Large transport vehicle violation detection method and system based on multi-dimensional data matching

Abstract

The invention belongs to the field of intelligent transportation, and discloses a large transport vehicle violation detection method and system based on multi-dimensional data matching, and the method comprises the steps: carrying out the data segmentation of the data information of the axle load and axle distance of a special multi-axle large vehicle measured by a dynamic weighing system, and forming a one-dimensional discrete data graph, and according to the formed one-dimensional discrete data graph, establishing a mapping relation between the axle load wheelbase and the total load of the large transport vehicle, and judging whether the large transport vehicle is overloaded or not based on the relation. An image of a large transport vehicle is collected through a highway camera, a point cloud model of the large transport vehicle is generated by adopting an SLAM algorithm, geometric parameters of the large transport vehicle are extracted, and the obtained point cloud model is matched with a preset contour template so as to judge whether the large transport vehicle is loaded illegally or not. According to the invention, one-dimensional data and two-dimensional data are fused in detection of overweight and illegal loading of the vehicle, so that violation detection of the large transport vehicle is realized.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent transportation, and in particular relates to a method and system for detecting violations of large-scale transport vehicles based on multi-dimensional data matching. Background Art

[0002] Large cargo transport vehicles are often seen in daily life. Large cargo transport vehicles are particularly common on highways because their particularity poses severe challenges to road safety, bridge load-bearing and traffic management. Strict management of large cargo transport vehicles is one of the important factors in improving road safety and traffic management. This includes detecting whether large cargo transport vehicles are overweight or illegally loaded. At present, traditional detection methods still mainly rely on manual verification and static weighing equipment to detect them, but traditional methods are inefficient and consume too much manpower and material resources. The use of dynamic weighing systems and visual detection to detect whether there are overweight and illegal loading behaviors can save manpower and material resources and greatly improve efficiency.

[0003] If the large transport trucks driving on the highway are overweight, it will pose a considerable threat to road safety and bridge load-bearing capacity. If the large transport trucks do not load the goods in accordance with regulations, if an accident occurs, it will cause great harm to other drivers driving on the road, shorten the service life of the bridge and increase the maintenance cost. Summary of the invention

[0004] In order to solve the problems existing in the prior art, the present invention provides a method and system for detecting violations of large-scale transport vehicles based on multi-dimensional data matching. The total load is calculated based on the relationship between the wheelbase and the number of axles measured by the dynamic weighing system to determine whether there is overweight behavior. The image information obtained by the camera is processed to obtain its point cloud model, and its outer contour is compared with the preset contour to see whether there is illegal loading behavior. The center of the vehicle in the image information is compared with the center of the road. If the difference is greater than the specified threshold, it means that there is a violation.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A method for detecting violations of large-scale transport vehicles based on multi-dimensional data matching, the method comprising:

[0007] The data information of the axle load and wheelbase of the multi-axle heavy-duty vehicle measured by the dynamic weighing system is segmented to form a one-dimensional discrete data graph, and a mapping relationship between the axle load and wheelbase of the heavy-duty transport vehicle and the total load is established according to the one-dimensional discrete data graph, and whether the heavy-duty transport vehicle is overweight is judged based on the mapping relationship;

[0008] Generate a point cloud model of the large-piece transport vehicle through the SLAM algorithm for the collected images of the large-piece transport vehicle. According to the point cloud model, extract the geometric parameters of the large-piece transport vehicle and the goods. According to the geometric parameters, identify whether the center offset of the large-piece transport vehicle exceeds the threshold;

[0009] Based on the overweight detection result and the illegal loading detection result of the large-piece transport vehicle, achieve the illegal detection of the large-piece transport vehicle.

[0010] Preferably, perform data segmentation processing on the data information of the axle load and axle distance of the extra-multi-axle large-piece vehicle measured by the dynamic weighing system to form a one-dimensional discrete data graph. According to the one-dimensional discrete data graph, establish a mapping relationship between the axle load and axle distance of the large-piece transport vehicle and the total load. Based on the mapping relationship, judge whether the large-piece transport vehicle has overweight behavior, including:

[0011] Process the one-dimensional information of the axle load and axle distance of the large-piece transport vehicle collected by the dynamic weighing system. Set the axle load as the ordinate y-axis and the axle distance as the abscissa x-axis, so as to perform one-dimensional data segmentation on the information collected by the dynamic weighing system;

[0012] Plot the information collected by the dynamic weighing system and processed in the one-dimensional discrete data image, and mark the discrete points with different positions with different symbols and plot them on the discrete data graph. Each different vehicle corresponds to a different interval block;

[0013] The number of interval blocks in the one-dimensional discrete data graph represents the number of large-piece transport vehicles, and the number of axles of the vehicle represents the number of discrete points in the interval block. Sum the quantities of all discrete points in each interval block along the y-axis direction, which is the total load of each extra-multi-axle large-piece vehicle. Then compare the obtained total load of the vehicle with the predetermined load to judge whether there is overweight behavior.

[0014] Preferably, generating a point cloud model of the large-piece transport vehicle through the SLAM algorithm for the collected images of the large-piece transport vehicle includes:

[0015] Let be the camera coordinate system, and the direction pointing to the optical center of the plane is the positive front of the Z-axis, be the optical center of the camera, and the spatial target point is projected through the optical center and falls on the physical imaging plane , and the imaging point is . Let the coordinate of the spatial point be , and the point projected on the plane is . Let the distance from the physical imaging plane to the small hole be ; According to the similarity relationship of triangles, we have:

[0016] ;

[0017] According to the similarity relationship of triangles, the spatial relationship between the spatial point and the corresponding pixel point is obtained. According to the spatial relationship between the spatial point and the corresponding pixel point, in the camera, what is finally obtained is the position of the pixel point;

[0018] Among them, according to the spatial relationship between the spatial point and the corresponding pixel point, in the camera, what is finally obtained is the position of the pixel point, including:

[0019] Set a fixed pixel plane on a physical imaging plane , and the pixel coordinates obtained on the pixel plane are ;

[0020] The definition method of the pixel coordinate system is: the origin is located at the upper left corner of the image, the x-axis direction is to the right, parallel to the x-axis direction, the y-axis direction is downward, parallel to the y-axis direction; there is a scaling factor and a translation of the origin between the pixel coordinates and the imaging plane; assuming that the pixel is scaled by times on the x-axis, and scaled by times on the y-axis, and the origin is translated by , then the coordinates of and the pixel coordinates

[0021] ;

[0022] According to the similarity relationship of triangles and the relationship between the coordinates of and the pixel coordinates ;

[0023] Then the homogeneous coordinate expression of the pixel plane coordinate system is:

[0024] ;

[0025] In the formula, the matrix composed of the intermediate quantities is called the internal parameter matrix of the camera. The internal parameter matrix projects the position of the reference point from the position in the camera's three-dimensional coordinate system to the position in the camera's internal pixel coordinate system, , , are the coordinates of the pixel coordinate system, , , are the coordinates of the spatial coordinate system; , is the translation of the origin relative to the center, .

[0026] Preferably, according to the point cloud model, geometric parameters of the large-piece transport vehicle and the cargo are extracted, including:

[0027] Taking any point in the point cloud model of the large-piece transport vehicle as the origin, the vertical direction corresponding to the vehicle body of the large-piece transport vehicle is the x-axis, the vehicle body direction is the y-direction, and the vertical direction is the z-direction to establish a coordinate axis;

[0028] According to the coordinate axis, the length, width, and height of the corresponding large-piece transport vehicle are obtained as follows: ; ; ; where are respectively the maximum values of all points in the point cloud in the x, y, and z directions are respectively the minimum values of all points in the point cloud in the x, y, and z directions; when the difference between the maximum and minimum values in the x direction is taken, the value obtained is the width of the large-piece transport vehicle, when the difference between the maximum and minimum values in the y direction is taken, the value obtained is the length of the large-piece transport vehicle, and when the difference between the maximum and minimum values in the z direction is taken, the value obtained is the height of the large-piece transport vehicle.

[0029] Preferably, according to the geometric parameters, it is identified whether the center offset of the large-piece transport vehicle exceeds the threshold, including:

[0030] Taking any point in the point cloud model of the large-piece transport vehicle as the origin, with the vehicle body direction as the y-axis and the direction perpendicular to the vehicle body as the x-axis to establish a coordinate system, the center position of the large-piece transport vehicle is: , where , correspond to the maximum and minimum values in the corresponding coordinate system respectively;

[0031] Identifying the value corresponding to the center of the highway;

[0032] Taking the difference between the identified center of the large-piece transport vehicle and the center of the highway and comparing the difference with a predetermined threshold:

[0033] Whether |μ' - μ| is less than or equal to σ, and when |μ' - μ| > σ, a serious offset phenomenon occurs in the large-piece transport vehicle, then the large-piece transport vehicle has a violation, where μ' is the actual center, μ is the theoretical center, and σ is the specified threshold.

[0034] Preferably, based on the overweight detection result of the large-piece transport vehicle and the illegal loading detection result of the large-piece transport vehicle, the illegal detection of the large-piece transport vehicle is realized, including:

[0035] The illegal detection of the large-piece transport vehicle consists of two parts: overweight detection and illegal vehicle loading. Among them, overweight detection is one-dimensional data detection, while illegal vehicle loading is the detection of two-dimensional images of the large-piece transport vehicle. By matching the one-dimensional and two-dimensional data, relevant information of the large-piece transport vehicle is obtained to determine whether there is any illegal behavior of the large-piece transport vehicle.

[0036] The present invention also provides an illegal detection system for large-piece transport vehicles based on multi-dimensional data matching. The system is used to implement the foregoing method. The system includes: an overweight detection module for large-piece transport vehicles, an illegal loading detection module for large-piece transport vehicles, and an illegal detection module for large-piece transport vehicles;

[0037] The overweight detection module for large-piece transport vehicles is used to perform data segmentation processing on the data information of the axle load and axle distance of the extra-axle large-piece vehicle measured by the dynamic weighing system to form a one-dimensional discrete data map. Based on the one-dimensional discrete data map, a mapping relationship between the axle load and axle distance of the large-piece transport vehicle and the total load is established, and based on the mapping relationship, it is determined whether the large-piece transport vehicle has an overweight behavior;

[0038] The illegal loading detection module for large-piece transport vehicles is used to generate a point cloud model of the large-piece transport vehicle through the SLAM algorithm for the collected images of the large-piece transport vehicle. Based on the point cloud model, the geometric parameters of the large-piece transport vehicle and the goods are extracted, and based on the geometric parameters, it is identified whether the center offset of the large-piece transport vehicle exceeds the threshold;

[0039] The illegal detection module for large-piece transport vehicles is used to realize the illegal detection of large-piece transport vehicles according to the overweight detection result of the large-piece transport vehicle and the illegal loading detection result of the large-piece transport vehicle.

[0040] Compared with the prior art, the beneficial effects of the present invention are:

[0041] The overweight and illegal loading detection system for large-piece goods transport vehicles proposed by the present invention can better manage large-piece transport vehicles on highways. Compared with the previous manual inspection of whether they are overloaded, the present invention is more efficient, saves time, and saves a part of the labor cost.

[0042] The present invention improves the safety factor of the road, which is also a manifestation of being responsible for the life safety of other drivers driving on the same highway, and can well prevent the danger of certain bridges collapsing due to the overweight of large-piece transport vehicles. It also increases the service life of highways and bridges and reduces the expenditure on maintenance costs. Description of the Drawings

[0043] To more clearly illustrate the technical solution of the present invention, the accompanying drawings required for use in the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other accompanying drawings can be obtained based on these drawings without creative efforts.

[0044] Figure 1 Schematic diagram of example measurement for an embodiment of the present invention;

[0045] Figure 2 Flowchart of the method for detecting overweight and illegal loading of large goods transportation vehicles in an embodiment of the present invention;

[0046] Figure 3 Schematic diagram of the flow of a method for detecting illegalities of large goods transportation vehicles based on multi-dimensional data matching in an embodiment of the present invention. Detailed implementation manners

[0047] The following will clearly and completely describe the technical solution in the embodiments of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0048] To make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific implementation manners.

[0049] Embodiment 1

[0050] As Figure 3 shown, the embodiment of the present invention provides a method for detecting illegalities of large goods transportation vehicles based on multi-dimensional data matching. The method includes:

[0051] Perform data segmentation processing on the data information of the axle load and wheelbase of a multi-axle large goods vehicle measured by a dynamic weighing system to form a one-dimensional discrete data graph. Based on the one-dimensional discrete data graph, establish a mapping relationship between the axle load and wheelbase of the large goods transport vehicle and the total load, and determine whether the large goods transport vehicle has overweight behavior based on the mapping relationship;

[0052] Generate a point cloud model of the large goods transport vehicle through the SLAM algorithm for the image of the large goods transport vehicle collected by the camera. Based on the point cloud model, extract the geometric parameters of the length, width, height, and cargo of the large goods transport vehicle, and identify whether the center offset of the large goods transport vehicle exceeds a threshold according to the geometric parameters;

[0053] Based on the overweight detection results of large-piece transport vehicles and the illegal loading detection results of large-piece transport vehicles, the illegal detection of large-piece transport vehicles is realized.

[0054] In this embodiment, data segmentation processing is performed on the data information of the axle load and axle distance of a special multi-axle large-piece vehicle measured by a dynamic weighing system to form a one-dimensional discrete data graph. Based on the one-dimensional discrete data graph, a mapping relationship between the axle load and axle distance of a large-piece transport vehicle and the total load is established. Based on the mapping relationship, it is determined whether a large-piece transport vehicle has overweight behavior, including:

[0055] Process the one-dimensional information of the axle load and axle distance of the large-piece transport vehicle collected by the dynamic weighing system. Classify and sort out the disordered and chaotic information about the axle load and axle distance of the large-piece transport vehicle obtained by the dynamic weighing system. Set the axle load as the vertical coordinate y-axis and the axle distance as the horizontal coordinate x-axis, so as to perform one-dimensional data segmentation on the disordered and chaotic information collected in the dynamic weighing system.

[0056] Plot the information collected by the dynamic weighing system and already processed in a one-dimensional discrete data image, and mark the discrete points with different positions with different symbols and plot them on the discrete data graph. The discrete points with different position relationships are the points in different positions in the data graph. Each different vehicle corresponds to a different interval block, and the interval blocks are divided according to different vehicles. And the points in each interval block correspond to the points after data segmentation processing.

[0057] In the one-dimensional discrete data graph, the number of interval blocks represents the number of large-piece transport vehicles, and the number of axles of the vehicle represents the number of discrete points in the interval block. Sum the number of all discrete points in each interval block along the y-axis direction, which is the total load of each special multi-axle large-piece vehicle. Let the total load be The load corresponding to each axle is Then ( ) and then compare the obtained total vehicle load with the specified load to determine whether there is overweight behavior.

[0058] In this embodiment, a point cloud model of a large-piece transport vehicle is generated for the collected large-piece transport vehicle image through the SLAM algorithm, including: converting the two-dimensional image information of the large-piece transport vehicle obtained by the camera through the SLAM algorithm, so as to convert the obtained information into the point cloud model of the large-piece transport vehicle, thereby obtaining the two-dimensional data information of the large-piece transport vehicle. Specifically:

[0059] Select the key frames with large amounts of information and reduce redundant information. To generate the point cloud model of the large-piece transport vehicle, the motion between multiple frames of images is required to infer its three-dimensional structure and then generate its point cloud model, so as to further obtain the two-dimensional data information of the large-piece transport vehicle.

[0060] The relevant principles and formulas involved are as follows:

[0061] Let be the camera coordinate system, and the direction pointing from the plane to the optical center is the positive front of the Z axis. be the optical center of the camera. In the real world, the spatial target point after passing through the optical center is projected and lands on the physical imaging plane , and the imaging point is . Let the coordinates of the spatial point be , and the point projected onto the plane is . Let the distance from the physical imaging plane to the small hole be (focal length); according to the similarity relationship of triangles, we have:

[0062] (1);

[0063] After organizing the above formula, we can get:

[0064] (2);

[0065] Equation (2) describes the spatial relationship between the spatial point and its pixel point, with the unit of meter (m). In the camera, what is finally obtained is the position of the pixel point. Therefore, it is also necessary to quantize the pixel plane on the imaging plane to describe the process of the sensor receiving light and converting it into pixel points. A fixed pixel plane is set on a physical imaging plane , and the pixel coordinates obtained on the pixel plane are .

[0066] The definition method of the pixel coordinate system is as follows: the origin is located at the upper left corner of the image, the axis points to the right and is parallel to the axis, and the axis points downward and is parallel to the axis. There is a scaling factor and a translation of the origin difference between the pixel coordinates and the imaging plane. Assume that the pixel is scaled by times on the axis and by times on the axis, and the origin is translated by The coordinates and pixel coordinates The relationship is as follows:

[0067] (3);

[0068] Substitute (2) into (3) and combine into Substitute into and combine to obtain:

[0069] (4);

[0070] Then the homogeneous coordinate expression of the pixel plane coordinate system is:

[0071] (5);

[0072] wherein, , , are the coordinates of the pixel coordinate system; , , are the coordinates of the space coordinate system; , is the translation of the origin relative to the center; .

[0073] In this formula, the matrix composed of the intermediate quantities is called the internal parameter matrix of the camera, and this matrix projects the reference point from its position in the camera's three-dimensional coordinate system to its position in the camera's internal pixel coordinate system.

[0074] The two-dimensional data information of the large-piece transport vehicle and the relevant information of the goods it transports are obtained from the information processed by the SLAM algorithm. The outer contour of the obtained large-piece transport vehicle is compared with the predetermined contour of the large-piece transport vehicle to see if its actual contour conforms to the predetermined contour. If not, there is a problem of illegal loading.

[0075] In this embodiment, according to the point cloud model, the geometric parameters of the large-piece transport vehicle are extracted, including:

[0076] From the original image information of the large-piece transport vehicle, its point cloud model is generated after being processed by the SLAM algorithm, and the relevant two-dimensional data information is obtained, so as to obtain the length, width, height of the vehicle and the geometric parameters of the goods from the two-dimensional data information.

[0077] After obtaining the point cloud model of the large-piece transport vehicle, which is composed of millions of points, taking any one of these points as the origin, the axis perpendicular to the vehicle body is the x-axis, the vehicle body direction is the y-axis, and the vertical direction is the z-axis to establish a coordinate axis. Thus, the length, width, and height of the vehicle are as follows: ( , ); ( , ); ( , ) where are the maximum values of all points in the x, y, and z directions in the point cloud respectively, while are the minimum values of all points in the x, y, and z directions in the point cloud respectively. The value obtained by taking the difference between the maximum and minimum values in the x direction is the width of the large-piece transport vehicle. The value obtained by taking the difference between the maximum and minimum values in the y direction is the length of the large-piece transport vehicle. The value obtained by taking the difference between the maximum and minimum values in the z direction is the height of the large-piece transport vehicle. When calculating the geometric parameters of the goods on the large-piece transport vehicle, only need to separate the point cloud model of the corresponding goods from the point cloud model, and take any one of these points as the origin, the axis perpendicular to the vehicle body is the x-axis, the vehicle body direction is the y-axis, and the vertical direction is the z-axis to establish a coordinate axis. Thus, the length, width, and height of the goods are as follows: ( , ); ( , ); ( , ) where are the maximum values of all points in the x, y, and z directions in the point cloud of the separated goods part respectively, while are the minimum values of all points in the x, y, and z directions in the point cloud respectively. The value obtained by taking the difference between the maximum and minimum values in the x direction is the width of the goods. The value obtained by taking the difference between the maximum and minimum values in the y direction is the length of the goods. The value obtained by taking the difference between the maximum and minimum values in the z direction is the height of the goods.

[0078] In this embodiment, according to the geometric parameters, identifying whether the center offset of the large-piece transport vehicle exceeds the threshold includes:

[0079] Identifying the center position of the large-piece transport vehicle from the image information of the large-piece transport vehicle obtained by the camera, and taking the difference between the identified center position of the large-piece transport vehicle and the established road center position. If the difference is greater than the specified threshold, it indicates that the large-piece transport vehicle has violated the regulations.

[0080] To find the center position of the large-piece transport vehicle, simply select any point in the point cloud model of the large-piece transport vehicle as the origin, take the vehicle body direction as the y-axis, and the direction perpendicular to the vehicle body as the x-axis to establish a coordinate system. The center position of the large-piece transport vehicle is: , where , respectively correspond to the maximum and minimum values in the corresponding coordinate system, and half of the sum of the two values is the center position of the large-piece transport vehicle. To find the center of the highway, the coordinate system set when finding the center position of the large-piece transport vehicle can be used, that is, the center of the highway and the center of the large-piece transport vehicle are in the same coordinate system. The two sides of the highway correspond to different x values, and the center value of the highway is: where , respectively correspond to the values on both sides of the highway.

[0081] Calculate the difference between the center of the identified large-piece transport vehicle and the center of the highway, and compare the difference with a predetermined threshold:

[0082] Whether |μ' - μ| is less than or equal to σ. When |μ' - μ| > σ, the large-piece transport vehicle shows a serious deviation phenomenon, and the large-piece transport vehicle has a violation. Here, μ' is the actual center, μ is the theoretical center, and σ is the specified threshold.

[0083] In this embodiment, according to the overweight detection result of the large-piece transport vehicle and the illegal loading detection result of the large-piece transport vehicle, the illegal detection of the large-piece transport vehicle is realized, including:

[0084] The illegal detection of the large-piece transport vehicle consists of two parts: overweight detection and illegal vehicle loading. Among them, overweight detection is one-dimensional data detection, and illegal vehicle loading is the detection of the two-dimensional image of the large-piece transport vehicle. One-dimensional data is the one-dimensional information obtained after processing in overweight detection, and two-dimensional data is the two-dimensional image information of the large-piece transport vehicle obtained through the camera. The total load of the large-piece transport vehicle is obtained from the processed one-dimensional data obtained, and it is judged whether there is an overweight behavior. The center position of the large-piece transport vehicle is obtained from the obtained two-dimensional image information to see if it coincides with the road center, and whether the loaded goods exceed the specified limit to judge whether there is an illegal loading behavior. Make them cooperate with each other to obtain information about whether the large-piece transport vehicle is overweight and whether it is illegally loaded, so as to judge whether there is an illegal behavior.

[0085] Embodiment 2

[0086] As Figure 1As shown in the figure, the present invention discloses a method for detecting violations of large-piece transportation vehicles based on multi-dimensional data matching, including determining whether the large-piece transport vehicle is overloaded and whether there are violations. Among them, determining whether it is overloaded includes the following steps:

[0087] Step 1: Perform one-dimensional data segmentation processing on the axle load and axle distance information of the large-piece vehicle measured by the dynamic weighing system to form a one-dimensional discrete data graph, which turns the originally chaotic and disordered information obtained from the dynamic weighing system into ordered information, which is beneficial to subsequent processing.

[0088] The specific representation of this one-dimensional data segmentation is the one-dimensional original data information of the axle load and axle distance of multiple consecutive special multi-axle large-piece vehicles collected originally. The one-dimensional original data of the axle load and axle distance specifically refers to the one-dimensional form data directly obtained from the dynamic weighing system, where the data of multiple vehicles are mixed and the distribution of the axle load and axle distance is in a disordered state. It is processed so that the axle distances are arranged in an increasing order one by one from left to right, and the axle load is distributed corresponding to each axle distance value based on the one-dimensional original data, thus forming two rows of data with the same dimension and the corresponding distribution of the axle load and axle distance.

[0089] Step 2: Establish a relationship between the one-dimensional data discrete graph obtained after preprocessing and the total load, axle load, and axle distance. Use mathematical expressions to represent the positional relationship between data discrete points, traverse and calculate the discrete points along the x-axis direction. Mark different symbols on the discrete points with different positional relationships and draw them on the one-dimensional data discrete graph.

[0090] The specific process of traversing and calculating the discrete points along the x-axis direction is to traverse along the direction of increasing axle distance values in the two-dimensional data, and use the slope expression to represent the change in the size of the axle load one by one from left to right. Mark different symbols on the discrete points with different positional relationships, and the specific drawing on the one-dimensional data discrete graph is that as the axle distance is distributed, there are different situations for the difference in axle load between adjacent two axles, that is, the value obtained by subtracting the current axle load from the axle load of the next adjacent axle is greater than or equal to 0 or less than a certain specific value. This is to distinguish the two situations and intuitively transform the problem into marking different symbols on data points with different positional relationships and drawing them.

[0091] Step 3: Divide the entire one-dimensional data discrete graph into several interval blocks. Draw a black straight line perpendicular to the x-axis between every two adjacent interval blocks as the dividing line of the interval blocks. The black straight line drawn perpendicular to the x-axis is actually the position for segmenting and identifying the data information of multiple consecutive special multi-axle large-piece transportation vehicles, and it is also for intuitively representing the segmentation and identification results of special multi-axle large-piece vehicles.

[0092] In the one-dimensional data discrete graph, the number of interval blocks represents the number of large-piece transport vehicles, while the number of axles of the vehicle represents the number of discrete points in the interval block. The sum of the quantities of all discrete points within each interval block along the y-axis direction is the total load of each extra-axle large-piece vehicle. Then, the obtained total vehicle load is compared with the specified load to determine whether there is overweight behavior. If there is overloading behavior, it will be identified and recorded; if there is no overloading behavior, it will not be recorded.

[0093] Step 4: Regarding whether there is illegal loading behavior of the vehicle, first, the relevant image information obtained from the camera is processed by the SLAM algorithm to generate the point cloud model of the large-piece transport vehicle, and the original image information is processed by the SLAM algorithm to obtain the two-dimensional data information of the large-piece transport vehicle.

[0094] According to the principle of similar triangles and the (focal length) of the camera, the relationship between the imaging points on the physical imaging plane and the pixel coordinates is obtained. And based on this relationship, the homogeneous coordinate expression of the pixel plane coordinate system is obtained, so as to project the reference point from its position in the camera three-dimensional coordinate system to its position in the camera internal pixel coordinate system.

[0095] Step 5: After the point cloud model of the large-piece transport vehicle is generated by the SLAM algorithm in Step 4, the original image information is transformed into the two-dimensional data information of the large-piece transport vehicle, and then the required information is extracted from the obtained two-dimensional data information.

[0096] In the generated point cloud model, taking any one point as the origin, the direction perpendicular to the vehicle body is the x-axis, the vehicle body direction is the y-axis, and the vertical direction is the z-axis to establish a coordinate axis. Thus, the length, width, and height of the vehicle are respectively: ( , ); ( , ); ( , ) where are respectively the maximum values of all points in the point cloud in the x, y, and z directions, while They are the minimum values of all points in the cloud of the store in the x, y, and z directions, respectively. When the difference between the maximum and minimum values in the x direction is obtained, it is the width of the large-piece transport vehicle. When the difference between the maximum and minimum values in the y direction is obtained, it is the length of the large-piece transport vehicle. When the difference between the maximum and minimum values in the z direction is obtained, it is the height of the large-piece transport vehicle. When calculating the geometric parameters of the goods of the large-piece transport vehicle, only need to separate the point cloud model of the corresponding goods in the point cloud model and repeat the process of calculating the geometric parameters of the large-piece transport vehicle to obtain the geometric parameters of its goods.

[0097] Obtain the geometric parameters such as the length, width, and height of the large-piece transport vehicle and the geometric parameters of its goods, and identify and record the information of the vehicle obtained and the information of the loaded goods.

[0098] Step 6: Generate a corresponding two-dimensional contour from the relevant geometric information of the large-piece transport vehicle obtained and the information of the goods it loads. The relevant two-dimensional contour information generated from the information obtained by the camera is the actual two-dimensional contour of the large-piece transport vehicle. Compare its actual two-dimensional contour with the preset two-dimensional contour of the vehicle. When its actual contour is less than or equal to the preset contour, the large-piece transport vehicle does not have a loading violation. Otherwise, there is a loading violation. At this time, it should be identified and recorded to facilitate finding the relevant vehicle and imposing penalties on it later.

[0099] Step 7: When the camera captures image information, to find the center position of the large-piece transport vehicle, just take any point in the point cloud model of the large-piece transport vehicle as the origin, use the vehicle body direction as the y-axis, and the direction perpendicular to the vehicle body as the x-axis to establish a coordinate system. The center position of the large-piece transport vehicle is: , where , They respectively correspond to the maximum and minimum values in the corresponding coordinate system. Half of the sum of the two values is the center position of the large-piece transport vehicle. The value corresponding to the center of the highway can also be found in the same coordinate system. Identify the edge contour information of the large-piece transport vehicle and the position of its actual center, and record the identified center position.

[0100] Step 8: The center position of the highway is always in the same position. When the large-piece transport vehicle is driving on it, its actual center may not coincide with the road center or even be far apart. At this time, the large-piece transport vehicle has a violation.

[0101] Based on a certain point on the highway as the origin to establish a coordinate system, the center of the road remains unchanged. The actual center of the identified large-piece transport vehicle may change. Obtain the specific value of the actual center of the identified vehicle in the coordinate system and calculate the difference between it and the value corresponding to the road center.

[0102] Step 9: In the coordinate system established with a certain point of the large-piece transport vehicle as the origin, corresponding values can be obtained for the actual center of the large-piece transport vehicle, and corresponding values also exist for its road center. The difference between the two values is calculated, and the absolute value of the obtained difference can be compared with the set threshold. If it is less than the preset threshold, there is no violation; otherwise, there is a violation. And this threshold is related to the outer contour of the large-piece transport vehicle.

[0103] When the large-piece transport vehicle is traveling on the highway, its actual center position is not always the same. When the deviation between its actual center and the road center is not large, there is no violation. Therefore, it is only necessary to ensure that the absolute value of the difference between the actual center of the vehicle and the road center is less than or equal to the threshold, which means there is no violation; otherwise, there is a violation. At this time, it is necessary to identify and record it to facilitate finding the relevant vehicle and imposing penalties and other actions later.

[0104] Using the above method, a new idea for a large-piece transport vehicle violation detection method based on multi-dimensional data matching is established. The flowchart of this method is as Figure 2 shown. This method can improve problems such as the low efficiency of detecting whether a large-piece transport vehicle is overloaded and illegally using labor. It can separately identify whether a large-piece transport vehicle is overloaded and illegally loaded, and can be applied to fields such as transportation and traffic control, having important theoretical significance and great practical application value.

[0105] Embodiment III

[0106] The embodiment of the present invention provides a large-piece transport vehicle violation detection system based on multi-dimensional data matching. The system is used to implement the foregoing method, and the system includes: a large-piece transport vehicle overweight detection module, a large-piece transport vehicle illegal loading detection module, and a large-piece transport vehicle violation detection module;

[0107] The large-piece transport vehicle overweight detection module is used to perform data segmentation processing on the data information of the axle load and axle distance of a special multi-axle large-piece vehicle measured by a dynamic weighing system to form a one-dimensional discrete data graph. Based on the one-dimensional discrete data graph, a mapping relationship between the axle load and axle distance of the large-piece transport vehicle and the total load is established, and based on the mapping relationship, it is judged whether the large-piece transport vehicle has an overweight behavior;

[0108] The large-piece transport vehicle illegal loading detection module is used to generate a point cloud model of the large-piece transport vehicle for the collected image of the large-piece transport vehicle through the SLAM algorithm. According to the point cloud model, the geometric parameters of the large-piece transport vehicle are extracted, and according to the geometric parameters, it is identified whether the center offset of the large-piece transport vehicle exceeds the threshold;

[0109] The oversize transport vehicle violation detection module is used to realize the oversize transport vehicle violation detection according to the overweight detection result of the oversize transport vehicle and the overloading detection result of the oversize transport vehicle.

[0110] The present invention can better manage oversize transport vehicles, reduce the accident rate caused by overweight and overloading, and this method is applicable to the real-time dynamic monitoring and management of oversize transport vehicles in scenarios such as highway toll stations and logistics hubs.

[0111] The above-described embodiments are only descriptions of the preferred embodiments of the present invention, and do not limit the scope of the present invention. Without departing from the design spirit of the present invention, various deformations and improvements made by those of ordinary skill in the art to the technical solutions of the present invention shall fall within the protection scope determined by the claims of the present invention.

Claims

1. A method for detecting violations of large-piece transportation vehicles based on multi-dimensional data matching, characterized in that, The method includes: Performing data segmentation processing on the data information of the axle loads and axle distances of special multi-axle large vehicles measured by a dynamic weighing system to form a one-dimensional discrete data graph. Based on the one-dimensional discrete data graph, establishing a mapping relationship between the axle load and axle distance of the large transport vehicle and the total load, and determining whether there is an overweight behavior of the large transport vehicle based on the mapping relationship; Generating a point cloud model of the large transport vehicle for the captured image of the large transport vehicle through the SLAM algorithm. Based on the point cloud model, extracting the geometric parameters of the large transport vehicle and the goods. Based on the geometric parameters, identifying whether the center offset of the large transport vehicle exceeds a threshold; Based on the overweight detection result of the large transport vehicle and the illegal loading detection result of the large transport vehicle, realizing the illegal detection of the large transport vehicle.

2. The method according to claim 1, characterized in that, Performing data segmentation processing on the data information of the axle loads and axle distances of special multi-axle large vehicles measured by a dynamic weighing system to form a one-dimensional discrete data graph. Based on the one-dimensional discrete data graph, establishing a mapping relationship between the axle load and axle distance of the large transport vehicle and the total load, and determining whether there is an overweight behavior of the large transport vehicle based on the mapping relationship, including: Processing the one-dimensional information of the axle load and axle distance of the large transport vehicle collected by the dynamic weighing system. Setting the axle load as the ordinate y-axis and the axle distance as the abscissa x-axis, so as to perform one-dimensional data segmentation on the information collected by the dynamic weighing system; Plotting the information collected by the dynamic weighing system and processed in a one-dimensional discrete data image, and respectively marking the discrete points with different positions with different symbols and plotting them on the discrete data graph. Each different vehicle corresponds to a different interval block; The number of interval blocks in the one-dimensional discrete data graph represents the number of large transport vehicles, and the number of axles of the vehicle represents the number of discrete points in the interval block. Summing the quantities of all discrete points in each interval block along the y-axis direction is the total vehicle load of each special multi-axle large vehicle. Then, comparing the obtained total vehicle load with the predetermined load to determine whether there is an overweight behavior.

3. The method according to claim 1, wherein Generating a point cloud model of the large transport vehicle for the captured image of the large transport vehicle through the SLAM algorithm, including: Let be the camera coordinate system, where the positive direction of the Z-axis points forward along the direction from the plane towards the optical center, be the optical center of the camera, and the spatial target point is projected through the optical center onto the physical imaging plane and the imaging point is . Let the coordinates of the spatial point be , and the point projected onto the plane is . Let the distance from the physical imaging plane to the small hole be ; According to the similarity relationship of triangles, we have: ； Obtaining the spatial relationship between the spatial point and the corresponding pixel point according to the triangle similarity relationship. According to the spatial relationship between the spatial point and the corresponding pixel point, in the camera, the position of the pixel point is finally obtained; Among them, according to the spatial relationship between the spatial point and the corresponding pixel point, in the camera, the position of the pixel point is finally obtained, including: Set a fixed pixel plane on a physical imaging plane , and the pixel coordinates obtained on the pixel plane are ; The definition method of the pixel coordinate system is: the origin is located at the upper left corner of the image, the x-axis direction is to the right, parallel to the x-axis direction, the y-axis direction is downward, parallel to the y-axis direction; there is a scaling factor and a translation of the origin between the pixel coordinates and the imaging plane; assuming that the pixel is scaled by times on the x-axis, scaled by times on the y-axis, and the origin is translated by relative to the center, then the relationship between the coordinates of and the pixel coordinates ； According to the similarity relationship of triangles and coordinates and pixel coordinates , the following is obtained: ; Then the homogeneous coordinate expression of the pixel plane coordinate system is as follows: ； Wherein, the intermediate quantity The matrix formed is called the internal parameter matrix of the camera. The internal parameter matrix projects the reference point From its position in the camera's three-dimensional coordinate system to its position in the camera's internal pixel coordinate system. , , Are the coordinates of the pixel coordinate system, , , Are the coordinates of the spatial coordinate system; , Are the translations of the origin relative to the center, .

4. The method according to claim 3, characterized in that, Based on the point cloud model, extracting the geometric parameters of the large transport vehicle and the goods, including: Taking any point in the point cloud model of the large transport vehicle as the origin, perpendicular to the body of the corresponding large transport vehicle as the x-axis, the body direction as the y-direction, and the vertical direction as the z-direction to establish a coordinate axis; According to the coordinate axes, the length, width, and height of the corresponding large-piece transport vehicle are obtained as follows: ; ; ; where are the maximum values of all points in the point cloud in the x, y, and z directions respectively are the minimum values of all points in the point cloud in the x, y, and z directions respectively; when the difference between the maximum and minimum values in the x direction is taken, the resulting value is the width of the large-piece transport vehicle, when the difference between the maximum and minimum values in the y direction is taken, the resulting value is the length of the large-piece transport vehicle, and when the difference between the maximum and minimum values in the z direction is taken, the resulting value is the height of the large-piece transport vehicle.

5. The method according to claim 4, wherein Based on the geometric parameters, identifying whether the center offset of the large transport vehicle exceeds a threshold, including: Take any point in the point cloud model of the large-piece transport vehicle as the origin, establish a coordinate system with the vehicle body direction as the y-axis and the direction perpendicular to the vehicle body as the x-axis. The central position of the large-piece transport vehicle is: , where , correspond to the maximum value and the minimum value in the corresponding coordinate system respectively; Identifying the value corresponding to the center of the highway; Taking the difference between the identified center of the large transport vehicle and the center of the highway and comparing the difference with a predetermined threshold: Whether |μ' - μ| is less than or equal to σ, and when |μ' - μ| > σ, a serious deviation phenomenon occurs in the large-piece transport vehicle, then the large-piece transport vehicle has a violation. Here, μ' is the actual center, μ is the theoretical center, and σ is the specified threshold.

6. The method according to claim 1, characterized in that, Based on the overweight detection result of the large-piece transport vehicle and the illegal loading detection result of the large-piece transport vehicle, illegal detection of the large-piece transport vehicle is realized, including: The illegal detection of the large-piece transport vehicle consists of two parts: overweight detection and illegal vehicle loading. Among them, overweight detection is one-dimensional data detection, and illegal vehicle loading is the detection of two-dimensional images of the large-piece transport vehicle. The one-dimensional and two-dimensional data are matched with each other to obtain relevant information of the large-piece transport vehicle and determine whether the large-piece transport vehicle has a violation.

7. A large-piece transportation vehicle violation detection system based on multi-dimensional data matching, the system is used to implement the method described in any one of claims 1-6, and is characterized in that, The system includes: a large-piece transport vehicle overweight detection module, a large-piece transport vehicle illegal loading detection module, and a large-piece transport vehicle illegal detection module; The large-piece transport vehicle overweight detection module is used to perform data segmentation processing on the data information of the axle load and axle distance of the extra-axle large-piece vehicle measured by the dynamic weighing system to form a one-dimensional discrete data map. Based on the one-dimensional discrete data map, a mapping relationship between the axle load and axle distance of the large-piece transport vehicle and the total load is established, and whether the large-piece transport vehicle has an overweight behavior is judged based on the mapping relationship; The large-piece transport vehicle illegal loading detection module is used to generate a point cloud model of the large-piece transport vehicle through the SLAM algorithm for the collected images of the large-piece transport vehicle. Based on the point cloud model, geometric parameters of the large-piece transport vehicle and the goods are extracted, and whether the center offset of the large-piece transport vehicle exceeds the threshold is identified according to the geometric parameters; The large-piece transport vehicle illegal detection module is used to realize the illegal detection of the large-piece transport vehicle according to the overweight detection result of the large-piece transport vehicle and the illegal loading detection result of the large-piece transport vehicle.

Images