CN112085843A - Method and device for extracting and measuring tunnel type target features in real time - Google Patents

Abstract

The invention provides a method and a device for extracting and measuring tunnel type target characteristics in real time, which solve the technical problem that the existing tunnel type target characteristics are difficult to extract and measure. The method comprises the following steps: eliminating interference point clouds to obtain effective point clouds; fitting the ground point cloud and calculating a normal vector of a fitting plane to solve a pitch angle; after ground point clouds are removed, the pitch angle of the residual effective point clouds is compensated; extracting characteristic point clouds in the effective point clouds, and projecting the characteristic point clouds on a projection plane to generate cross-section point clouds; identifying boundary point clouds in the cross-section point clouds, and distinguishing the cross-section point clouds by taking the boundary point clouds as boundary points; constructing a mathematical description model of the tunnel type target section according to the distinguished section point cloud; and calculating the height and width of the tunnel-type target according to the mathematical description model. The method can extract the tunnel type target characteristics in real time in the vehicle running process, simultaneously realize the mathematical description and the height and width measurement of the tunnel type target characteristics, meet the advanced prejudgment requirement and assist the vehicle driving.

Description

A method and device for real-time extraction and measurement of tunnel-like target features

technical field

The invention relates to the technical field of intelligent driving, in particular to a method and device for real-time extraction and measurement of tunnel-like target features.

Background technique

With the continuous development of science and technology, as well as the development of automobile manufacturing and information technology, people's means of travel have gradually changed from horse-drawn carriages to bicycles, fuel vehicles, and electric vehicles. Now intelligent driving has been put on a new agenda.

In the road conditions of vehicles, the target height-limited road infrastructure such as culverts, tunnels, and bridges is an important factor affecting vehicle traffic, and it is also a research hotspot in the field of intelligent driving. Realizing the detection and extraction of the features of tunnel-like target height-limiting facilities can effectively improve the safety and reliability of vehicle driving. However, it is not easy to extract and measure the features of tunnel-like targets, and the prior art also lacks the technology for extracting and measuring the features of tunnel-like targets.

SUMMARY OF THE INVENTION

In view of the above problems, embodiments of the present invention provide a method for real-time extraction and measurement of tunnel-like target features, which solves the technical problem that the existing tunnel-like target features are not easy to extract and measure.

In a first aspect, the present invention provides a method for real-time extraction and measurement of tunnel-like target features, including:

Eliminate the interfering point cloud in the 3D point cloud collected by the multi-line lidar to obtain an effective point cloud;

Select the point cloud with the smallest elevation in the effective point cloud to form the ground point cloud, fit the ground point cloud to the plane and calculate the normal vector of the fitted plane to solve the pitch angle;

Compensate the pitch angle for the remaining valid point cloud after removing the ground point cloud;

Extract the point cloud with the upward emission angle in the effective point cloud after pitch angle compensation to form the characteristic point cloud of the tunnel-like target, and project the characteristic point cloud on the projection plane parallel to the axle and perpendicular to the advancing direction to generate the cross-section point cloud;

Identify the boundary point cloud in the section point cloud, and use the boundary point cloud as the boundary point to divide the section point cloud into top boundary point cloud, left boundary point cloud and right boundary point cloud;

Construct the mathematical description model of tunnel-like target section according to the differentiated section point cloud;

Calculate the height and width of the tunnel-like target according to the mathematical description model.

In one embodiment, obtaining an effective point cloud by eliminating the interference point cloud in the three-dimensional point cloud collected by the multi-line laser radar includes:

Determining unnecessary point clouds not in the area in front of the vehicle;

Determine the point cloud whose data value is not NAF as invalid point cloud;

Determining a point cloud with no adjacent point cloud around it as an isolated point cloud;

Valid point clouds are obtained by excluding the redundant point clouds, invalid point clouds and isolated point clouds.

In one embodiment, selecting the point cloud with the smallest elevation in the valid point cloud to form a ground point cloud, performing plane fitting on the ground point cloud and calculating the normal vector of the fitted plane to solve the pitch angle includes:

The effective point cloud data is spatially rasterized, and each grid is a cuboid with the same bottom surface and a height that varies according to the spatial distribution of the point cloud;

The point clouds in each grid are sorted by elevation, and the lowest point cloud in each grid is compared with the lowest point cloud in the adjacent grid to obtain the point cloud with the lowest elevation, and the distance from the point cloud with the lowest elevation is less than the threshold range The point cloud inside is regarded as the ground point cloud;

Fit the ground point cloud to the plane, find the normal vector of the fitted plane, and solve the pitch angle according to the normal vector of the fitted plane.

In one embodiment, the point cloud with the upward emission angle in the effective point cloud after pitch angle compensation is extracted to form the feature point cloud of the tunnel-like target, and the feature point cloud is projected on a projection plane parallel to the axle and perpendicular to the advancing direction The point cloud generated on the cross section includes:

Set the XOY plane of the multi-line lidar coordinate system as the radar plane;

Calculate the angle θ between the single-beam laser point cloud and the radar plane in the effective point cloud

According to the radar installation angle and the radar vertical field of view parameters, the single-beam laser point cloud with the included angle θ between 10° and 15° is selected as the feature point cloud;

The feature point cloud is projected onto a plane perpendicular to the axle to generate the section point cloud of the tunnel-like target.

In one embodiment, identifying the boundary point cloud in the cross-section point cloud, and using the boundary point cloud as the boundary point to divide the cross-section point cloud into a top boundary point cloud, a left boundary point cloud and a right boundary point cloud includes:

Select the upper, middle, and lower adjacent point clouds from the cross-section point cloud to form a scanning triangle, and scan the cross-section point cloud into left and right parts clockwise and counterclockwise respectively from the front of the vehicle; take the middle point cloud as the vertex , calculate the angle of the vertices of the scanned triangle;

Take the point at which the angle value of the vertices of the swept triangle changes abruptly as the boundary point cloud in the section point cloud;

Take the point cloud before the boundary point as the top boundary point cloud of the tunnel target, the point cloud after the left half boundary point cloud as the left boundary point cloud, and the point cloud after the right half boundary point cloud as the right boundary point cloud;

Save the top boundary point cloud, left boundary point cloud, and right boundary point cloud in the top boundary container, left boundary container, and right boundary container, respectively.

In one embodiment, the mathematical description model for constructing the tunnel-like target section according to the differentiated section point cloud includes:

Perform quadratic curve fitting on the top boundary point cloud to obtain the top boundary boundary curve parameters U(a ₁ ,b ₁ ,c), U(a ₁ ,b ₁ ,c) represents the top edge quadratic curve coefficient, and construct the top edge Quadratic curve model y=a ₁ x ² +b ₁ x+c;

Perform straight line fitting on the left boundary point cloud and the right boundary point cloud to obtain the left and right boundary curve parameters L(a ₂ , b ₂ ) and R(a ₃ , b ₃ ), where L(a ₂ , b ₂ ) represents The left straight line coefficient; R(a ₃ , b ₃ ) the right straight line coefficient, to construct the left straight line model y=a ₂ x+b ₂ and the right straight line model y=a ₃ x+b ₃ .

In one embodiment, the calculating the height and width of the tunnel-like target according to the mathematical description model includes:

According to the left and right boundary line model, according to the formula width=fabs(L_b-R_b), calculate the target width;

In the formula: L_b represents the coefficient of the linear constant term of the left boundary; R_b represents the coefficient of the linear constant term of the right boundary, respectively.

计算目标高度；According to the top boundary curve model, according to the formula:

Calculate the target height;

In the formula: h ₀ is the height of the multi-line lidar from the ground.

In a second aspect, the present invention provides a device for real-time extraction and measurement of tunnel-like target features, the device comprising:

Point cloud preprocessing module: Eliminate the interference point cloud in the 3D point cloud collected by the multi-line lidar, and obtain an effective point cloud;

Slope calculation module: used to select the point cloud with the smallest elevation in the effective point cloud to form the ground point cloud, fit the ground point cloud to the plane and calculate the normal vector of the fitted plane to solve the pitch angle;

Compensation module: used to compensate the pitch angle for the remaining valid point cloud after removing the ground point cloud;

Extraction module: It is used to extract the point cloud with the upward emission angle in the effective point cloud after pitch angle compensation to form the feature point cloud of the tunnel-like target, and project the feature point cloud on the projection plane parallel to the axle and perpendicular to the advancing direction to generate Section point cloud;

Classification module: used to identify the boundary point cloud in the section point cloud, and use the boundary point cloud as the boundary point to divide the section point cloud into top boundary point cloud, left boundary point cloud and right boundary point cloud;

Building block: used to construct a mathematical description model of the tunnel-like target section based on the differentiated section point cloud;

Calculation module: used to calculate the height and width of tunnel-like targets according to the mathematical description model.

In a third aspect, the present invention provides an electronic device, comprising:

processor, memory, interface for communicating with the gateway;

The memory is used for storing programs and data, and the processor invokes the program stored in the memory to execute the method for real-time extraction and measurement of tunnel-like target features provided in any one of the first aspects.

In a fourth aspect, the present invention provides a computer-readable storage medium, where the computer-readable storage medium includes a program that, when executed by a processor, is used to execute the tunnel-type target feature real-time real-time feature provided in any one of the first aspects. Extraction and measurement methods.

As can be seen from the above description, the embodiments of the present invention provide a method and device for real-time extraction and measurement of tunnel-like target features. The present invention can extract tunnel-like target features in real-time during vehicle driving, and simultaneously realize the mathematical description of tunnel-like target features and The measurement of the height and width of tunnel-like targets meets the needs of intelligent driving in advance prediction, assists vehicle driving, and improves the safety and reliability of vehicle driving.

Description of drawings

1 is a schematic flowchart of a method for real-time extraction and measurement of tunnel-like target features provided by an embodiment of the present invention;

2 is a schematic diagram of identifying the boundary point cloud in the cross-section point cloud in an implementation of the present invention;

3 is a schematic structural diagram of an apparatus for real-time extraction and measurement of tunnel-like target features provided by an embodiment of the present invention;

FIG. 4 is a schematic structural diagram of an electronic device in an embodiment of the present invention.

Detailed ways

In order to make the objectives, technical solutions and advantages of the present invention clearer and more comprehensible, the present invention will be further described below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only some, but not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

Figure 1 shows a real-time extraction and measurement method of tunnel-like target features according to an embodiment of the present invention. In Figure 1, the method includes:

S101: Eliminate the interfering point cloud in the three-dimensional point cloud collected by the multi-line lidar to obtain an effective point cloud;

Specifically, the multi-line lidar is installed on the roof, and the 3D point cloud information in front of the vehicle is collected by the multi-line lidar, but the 3D point cloud collected by the multi-line lidar is mixed with redundant point clouds, invalid point clouds or isolated point clouds. For interfering point clouds such as point clouds, in this step, the 3D point clouds collected by the multi-line lidar are preprocessed to eliminate the interfering point clouds, obtain valid point clouds, reduce the amount of subsequent calculation, and improve the accuracy of the collected point clouds.

S102: Select the point cloud with the smallest elevation in the effective point cloud to form a ground point cloud, perform plane fitting on the ground point cloud and calculate the normal vector of the fitted plane to solve the pitch angle;

Specifically, the preprocessed point cloud contains the ground point cloud and the point cloud of tunnel-like target features above the ground. The actual road that the vehicle travels on generally has a slope, resulting in a certain gap between the multi-line lidar measurement plane and the road plane. Therefore, it is necessary to calculate the slope of the road (that is, the pitch angle), and fit the ground point cloud to obtain the fitted plane for the ground plane. By calculating the normal vector of the fitted plane Further, the slope of the road can be known.

S103: Compensate the pitch angle to the remaining valid point cloud after removing the ground point cloud;

Specifically, the pitch angle compensation is performed on the remaining valid point cloud after removing the ground point cloud according to the road slope obtained in the previous step, and the pitch angle compensation is performed on the valid point cloud obtained by the multi-line lidar to reduce the multi-line lidar. Measure the angle between the plane and the road plane, thereby reducing the measurement error.

S104: Extract the point cloud with the upward emission angle in the effective point cloud after the pitch angle compensation to form the feature point cloud of the tunnel-like target, and project the feature point cloud on the projection plane parallel to the axle and perpendicular to the advancing direction to generate a section point cloud ;

Specifically, since the multi-line lidar is installed on the roof, the multi-line lidar is at a certain distance from the ground, so the radar plane of the multi-line lidar is also higher than the ground, and the point cloud below the radar plane detects ground information. It is not the feature information of tunnel targets; the point cloud information located above the radar plane is an effective feature point cloud to describe the characteristics of tunnel targets. The feature point cloud contains the three-dimensional information of the tunnel target in front of the vehicle. Projected on a plane parallel to the axle and perpendicular to the forward direction, so that the section point cloud containing the section information of the tunnel-like target is integrated on this plane.

S105: Identify the boundary point cloud in the section point cloud, and use the boundary point cloud as the boundary point to divide the section point cloud into a top boundary point cloud, a left boundary point cloud and a right boundary point cloud;

Specifically, it can be understood that the cross-section of a tunnel-like target consists of two boundaries and a top edge connecting the two boundaries. How to extract or identify the feature information of the tunnel-like target represented by each part in the cross-section point cloud requires the opposite side The boundary and the top boundary are distinguished, and there is an obvious boundary between the side boundary and the top boundary. Through the boundary point cloud, it can be judged which cross-section point cloud belongs to the top boundary point cloud, and which belongs to the left boundary point cloud and the right boundary point cloud.

S106: construct a mathematical description model of the tunnel-like target section according to the differentiated section point cloud;

Specifically, according to the top boundary point cloud, the left boundary point cloud and the right boundary point cloud obtained in the previous step, the mathematical description model of the geometric description of the tunnel-like target can be constructed by ceres curve fitting.

S107: Calculate the height and width of the tunnel-like target according to the mathematical description model.

Specifically, according to the mathematical description model of the geometric description of the tunnel-like target, the width of the tunnel-like target can be measured by calculating the distance between the left boundary point cloud and the right boundary point cloud, and the tunnel can be measured by calculating the height of the highest point of the top boundary point cloud. The height of the class target.

In this embodiment, tunnel-like target features can be extracted in real time during vehicle driving, and at the same time, the mathematical description of tunnel-like target features and the measurement of the height and width of tunnel-like targets can be realized, which satisfies the needs of intelligent driving in advance prediction and assists Vehicle driving improves the safety and reliability of vehicle driving.

Based on the above embodiment, as a preferred embodiment, step 101 specifically includes the following steps:

Determining unnecessary point clouds not in the area in front of the vehicle;

Specifically, in order to meet the advance prediction of vehicle driving, point clouds that are not within a range of 30m*20m ahead of the vehicle are determined as redundant point clouds.

Determine the point cloud whose data value is not NAF as invalid point cloud;

Specifically, NAF is a non-adjacent representation type, and a point cloud whose point cloud data value type does not belong to NAF is determined as an invalid point cloud.

Determining a point cloud with no adjacent point cloud around it as an isolated point cloud;

Specifically, some point clouds in the point cloud collected by multi-line lidar are far away from adjacent point clouds, and cannot establish contact with adjacent point clouds, which affects the results of the clustering algorithm. The point cloud larger than 0.3m is regarded as an isolated point cloud.

Valid point clouds are obtained by excluding the redundant point clouds, invalid point clouds and isolated point clouds.

In this embodiment, the point cloud information collected by the multi-line lidar is preprocessed, and the redundant point cloud, invalid point cloud and isolated point cloud in the collected point cloud are eliminated, and these interference point clouds do not participate in the subsequent The feature description of tunnel objects and the measurement of width and height can improve the accuracy and accuracy of feature description and width and height of tunnel objects.

Based on the above embodiment, as a preferred embodiment, step 102 specifically includes the following steps:

The effective point cloud data is spatially rasterized, and each grid is a cuboid with the same bottom surface and a height that varies according to the spatial distribution of the point cloud;

The point clouds in each grid are sorted by elevation, and the lowest point cloud in each grid is compared with the lowest point cloud in the adjacent grid to obtain the point cloud with the lowest elevation, and the distance from the point cloud with the lowest elevation is less than the threshold range The point cloud inside is regarded as the ground point cloud;

Fit the ground point cloud to the plane, find the normal vector of the fitted plane, and solve the pitch angle according to the normal vector of the fitted plane.

In this embodiment, due to the existing slope of the actual road, there is an included angle between the radar plane of the multi-line lidar and the road plane, resulting in an error between the measured height of the tunnel-like target and the actual height of the tunnel. Therefore, it is necessary to compensate the pitch angle of the point cloud to reduce the generation of this error. The road gradient needs to be measured when compensating the point cloud, and the road gradient can be obtained quickly through the above steps.

Based on the above embodiment, as a preferred embodiment, step 104 specifically includes the following steps:

Set the XOY plane of the multi-line lidar coordinate system as the radar plane;

Specifically, the multi-line radar coordinate system is: the radar body is the origin, the front of the radar body is the Y axis, the right side of the radar body is the X axis, and the Z axis is upward, and the XOY plane of the multi-line lidar coordinate system is set as the radar plane.

Calculate the angle θ between the single-beam laser point cloud and the radar plane in the effective point cloud

In the formula: p _ix , p _iy , and p _iz represent the coordinate values of the i-th point cloud, respectively.

According to the radar installation angle and the radar vertical field of view parameters, the single-beam laser point cloud with the included angle θ between 10° and 15° is selected as the feature point cloud;

The feature point cloud is projected onto a plane perpendicular to the axle to generate the section point cloud of the tunnel-like target.

In this embodiment, it is determined whether the emission angle of the single laser beam is above the radar plane by calculating the angle between the single beam of laser light and the radar plane, and from the remaining effective point cloud after compensation, the angle of the emission angle above the radar plane is extracted. The single-beam laser point cloud excludes the ground point cloud collected by the multi-line laser radar, which reduces the amount of calculation and eliminates the influence of the ground point cloud on the measurement results. After excluding the ground point cloud, the characteristic point cloud of the tunnel-like target above the radar plane is obtained, and the characteristic point cloud is projected on the plane perpendicular to the axle to obtain the cross-section point cloud of the tunnel-like target, and then obtain the information of the tunnel-like target cross-section.

Based on the above embodiment, as a preferred embodiment, step 105 specifically includes the following steps:

Select the upper, middle, and lower adjacent point clouds from the cross-section point cloud to form a scanning triangle, and scan the cross-section point cloud into left and right parts clockwise and counterclockwise respectively from the front of the vehicle; take the middle point cloud as the vertex , calculate the angle of the vertices of the scanned triangle;

Specifically, the cross-section point cloud is actually consistent with the cross-section of the tunnel-like target. Dividing the cross-section point cloud into two parts for scanning can improve the recognition speed. The shape of the cross-section point cloud is shown in Figure 2. Three adjacent point clouds A, B, and C in the cross-section point cloud are selected, and point cloud B is selected as the vertex. The angle value of vertex B can be calculated according to the triangular cosine theorem.

Take the point at which the angle value of the vertices of the swept triangle changes abruptly as the boundary point cloud in the section point cloud;

Specifically, it can be understood that the cross-section of the tunnel-type target actually consists of left and right side edges and a top edge, and the connection between the left and right side edges and the top edge is actually a boundary point. The angle value of vertex B of the scanning triangle will also change during the scanning process. When the junction between the top edge and the side edge is scanned, the angle value of the vertex must change abruptly, and the point cloud with sudden change in angle value can be identified as the boundary point. cloud.

Take the point cloud before the boundary point as the top boundary point cloud of the tunnel target, the point cloud after the left half boundary point cloud as the left boundary point cloud, and the point cloud after the right half boundary point cloud as the right boundary point cloud;

Save the top boundary point cloud, left boundary point cloud, and right boundary point cloud in the top boundary container U_lanes, the left boundary container Left_lanes, and the right boundary container Right_lanes, respectively.

Specifically, the above-mentioned top boundary container U_lanes, left boundary container Left_lanes and right boundary container Right_lanes are actually point cloud arrays, which are used to store the classified point clouds and form a point cloud set of corresponding attributes, which is convenient for the top boundary point cloud, The left boundary point cloud and the right boundary point cloud are clustered and fitted.

In this embodiment, the acquired cross-section point cloud is scanned by triangles to identify its boundary point cloud, and then the cross-section point cloud is classified by the boundary point cloud, and the classified cross-section point cloud is saved in the boundary container corresponding to each category Through the above steps, the rapid identification of the cross-section data of the tunnel-type target is realized, and the cross-section data of the tunnel-type target is classified and stored, so as to realize the classification and extraction of the characteristics of the tunnel-type target, and facilitate the cross-section data of the tunnel-type target. analysis.

Based on the above embodiment, as a preferred embodiment, step 106 specifically includes the following steps:

Perform quadratic curve fitting on the top boundary point cloud to obtain the top boundary boundary curve parameters U(a ₁ ,b ₁ ,c), U(a ₁ ,b ₁ ,c) represents the top edge quadratic curve coefficient, and construct the top edge Quadratic curve model y=a ₁ x ² +b ₁ x+c;

It can be understood that the top boundary of tunnel targets is mostly in the form of a parabola, so the top boundary point cloud is fitted with a ceres curve fitting method to perform quadratic curve fitting, and a quadratic curve model is constructed for this purpose.

Perform straight line fitting on the left boundary point cloud and the right boundary point cloud to obtain the left and right boundary curve parameters L(a ₂ , b ₂ ) and R(a ₃ , b ₃ ), where L(a ₂ , b ₂ ) represents Left straight line coefficient; R(a ₃ , b ₃ ) right straight line coefficient, construct the left straight line model y=a ₂ x+b ₂ and the right straight line model y=a ₃ x+b ₃ ;

It can be understood that most of the left and right sides of the tunnel targets are in the form of straight lines, so the side boundaries are fitted with straight lines, and a straight line model is constructed for this purpose.

In this embodiment, through the mathematical fitting method, an accurate mathematical description of the tunnel-like target is realized, and an accurate external operating environment description data source is provided for the intelligent driving of the vehicle.

Based on the above embodiment, as a preferred embodiment, step 107 specifically includes the following steps:

According to the left and right boundary line model, according to the formula width=fabs(L_b ₂ -R_b ₃ ), calculate the target width;

In the formula: L_b ₂ represents the coefficient of the linear constant term of the left boundary; R_b ₃ respectively represents the coefficient of the linear constant term of the right boundary.

计算目标高度；According to the top boundary curve model, according to the formula:

Calculate the target height;

In the formula: h ₀ is the height of the multi-line lidar from the ground.

It can be understood that the height of the multi-line lidar from the ground is known data, which is mainly determined by the height of the vehicle.

In this embodiment, the measurement of the width and height of the tunnel-like target is realized by the calculation of the parameters in the mathematical description formed by the tunnel-like target, which improves the perception of the tunnel-like target by the intelligent driving, and provides the driving safety of the vehicle. Assure.

To sum up, the present invention realizes real-time measurement of the height and width of tunnel-like targets, and the data update frequency reaches 10HZ; it has the capability of advanced measurement, and the measurement accuracy can reach 0.5m; the use conditions are not affected by changes in ambient light, and in backlight, night and other environments can work normally.

Based on the same inventive concept, an embodiment of the present application also provides a device for real-time extraction and measurement of tunnel-like target features, which can be used to implement the method for real-time extraction and measurement of tunnel-like target features described in the above embodiments, as follows: example. Since the principle of a real-time extraction and measurement device for tunnel-like target features is similar to a method for real-time extraction and measurement of tunnel-like target features, the implementation of a real-time extraction and measurement device for tunnel-like target features can refer to the implementation of the method, repeating will not be repeated here. As used below, the term "module" may be a combination of software and/or hardware that implements a predetermined function. Although the apparatus for real-time extraction and measurement of tunnel-like target features described in the following embodiments is preferably implemented in software, implementation in hardware or a combination of software and hardware is also possible and conceivable.

As shown in FIG. 2, an embodiment of the present invention provides a real-time extraction and measurement device for tunnel-type target features, the device includes:

Point cloud preprocessing module 201: Eliminate the interfering point cloud in the three-dimensional point cloud collected by the multi-line lidar, and obtain an effective point cloud;

Slope calculation module 202: used to select the point cloud with the smallest elevation in the valid point cloud to form a ground point cloud, perform plane fitting on the ground point cloud and calculate the normal vector of the fitted plane to solve the pitch angle;

Compensation module 203: used to compensate the pitch angle for the remaining valid point cloud after removing the ground point cloud;

Extraction module 204: for extracting the point cloud with the upward emission angle in the effective point cloud after pitch angle compensation to form the characteristic point cloud of the tunnel-like target, and projecting the characteristic point cloud on the projection plane parallel to the axle and perpendicular to the advancing direction Generate section point cloud;

Classification module 205: used to identify the boundary point cloud in the cross-section point cloud, and use the boundary point cloud as the boundary point to classify the cross-section point cloud into a top boundary point cloud, a left boundary point cloud and a right boundary point cloud;

Building module 206: for constructing a mathematical description model of the tunnel-like target section according to the differentiated section point cloud;

Calculation module 207: used to calculate the height and width of the tunnel-like target according to the mathematical description model.

Embodiments of the present application also provide specific implementations of an electronic device that can implement all the steps in the method for real-time extraction and measurement of tunnel-like target features in the foregoing embodiment. Referring to FIG. 3 , the electronic device 300 specifically includes the following contents:

processor 310, memory 320, communication unit 330 and bus 340;

Among them, the processor 310, the memory 320, and the communication unit 330 communicate with each other through the bus 340; the communication unit 330 is used to realize the information transmission between the server-side equipment and the terminal equipment and other related equipment.

The processor 310 is configured to call the computer program in the memory 320, and when the processor executes the computer program, all steps in the method for real-time extraction and measurement of tunnel-like target features in the foregoing embodiment are implemented.

Those of ordinary skill in the art should understand that the memory can be, but is not limited to, random access memory (Random Access Memory, referred to as: RAM), read only memory (Read Only Memory, referred to as: ROM), programmable read only memory (Programmable Read Only Memory) -Only Memory, referred to as: PROM), Erasable Programmable Read-Only Memory (referred to as: EPROM), Electrically Erasable Programmable Read-Only Memory (Electric Erasable Programmable Read-Only Memory, referred to as: EEPROM) and so on. The memory is used to store the program, and the processor executes the program after receiving the execution instruction.

Further, the software programs and modules in the above-mentioned memory may also include an operating system, which may include various software components and/or drivers for managing system tasks (such as memory management, storage device control, power management, etc.), and may Intercommunicate with various hardware or software components to provide the operating environment for other software components.

The processor may be an integrated circuit chip with signal processing capability. The aforementioned processor may be a general-purpose processor, including a central processing unit (Central Processing Unit, CPU for short), a network processor (Network Processor, NP for short). The methods, steps, and logic block diagrams disclosed in the embodiments of this application can be implemented or executed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.

The present application further provides a computer-readable storage medium, where the computer-readable storage medium includes a program, which, when executed by a processor, is used to perform real-time extraction of tunnel-like target features provided by any of the foregoing method embodiments and measurement methods.

Those of ordinary skill in the art should understand that all or part of the steps of implementing the above method embodiments may be completed by program instructions related to hardware. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the steps including the above method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk and other media that can store program codes, and the specific medium type is not limited in this application. .

The above description is only a preferred embodiment of the present invention, but the protection scope of the present invention is not limited to this. Substitutions should be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims (10)

1. A real-time extraction and measurement method for tunnel type target features is characterized by comprising the following steps:

eliminating interference point clouds in the three-dimensional point clouds collected by the multi-line laser radar to obtain effective point clouds;

selecting the point cloud with the minimum height range in the effective point cloud to form ground point cloud, performing plane fitting on the ground point cloud, and calculating a normal vector of a fitting plane to solve a pitch angle;

after ground point clouds are removed, the pitch angle of the residual effective point clouds is compensated;

extracting point clouds with upward emission angles in the effective point clouds after pitch angle compensation to form a characteristic point cloud of a tunnel type target, and projecting the characteristic point cloud on a projection plane which is parallel to the axle and vertical to the advancing direction to generate a section point cloud;

identifying boundary point clouds in the section point clouds, and distinguishing the section point clouds into top boundary point clouds, left boundary point clouds and right boundary point clouds by taking the boundary point clouds as boundary points;

constructing a mathematical description model of the tunnel type target section according to the distinguished section point cloud;

and calculating the height and width of the tunnel-type target according to the mathematical description model.

2. The method for extracting and measuring the target feature of the tunnel type according to claim 1, wherein the step of eliminating the interference point cloud in the three-dimensional point cloud collected by the multi-line laser radar to obtain the effective point cloud comprises the following steps:

determining an area not in front of the vehicle as an redundant point cloud;

determining the point cloud with the data value not being NAF as invalid point cloud;

determining a point cloud without adjacent point clouds around as an isolated point cloud;

and removing the redundant point clouds, the invalid point clouds and the isolated point clouds to obtain the valid point clouds.

3. The method for extracting and measuring the target features of the tunnel type according to claim 1, wherein the selecting the point cloud with the minimum height in the effective point clouds to form a ground point cloud, and performing plane fitting on the ground point cloud and calculating a normal vector solution pitch angle of a fitting plane comprises:

rasterizing the effective point cloud data space, wherein each grid is a cuboid with the same bottom surface and the height changed according to the point cloud space distribution;

the point clouds in each grid are sorted according to the elevation, the point cloud with the lowest elevation in each grid is compared with the point cloud with the lowest elevation in the adjacent grid, the point cloud with the lowest elevation is obtained, and the point cloud with the lowest elevation and the distance smaller than the threshold range is used as the ground point cloud;

and performing plane fitting on the ground point cloud, solving a normal vector of a fitting plane, and solving the pitch angle according to the normal vector of the fitting plane.

4. The method for extracting and measuring the characteristics of the tunnel-like target in real time according to claim 1, wherein the extracting of the point clouds with upward emission angles from the effective point clouds after the pitch angle compensation to form the characteristic point clouds of the tunnel-like target, and the projecting of the characteristic point clouds on the projection plane parallel to the axle and perpendicular to the advancing direction to generate the cross-section point clouds comprises:

determining an XOY plane of a multiline laser radar coordinate system as a radar plane;

calculating the included angle theta between the single-beam laser point cloud and the radar plane in the effective point cloud

Screening single-beam laser point clouds with an included angle theta of 10-15 degrees as characteristic point clouds according to the radar installation angle and the radar vertical field angle parameter;

and projecting the characteristic point cloud to a plane vertical to the axle to generate the cross-section point cloud of the tunnel type target.

5. The method for extracting and measuring characteristics of a tunnel-like target in real time according to claim 1, wherein the step of identifying a boundary point cloud in the cross-section point cloud, and the step of dividing the cross-section point cloud into a top boundary point cloud, a left boundary point cloud and a right boundary point cloud by using the boundary point cloud as a boundary point comprises the steps of:

selecting upper, middle and lower adjacent point clouds from the cross-section point clouds to form a scanning triangle, and dividing the cross-section point clouds into a left part and a right part from the right front of the vehicle according to clockwise and anticlockwise directions for scanning; calculating the angle of the vertex of the scanning triangle by taking the intermediate point cloud as the vertex;

taking the point with the abrupt change of the angle value of the scanning triangle vertex as the boundary point cloud in the cross-section point cloud;

taking the point cloud before the boundary point as the top boundary point cloud of the tunnel type target, taking the point cloud after the boundary point cloud of the left half part as the left boundary point cloud, and taking the point cloud after the boundary point cloud of the right half part as the right boundary point cloud;

and respectively storing the top boundary point cloud, the left boundary point cloud and the right boundary point cloud in a top boundary container, a left boundary container and a right boundary container.

6. The method for extracting and measuring the characteristics of the tunnel-type target in real time according to claim 1, wherein the step of constructing a mathematical description model of the cross section of the tunnel-type target according to the differentiated cross-section point cloud comprises the following steps:

performing quadratic curve fitting on the top boundary point cloud to obtain a top boundary curve parameter U (a)₁,b₁,c)，U(a₁,b₁And c) representing the coefficient of the top edge secondary curve, and constructing a top edge secondary curve model y as a₁x²+b₁x+c；

Performing straight line fitting on the left boundary point cloud and the right boundary point cloud to obtain a left boundary curve parameter L and a right boundary curve parameter L (a)₂,b₂) And R (a)₃,b₃) Wherein L (a)₂,b₂) Represents the left straight line coefficient;R(a₃,b₃) And constructing a left straight line model y as a by using the right straight line coefficient₂x+b₂And the right straight line model y ═ a₃x+b₃。

7. The method for extracting and measuring the characteristics of the tunnel-like target in real time according to claim 6, wherein the step of calculating the height and the width of the tunnel-like target according to the mathematical description model comprises the following steps:

calculating a target width according to a formula width, namely fabs (L _ b-R _ b) according to the left and right boundary straight line model;

in the formula: l _ b represents a left boundary linear constant term coefficient; r _ b respectively represents a right boundary linear constant term coefficient;

according to the top boundary curve model and the formula:

calculating the target height;

in the formula: h is₀The height of the multiline laser radar from the ground is adopted.

8. The utility model provides a real-time extraction of tunnel class target feature and measuring device which characterized in that, the device includes:

a point cloud preprocessing module: eliminating interference point clouds in the three-dimensional point clouds collected by the multi-line laser radar to obtain effective point clouds;

a gradient calculation module: the system is used for selecting the point cloud with the minimum height in the effective point cloud to form ground point cloud, performing plane fitting on the ground point cloud, and calculating a normal vector of a fitting plane to solve a pitch angle;

a compensation module: the system is used for compensating a pitch angle for the residual effective point cloud after the ground point cloud is removed;

an extraction module: the system is used for extracting point clouds with upward emission angles from the effective point clouds after pitch angle compensation to form a characteristic point cloud of a tunnel type target, and projecting the characteristic point cloud on a projection plane which is parallel to an axle and vertical to the advancing direction to generate a section point cloud;

a classification module: the system is used for identifying boundary point clouds in the section point clouds and distinguishing the section point clouds into top boundary point clouds, left boundary point clouds and right boundary point clouds by taking the boundary point clouds as boundary points;

constructing a module: the mathematical description model is used for constructing a tunnel type target section according to the distinguished section point cloud;

a calculation module: the method is used for calculating the height and the width of the tunnel-type target according to the mathematical description model.

9. An electronic device, comprising:

a processor, a memory, an interface to communicate with a gateway;

the memory is used for storing programs and data, and the processor calls the programs stored in the memory to execute the tunnel class target feature real-time extraction and measurement method of any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that the computer-readable storage medium comprises a program which, when executed by a processor, is configured to perform the method for real-time extraction and measurement of target features of tunnel classes of any of claims 1 to 7.

Images