Railway clearance automated measurement method and system based on laser point cloud

Abstract

The present invention relates to the technical field of railway clearance measurement, and proposes a railway clearance automation measurement method and system based on laser point cloud, including: in the process of railway clearance measurement, collecting overall point cloud data and local point cloud data; marking the first adjacent collection time and target point cloud data, obtaining the convex significant weight and adjacent curvature of the target point cloud data, and obtaining the local curvature ratio of the first adjacent collection time; according to the angle between the vectors composed of the corresponding point cloud data in the adjacent collection time and the convex significant weight of the target point cloud data, obtaining the local height angle ratio of the first adjacent collection time, and obtaining the optimal track length of the first adjacent collection time; performing railway segmentation processing according to the optimal track length of all collection times in the process of railway clearance measurement, combining local point cloud data, and obtaining the railway clearance automation measurement result. The present invention solves the problem that the unreasonable setting of railway segment length leads to inaccurate railway clearance measurement results.

Description

Technical Field

[0001] The invention relates to the technical field of railway clearance measurement, and in particular to a railway clearance automatic measurement method and system based on laser point cloud. Background Art

[0002] Automated railway clearance measurement is one of the measurement tasks of daily railway maintenance. It can prevent the vehicle driving space along the railway from being blocked by obstacles and ensure the safety of railway travel. Generally, an automated measuring vehicle is used to realize automated railway clearance measurement. The laser scanner installed on the measuring vehicle is used to scan the point cloud of objects around the railway along the way. By analyzing the spatial relationship between the object point cloud and the railway track, it is determined whether there are obstacles within the spatial limit of the railway. In the process of automated railway clearance measurement, due to the curvature of the railway track, the railway needs to be segmented, and each railway segment is analyzed separately. The limit model is placed along the centerline direction of the railway segment to determine whether there is an over-limit problem in the railway segment.

[0003] Generally, a fixed constant is used as the value of the railway segment length to perform railway segment processing. In locations where the railway direction does not change, the value of the railway segment length is too small, which will lead to a waste of computing resources; in locations where the railway direction changes, the value of the railway segment length is too small and cannot be accurately determined, which can easily lead to inaccurate judgment of the over-limit problem of the railway segment, affecting the results of the railway limit automation measurement. Summary of the invention

[0004] The present invention provides a railway clearance automatic measurement method and system based on laser point cloud to solve the problem that the unreasonable setting of railway segment length leads to inaccurate railway clearance measurement results. The technical solutions adopted are as follows:

[0005] In a first aspect, an embodiment of the present invention provides a method for automated railway clearance measurement based on laser point cloud, the method comprising the following steps:

[0006] In the process of railway clearance measurement, the overall point cloud data is collected at the initial position of the 3D full-element collection vehicle, and the local point cloud data is collected at each collection moment during the movement process. According to the positions and positional relationships of the point cloud data in the overall point cloud data and the local point cloud data at two adjacent collection moments, the left railway point cloud, the right railway point cloud and the potential obstacle point cloud are divided from the local point cloud data, and the matching point cloud data in the local point cloud data at two adjacent collection moments are obtained;

[0007] Record any collection moment in the process of railway limit measurement as the first adjacent collection moment, record any point cloud data in the local point cloud data of the first adjacent collection moment as the target point cloud data, obtain the convex significant weight of the target point cloud data according to the distance and spatial position relationship between the point cloud data in the local point cloud data of the first adjacent collection moment and the target point cloud data, determine the adjacent curvature of the target point cloud data according to the position difference of the matching point cloud data in the adjacent collection moments of the first adjacent collection moment, and obtain the local curvature ratio of the first adjacent collection moment by combining the convex significant weights of all point cloud data at the first adjacent collection moment;

[0008] According to the angle between the vectors formed by the corresponding point cloud data at the adjacent acquisition moments of the first adjacent acquisition moment, the projection angle of the target point cloud data is determined, and the local height angle ratio at the first adjacent acquisition moment is obtained in combination with the convex saliency weights of all point cloud data at the first adjacent acquisition moment, and the optimal track length at the first adjacent acquisition moment is obtained in combination with the local curvature ratio at the first adjacent acquisition moment;

[0009] The railway is segmented according to the optimal track length at all acquisition times during the railway clearance measurement process, and the railway clearance automated measurement results are obtained by combining the left railway point cloud, the right railway point cloud and the potential obstacle point cloud.

[0010] Furthermore, the method of collecting the overall point cloud data at the initial position of the 3D full-element collection vehicle and collecting the local point cloud data at each collection moment during the movement process includes the following specific methods:

[0011] The initial position of the displacement sensor of the 3D full-factor acquisition vehicle is used as the coordinate origin, the direction perpendicular to the ground where the 3D full-factor acquisition vehicle is located is used as the positive direction of the z-axis, the initial movement direction of the 3D full-factor acquisition vehicle is used as the positive direction of the x-axis, and the y-axis is established perpendicular to the xoz plane to establish a first spatial rectangular coordinate system. At the initial position of the 3D full-factor acquisition vehicle, the 3D full-factor acquisition vehicle is used to collect the overall point cloud data, and the coordinate system of the overall point cloud data is the first spatial rectangular coordinate system;

[0012] For each collection moment, the position of the displacement sensor carried by the 3D full-factor collection vehicle at the collection moment is used as the coordinate origin, the direction perpendicular to the ground where the 3D full-factor collection vehicle is located is used as the positive direction of the z-axis, the movement direction of the 3D full-factor collection vehicle at the collection moment is used as the positive direction of the x-axis, and the y-axis is established perpendicular to the xoz plane to establish a second spatial rectangular coordinate system. The point cloud obtained at the collection moment is recorded as the local point cloud data at the collection moment.

[0013] Further, the method of dividing the left railway point cloud, the right railway point cloud and the potential obstacle point cloud from the local point cloud data, and obtaining the matching point cloud data in the local point cloud data at two adjacent acquisition moments includes the following specific methods:

[0014] The coordinate value of the y-axis in the local point cloud data is All point clouds within the range are recorded as left railway point clouds, and the coordinates of the y-axis are taken in All point clouds within the range are recorded as right railway point clouds, where is the distance between two parallel rails; the remaining point cloud data after removing the left rail point cloud and the right rail point cloud from the local point cloud data is recorded as the potential obstacle point cloud;

[0015] When the point cloud data contained in the first adjacent acquisition moment and the second adjacent acquisition moment correspond to the same point cloud data in the overall point cloud number, the point cloud data contained in the local point cloud data of the first adjacent acquisition moment and the second adjacent acquisition moment is recorded as matching point cloud data, and the second adjacent acquisition moment is the next adjacent acquisition moment after the first adjacent acquisition moment.

[0016] Further, the method of obtaining the convex saliency weight of the target point cloud data according to the distance and spatial position relationship between the point cloud data in the local point cloud data at the first adjacent acquisition moment and the target point cloud data includes the following specific methods:

[0017] The three point cloud data in the local point cloud data that are closest to the target point cloud data and are not collinear are recorded as the adjacent point cloud data of the target point cloud data, and the plane determined by the three adjacent point cloud data of the target point cloud data is recorded as the adjacent plane of the target point cloud data;

[0018] When the straight line passing through the target point cloud data and the displacement sensor position passes through the adjacent plane of the target point cloud data, the concave-convex mark constant of the target point cloud data is recorded as -1; when the straight line passing through the target point cloud data and the displacement sensor position is parallel to the adjacent plane of the target point cloud data, the concave-convex mark constant of the target point cloud data is recorded as 1, where the displacement sensor position is the displacement sensor position carried by the 3D full-factor acquisition vehicle at the first adjacent acquisition moment;

[0019] The area of ​​the triangle formed by the three adjacent point cloud data of the target point cloud data is recorded as the adjacent triangle area of ​​the target point cloud data;

[0020] The distance between the target point cloud data and the adjacent plane of the target point cloud data is recorded as the adjacent distance of the target point cloud data;

[0021] According to the concave-convex sign constant, adjacent distance and adjacent triangle area of ​​the target point cloud data, the convex saliency weight of the target point cloud data is obtained.

[0022] Furthermore, the convex saliency weight of the target point cloud data is obtained according to the concave-convex sign constant, the adjacent distance and the adjacent triangle area of ​​the target point cloud data, including the specific method of:

[0023] The ratio of the proximity distance of the target point cloud data to the adjacent triangle area is recorded as the first ratio of the target point cloud data, and the product of the concave-convex sign constant of the target point cloud data and the first ratio is recorded as the convex saliency of the target point cloud data;

[0024] The normalized value of the convex saliency of the target point cloud data at the first adjacent acquisition moment is recorded as the convex saliency weight of the target point cloud data.

[0025] Further, the method of determining the adjacent curvature of the target point cloud data according to the position difference of the matching point cloud data in the adjacent acquisition moments of the first adjacent acquisition moment, and obtaining the local curvature ratio of the first adjacent acquisition moment in combination with the convex saliency weights of all point cloud data at the first adjacent acquisition moment includes the following specific methods:

[0026] The point cloud data corresponding to the target point cloud data in the point cloud data collected at the previous adjacent collection time and the next adjacent collection time of the first adjacent collection time are recorded as the first adjacent point cloud data and the second adjacent point cloud data of the target point cloud data respectively;

[0027] Calculate the curvature of the target point cloud data according to the coordinates of the target point cloud data, the first adjacent point cloud data and the second adjacent point cloud data in the spatial rectangular coordinate system where the target point cloud data is located, and record it as the adjacent curvature of the target point cloud data;

[0028] Use the convex saliency weights of all point cloud data at the first adjacent collection time to perform weighted summation on the adjacent curvatures of the corresponding point cloud data, and obtain the rail curvature of the position of the 3D full-element collection vehicle at the first adjacent collection time;

[0029] The ratio of the rail curvature corresponding to the first adjacent collection moment to the first preset constant is recorded as the local curvature ratio of the first adjacent collection moment.

[0030] Further, the projection angle of the target point cloud data is determined, and the convex saliency weights of all point cloud data at the first adjacent acquisition moment are combined to obtain the local height angle ratio at the first adjacent acquisition moment, including the specific method of:

[0031] The vector pointing from the first adjacent point cloud data to the target point cloud data is recorded as the first vector of the target point cloud data; the vector pointing from the target point cloud data to the second adjacent point cloud data is recorded as the second vector of the target point cloud data; the angle between the projection vectors of the first vector of the target point cloud data and the second vector on the xoy plane in the second spatial rectangular coordinate system where the target point cloud data is located is recorded as the projection angle of the target point cloud data;

[0032] The projection angles of the corresponding point cloud data are weighted summed using the convex saliency weights of all point cloud data at the first adjacent collection time to obtain the weighted projection angle of the position of the 3D full-element collection vehicle at the first adjacent collection time;

[0033] The ratio of the weighted projection angle corresponding to the first adjacent acquisition moments to the second preset constant is recorded as the local height angle ratio of the first adjacent acquisition moments.

[0034] Furthermore, the specific method for obtaining the optimal track length of the first adjacent acquisition moments is:

[0035] The sum of the local curvature ratio and the local height angle ratio at the first adjacent collection time is recorded as the road condition complexity at the first adjacent collection time;

[0036] The ratio of the third preset constant to the road condition complexity at the first adjacent collection time is recorded as the optimal track length at the first adjacent collection time.

[0037] Furthermore, the railway segmentation processing is performed according to the optimal track length at all acquisition moments in the process of railway clearance measurement, and the railway clearance automation measurement result is obtained by combining the left railway point cloud, the right railway point cloud and the potential obstacle point cloud, including the specific method of:

[0038] Starting from the second collection moment in the process of railway limit measurement, when the distance moved by the 3D full-element collection vehicle is equal to or less than the railway segment length of the 3D full-element collection vehicle at the collection moment, a segmentation point is set at the location of the 3D full-element collection vehicle; starting from the next adjacent collection moment corresponding to the segmentation point, when the distance moved by the 3D full-element collection vehicle from the segmentation point is equal to or less than the railway segment length of the 3D full-element collection vehicle at the collection moment, a segmentation point is set at the location of the 3D full-element collection vehicle, and the above steps are repeated to divide all railway tracks in the process of railway limit measurement into different segmentation points;

[0039] At each segmentation point, a plane perpendicular to the movement direction of the 3D full-element acquisition vehicle at the segmentation point is taken. Each segmentation point corresponds to a plane. These planes divide the overall point cloud data into multiple segmentation point cloud data sets. For each segmentation point cloud data set, a straight line fitting is performed on the left railway point cloud contained in the segmentation point cloud data set to obtain the left railway straight line. A straight line fitting is performed on the right railway point cloud contained in the segmentation point cloud data set to obtain the right railway straight line, and the center lines of the left railway straight line and the right railway straight line are obtained.

[0040] The center line of the railway overrun model is overlapped with the center lines of the left railway straight line and the right railway straight line. When the railway overrun model coincides with the potential obstacle point cloud in the segmented point cloud dataset, it is determined that the railway limit is exceeded. When no coincidence occurs, it is determined that the railway limit is not exceeded.

[0041] In the second aspect, an embodiment of the present invention also provides a railway clearance automation measurement system based on laser point cloud, including a memory, a processor, and a computer program stored in the memory and running on the processor, and when the processor executes the computer program, the steps of any one of the above methods are implemented.

[0042] The beneficial effects of the present invention are:

[0043] This application collects overall point cloud data and local point cloud data at each collection moment during the movement process. The overall point cloud data is the point cloud data of all railway tracks, interference objects on the railway side and other objects for limit measurement. The local point cloud data is the point cloud data collected at the location of the three-dimensional full-element collection vehicle at the collection moment. The point cloud data corresponding to the tracks on both sides and potential obstacles are divided from the local point cloud data. In order to screen point cloud data with better matching effects from the point cloud matching results and obtain more accurate railway tracks, the confidence of the point cloud data corresponding to these positions is evaluated based on the characteristics of the convex parts of the surfaces of objects such as buildings, which have significant features, stable scanning data and are not easily blocked, to obtain the convex significance weight of the target point cloud data, and the position difference of the matched point cloud data in adjacent collection moments is combined. , evaluate the curvature of the rails and the complexity of the railway conditions at the location of the 3D full-element data collection vehicle at each collection moment, and obtain the local curvature ratio; further, according to the height difference in the vertical direction during the railway laying process, continue to analyze the railway conditions, obtain the local height angle ratio at each collection moment, and determine the optimal track length at the collection moment in combination with the local curvature ratio at the collection moment, and perform railway segmentation processing according to the optimal track length at all collection moments in the process of railway limit measurement, and obtain the optimal value of the railway segment length according to the actual railway conditions, so as to solve the problem of unreasonable setting of the railway segment length and inaccurate railway limit measurement results. Finally, combine the left railway point cloud, the right railway point cloud and the potential obstacle point cloud to obtain the railway limit automation measurement results, and improve the accuracy of railway limit measurement. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0045] Figure 1 A schematic flow chart of a method for automated railway clearance measurement based on laser point cloud provided by an embodiment of the present invention;

[0046] Figure 2 A schematic diagram of the configuration of a displacement sensor provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] See also Figure 1 , which shows a flow chart of a railway clearance automation measurement method based on laser point cloud provided by an embodiment of the present invention, the method comprising the following steps:

[0049] Step S001, during the railway clearance measurement, the overall point cloud data is collected at the initial position of the 3D full-element collection vehicle, and the local point cloud data is collected at each collection moment during the movement process. According to the position and position relationship of the point cloud data in the overall point cloud data and the local point cloud data at two adjacent collection moments, the left railway point cloud, the right railway point cloud and the potential obstacle point cloud are divided from the local point cloud data, and the matching point cloud data in the local point cloud data at two adjacent collection moments are obtained.

[0050] The displacement sensor carried by the 3D full-factor acquisition vehicle is used to locate the position of the 3D full-factor acquisition vehicle. The initial position of the displacement sensor of the 3D full-factor acquisition vehicle is used as the coordinate origin. The direction perpendicular to the ground where the 3D full-factor acquisition vehicle is located is used as the positive direction of the z-axis. The initial movement direction of the 3D full-factor acquisition vehicle is used as the positive direction of the x-axis. The y-axis is established perpendicular to the xoz plane to establish the first spatial rectangular coordinate system. The positioning coordinates in the first spatial rectangular coordinate system collected by the displacement sensor are the position of the 3D full-factor acquisition vehicle. Among them, the displacement sensor of the 3D full-factor acquisition vehicle is located at the geometric plane center of the 3D full-factor acquisition vehicle in the xoy plane in the spatial rectangular coordinate system, and at half the height of the railway track.

[0051] The schematic diagram of the displacement sensor setting is as follows Figure 2 As shown, in Figure 2In the figure, A is a 3D full-factor data acquisition vehicle, B is a displacement sensor, C is a rail, and the ground is under the rail. A, B, and C are all cross-sectional diagrams. 2L is the distance between the two rails, L is half of the distance between the two rails, 2Hi is the height of the rail, and Hi is half of the height of the rail. The 3D full-factor data acquisition vehicle is traveling on the rail, and the displacement sensor is connected to the 3D full-factor data acquisition vehicle. The displacement sensor of the 3D full-factor data acquisition vehicle is located at the center of the geometric plane of the 3D full-factor data acquisition vehicle in the xoy plane in the spatial rectangular coordinate system, and at half the height of the railway track.

[0052] In the process of railway limit measurement, the laser point cloud camera carried by the three-dimensional full-element collection vehicle is used to collect the overall point cloud data. The coordinate system of the overall point cloud data is the first space rectangular coordinate system.

[0053] It can be understood that the overall point cloud data includes all point clouds at all acquisition moments in the point cloud data acquisition process, and the overall point cloud data includes multiple point clouds.

[0054] Preferably, in one embodiment of the present application, when collecting the overall point cloud data, the data sampling frequency in this embodiment is 200 Hz. In the actual application process, as other implementation methods, the implementer can determine the sampling rate according to the actual situation, and this application does not impose any special restrictions.

[0055] For each collection moment, the position of the displacement sensor carried by the 3D full-factor collection vehicle at the collection moment is taken as the coordinate origin, the direction perpendicular to the ground where the 3D full-factor collection vehicle is located is taken as the positive direction of the z-axis, the movement direction of the 3D full-factor collection vehicle at the collection moment is taken as the positive direction of the x-axis, and the y-axis is established perpendicular to the xoz plane to establish a second spatial rectangular coordinate system. The positioning coordinates in the second spatial rectangular coordinate system collected by the displacement sensor are the position of the 3D full-factor collection vehicle, and the point cloud obtained at the collection moment is recorded as the local point cloud data at the collection moment.

[0056] It can be understood that the local point cloud data includes the point cloud data collected at the corresponding collection moment, and each collection moment corresponds to a local point cloud data.

[0057] For any local point cloud data, the coordinate value of the y-axis is All point clouds within the range are recorded as left railway point clouds, and the coordinates of the y-axis are taken as All point clouds within the range are recorded as right railway point clouds. The remaining point cloud data after removing the left railway point cloud and the right railway point cloud from the local point cloud data is recorded as potential obstacle point cloud. The point cloud data contained in the potential obstacle point cloud corresponds to the point cloud data involved in the subsequent railway clearance measurement and judgment.

[0058] in, is the distance between two parallel rails. The value is 1.6 meters.

[0059] In order to obtain the point cloud corresponding to the same object position in the local point cloud data of two adjacent acquisition moments, point cloud matching is performed on the local point cloud data of two adjacent acquisition moments. The local point cloud data of two adjacent acquisition moments are respectively recorded as the local point cloud data of the first adjacent acquisition moment and the second adjacent acquisition moment, wherein the first adjacent acquisition moment is any acquisition moment, and the first adjacent acquisition moment is the previous adjacent acquisition moment of the second adjacent acquisition moment. The point cloud data corresponding to the point cloud data contained in the local point cloud data of the first adjacent acquisition moment in the overall point cloud number is obtained, and the point cloud data corresponding to the point cloud data contained in the local point cloud data of the second adjacent acquisition moment in the overall point cloud number is obtained. When the point cloud data corresponding to the point cloud data contained in the first adjacent acquisition moment and the second adjacent acquisition moment in the overall point cloud number are the same point cloud data, the point cloud data contained in the local point cloud data of the first adjacent acquisition moment and the second adjacent acquisition moment are recorded as matching point cloud data.

[0060] When the point cloud data contained in the local point cloud data cannot find corresponding point cloud data in the overall point cloud number, the point cloud data in the overall point cloud number that is closest to the corresponding position of the point cloud data contained in the local point cloud data in the overall point cloud number is used as the point cloud data corresponding to the point cloud data contained in the local point cloud data in the overall point cloud number.

[0061] It can be understood that when the point cloud data contained in the local point cloud data of the first adjacent acquisition moment and the second adjacent acquisition moment cannot obtain matching point cloud data, the point cloud data contained in the local point cloud data that cannot obtain matching point cloud data will be deleted, and these point cloud data correspond to noise data; since the last acquisition moment does not have a subsequent adjacent acquisition moment, and the first acquisition moment does not have a previous adjacent acquisition moment, the first acquisition moment and the last acquisition moment are not analyzed.

[0062] At this point, the point cloud matching results of the point cloud data included in the local point cloud data at the first adjacent collection time and the second adjacent collection time are obtained. The point cloud matching results include the movement information of the 3D full-element collection vehicle and the track information of the railway.

[0063] At this point, the matching point cloud data in the local point cloud data at two adjacent acquisition moments are obtained.

[0064] Step S002, record any collection moment in the process of railway limit measurement as the first adjacent collection moment, record any point cloud data in the local point cloud data of the first adjacent collection moment as the target point cloud data, obtain the convex significant weight of the target point cloud data according to the distance and spatial position relationship between the point cloud data in the local point cloud data of the first adjacent collection moment and the target point cloud data, determine the adjacent curvature of the target point cloud data according to the position difference of the matching point cloud data in the adjacent collection moments of the first adjacent collection moment, and obtain the local curvature ratio of the first adjacent collection moment in combination with the convex significant weights of all point cloud data of the first adjacent collection moment.

[0065] In order to control and adjust the railway segmentation processing according to the railway conditions, adopt more detailed segmentation processing in complex road conditions to improve the railway measurement accuracy, and adopt more comprehensive segmentation processing in simple road conditions to improve the railway measurement speed, it is necessary to determine the curvature of the railway at different positions based on the movement information of the three-dimensional full-element acquisition vehicle and the trajectory information of the railway, so as to provide a more accurate basis for the railway segmentation processing to determine the value of the railway segment length.

[0066] It should be noted that the point cloud matching results of the point cloud data contained in the local point cloud data at two adjacent acquisition moments are only the matching results obtained based on the point cloud positions. It is also necessary to screen the point cloud data with better matching effects to obtain a more accurate railway track.

[0067] When performing point cloud laser scanning, the convex parts of the surface of objects such as buildings have significant features and the scanned data is stable. When the 3D full-element collection vehicle is scanning at different positions, the point cloud data corresponding to the convex parts are not easily blocked. Therefore, the point cloud matching results of the point cloud data corresponding to the convex parts are more likely to correspond to the same position of the same object, and the point cloud matching results can more accurately reflect the movement of the 3D full-element collection vehicle on the railway.

[0068] Any point cloud data in the local point cloud data at the first adjacent acquisition time is recorded as the target point cloud data, and the three point cloud data in the local point cloud data that are closest to the target point cloud data and are not collinear are recorded as the adjacent point cloud data of the target point cloud data. The plane determined by the three adjacent point cloud data of the target point cloud data is recorded as the adjacent plane of the target point cloud data. When the straight line passing through the target point cloud data and the position of the displacement sensor carried by the three-dimensional full-element collection vehicle at the first adjacent acquisition time passes through the adjacent plane of the target point cloud data, the target point cloud data is recorded as concave data; when the straight line passing through the target point cloud data and the position of the displacement sensor carried by the three-dimensional full-element collection vehicle at the first adjacent acquisition time is parallel to the adjacent plane of the target point cloud data, the target point cloud data is recorded as convex data. When the target point cloud data is concave data, the concave-convex mark constant of the target point cloud data is recorded as -1; when the target point cloud data is convex data, the concave-convex mark constant of the target point cloud data is recorded as 1.

[0069] The adjacent plane of the target point cloud data is the plane where the building or other object closest to the target point cloud data is located.

[0070] The area of ​​the triangle formed by the three adjacent point cloud data of the target point cloud data is recorded as the adjacent triangle area of ​​the target point cloud data. The distance between the target point cloud data and the adjacent plane of the target point cloud data is recorded as the adjacent distance of the target point cloud data.

[0071] The convexity of the target point cloud data is obtained according to the concave-convex sign constant, the adjacent distance and the adjacent triangle area of ​​the target point cloud data.

[0072] Preferably, as an embodiment of the present application, the ratio of the proximity distance of the target point cloud data to the adjacent triangle area is recorded as the first ratio of the target point cloud data, and the product of the convexity mark constant of the target point cloud data and the first ratio is recorded as the convexity of the target point cloud data.

[0073] The same method can be used to obtain the convex saliency of any point cloud data at any acquisition time.

[0074] Specifically, the calculation formula for the convex saliency of any point cloud data at any collection time is:

[0075]

[0076] In the formula, Indicates The first collection time The convex saliency of point cloud data; Indicates The first collection time The concave and convex mark constant of point cloud data; Indicates The first collection time The proximity distance of point cloud data; Indicates The first collection time The area of ​​the adjacent triangle of the point cloud data.

[0077] in, Indicates The first collection time The first ratio of point cloud data.

[0078] When the proximity distance of the target point cloud data is larger and the area of ​​the adjacent triangle is smaller, the convex feature of the object corresponding to the position of the target point cloud data is more significant, the data information of the target point cloud data is more stable, and the point cloud matching result corresponding to the target point cloud data can more accurately reflect the movement of the three-dimensional full-factor collection vehicle on the railway. At this time, the convex significance of the target point cloud data is greater.

[0079] The convex saliency of all point cloud data at the first adjacent acquisition moment is normalized to obtain the convex saliency weight of all point cloud data at the first adjacent acquisition moment.

[0080] Preferably, as an embodiment of the present application, the minimum value of the convex significance of all point cloud data at the first adjacent acquisition moment is recorded as the minimum convex significance; when the minimum convex significance is less than 0, the sum of the convex significance of the point cloud data at the first adjacent acquisition moment and the absolute value of the minimum convex significance is recorded as the first sum of the point cloud data, completing the non-negative processing of the convex significance of all point cloud data at the first adjacent acquisition moment; when the minimum convex significance is greater than or equal to 0, the convex significance of the point cloud data at the first adjacent acquisition moment is directly recorded as the first sum of the point cloud data. The cumulative sum of the first sum of all point cloud data at the first adjacent acquisition moment is recorded as the second sum of the first adjacent acquisition moment. The ratio of the first sum of the point cloud data to the second sum of the first adjacent acquisition moment is recorded as the convex significance weight of the point cloud data.

[0081] The convex saliency weight of point cloud data is positively correlated with the convex saliency of point cloud data. When the convex saliency weight of point cloud data is larger, the point cloud matching result corresponding to the point cloud data can more accurately reflect the movement of the 3D full-element collection vehicle on the railway.

[0082] Point cloud data corresponding to the target point cloud data in the point cloud data collected at the previous adjacent collection time and the next adjacent collection time of the first adjacent collection time are respectively recorded as first adjacent point cloud data and second adjacent point cloud data of the target point cloud data.

[0083] According to the spatial position changes of the target point cloud data, the first adjacent point cloud data and the second adjacent point cloud data relative to the origin of the corresponding spatial rectangular coordinate system, the movement of the three-dimensional full-element collection vehicle on the railway at the collection time adjacent to the first adjacent collection time can be obtained.

[0084] Since the target point cloud data and the first adjacent point cloud data and the second adjacent point cloud data are matching point cloud data, they theoretically correspond to the same object position at different acquisition times. The coordinates of the first adjacent point cloud data and the second adjacent point cloud data of the target point cloud data in the spatial rectangular coordinate system where the target point cloud data is located are obtained, and the curvature of the target point cloud data is calculated according to the coordinates of the target point cloud data, the first adjacent point cloud data and the second adjacent point cloud data in the spatial rectangular coordinate system where the target point cloud data is located, and the curvature of the target point cloud data is recorded as the adjacent curvature of the target point cloud data.

[0085] Calculating the curvature based on the coordinates of a point is a well-known technique and will not be described in detail.

[0086] The convex significant weights of all point cloud data at the first adjacent collection moment are used to perform weighted summation on the adjacent curvatures of the corresponding point cloud data to obtain the rail curvature of the position of the three-dimensional full-element collection vehicle at the first adjacent collection moment. The ratio of the rail curvature corresponding to the first adjacent collection moment to the first preset constant is recorded as the local curvature ratio of the first adjacent collection moment.

[0087] Among them, the value of the first preset constant is 0.002. In actual application, as other implementation methods, the implementer can decide the value of the first preset constant according to actual conditions, and this application does not impose any special restrictions.

[0088] When the convex significant weight of all point cloud data at the first adjacent collection moment is greater and the adjacent curvature is greater, the local curvature ratio at the first adjacent collection moment is greater. At this time, the greater the curvature of the rail at the location of the 3D full-element collection vehicle at the first adjacent collection moment, the more complex the railway conditions, and the smaller the railway segment length should be selected for the rail at this location.

[0089] The same method can be used to obtain the local curvature ratio at any sampling time in the process of railway clearance measurement.

[0090] At this point, the local curvature ratio at any acquisition time in the process of railway clearance measurement is obtained.

[0091] Step S003, according to the angle between the vectors formed by the corresponding point cloud data in the adjacent acquisition moments of the first adjacent acquisition moment, determine the projection angle of the target point cloud data, combine the convex saliency weights of all the point cloud data in the first adjacent acquisition moment, obtain the local height angle ratio of the first adjacent acquisition moment, combine the local curvature ratio of the first adjacent acquisition moment, and obtain the optimal track length of the first adjacent acquisition moment.

[0092] When the railway condition is more complicated, a smaller railway segment length should be selected, and a more detailed segmentation process should be adopted to improve the railway measurement accuracy. Furthermore, it is necessary to continue to analyze the railway condition based on the vertical height difference during the railway laying process.

[0093] The vector pointing from the first adjacent point cloud data to the target point cloud data is recorded as the first vector of the target point cloud data, the vector pointing from the target point cloud data to the second adjacent point cloud data is recorded as the second vector of the target point cloud data, and the angle between the first vector of the target point cloud data and the projection vector of the second vector on the xoy plane in the second spatial rectangular coordinate system where the target point cloud data is located is recorded as the projection angle of the target point cloud data.

[0094] The projection angles of the corresponding point cloud data are weightedly summed using the convex saliency weights of all point cloud data at the first adjacent collection moment to obtain the weighted projection angle of the 3D full-element collection vehicle at the first adjacent collection moment. The ratio of the weighted projection angle corresponding to the first adjacent collection moment to the second preset constant is recorded as the local height angle ratio of the first adjacent collection moment.

[0095] Among them, the value of the second preset constant is 27. In actual application, as other implementation methods, the implementer can decide the value of the second preset constant according to actual conditions, and this application does not impose any special restrictions.

[0096] The optimal track length at the first adjacent acquisition moments is determined according to the local curvature ratio and the local height angle ratio at the first adjacent acquisition moments.

[0097] Preferably, as an embodiment of the present application, the sum of the local curvature ratio and the local height angle ratio at the first adjacent collection moments is recorded as the road condition complexity at the first adjacent collection moments, and the ratio of the third preset constant to the road condition complexity at the first adjacent collection moment is recorded as the optimal track length at the first adjacent collection moment.

[0098] Among them, the value of the third preset constant is 27. In actual application, as other implementation methods, the product of the local curvature ratio and the local height angle ratio of the first adjacent collection moments is recorded as the road condition complexity of the first adjacent collection moments, and the ratio of the third preset constant to the road condition complexity of the first adjacent collection moment is recorded as the optimal track length of the first adjacent collection moment. The implementer can decide the value of the third preset constant according to the actual situation, and this application does not impose any special restrictions.

[0099] When the road condition complexity at the first adjacent collection moment is greater, the road condition at the railway track position where the three-dimensional full-element collection vehicle is at the first adjacent collection moment is more complex. At this railway track position, a smaller railway segment length value should be selected and more detailed segment processing should be adopted to improve the railway measurement accuracy. It can be understood that at this time, the optimal track length at the first adjacent collection moment is the larger.

[0100] The same method can be used to obtain the optimal track length at any collection time during the railway clearance measurement process.

[0101] The optimal track length at the collection time in the process of railway limit measurement is used as the value of the railway segment length at the collection time position of the three-dimensional full-element collection vehicle. In particular, when the optimal track length at the collection time is greater than 1, the value of the railway segment length at the collection time position of the three-dimensional full-element collection vehicle is set to the fourth preset constant to avoid the calculation of the railway limit measurement being too complicated due to the too short railway segment length value. When the optimal track length at the collection time is equal to 0, the value of the railway segment length at the collection time position of the three-dimensional full-element collection vehicle is set to the fifth preset constant.

[0102] Among them, the value of the fourth preset constant is 1 meter, and the value of the fifth preset constant is 100 meters. In actual application, as other implementation methods, the implementer can decide the values ​​of the fourth preset constant and the fifth preset constant according to actual conditions, and this application does not impose any special restrictions.

[0103] So far, in the process of obtaining railway limit measurement, the value of the railway segment length at any collection time position of the three-dimensional full-element collection vehicle is obtained.

[0104] Step S004, performing railway segmentation processing according to the optimal track length at all acquisition times during the railway clearance measurement process, combining the left railway point cloud, the right railway point cloud and the potential obstacle point cloud to obtain the railway clearance automatic measurement result.

[0105] During the railway limit measurement process, starting from the second collection moment in the railway limit measurement process, when the distance moved by the 3D full-element collection vehicle is equal to or less than the railway segment length of the 3D full-element collection vehicle at the collection moment, a segmentation point is set at the location of the 3D full-element collection vehicle; starting from the next adjacent collection moment corresponding to the segmentation point, when the distance moved by the 3D full-element collection vehicle from the segmentation point is equal to or less than the railway segment length of the 3D full-element collection vehicle at the collection moment, a segmentation point is set at the location of the 3D full-element collection vehicle, and the above steps are repeated to divide all railway tracks in the railway limit measurement process into different segmentation points.

[0106] At each segmentation point, a plane perpendicular to the movement direction of the 3D full-element collection vehicle at the segmentation point is taken. Each segmentation point corresponds to a plane, and these planes divide the overall point cloud data into multiple segmentation point cloud data sets. For each segmentation point cloud data set, the least squares method is used to perform straight line fitting on the left railway point cloud contained in the segmentation point cloud data set to obtain the left railway straight line, and the right railway point cloud contained in the segmentation point cloud data set is performed to perform straight line fitting to obtain the right railway straight line. The left railway straight line and the right railway straight line correspond to the two rails of the railway between the two segmentation points. The center lines of the left railway straight line and the right railway straight line are obtained. The center line of the railway overrun model is overlapped with the center lines of the left railway straight line and the right railway straight line, and the railway overrun model is compared with the potential obstacle point cloud in the segmentation point cloud data set. When overlap occurs, it is determined that the railway limit exceeds the limit, and when no overlap occurs, it is determined that the railway limit does not exceed the limit.

[0107] The center line is geometrically located in the middle of two given straight lines, namely the left railway straight line and the right railway straight line. The definition and calculation of the center line are well-known technologies and will not be described in detail.

[0108] It can be understood that the railway overrun model is a three-dimensional model constructed based on the volume of railway vehicles. By aligning the railway overrun model with the center lines of the two rails of the railway, the movement of vehicles on the railway can be simulated. If the potential obstacle point cloud coincides with the railway overrun model, the vehicle will collide with objects such as buildings corresponding to the potential obstacle point cloud when traveling on the railway. At this time, the railway limit exceeds the limit, and objects such as buildings near the railway need to be cleared. The railway building limit comprehensive algorithm can be used to obtain the railway overrun model. The acquisition of the railway overrun model is a well-known technology and will not be repeated here.

[0109] At this point, the railway clearance automation measurement results are obtained.

[0110] Based on the same inventive concept as the above method, an embodiment of the present invention also provides an automated railway clearance measurement system based on laser point cloud, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, the steps of any one of the above automated railway clearance measurement methods based on laser point cloud are implemented.

[0111] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. The railway clearance automatic measurement method based on laser point cloud is characterized by: The method comprises the following steps: In the process of railway clearance measurement, the overall point cloud data is collected at the initial position of the 3D full-element collection vehicle, and the local point cloud data is collected at each collection moment during the movement process. According to the positions and positional relationships of the point cloud data in the overall point cloud data and the local point cloud data at two adjacent collection moments, the left railway point cloud, the right railway point cloud and the potential obstacle point cloud are divided from the local point cloud data, and the matching point cloud data in the local point cloud data at two adjacent collection moments are obtained; Record any collection moment in the process of railway limit measurement as the first adjacent collection moment, record any point cloud data in the local point cloud data of the first adjacent collection moment as the target point cloud data, obtain the convex significant weight of the target point cloud data according to the distance and spatial position relationship between the point cloud data in the local point cloud data of the first adjacent collection moment and the target point cloud data, determine the adjacent curvature of the target point cloud data according to the position difference of the matching point cloud data in the adjacent collection moments of the first adjacent collection moment, and obtain the local curvature ratio of the first adjacent collection moment by combining the convex significant weights of all point cloud data at the first adjacent collection moment; According to the angle between the vectors formed by the corresponding point cloud data at the adjacent acquisition moments of the first adjacent acquisition moment, the projection angle of the target point cloud data is determined, and the local height angle ratio at the first adjacent acquisition moment is obtained in combination with the convex saliency weights of all point cloud data at the first adjacent acquisition moment, and the optimal track length at the first adjacent acquisition moment is obtained in combination with the local curvature ratio at the first adjacent acquisition moment; The railway is segmented according to the optimal track length at all acquisition times during the railway clearance measurement process, and the railway clearance automatic measurement results are obtained by combining the left railway point cloud, the right railway point cloud and the potential obstacle point cloud.

2. The railway clearance automatic measurement method based on laser point cloud according to claim 1 is characterized in that: The method of collecting the overall point cloud data at the initial position of the 3D full-element acquisition vehicle and collecting the local point cloud data at each acquisition moment during the movement process includes: The initial position of the displacement sensor of the 3D full-factor acquisition vehicle is used as the coordinate origin, the direction perpendicular to the ground where the 3D full-factor acquisition vehicle is located is used as the positive direction of the z-axis, the initial movement direction of the 3D full-factor acquisition vehicle is used as the positive direction of the x-axis, and the y-axis is established perpendicular to the xoz plane to establish a first spatial rectangular coordinate system. At the initial position of the 3D full-factor acquisition vehicle, the 3D full-factor acquisition vehicle is used to collect the overall point cloud data, and the coordinate system of the overall point cloud data is the first spatial rectangular coordinate system; For each collection moment, the position of the displacement sensor carried by the 3D full-factor collection vehicle at the collection moment is used as the coordinate origin, the direction perpendicular to the ground where the 3D full-factor collection vehicle is located is used as the positive direction of the z-axis, the movement direction of the 3D full-factor collection vehicle at the collection moment is used as the positive direction of the x-axis, and the y-axis is established perpendicular to the xoz plane to establish a second spatial rectangular coordinate system. The point cloud obtained at the collection moment is recorded as the local point cloud data at the collection moment.

3. The railway clearance automatic measurement method based on laser point cloud according to claim 1 is characterized in that: The method of dividing the left railway point cloud, the right railway point cloud and the potential obstacle point cloud from the local point cloud data and obtaining the matching point cloud data in the local point cloud data at two adjacent acquisition moments includes: The coordinate value of the y-axis in the local point cloud data is All point clouds within the range are recorded as left railway point clouds, and the coordinates of the y-axis are taken in All point clouds within the range are recorded as right railway point clouds, where is the distance between two parallel rails; the remaining point cloud data after removing the left rail point cloud and the right rail point cloud from the local point cloud data is recorded as the potential obstacle point cloud; When the point cloud data contained in the first adjacent acquisition moment and the second adjacent acquisition moment correspond to the same point cloud data in the overall point cloud number, the point cloud data contained in the local point cloud data of the first adjacent acquisition moment and the second adjacent acquisition moment is recorded as matching point cloud data, and the second adjacent acquisition moment is the next adjacent acquisition moment after the first adjacent acquisition moment.

4. The railway clearance automatic measurement method based on laser point cloud according to claim 1 is characterized in that: The specific method of obtaining the convex saliency weight of the target point cloud data according to the distance and spatial position relationship between the point cloud data in the local point cloud data at the first adjacent acquisition moment and the target point cloud data is as follows: The three point cloud data in the local point cloud data that are closest to the target point cloud data and are not collinear are recorded as the adjacent point cloud data of the target point cloud data, and the plane determined by the three adjacent point cloud data of the target point cloud data is recorded as the adjacent plane of the target point cloud data; When the straight line passing through the target point cloud data and the displacement sensor position passes through the adjacent plane of the target point cloud data, the concave-convex mark constant of the target point cloud data is recorded as -1; When the straight line passing through the target point cloud data and the displacement sensor position is parallel to the adjacent plane of the target point cloud data, the concave-convex sign constant of the target point cloud data is recorded as 1, wherein the displacement sensor position is the displacement sensor position carried by the 3D full-factor acquisition vehicle at the first adjacent acquisition moment; The area of ​​the triangle formed by the three adjacent point cloud data of the target point cloud data is recorded as the adjacent triangle area of ​​the target point cloud data; The distance between the target point cloud data and the adjacent plane of the target point cloud data is recorded as the adjacent distance of the target point cloud data; According to the concave-convex sign constant, adjacent distance and adjacent triangle area of ​​the target point cloud data, the convex saliency weight of the target point cloud data is obtained.

5. The railway clearance automatic measurement method based on laser point cloud according to claim 4 is characterized in that: The specific method of obtaining the convex saliency weight of the target point cloud data according to the concave-convex sign constant, the adjacent distance and the adjacent triangle area of ​​the target point cloud data includes: The ratio of the proximity distance of the target point cloud data to the adjacent triangle area is recorded as the first ratio of the target point cloud data, and the product of the concave-convex sign constant of the target point cloud data and the first ratio is recorded as the convex saliency of the target point cloud data; The normalized value of the convex saliency of the target point cloud data at the first adjacent acquisition moment is recorded as the convex saliency weight of the target point cloud data.

6. The railway clearance automatic measurement method based on laser point cloud according to claim 1 is characterized in that: The method of determining the adjacent curvature of the target point cloud data according to the position difference of the matching point cloud data in the adjacent acquisition moments of the first adjacent acquisition moment, and obtaining the local curvature ratio of the first adjacent acquisition moment by combining the convex saliency weights of all point cloud data at the first adjacent acquisition moment includes the following specific methods: The point cloud data corresponding to the target point cloud data in the point cloud data collected at the previous adjacent collection time and the next adjacent collection time of the first adjacent collection time are recorded as the first adjacent point cloud data and the second adjacent point cloud data of the target point cloud data respectively; Calculate the curvature of the target point cloud data according to the coordinates of the target point cloud data, the first adjacent point cloud data and the second adjacent point cloud data in the spatial rectangular coordinate system where the target point cloud data is located, and record it as the adjacent curvature of the target point cloud data; Use the convex saliency weights of all point cloud data at the first adjacent collection time to perform weighted summation on the adjacent curvatures of the corresponding point cloud data, and obtain the rail curvature of the position of the 3D full-element collection vehicle at the first adjacent collection time; The ratio of the rail curvature corresponding to the first adjacent collection moment to the first preset constant is recorded as the local curvature ratio of the first adjacent collection moment.

7. The railway clearance automatic measurement method based on laser point cloud according to claim 1 is characterized in that: The method of determining the projection angle of the target point cloud data and combining the convex saliency weights of all point cloud data at the first adjacent acquisition time to obtain the local height angle ratio at the first adjacent acquisition time includes: The vector pointing from the first adjacent point cloud data to the target point cloud data is recorded as the first vector of the target point cloud data; the vector pointing from the target point cloud data to the second adjacent point cloud data is recorded as the second vector of the target point cloud data; the angle between the projection vectors of the first vector of the target point cloud data and the second vector on the xoy plane in the second spatial rectangular coordinate system where the target point cloud data is located is recorded as the projection angle of the target point cloud data; The projection angles of the corresponding point cloud data are weighted summed using the convex saliency weights of all point cloud data at the first adjacent collection time to obtain the weighted projection angle of the position of the 3D full-element collection vehicle at the first adjacent collection time; The ratio of the weighted projection angle corresponding to the first adjacent acquisition moments to the second preset constant is recorded as the local height angle ratio of the first adjacent acquisition moments.

8. The railway clearance automatic measurement method based on laser point cloud according to claim 1 is characterized in that: The specific method for obtaining the optimal track length at the first adjacent acquisition time is: The sum of the local curvature ratio and the local height angle ratio at the first adjacent collection time is recorded as the road condition complexity at the first adjacent collection time; The ratio of the third preset constant to the road condition complexity at the first adjacent collection time is recorded as the optimal track length at the first adjacent collection time.

9. The railway clearance automatic measurement method based on laser point cloud according to claim 1 is characterized in that: The railway segmentation processing is performed according to the optimal track length at all acquisition moments in the process of railway clearance measurement, and the railway clearance automation measurement result is obtained by combining the left railway point cloud, the right railway point cloud and the potential obstacle point cloud, including the specific method of: Starting from the second collection moment in the process of railway limit measurement, when the distance moved by the 3D full-element collection vehicle is equal to or less than the railway segment length of the 3D full-element collection vehicle at the collection moment, a segmentation point is set at the location of the 3D full-element collection vehicle; starting from the next adjacent collection moment corresponding to the segmentation point, when the distance moved by the 3D full-element collection vehicle from the segmentation point is equal to or less than the railway segment length of the 3D full-element collection vehicle at the collection moment, a segmentation point is set at the location of the 3D full-element collection vehicle, and the above steps are repeated to divide all railway tracks in the process of railway limit measurement into different segmentation points; At each segmentation point, a plane perpendicular to the movement direction of the 3D full-element acquisition vehicle at the segmentation point is taken. Each segmentation point corresponds to a plane. These planes divide the overall point cloud data into multiple segmentation point cloud data sets. For each segmentation point cloud data set, a straight line fitting is performed on the left railway point cloud contained in the segmentation point cloud data set to obtain the left railway straight line. A straight line fitting is performed on the right railway point cloud contained in the segmentation point cloud data set to obtain the right railway straight line, and the center lines of the left railway straight line and the right railway straight line are obtained. The center line of the railway overrun model is overlapped with the center lines of the left railway straight line and the right railway straight line. When the railway overrun model coincides with the potential obstacle point cloud in the segmented point cloud dataset, it is determined that the railway limit is exceeded. When no coincidence occurs, it is determined that the railway limit is not exceeded.

10. A railway clearance automatic measurement system based on laser point cloud, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 9 are implemented.

Images