A method, device and storage medium for filtering point cloud of complex tunnel based on tunnel geometric features

Abstract

The present invention discloses a method, device and storage medium for filtering point clouds of complex-shaped tunnels based on tunnel geometric features. The point cloud filtering method includes: adjusting the deflection of the overall tunnel point cloud; dividing the original tunnel point cloud into tunnel segmented point clouds with equal thickness along the Y-axis direction of the global coordinate system, correcting the deflection angle and orientation of the segmented point clouds, and generating corresponding cross-sectional point clouds; using a convex hull algorithm to preliminarily identify the contour of the tunnel cross-sectional point cloud, and sequentially calculating the distances of all non-nominal contour points to the nominal contour of the tunnel cross-sectional area; comparing the relative size of the distances of the non-nominal contour points to the nominal contour of the tunnel cross-sectional area with the distance threshold, and performing secondary screening on the non-nominal contour points of the tunnel cross-sectional point cloud to realize the recognition of the tunnel segmented contour point cloud; sequentially filtering each segmented point cloud of the tunnel, and fusing the segmented contour point cloud, and finally obtaining the contour point cloud of the tunnel. The present invention realizes the recognition of the contour point cloud of the irregular-section tunnel, and improves the calculation accuracy of the point cloud filtering of complex-shaped tunnels.

Description

Technical Field

[0001] The present invention belongs to the technical field of tunnel point cloud data processing, and relates to a method, a device and a storage medium for filtering complex-shaped tunnel point clouds based on tunnel geometric features. Background Art

[0002] Three-dimensional laser scanning technology (LiDAR technology) has the characteristics of fast scanning technology, high scanning accuracy, non-contact and not affected by the scanning environment, making it widely used in tunnel monitoring measurement and monitoring and early warning. Since the massive tunnel point cloud obtained by LiDAR technology inevitably has a large number of non-tunnel contour points, that is, interference points, which seriously affects its application in the field of tunnel engineering, before further application, the tunnel point cloud needs to be filtered, that is, the contour point cloud of the tunnel point cloud is distinguished from the non-contour point cloud. However, tunnel point cloud filtering involves the calculation of massive points, and the filtering accuracy has a great impact on the further application of tunnel point clouds. High-precision filtering of tunnel point clouds is the premise and basis for its application in the field of tunnel engineering.

[0003] Existing tunnel point cloud filtering methods mainly target circular or elliptical cross-section tunnels, and use quadratic curve equations, such as circle equations and ellipse equations, to fit the plane point cloud after the tunnel segment point cloud is projected, and determine whether it is a noise point based on the distance from the plane point cloud to the fitting center, thereby achieving filtering. However, this type of method is only applicable to tunnels with circular or elliptical cross-sections, and has great limitations when filtering point clouds of semicircular arches, three-center arches, and irregular cross-section tunnels.

[0004] In recent years, machine learning has been widely used in the field of underground engineering, but this method requires comprehensive training samples to be prepared in advance. At the same time, the training samples must also cover all the morphological characteristics of the tunnel. This makes it very difficult to achieve high-precision filtering effects for irregular cross-section tunnel point clouds. Summary of the invention

[0005] In order to solve the above problems, the present invention provides a method, device and storage medium for filtering complex-shaped tunnel point clouds based on tunnel geometric features, which realizes the recognition of irregular-section tunnel contour point clouds, improves the accuracy of tunnel point cloud filtering, and solves the limitation problem that existing tunnel point cloud filtering methods cannot effectively perform irregular-shaped tunnel point cloud filtering.

[0006] The technical solution adopted by the present invention is a method for filtering point clouds of complex-shaped tunnels based on tunnel geometric features, comprising:

[0007] Through the skew correction of the overall tunnel point cloud, the overall extension direction of the tunnel point cloud is kept consistent with the Y-axis direction of the global coordinate system; the original tunnel point cloud is divided into tunnel segment point clouds with equal thickness along the Y-axis direction of the global coordinate system, and the corresponding cross-sectional point cloud is generated by correcting the deflection angle and orientation of the segment point cloud; the contour of the tunnel cross-sectional point cloud is preliminarily identified by the convex hull algorithm, and the distance from the non-nominal contour point to the nominal contour of the tunnel cross-sectional point cloud is calculated in turn; by comparing the relative size of the distance from the non-nominal contour point to the nominal contour of the tunnel cross-sectional point cloud with the distance threshold, the non-nominal contour points of the tunnel cross-sectional point cloud are secondary screened, thereby realizing the recognition of the tunnel cross-sectional point cloud contour; the segmented point clouds of the tunnel are filtered in turn, and the irregular section tunnel contour point cloud is finally recognized by the fusion of the segmented tunnel contour point clouds.

[0008] Further, the method specifically includes the following steps:

[0009] S1, the original point cloud of the deep irregular cross-section tunnel is obtained by the ground 3D laser scanning system;

[0010] S2, based on the characteristics of the point cloud obtained by the ground 3D laser scanning system, the tunnel point cloud is projected to the XOY plane, where X represents the tunnel transverse direction and Y represents the tunnel longitudinal direction, and the angle between the tunnel point cloud and the Y axis of the global coordinate system is calculated; the tunnel point cloud is rotated to make the overall extension direction of the tunnel point cloud consistent with the Y axis direction of the global coordinate system;

[0011] S3, along the Y-axis direction of the global coordinate system, the tunnel point cloud is divided into a number of thin slices of tunnel segment point clouds T with thickness t. i ;

[0012] S4, calculate the tunnel segment point cloud T through the two-dimensional convex hull algorithm i Convex polygons and vertices of all projection points in the XOY plane;

[0013] S5, based on the feature with the smallest angle corresponding to the corner, identify the four corner points of the convex polygon of the projection point of the tunnel segment point cloud in the XOY plane, calculate the average deflection angle by calculating the angle between the short side segments on both sides and the Y axis, and then calculate the corresponding orientation;

[0014] S6, based on the deflection angle θ of the tunnel segment point cloud i and orientation, for the tunnel segment point cloud T i The coordinates are rotated and corrected to make the orientation of the tunnel segment point cloud consistent with the Y axis of the global coordinate system. The tunnel segment point cloud after coordinate transformation is projected along the Y axis of the global coordinate system to obtain the tunnel segment point cloud T i Corresponding to the tunnel cross-section point cloud T in the XOZ plane i ′;

[0015] S7, calculate the tunnel cross-section point cloud T using the two-dimensional convex hull algorithmi The convex polygon corresponding to ′ is defined as the nominal contour point of the tunnel, and the tunnel cross-section point cloud T i The remaining points of ′ are regarded as non-nominal contour points; the distance from each non-nominal contour point in the tunnel section point cloud to the nominal contour point of the tunnel is calculated;

[0016] S8, by comparing the distance from the non-nominal contour point to the nominal contour of the tunnel section with the distance threshold d threshold The relative size of the tunnel cross-section point cloud is used to perform secondary screening of the non-nominal contour points, thereby realizing the recognition of the tunnel cross-section point cloud contour;

[0017] S9, filtering each segment point cloud of the tunnel in turn, and fusing the contour point cloud and non-contour point cloud of each segment point cloud of the tunnel respectively, to obtain the contour point cloud and non-contour point cloud of the tunnel, thereby realizing the filtering of the irregular section tunnel point cloud.

[0018] Furthermore, in S2, the line connecting the midpoints of the sections at both ends of the tunnel extension direction is used as a reference to calculate the angle between the line connecting the midpoints of the two sections and the Y axis of the global coordinate system, and the tunnel point cloud is rotated according to formula (1):

[0019]

[0020] Where α is the angle between the tunnel point cloud and the Y axis of the global coordinate system, (X, Y, Z) is the coordinate of the tunnel point cloud before coordinate transformation, and (X new ,Y new ,Z new ) is the coordinate of the tunnel point cloud after coordinate transformation.

[0021] Furthermore, in S4, point P is calculated according to formula (2): j The angle of , from which the angles of all vertices are calculated;

[0022]

[0023] P j The vertices of the convex polygon of all projection points in the XOY plane are sorted in counterclockwise order, P j The previous point is P j-1 , P j The last point is P j+1 .

[0024] Further, in S5, the corner points corresponding to the tunnel segment projection point cloud are located at the four corners. The four corner points are numbered according to the relative sizes of the coordinate components of the XY coordinates. The four corner points are numbered in a counterclockwise direction, starting from the lower left corner vertex. i1 , P i2 , P i3 , Pi4 According to formula (3) and formula (4), the deflection angle θ between the short sides of the left and right sides of the tunnel segment projection point cloud and the Y axis of the global coordinate system is calculated respectively. il ,θ ir ;

[0025]

[0026]

[0027]

[0028] According to formula (5), the tunnel segment point cloud T is calculated. i The average deflection angle θ with the Y axis of the global coordinate system i .

[0029] Furthermore, in S7, the tunnel section non-nominal contour point m i Calculation steps for the distance to the nominal tunnel profile:

[0030] First, the non-nominal contour point m of the tunnel is calculated according to the point-to-line distance calculation method. i To the line segment l that constitutes the nominal tunnel profile j The distance d ij ;

[0031] Then, the non-nominal contour points m of the tunnel section are calculated in sequence i To form the tunnel nominal contour segment l j+1 The distance d ij+1 , j = 1, 2, 3, ...;

[0032] Finally, according to the calculated tunnel section non-nominal contour point m i Distance d from all segments to the nominal tunnel profile ij , the minimum value is the non-nominal contour point m of the tunnel section i Distance d to the nominal tunnel profile i ,

[0033] i.e. i =min{d ij ,j=1,2,3,…}.

[0034] Furthermore, the method for calculating the distance from a point to a line segment is divided into the following three cases:

[0035] 1) If the point of intersection of the perpendicular line between a point and a line segment and the line on which the line segment lies lies on the line segment, then the distance from the point to the line is the distance from the point to the line segment;

[0036] 2) If the intersection point is on one side of the line segment, then the distance between the point and the endpoint of that side of the line segment is the distance from the point to the line segment;

[0037] 3) If the intersection point is on the other side of the line segment, then the distance between the point and the endpoint of that side of the line segment is the distance from the point to the line segment.

[0038] Furthermore, in S8, the distance threshold d from the non-nominal contour point of the tunnel section to the nominal contour of the tunnel section is comprehensively determined by the actual distance between the tunnel point cloud contour and the different auxiliary facilities inside the tunnel. threshold , distance d i ≤d threshold is determined as a contour point, and d i ﹥d threshold Determined as non-contour point.

[0039] A complex morphology tunnel point cloud filtering device based on tunnel geometric features, comprising:

[0040] a memory for storing instructions executable by a processor; and

[0041] The processor is used to execute the instructions to implement the above-mentioned complex morphology tunnel point cloud filtering method based on tunnel geometric features.

[0042] A computer readable medium storing computer program codes, wherein the computer program codes, when executed by a processor, implement the above-mentioned method for filtering a complex morphology tunnel point cloud based on tunnel geometric features.

[0043] The beneficial effects of the present invention are:

[0044] The present invention is a method for filtering complex morphological tunnel point clouds based on the geometric features of the tunnel. Based on the geometric features of the outer contour of the irregular section tunnel point cloud, the deflection angle and orientation of the tunnel segment point cloud are calculated by identifying the convex polygon of the tunnel segment point cloud projection and calculating its inner vertex angle and feature analysis, and the deflection angle and orientation of the tunnel segment point cloud are corrected to obtain the tunnel section point cloud corresponding to the irregular section tunnel segment point cloud, which effectively improves the accuracy of the tunnel segment point cloud generating the section point cloud. Based on the contour features of the tunnel section point cloud, the convex polygon surrounding all points of the tunnel section is identified by a two-dimensional convex hull algorithm, and the tunnel section contour points are preliminarily identified. A distance calculation method between the non-nominal contour points of the tunnel section and the nominal contour of the tunnel is established, and the distance of the non-nominal contour point cloud of the tunnel section is calculated in sequence. Then, the tunnel contour is finely identified by a determined distance threshold, and the filtering of the irregular section tunnel segment point cloud is realized. Finally, the contour point cloud of the tunnel point cloud is calculated in sequence, and then the tunnel segment point cloud is fused to obtain the overall contour point cloud of the tunnel.

[0045] The irregular cross-section tunnel point cloud filtering method involved in the present invention breaks through the limitation that the existing tunnel point cloud filtering method is only applicable to regular cross-sections such as circles and ellipses, and improves the calculation accuracy and operation efficiency of irregular cross-section tunnel point cloud filtering. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] 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 work.

[0047] Figure 1 It is a schematic diagram of calculating the inclination angle of the overall point cloud of the tunnel in an embodiment of the present invention.

[0048] Figure 2 Schematic diagram of point-to-line segment distance calculation in an embodiment of the present invention, wherein (a), (b), and (c) correspond to three situations respectively.

[0049] Figure 3 It is a point cloud of a deep tunnel with irregular cross-section in an embodiment of the present invention.

[0050] Figure 4 It is a flow chart of point cloud filtering of an irregular cross-section tunnel according to an embodiment of the present invention.

[0051] Figure 5 It is a segmented point cloud of a deep tunnel with irregular cross-section in an embodiment of the present invention.

[0052] Figure 6 It is the projection of the tunnel segment point cloud on the XOY plane and its convex polygon in the embodiment of the present invention.

[0053] Figure 7 It is the identification of the corner points of the convex polygon of the tunnel segment projection point cloud in the embodiment of the present invention.

[0054] Figure 8 It is the tunnel segment projection point cloud and its nominal contour recognition in the embodiment of the present invention.

[0055] Fig. 9 It is the distribution of non-nominal contour points of the tunnel cross-section point cloud and their distances from the nominal contour in the embodiment of the present invention.

[0056] Fig.10 This is the detailed identification of the nominal contour of the tunnel segment projection point cloud in the embodiment of the present invention.

[0057] Fig.11 It is the accurate identification of the segmented point cloud contour of the tunnel in the embodiment of the present invention.

[0058] Fig.12 It is the point cloud contour recognition of a deep irregular cross-section tunnel in an embodiment of the present invention. DETAILED DESCRIPTION

[0059] The following will be combined with 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 making creative work are within the scope of protection of the present invention.

[0060] A method for filtering complex tunnel point clouds based on tunnel geometric features can be applied to Figure 3 In the environment shown; Figure 4 As shown, the following steps are included:

[0061] 1. A ground-based 3D laser scanning system is used to collect 3D surface coordinate points of the tunnel surface. The embodiment of the present invention obtains the original point cloud of the irregular cross-section tunnel as follows: Figure 3 As shown, the format of the original point cloud of the tunnel is converted to obtain a universal point cloud format, and the original point cloud of the tunnel is sparsely processed to reduce the point cloud density and improve the calculation efficiency of the tunnel point cloud while ensuring the accuracy.

[0062] 2. According to the characteristics of the point cloud obtained by the ground 3D laser scanning system, the tunnel point cloud is projected onto the XOY plane (X is the tunnel transverse direction, Y is the tunnel longitudinal direction), and the line connecting the midpoints of the sections at both ends of the tunnel extension direction is used as a reference to calculate the angle between the line connecting the two midpoints and the Y axis of the global coordinate system, such as Figure 1 As shown; and according to formula (1), the tunnel point cloud is rotated and transformed so that the overall extension direction of the tunnel point cloud is consistent with the Y-axis direction of the global coordinate system.

[0063]

[0064] Where α is the angle between the tunnel point cloud and the Y axis of the global coordinate system, (X, Y, Z) is the coordinate of the tunnel point cloud before coordinate transformation, and (X new ,Y new ,Z new ) is the coordinate of the tunnel point cloud after coordinate transformation.

[0065] 3. Usually, the tunnel point cloud filtering or denoising requires first segmenting the tunnel point cloud, and then obtaining the cross-sectional projection point cloud from the tunnel segment point cloud. However, since the actual axis of the irregular cross-section tunnel is unknown, if the indeterminacy of the deflection angle of the actual axis is taken into account when calculating the normal vector of the tunnel segment point cloud, it will inevitably lead to repeated calculation or non-calculation of local points of adjacent segment point clouds of the tunnel. Therefore, for irregular morphological tunnel point clouds, the method of calculating the segment normal cannot fully consider the irregular cross-section tunnel point cloud.

[0066] In order to overcome this problem, the embodiment of the present invention first obtains the irregular section tunnel segment point cloud along the extension direction of the tunnel overall point cloud. In order to obtain the irregular section tunnel segment point cloud, the tunnel point cloud is divided into a number of thin slices of tunnel segment point clouds T with a thickness of t along the extension direction of the tunnel point cloud, that is, the Y-axis direction of the global coordinate system. i In the embodiment, the tunnel point cloud with a tunnel section length of 30 m is divided into 150 tunnel segment point clouds along the positive direction of the Y axis of the global coordinate system according to the thickness t=0.2 m. Figure 5 As shown in the figure, the tunnel segment point cloud is obtained along the positive direction of the Y axis of the global coordinate system where the tunnel point cloud is located. At the same time, the actual axis of the irregular section tunnel is not a straight line. Therefore, it is inevitable that there is a certain deviation from the Y axis. Therefore, it is necessary to calculate the deflection angle and orientation of the irregular section tunnel segment point cloud, and correct the tunnel segment point cloud for the deviation between the segment point cloud and the overall extension direction of the tunnel, and then obtain the cross-sectional point cloud corresponding to the irregular section tunnel segment point cloud by projection.

[0067] 4. In order to calculate the inclination of the irregular section tunnel segment point cloud, a method for calculating the inclination angle of the tunnel segment point cloud is established. i Project to the XOY plane to get the tunnel segment point cloud T i The projection points in the XOY plane can be surrounded by the convex polygon composed of the outermost projection points. To this end, the two-dimensional convex hull algorithm is used to calculate the tunnel segment point cloud T i The convex polygon and its vertices of all projection points in the XOY plane are as follows Figure 6 As shown, there are 21 vertices.

[0068] Before calculating the deflection angle of the segmented point cloud, the four corner points of the convex polygon at the corner of the tunnel segmented point cloud must be obtained first, and the average deflection angle is calculated by calculating the angle between the short side segments on both sides and the Y axis, so as to consider the influence of the segmented point cloud deflection. The angle corresponding to each vertex of the convex polygon is calculated in turn by using the convex polygon inner vertex calculation method, and the corner points of the convex polygon of the tunnel segmented point cloud projection point in the XOY plane are identified based on the feature of the minimum angle corresponding to the corner, such as Figure 7 As shown, Figure 7 The points surrounded by the blue circle are the tunnel segment point cloud T i The corresponding corner point.

[0069] Specifically, first, taking the center of the convex polygon as the reference point, calculate the angles of all vertices with the positive direction of the X-axis and arrange them in counterclockwise order according to the angles; secondly, calculate point P j When the angle is , the point P in front and behind it in counterclockwise order j-1 , P j+1For reference, the angle of the point is calculated according to formula (2), which is the angle of the vertex, and the angles of all vertices can be calculated from this. The vertex angle is the smallest at the corner, close to 90 degrees, while the angles of other parts are larger, close to 180 degrees. Thus, the four corner points corresponding to the projection points of the tunnel segment point cloud can be obtained.

[0070]

[0071] refer to Figure 6 , P j The vertices of the convex polygon of all projection points in the XOY plane are sorted in counterclockwise order, P j The previous point is P j-1 , P j The last point is P j+1 .

[0072] 5. According to the relative position relationship of the coordinate components of the four corner points in the plane, calculate the deflection angle of the tunnel segment point cloud, and then calculate its orientation. Figure 7 It can be seen that the corner points corresponding to the tunnel segment projection point cloud are located at the four corners. According to the relative size of the coordinate components of the four corner points in the XY coordinates, they are numbered respectively. Taking the lower left corner vertex as the starting point, the four corner points are recorded as P in the counterclockwise direction. i1 , P i2 , P i3 , P i4 According to formula (3) and formula (4), the deflection angle θ between the left and right hypotenuses of the tunnel segment projection point cloud and the Y axis of the global coordinate system is calculated respectively. il ,θ ir , whose average value is the tunnel segment point cloud T i The deflection angle θ with respect to the Y axis of the global coordinate system i (Formula 5); The deflection angle θ of the tunnel segment point cloud is calculated from this i = 2.62°. Then, by calculating the sum of the left and right vectors The cross product with the positive Y-axis of the global coordinate system. When the cross product is positive, the deflection angle θ i is negative, otherwise, the deflection angle θ i is positive; since the calculated cross product is positive, the deflection angle is negative.

[0073]

[0074]

[0075]

[0076] 6. Deflection angle θ based on tunnel segment point cloud i and orientation, the tunnel segment point cloud T is obtained by formula (1) iThe coordinates are rotated and corrected (i.e., the coordinates are rotated). At this time, the orientation of the tunnel segment point cloud is consistent with the Y axis of the global coordinate system. The tunnel segment point cloud after coordinate transformation is projected along the Y axis of the global coordinate system to obtain the tunnel segment point cloud T i Corresponding to the tunnel cross-section point cloud T in the XOZ plane i ',like Figure 8 shown.

[0077] 7. Due to the tunnel cross-section point cloud T around the XOZ plane i The convex polygon of ′ must be the tunnel section contour point cloud, so the two-dimensional convex hull algorithm is used to calculate the tunnel section point cloud T i The recognition result of the convex polygon corresponding to ′ is as follows: Figure 8 The blue points in the figure are defined as nominal contour points of the tunnel. Due to the irregular characteristics of the tunnel section, the local contour points of the tunnel section have concave features, and the convex polygon of the tunnel section point cloud is only a part of the tunnel section contour point cloud. Therefore, the two-dimensional convex hull algorithm cannot accurately identify the entire contour of the irregular section tunnel point cloud, but its calculation results can be used as the basis for further calculations.

[0078] The convex polygons identified by the two-dimensional convex hull algorithm and their corresponding vertices are regarded as the nominal contour of the tunnel section and the nominal contour points of the tunnel section, respectively, and the tunnel section point cloud T i The remaining points of ′ are regarded as non-nominal contour points, such as Figure 8 According to the distance calculation method from the non-nominal contour point of the tunnel section to the nominal contour of the tunnel section, by calculating the distance from each non-nominal contour point in the tunnel section point cloud to the nominal contour point of the tunnel, the distance distribution from each non-nominal contour point in the tunnel section point cloud to the nominal contour of the tunnel section is obtained, as shown in Fig. 9 shown.

[0079] Calculate the distance from a point to a line segment in a plane, and establish a method for calculating the distance from a point to a line segment in a plane. Unlike the calculation of the distance from a point to a straight line, the calculation of the distance from a point to a line segment requires first determining the positional relationship between the point and the line segment before completing the calculation of the distance from the point to the line segment. Based on the relative relationship between the perpendicular line between the point and the line segment and the intersection of the line segment, the calculation of the distance from a point to a line segment can be divided into three situations: 1) If the intersection of the perpendicular line between the point and the line segment and the line segment is on the line segment, then the distance from the point to the line is the distance from the point to the line segment; 2) If the intersection is on one side of the line segment, then the distance between the point and the endpoint of that side of the line segment is the distance from the point to the line segment; 3) If the intersection is on the other side of the line segment, then the distance between the point and the endpoint of that side of the line segment is the distance from the point to the line segment. For example, to calculate point P i To Segment First, calculate the vector according to formula (6):

[0080]

[0081] Then, calculate the dot product separately like Then d=|P i B i |, see Figure 2 (b) in Figure ; if Then d=|P i B i+1 |See Figure 2 (c) in Figure 1, otherwise See Figure 2 Figure (a) in .

[0082] The nominal contour points and non-nominal contour points of the tunnel section point cloud are obtained through preliminary identification of the tunnel contour. The adjacent points that make up the non-nominal contour points are connected to form the nominal contour of the tunnel, which is a closed polygon composed of segmented line segments. i The distance to the nominal tunnel profile can be calculated as follows:

[0083] First, the non-nominal contour point m of the tunnel is calculated according to the point-to-line distance calculation method. i To the line segment l that constitutes the nominal tunnel profile j The distance d ij ;

[0084] Then, the non-nominal contour points m of the tunnel section are calculated in sequence i To form the tunnel nominal contour segment l j+1 The distance d ij+1 , j = 1, 2, 3, ...;

[0085] Finally, according to the calculated tunnel section non-nominal contour point m i Distance d from all segments to the nominal tunnel profile ij , the minimum value is the non-nominal contour point m of the tunnel section i Distance d to the nominal tunnel profile i ,

[0086] i.e. i =min{d ij ,j=1,2,3,…};

[0087] The distances from all non-tunnel nominal contour points to the tunnel nominal contour can be calculated in the above manner.

[0088] 8. Comprehensively determine the distance threshold d from the non-nominal contour point of the tunnel section to the nominal contour of the tunnel section through the actual distance between the tunnel point cloud contour and the various ancillary facilities inside the tunnel. threshold= 0.03m, by comparing the distance d from the non-nominal contour point to the nominal contour of the tunnel section i With distance threshold d threshold The relative size of the non-nominal contour points of the tunnel cross-section point cloud can be secondary screened (the nominal contour of the tunnel is defined according to the convex polygon identified by the convex hull algorithm. The closer the distance between the non-nominal contour point and the nominal contour is, the greater the possibility that it is a contour point. In the embodiment, the distance d is specified i ≤d threshold is determined as a contour point, and d i ﹥d threshold Determined as non-contour points), thereby achieving accurate recognition of the tunnel section point cloud contour ( Fig.10 ). Finally, the contour points and non-contour points of the irregular cross-section tunnel segment point cloud are distinguished, such as Fig.11 As shown, accurate recognition of the tunnel segment point cloud contour is achieved.

[0089] 9. Through the above steps, the contour point cloud and non-contour point cloud of the irregular cross-section tunnel segment point cloud can be accurately identified. For each segment point cloud of the tunnel point cloud, repeating the above steps can obtain the contour point cloud and non-contour point cloud of each tunnel segment point cloud in turn.

[0090] The contour point cloud and non-contour point cloud of each segment point cloud of the tunnel are fused separately to obtain the contour point cloud and non-contour point cloud of the tunnel respectively, so as to realize the accurate filtering of the point cloud of the tunnel with irregular section (complex shape) ( Fig.12 ). The above-mentioned calculation process of filtering the irregular cross-section tunnel point cloud is easy to implement in a programmable manner, which can greatly improve the calculation accuracy and operation efficiency of the irregular cross-section tunnel point cloud filtering.

[0091] The embodiment of the present invention takes into account the situation that the actual axis of the deep-buried tunnel is not a straight line. By correcting the skew angle of the tunnel point cloud and the deflection angle of the segmented point cloud, and based on the geometric characteristics of the tunnel, the nominal contour of the irregular cross-section tunnel is obtained. On this basis, by establishing a distance calculation method between the tunnel non-contour point cloud and the contour point cloud and its threshold determination, accurate identification based on the contour point cloud of the irregular cross-section tunnel is achieved; the existing tunnel point cloud filtering technology cannot achieve this effect. The irregular cross-section tunnel point cloud filtering method involved in the embodiment of the present invention is suitable for urban subway tunnels, mountain tunnels, submarine tunnels, deep mine tunnels, etc.

[0092] If the complex morphology tunnel point cloud filtering method based on tunnel geometric features described in the embodiment of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the complex morphology tunnel point cloud filtering method based on tunnel geometric features described in the embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, ROM, RAM, disk or CD-ROM and other media that can store program codes.

[0093] The above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims

1. A method for filtering point clouds of complex tunnels based on tunnel geometric features, characterized in that: The specific steps include: S1, the original point cloud of the deep irregular cross-section tunnel is obtained by the ground 3D laser scanning system; S2, based on the characteristics of the point cloud obtained by the ground 3D laser scanning system, the tunnel point cloud is projected to the XOY plane, where X represents the tunnel transverse direction and Y represents the tunnel longitudinal direction, and the angle between the tunnel point cloud and the Y axis of the global coordinate system is calculated; the tunnel point cloud is rotated to make the overall extension direction of the tunnel point cloud consistent with the Y axis direction of the global coordinate system; S3, along the Y-axis direction of the global coordinate system, the tunnel point cloud is divided into a number of thin slices of tunnel segment point clouds T with thickness t. i ; S4, calculate the tunnel segment point cloud T through the two-dimensional convex hull algorithm i Convex polygons and vertices of all projection points in the XOY plane; S5, based on the feature with the smallest angle corresponding to the corner, identify the four corner points of the convex polygon of the projection point of the tunnel segment point cloud in the XOY plane, calculate the average deflection angle by calculating the angle between the short side segments on both sides and the Y axis, and then calculate the corresponding orientation; S6, based on the deflection angle θ of the tunnel segment point cloud i and orientation, for the tunnel segment point cloud T i The coordinates are rotated and corrected to make the orientation of the tunnel segment point cloud consistent with the Y axis of the global coordinate system. The tunnel segment point cloud after coordinate transformation is projected along the Y axis of the global coordinate system to obtain the tunnel segment point cloud T i Corresponding to the tunnel cross-section point cloud T in the XOZ plane i ′; S7, calculate the tunnel cross-section point cloud T using the two-dimensional convex hull algorithm i The convex polygon corresponding to ′ is defined as the nominal contour point of the tunnel, and the tunnel cross-section point cloud T i The remaining points of ′ are regarded as non-nominal contour points; Calculate the distance from each non-nominal contour point in the tunnel cross-section point cloud to the nominal contour point of the tunnel; S8, by comparing the distance from the non-nominal contour point to the nominal contour of the tunnel section with the distance threshold d threshold The relative size of the non-nominal contour points of the tunnel section point cloud is secondary screened to realize the recognition of the tunnel section point cloud contour; S9, filtering each segment point cloud of the tunnel in turn, and fusing the contour point cloud and non-contour point cloud of each segment point cloud of the tunnel respectively, to obtain the contour point cloud and non-contour point cloud of the tunnel, thereby realizing the filtering of the irregular section tunnel point cloud.

2. According to claim 1, a method for filtering complex morphology tunnel point clouds based on tunnel geometric features, characterized in that: In S2, the line connecting the midpoints of the sections at both ends of the tunnel extension direction is used as a reference to calculate the angle between the line connecting the midpoints of the two sections and the Y axis of the global coordinate system, and the tunnel point cloud is rotated according to formula (1): Where α is the angle between the tunnel point cloud and the Y axis of the global coordinate system, (X, Y, Z) is the coordinate of the tunnel point cloud before coordinate transformation, and (X new ,Y new ,Z new ) is the coordinate of the tunnel point cloud after coordinate transformation.

3. According to claim 1, a method for filtering complex morphology tunnel point clouds based on tunnel geometric features, characterized in that: In S4, point P is calculated according to formula (2): j The angle of , from which the angles of all vertices are calculated; P j The vertices of the convex polygon of all projection points in the XOY plane are sorted in counterclockwise order, P j The previous point is P j-1 , P j The last point is P j+1 .

4. According to claim 1, a method for filtering complex morphological tunnel point clouds based on tunnel geometric features is characterized in that: In S5, the corner points corresponding to the tunnel segment projection point cloud are located at the four corners. The four corner points are numbered according to the relative sizes of the coordinate components of the XY coordinates. The four corner points are numbered in a counterclockwise direction, starting from the lower left corner vertex. i1 , P i2 , P i3 , P i4 According to formula (3) and formula (4), the deflection angle θ between the short sides of the left and right sides of the tunnel segment projection point cloud and the Y axis of the global coordinate system is calculated respectively. il ,θ ir ; According to formula (5), the tunnel segment point cloud T is calculated i The average deflection angle θ with the Y axis of the global coordinate system i .

5. According to claim 1, a method for filtering complex morphology tunnel point clouds based on tunnel geometric features, characterized in that: In S7, the tunnel section non-nominal contour point m i Calculation steps for the distance to the nominal tunnel profile: First, the non-nominal contour point m of the tunnel is calculated according to the point-to-line segment distance calculation method. i To the line segment l that constitutes the nominal tunnel profile j The distance d ij ; Then, the non-nominal contour points m of the tunnel section are calculated in sequence i To form the tunnel nominal contour segment l j+1 The distance d ij+1 , j = 1, 2, 3, ...; Finally, according to the calculated tunnel section non-nominal contour point m i Distance d from all segments to the nominal tunnel profile ij , the minimum value is the non-nominal contour point m of the tunnel section i Distance d to the nominal tunnel profile i , That is, d i = min{d ij , j = 1, 2, 3, …}.

6. According to claim 5, a method for filtering complex morphology tunnel point clouds based on tunnel geometric features is characterized in that: The calculation method of the distance from a point to a line segment is divided into the following three cases: 1) If the point of intersection of the perpendicular line between a point and a line segment and the line on which the line segment lies lies on the line segment, then the distance from the point to the line is the distance from the point to the line segment; 2) If the intersection point is on one side of the line segment, then the distance between the point and the endpoint of that side of the line segment is the distance from the point to the line segment; 3) If the intersection point is on the other side of the line segment, then the distance between the point and the endpoint of that side of the line segment is the distance from the point to the line segment.

7. According to claim 1, a method for filtering complex morphology tunnel point clouds based on tunnel geometric features, characterized in that: In S8, the distance threshold d from the non-nominal contour point of the tunnel section to the nominal contour of the tunnel section is comprehensively determined by the actual distance between the tunnel point cloud contour and the different auxiliary facilities inside the tunnel. threshold , distance d i ≤d threshold is determined as a contour point, and d i ﹥d threshold Determined as non-contour point.

8. A complex morphology tunnel point cloud filtering device based on tunnel geometric features, characterized in that: include: a memory for storing instructions executable by a processor; as well as A processor is used to execute the instructions to implement the complex morphology tunnel point cloud filtering method based on tunnel geometric features as described in any one of claims 1 to 7.

9. A computer-readable medium storing computer program codes, wherein the computer program codes, when executed by a processor, implement the method for filtering complex morphology tunnel point clouds based on tunnel geometric features as claimed in any one of claims 1 to 7.

Images