the prior art, the present invention has the following beneficial effects:

[0047] 1. Based on the use of binocular cameras to obtain depth information to detect the flatness of the road surface, and after the flatness detection is passed, the particle sizes of the particles are calculated based on the plan view of the vertical perspective of the road surface, and the second particle grading information of the actual road surface is extracted. The difference between the second particle grading information and the corresponding original particle grading information is combined with the measurement ratio to obtain the density. It has the advantage of being able to intuitively reflect the density. It does not require a large sample set and powerful computing power to achieve edge computing, so that without the need for high computing power and training samples, it can achieve compaction while simultaneously measuring the flatness and density during the road compaction operation.

[0048] 2. Traditional methods can usually only detect flatness or density separately. The present invention uses the same system to simultaneously detect the flatness and density of the road surface, which simplifies the detection process, improves efficiency, and realizes the reuse of detection equipment.

[0049] 3. The designed mounting platform can compensate for the impact of the roller body vibration on the camera during operation, and can stably collect high-quality images during roller operation.

[0050] 4. The original particle grading information in the form of a sequence is obtained by using a particle size distribution histogram, and the second particle grading information is obtained by equal division. The lengths of the two are equal, which makes it easy to calculate the difference, reduces the requirements for computing power, and can obtain the accurate difference between the two. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 is a schematic diagram of a roller platform carried by the present invention;

[0052] Figure 2 It is a structural schematic diagram of the present invention;

[0053] Figure 3 is a schematic diagram of a first support frame;

[0054] Figure 4 This is a schematic diagram of the camera bracket;

[0055] Figure 5 It is a flow chart of road surface density detection of the present invention;

[0056] Figure 6 It is a point cloud effect diagram of the road surface flatness of the present invention;

[0057] Figure 7 This is a road surface density detection effect diagram of the present invention;

[0058] Figure 8 It is a schematic diagram of the main steps of the present invention;

[0059] Wherein: 1. first support frame, 2. second servo, 3. base plate, 4. first servo, 5. camera bracket, 6. acceleration sensor, 7. binocular camera, 8. mounting seat, 1-1. first mounting plate, 1-2. second support plate, 3-1. roller table, 5-1. first connecting piece, 5-2. first connecting plate, 5-3. second connecting plate, 9-1. driving gear, 9-2. driven gear, 9-3. rotating shaft, 101. roller platform, 102. detection device. DETAILED DESCRIPTION

[0060] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0061] A visual inspection device for road surface flatness and compactness mounted on a roller platform, such as Figure 1 As shown, including:

[0062] A binocular camera and a mounting platform for mounting the binocular camera, wherein the mounting platform is fixed on the roller platform;

[0063] The controller is connected to the binocular camera to obtain images taken by the binocular camera, etc.

[0064] like Figure 2 As shown, in most embodiments, the mounting station includes:

[0065] A first support frame 1 is fixed to the roller platform through a plurality of pillars;

[0066] The second steering gear 2 is fixed on the first support frame 1, and the output shaft is arranged vertically;

[0067] A base plate 3 and a first steering gear 4 arranged on the base plate 3, wherein the base plate 3 is arranged horizontally, located above the first support frame 1 and the bottom of the base plate 3 is connected to the output shaft of the second steering gear 2, wherein the output shaft of the first steering gear 4 and the output shaft of the second steering gear 2 are arranged vertically;

[0068] The camera bracket 5 and the acceleration sensor 6 arranged on the camera bracket 5 , the acceleration sensor 6 and the binocular camera 7 are all fixed on the camera bracket 5 and move synchronously with the movement of the camera bracket 5 , and the camera bracket 5 is connected to the output shaft of the first servo 4 .

[0069] like Figure 3 As shown, in this embodiment, the first support frame 1 includes:

[0070] A first mounting plate 1-1, on which the second steering gear 2 is mounted;

[0071] The second support plate 1 - 2 has a through hole at its center for the output shaft of the second steering gear 2 to pass through.

[0072] In addition, the mounting platform further includes a mounting seat 8, which is disposed on the substrate 3, and the first steering gear 4 is disposed on the substrate 3 through the mounting seat 8. Furthermore, the controller connects the second steering gear 2, the first steering gear 4 and the acceleration sensor 6, obtains the vibration data collected by the acceleration sensor 6, and analyzes the frequency, amplitude and direction of the vibration; calculates the compensation angle according to the frequency, amplitude and direction of the vibration, and decomposes the compensation angle into the heading angle and azimuth angle, and controls the actions of the second steering gear 2 and the first steering gear 4 respectively.

[0073] In this embodiment, the controller is specifically a host computer, the camera bracket 5 and the base plate 3 constitute an active adjustable bracket, and the binocular camera specifically adopts a pair of industrial cameras, which are responsible for collecting road surface images after the roller operation and transmitting them to the host computer for analysis in real time. The active adjustable bracket can flexibly adjust the angle and spacing of the camera. The host computer is responsible for real-time analysis of the collected image information, and analyzes the flatness by calculating the depth information of the road surface according to the binocular stereoscopic vision principle, and analyzes the road surface density according to image processing technology. In addition, a two-dimensional turntable is composed of the second steering gear 2 and the first steering gear 4 as the main components, and an active vibration isolation system is formed by combining the acceleration sensor 6, etc. The acceleration sensor 6 is installed on the camera bracket 5 and is responsible for collecting the vibration acceleration of the camera during the roller operation. The host computer is responsible for real-time analysis of the camera vibration acceleration, calculating the compensation value, and sending control instructions to the two steering gears of the two-dimensional turntable. The two-dimensional turntable is responsible for receiving the control instructions of the host computer and executing the corresponding rotation action to perform active vibration isolation to offset the impact of the vibration of the roller operation on the camera.

[0074] The industrial camera in this embodiment has anti-vibration performance and high frame rate characteristics, which can reduce the impact of vibration on image quality during the operation of the roller, improve the stability and accuracy of image acquisition, and thus ensure the reliability of working on uneven roads. The movable adjustable bracket is equipped with a multi-degree-of-freedom motion structure, allowing the operator to adjust the camera angle according to different road conditions and environmental conditions to obtain the best image acquisition angle and ensure the accuracy of road surface detection.

[0075] In addition, the system of the present application can automatically record and save the analysis results and pavement density parameters, generate a real-time density distribution map, and transmit it to the cloud via a wireless network to facilitate real-time monitoring and remote management of the construction site.

[0076] In this embodiment, the main body of the base plate 3 is rectangular, and roller tables 3-1 are arranged on the two wide sides. The long side direction of the rectangle is perpendicular to the visual axis of the camera 7. The camera bracket 5 includes a first connecting member 5-1, a first connecting plate 5-2 and a second connecting plate 5-3. In addition, the mounting platform also includes a transmission mechanism, which includes a driving gear 9-1, a driven gear 9-2 and a rotating shaft 9-3. The two ends of the rotating shaft 9-3 are rotatably mounted in the two roller tables 3-1 respectively. The axis of the rotating shaft 9-3 is perpendicular to the visual axis of the camera. The driving gear 9-1 is arranged on the output shaft of the first servo 4 and is connected to the driven gear 9-2. Meshing, the driven gear 9-2 is arranged on the rotating shaft 9-3 and is coaxially arranged with the rotating shaft 9-3. The two first connecting members 5-1 have the same shape and are both arranged on the rotating shaft 9-3. The two ends of the first connecting plate 5-2 are respectively fixed on the two first connecting members 5-1. There are two second connecting plates 5-3, which correspond to two industrial cameras respectively. Each industrial camera is connected to the corresponding second connecting plate 5-3. The two second connecting plates 5-3 are both installed on the first connecting plate 5-2. In addition, the acceleration sensor 6 is installed on the first connecting plate 5-2 and is located in the middle of the two second connecting plates 5-3. In the process of density detection, the light reflected by the surface in the image, the texture characteristics, the depth information, etc. are crucial to the analysis algorithm. If the camera vibrates, it may cause unnecessary dynamic blur in the image acquisition process, thereby affecting the judgment of the image processing algorithm on the road density. The above design ensures the stability of the camera by accurately monitoring and compensating for the vibration of the camera, thereby improving the accuracy of the image-based density analysis. By placing the accelerometer on the camera bracket, the actual vibration of the camera can be monitored more accurately, so that the steering system can compensate for these vibrations more effectively. If the accelerometer is installed on the vehicle, the servo may misjudge the vibration state of the camera due to the difference in synchronization between the vehicle vibration and the camera vibration, affecting the compensation effect. The accelerometer is directly installed on the camera bracket, which can feedback the movement state of the camera more quickly. The steering system can respond quickly and adjust the camera's posture in time to avoid image distortion caused by vibration. This quick response is especially important for quickly and accurately analyzing image data in density detection.

[0077] In addition, if Figures 5 to 8 As shown, the controller is configured to perform the following steps:

[0078] Step S1: obtaining a binocular image of a road surface captured by a binocular camera, and performing correction based on calibration parameters to obtain a corrected binocular image;

[0079] Step S2: obtaining a depth map based on the corrected binocular image, and obtaining the flatness of the road surface based on the depth map;

[0080] Step S3: determining whether the flatness of the road surface is within a pre-configured first flatness range, if yes, executing step S4;

[0081] Step S4: calculating the particle size based on the plane view of the road surface at a vertical viewing angle, and obtaining second particle grading information;

[0082] Step S5: Calculate the difference between the second particle gradation information and the original particle gradation information, and determine the density in combination with the metric ratio obtained based on the aggregate information.

[0083] Based on the use of binocular cameras to obtain depth information to detect the flatness of the road surface, and after the flatness detection is passed, the particle sizes of the particles are calculated based on the plan view of the vertical perspective of the road surface, and the second particle grading information of the actual road surface is extracted. The difference between the second particle grading information and the corresponding original particle grading information is combined with the measurement ratio to obtain the density. This has the advantage of being able to intuitively reflect the density, and edge computing can be achieved without a large sample set and powerful computing power. Therefore, without the need for high computing power and training samples, the road surface can be compacted while the flatness and density are measured synchronously.

[0084] In this embodiment, the field of view needs to be adjusted first. Specifically, the movable camera bracket needs to be adjusted according to the field of view angle, resolution, working environment and other requirements of the imaging system so that the field of view of the industrial camera covers the road surface rolled by the roller. After adjusting the camera angle, the camera is calibrated using a chessboard to obtain the camera's internal and external parameters.

[0085] In addition, step S2 comprises:

[0086] Step S2-1: Based on the corrected binocular image, key points are matched by key point matching, and based on the key positions in the corrected binocular image, a disparity map is calculated;

[0087] Step S2-2: Filling holes in the obtained disparity map using a neighborhood interpolation method, and performing depth calculation based on the disparity principle to obtain a depth map;

[0088] Step S2-3: Generate three-dimensional point cloud data according to the depth map, and perform a road surface flatness analysis on the three-dimensional point cloud data to obtain the road surface flatness.

[0089] In this embodiment, the plan view of the road surface at a vertical viewing angle is obtained through the corrected binocular image. Of course, in other embodiments, the plan view of the road surface at a vertical viewing angle can be directly obtained by adjusting the visual axis of the binocular camera to a vertical state.

[0090] In this embodiment, step S4 includes:

[0091] Step S4-1: Use the histogram equalization method to process the vertical perspective plan view of the road surface to obtain a first intermediate image to increase the contrast, so that the pore part is easier to distinguish from the background, and the mesopores and micropores become more obvious.

[0092] Step S4-2: After filtering the first intermediate image, a threshold segmentation method is used to segment the image to obtain particles in the image;

[0093] Among them, considering that the noise after image enhancement is more similar to salt and pepper noise, the median filter has a slight blur and a high edge retention, so the median filter is adopted.

[0094] The threshold segmentation process is as follows: first, the image is binarized, and then the initial threshold is obtained using the Otsu method. The initial threshold is used to divide the two groups of gray values ​​into Gaussian probability density functions, and the intersection of the two functions is continuously taken as the new threshold until the iterative threshold no longer changes. The image is then threshold segmented using the local adaptive threshold method. The mathematical expression of the average grayscale Δm between classes is as follows:

[0095] Δm=|m1(T)-m2(T)|

[0096]

[0097] Where: m1(T) is the average gray value of the class below the threshold T (foreground class), m2(T) is the average gray value of the class above the threshold T (background class), m1(k) is the weighted average gray value of the foreground class, P(·) is the normalized probability distribution function of gray level i, C1 is the normalization constant of the foreground class, i is the gray level in the image, and C2 is the normalization constant of the background class.

[0098] Step S4-3: After performing morphological processing on the particles in the image to remove burrs and isolated spots on the particle surface, the particle size of each particle is calculated;

[0099] Among them, the morphological processing process mainly fills holes in the binary image, and then performs opening and closing operations to perform morphological processing on the image. According to experiments, the single opening operation has the best effect, which can smooth the image contour and remove burrs and isolated spots. After marking the foreground and background of the image, the watershed algorithm is used to segment the adhesion soil particles.

[0100] The particle size calculation can use the function CC = bwconncomp (K, Conn) to calculate the size of the connected area in the binary image, and then use the function stats1 = regionprops (CC1, 'EquivDiameter') to measure the approximate diameter in each connected area, which represents the particle size of each filler particle in the image.

[0101] Step S4-4: obtaining second particle gradation information based on the proportions of particles with different particle sizes, specifically including:

[0102] Step S4-4-1: sort all particles from small to large in size and divide them into 10 equal parts;

[0103] Step S4-4-2: The minimum value of the particle size, the particle size of each segmentation point, and the maximum value of the particle size are sequenced as the second particle grading information.

[0104] Similarly, the original particle gradation information is obtained based on the original gradation of the pavement filler. The acquisition process includes:

[0105] Draw a distribution histogram based on the original gradation of the pavement filler and divide it into 10 equal parts;

[0106] The sequence of the minimum particle size, the particle size of each segmentation point, and the maximum particle size is taken as the original particle grading information.

[0107] In this embodiment, the second particle size distribution information obtained is:

[0108]

[0109] Where: FV2 is the second particle grading information;

[0110] The original particle size distribution information is:

[0111]

[0112] Among them: FV0 is the second particle grading information,

[0113] The difference between the second particle gradation information and the original particle gradation information is specifically:

[0114] (FV2-FV0) / FV0

[0115] In this way, the density can be obtained by combining the measurement ratio.

[0116] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program code.

Claims

1. A road surface flatness and compactness visual detection device mounted on a roller platform, comprising: A binocular camera and a mounting platform for mounting the binocular camera, wherein the mounting platform is fixed on the roller platform; It is characterized by further comprising: The controller, connected to the binocular camera, is configured to perform the following steps: Step S1: obtaining a binocular image of a road surface captured by a binocular camera, and performing correction based on calibration parameters to obtain a corrected binocular image; Step S2: obtaining a depth map based on the corrected binocular image, and obtaining the flatness of the road surface based on the depth map; Step S3: determining whether the flatness of the road surface is within a pre-configured first flatness range, if yes, executing step S4; Step S4: calculating the particle size based on the plane view of the road surface at a vertical viewing angle, and obtaining second particle grading information; Step S5: Calculate the difference between the second particle gradation information and the original particle gradation information, and determine the density in combination with the metric ratio obtained based on the aggregate information.

2. A road surface flatness and compactness visual inspection device mounted on a roller platform according to claim 1, characterized in that: The step S2 comprises: Step S2-1: Based on the corrected binocular image, key points are matched by key point matching, and based on the key positions in the corrected binocular image, a disparity map is calculated; Step S2-2: Filling holes in the obtained disparity map using a neighborhood interpolation method, and performing depth calculation based on the disparity principle to obtain a depth map; Step S2-3: Generate three-dimensional point cloud data according to the depth map, and perform a road surface flatness analysis on the three-dimensional point cloud data to obtain the road surface flatness.

3. The visual inspection device for road surface flatness and compactness mounted on a road roller platform according to claim 2, characterized in that: The plan view of the road surface at a vertical viewing angle is obtained through the corrected binocular image.

4. The visual inspection device for road surface flatness and compactness mounted on a road roller platform according to claim 2, characterized in that: The step S4 comprises: Step S4-1: using a histogram equalization method to process the vertical viewing angle plan view of the road surface to obtain a first intermediate image to increase contrast; Step S4-2: After filtering the first intermediate image, a threshold segmentation method is used to segment the image to obtain particles in the image; Step S4-3: After performing morphological processing on the particles in the image to remove burrs and isolated spots on the particle surface, the particle size of each particle is calculated; Step S4-4: obtaining second particle grading information based on the proportions of particles with different particle sizes.

5. The visual inspection device for road surface flatness and compactness mounted on a road roller platform according to claim 4, characterized in that: The step S4-4 specifically includes: Step S4-4-1: sort all particles from small to large in size and divide them into 10 equal parts; Step S4-4-2: The minimum value of the particle size, the particle size of each segmentation point, and the maximum value of the particle size are sequenced as the second particle grading information.

6. The visual inspection device for road surface flatness and compactness mounted on a road roller platform according to claim 5, characterized in that: The original particle gradation information is obtained based on the original gradation of the pavement filler, and the acquisition process includes: Draw a distribution histogram based on the original gradation of the pavement filler and divide it into 10 equal parts; The sequence of the minimum particle size, the particle size of each segmentation point, and the maximum particle size is taken as the original particle grading information.

7. The visual inspection device for road surface flatness and compactness mounted on a road roller platform according to claim 1, characterized in that: The mounting platform comprises: A first support frame (1) is fixed on the roller platform via a plurality of pillars; The second steering gear (2) is fixed on the first support frame (1), and the output shaft is arranged vertically; A base plate (3) and a first steering gear (4) arranged on the base plate (3), wherein the base plate (3) is arranged horizontally, located above the first support frame (1) and the bottom is connected to the output shaft of the second steering gear (2), wherein the output shaft of the first steering gear (4) and the output shaft of the second steering gear (2) are arranged vertically; A camera bracket (5) and an acceleration sensor (6) arranged on the camera bracket (5), wherein the acceleration sensor (6) and the binocular camera (7) are both fixed on the camera bracket (5) and move synchronously with the movement of the camera bracket (5), and the camera bracket (5) is connected to the output shaft of the first steering gear (4).

8. The visual inspection device for road surface flatness and compactness mounted on a road roller platform according to claim 7, characterized in that: The first support frame (1) comprises: A first mounting plate (1-1), wherein the second steering gear (2) is mounted on the first mounting plate (1-1); The second support plate (1-2) has a through hole at its center for the output shaft of the second steering gear (2) to pass through.

9. The visual inspection device for road surface flatness and compactness mounted on a road roller platform according to claim 7, characterized in that: The mounting platform also includes a mounting seat (8), the mounting seat (8) is arranged on the base plate (3), and the first steering gear (4) is arranged on the base plate (3) through the mounting seat (8).

10. The visual inspection device for road surface flatness and compactness mounted on a road roller platform according to claim 7, characterized in that: The controller is connected to the second steering engine (2), the first steering engine (4) and the acceleration sensor (6), and is configured to further perform the following steps: Acquire vibration data collected by the acceleration sensor (6), and analyze to obtain the frequency, amplitude and direction of the vibration; The compensation angle is calculated according to the frequency, amplitude and direction of the vibration, and the compensation angle is decomposed into a heading angle and an azimuth angle, and the actions of the second steering engine (2) and the first steering engine (4) are controlled respectively.

Images