Target recognition method and target recognition device

Abstract

The present invention provides a target recognition method and a target recognition device. The method includes: processing a point cloud image to obtain a static point cloud set; processing the point clouds in the static point cloud set to obtain at least one subset. For any subset, all the point clouds in the subset are located on the same straight line, and the subset includes all the point clouds in the static point cloud set that are located on this straight line; determining a target straight line according to the at least one subset; obtaining a feature point cloud set for representing the point cloud distribution feature of the target straight line according to the target straight line and the static point cloud set, and calculating the point cloud distribution feature of the target straight line; judging whether there is a building according to the point cloud distribution feature of the target straight line; if there is a building, removing the targets located outside the building according to the position of the building. The present invention can effectively remove the false targets outside the building, thereby improving the target recognition accuracy of the vehicle-mounted radar.

Description

Technical Field

[0001] This invention relates to the field of radar detection technology, and in particular to a target identification method and a target identification device. Background Technology

[0002] Currently, millimeter-wave radars offer several advantages. Firstly, their higher operating frequency allows for easy acquisition of a wider bandwidth. Secondly, their shorter wavelength enables smaller antennas, facilitating convenient installation on vehicles. Furthermore, the shorter wavelength makes them more sensitive to speed. Therefore, millimeter-wave radars are widely used in autonomous driving and intelligent driver assistance systems.

[0003] The primary function of millimeter-wave radar in vehicle driving systems is target detection and tracking, effectively acquiring information such as the target's distance, angle, and speed. In parking mode, due to the complex environment of a garage, building targets also generate echo signals. These echo signals contain multipath information, causing the radar to detect some false targets, thus reducing the radar's target detection accuracy. Summary of the Invention

[0004] In view of this, the present invention provides a target recognition method and a target recognition device, which can solve the problem of low target detection accuracy of radar during parking.

[0005] In a first aspect, embodiments of the present invention provide a target recognition method, comprising:

[0006] The point cloud image is processed to obtain a static point cloud set, wherein the static point cloud set includes all point clouds in the point cloud image that are in a static state, and the point cloud image is obtained by processing the echo signal received by the radar installed on the vehicle.

[0007] The point clouds in the static point cloud set are processed to obtain at least one subset. For any subset, all point clouds in the subset are located on the same straight line, and the subset includes all point clouds in the static point cloud set that are located on the straight line. The number of point clouds in each subset of the at least one subset is greater than 2.

[0008] Determine the target line based on at least one subset;

[0009] Based on the target straight line and the static point cloud set, a feature point cloud set is obtained to represent the point cloud distribution characteristics of the target straight line, and the point cloud distribution characteristics of the target straight line are calculated, wherein the point cloud distribution characteristics include the length of the line segment corresponding to the target straight line;

[0010] Based on the point cloud distribution characteristics of the target straight line, determine whether a building exists;

[0011] If a building exists, targets located outside the building are removed based on the building's location.

[0012] In one possible implementation, processing the point cloud image to obtain a static point cloud set includes:

[0013] For any point cloud in the point cloud image, if the point cloud satisfies a first preset formula, then the point cloud is determined to be a stationary point cloud. The first preset formula is:

[0014] |V car ×cos(θ p )+V p |≤ε

[0015] Among them, V car θ is used to represent the vehicle speed. p V is used to represent the angle of the point cloud. p The velocity of the point cloud is used to represent the speed of the point cloud, and ε is used to represent the first preset value;

[0016] Obtain all point clouds in the point cloud image that satisfy the first preset formula to obtain the static point cloud set.

[0017] In one possible implementation, the static point cloud set includes s point clouds, and the process of processing the point clouds in the static point cloud set to obtain at least one subset includes:

[0018] For each stationary point cloud P in the set of stationary point clouds j Based on n preset different angle values, construct n straight lines passing through the stationary point cloud in the xy coordinate system, and determine the perpendicular distance of each line to the origin. Obtain the perpendicular distance corresponding to each angle value, and construct P. j n sets of polar coordinate parameters (θ) in the polar coordinate system based on Hough transform i , ρ i ), where for any angle value, the angle value is greater than or equal to 0 degrees and the angle value is less than 180 degrees, where 1≤i≤n, 1≤j≤s;

[0019] For the angle θ among the n different angle values i Traverse the polar coordinate parameters of each point cloud in the s point clouds corresponding to the angle value, and obtain all point clouds that intersect at the same point in the polar coordinate system at the angle value, forming a subset;

[0020] By iterating through each of the n different angle values, at least one subset is obtained.

[0021] In one possible implementation, the angle value θ among the n different angle values...i By iterating through the s point clouds, the polar coordinate parameters corresponding to each point cloud at that angle value are obtained, resulting in all point clouds intersecting at the same point in the polar coordinate system at that angle value, including:

[0022] In the polar coordinate system, when the angle value is θ i If the absolute value of the difference between the two ρ values ​​of the corresponding curves of two point clouds in the polar coordinate system is less than or equal to the second preset value, then it is determined that the curves corresponding to the two point clouds intersect at the same point in the polar coordinate system at that angle value.

[0023] In one possible implementation, if the at least one subset includes only one subset, then determining the target line based on the at least one subset includes:

[0024] The target straight line is obtained by performing linear fitting on the point cloud in the subset;

[0025] If the at least one subset includes multiple subsets, then determining the target line based on the at least one subset includes:

[0026] Among the plurality of subsets, determine the subset with the largest number of point clouds, and perform linear fitting on the point clouds in the subset with the largest number of point clouds to obtain the target straight line;

[0027] Alternatively, the subsets can be sorted in descending order of the number of point clouds, the first m subsets can be selected, and the point clouds in the m subsets can be linearly fitted sequentially to obtain the target line, wherein the target line includes m lines.

[0028] Alternatively, among the plurality of subsets, determine y subsets with a point cloud quantity greater than or equal to a third preset value, and sequentially perform linear fitting on the point clouds in the y subsets to obtain the target straight line, wherein the target straight line includes y straight lines.

[0029] In one possible implementation, obtaining a feature point cloud set representing the point cloud distribution characteristics of the target line based on the target line and the static point cloud set, and calculating the point cloud distribution characteristics of the target line includes:

[0030] Calculate the Euclidean distance between each point cloud in the static point cloud set and the target line in sequence;

[0031] Obtain all point clouds whose Euclidean distance to the target line is less than or equal to a fourth preset value to obtain the feature point cloud set. In the feature point cloud set, the point clouds whose Euclidean distance to the target line is greater than zero and less than or equal to the fourth preset value are the point clouds located near the target line.

[0032] Based on the feature point cloud set, calculate the point cloud distribution characteristics of the target straight line.

[0033] In one possible implementation, calculating the point cloud distribution features of the target straight line based on the feature point cloud set includes:

[0034] Calculate the difference between the maximum and minimum y-coordinate values ​​in the feature point cloud set to obtain the length of the line segment corresponding to the target line;

[0035] Alternatively, the length of the line segment corresponding to the target line can be obtained by calculating the difference between the maximum and minimum x-coordinate values ​​in the feature point cloud set.

[0036] In one possible implementation, the point cloud distribution features of the target straight line further include the density features of the line segments corresponding to the target straight line, and the step of calculating the point cloud distribution features of the target straight line based on the feature point cloud set includes:

[0037] The line segment corresponding to the target straight line is divided into a predetermined number of continuous but non-overlapping sub-segments.

[0038] Based on the feature point cloud set, the number of point clouds contained in each sub-segment is determined. The number of point clouds contained in each sub-segment is used to represent the density characteristics of the line segment corresponding to the target line. For any sub-segment, the number of point clouds contained in the sub-segment includes the point cloud located in the line segment and the point cloud located near the sub-segment.

[0039] In one possible implementation, determining whether a building exists based on the point cloud distribution characteristics of the target straight line includes:

[0040] If the number of point clouds on the target straight line is greater than or equal to a fifth preset value, and the length of the line segment corresponding to the target straight line is greater than or equal to a preset length, and the number of point clouds contained in each sub-line segment is greater than or equal to a sixth preset value, then it is determined that a building exists.

[0041] In a second aspect, embodiments of the present invention provide a target identification device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method as described in the first aspect or any possible implementation thereof.

[0042] The beneficial effects of the embodiments of the present invention compared with the prior art are as follows:

[0043] This invention provides a static point cloud set by processing point cloud images. In a parking environment, the spatial point cloud distribution of targets such as buildings, walls, and pillars is closest to a straight line. By analyzing the static point cloud set, a straight line is obtained. By analyzing the characteristics of the straight line, it is determined whether a building, such as a wall, exists. If it does, targets located outside the building, i.e., false targets, are removed from all identified targets, thereby improving the radar's identification accuracy in a parking environment. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1 This is a flowchart illustrating the implementation of a target recognition method provided in an embodiment of the present invention;

[0046] Figure 2 This is a schematic diagram illustrating the correspondence between point cloud velocity and vehicle velocity, provided by an embodiment of the present invention;

[0047] Figure 3 This is a flowchart illustrating the implementation of another target recognition method provided in an embodiment of the present invention;

[0048] Figure 4 This is a schematic diagram of the mapping relationship between points and lines in two coordinate spaces based on the Hough transform, provided by an embodiment of the present invention.

[0049] Figure 5 This is a schematic diagram illustrating the parameter transformation correspondence between a straight line based on Hough transform in rectangular coordinates and polar coordinates, provided by an embodiment of the present invention.

[0050] Figure 6 This is a schematic diagram of the target recognition device provided in an embodiment of the present invention;

[0051] Figure 7 This is a schematic diagram of the target recognition device provided in an embodiment of the present invention. Detailed Implementation

[0052] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of the invention. However, those skilled in the art will understand that the invention can be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

[0053] To make the objectives, technical solutions, and advantages of the present invention clearer, specific embodiments will be described below in conjunction with the accompanying drawings.

[0054] See Figure 1 The diagram illustrates a flowchart of a target recognition method provided by an embodiment of the present invention, which is described in detail below:

[0055] In step 101, the point cloud image is processed to obtain a static point cloud set.

[0056] The stationary point cloud set includes all point clouds in a stationary state in the point cloud image, which is obtained by processing the echo signals received by the radar installed on the vehicle.

[0057] In this embodiment of the invention, the radar installed on the vehicle can be a millimeter-wave radar or other types of radar; this embodiment does not limit the type. Taking millimeter-wave radar as an example, millimeter-wave radar point cloud imaging refers to obtaining a set of data points on the surface of an object through a measuring instrument, typically represented using three-dimensional coordinates. First, the echo signal received by the millimeter-wave radar is processed to extract the target point's position, velocity, and angular information such as azimuth and elevation angles. Finally, three-dimensional imaging processing is performed. After three-dimensional imaging, the coordinate values ​​of each point in the three-dimensional coordinate system are obtained, ultimately yielding the point cloud image corresponding to that frame of echo signal.

[0058] Figure 2 This invention provides a schematic diagram illustrating the correspondence between point cloud velocity and vehicle velocity, as shown in the embodiment of the invention. Figure 2 For any point cloud in the point cloud image, if the point cloud satisfies a first preset formula, then the point cloud is determined to be a stationary point cloud. The first preset formula is:

[0059] |V car ×cos(θ p )+ p |≤ε

[0060] Among them, V car θ is used to represent the speed of a vehicle. p V is used to represent the angle of the point cloud. p The velocity of the point cloud is used to represent the speed of the point cloud, and ε is used to represent the first preset value;

[0061] Obtain all point clouds in the point cloud image that satisfy the first preset formula to obtain a static point cloud set.

[0062] In step 102, the point clouds in the static point cloud set are processed to obtain at least one subset. For any subset, all point clouds in the subset are located on the same straight line, and the subset includes all point clouds in the static point cloud set that are located on the same straight line.

[0063] Optionally, in this step, at least one subset can be obtained through Hough transform, or at least one subset required in this step can be obtained through any of the existing line extraction methods. This embodiment of the invention does not limit this.

[0064] By using the line extraction method, all point clouds in the static point cloud set that are on the same straight line are obtained, and for each straight line, the number of point clouds on that straight line is greater than 2. The point clouds that are on the same straight line are placed into a subset, which is a subset in this step. In this step, all subsets that meet the above conditions are obtained by using the line extraction method.

[0065] In step 103, the target line is determined based on at least one subset.

[0066] In one possible implementation, if at least one subset in step 102 includes only one subset, then the point cloud in that subset is linearly fitted to obtain the target straight line.

[0067] In another possible implementation, if at least one subset in step 103 comprises multiple subsets, then:

[0068] In one alternative implementation, the subset with the largest number of point clouds is determined from multiple subsets, and the point clouds in the subset with the largest number of point clouds are linearly fitted to obtain the target straight line.

[0069] In one optional implementation, multiple subsets are sorted in descending order of the number of point clouds, and the first m subsets are selected. Linear fitting is then performed sequentially on the point clouds in each of the m subsets to obtain a target straight line, where the target straight line comprises m straight lines. For example, at least one subset in step 102 comprises four subsets: subset 1 (48 point clouds), subset 2 (46 point clouds), subset 3 (35 point clouds), and subset 4 (12 point clouds). In this embodiment, the two subsets with the largest number of point clouds, namely subset 1 and subset 2, are selected, and linear fitting is performed on the point clouds in subset 1 and subset 2 respectively to obtain two straight lines.

[0070] In one optional implementation, y subsets with a point cloud count greater than or equal to a third preset value are determined from multiple subsets. Linear fitting is then performed sequentially on the point clouds in each of the y subsets to obtain a target straight line, where the target straight line comprises y straight lines. For example, at least one subset in step 102 comprises five subsets: subset 1 (48 point clouds), subset 2 (46 point clouds), subset 3 (35 point clouds), subset 4 (12 point clouds), and subset 5 (4 point clouds). In this embodiment, subsets with a point cloud count greater than or equal to 40, namely subset 1 and subset 2, are selected. Linear fitting is then performed on the point clouds in subset 1 and subset 2 respectively to obtain two straight lines.

[0071] In this embodiment of the invention, the point cloud in the subset can be linearly fitted by the least squares method, or other methods can be used to linearly fit the point cloud in the subset. This embodiment of the invention does not limit the specific method used.

[0072] Least squares is a linear regression algorithm whose basic principle is to minimize variance. Let's take a two-dimensional array as an example, with n sets of data [(x1, y1), (x2, y2), ..., (x...]. n y n The linear expression y = ax + b is formed, and the least squares method represents solving for the optimal solution of Formula 1, which is:

[0073]

[0074] This problem is an unconstrained optimization problem. Taking the derivative with respect to (a, b) respectively, we can obtain the slope a and the intercept b:

[0075]

[0076]

[0077] Assume the subset selected in this step is Q. M Q M Substituting the point cloud data into the above formula, we can solve for (a, b) and obtain the equation of the straight line y = a. q x+b q .

[0078] In step 104, a feature point cloud set is obtained to represent the point cloud distribution characteristics of the target line based on the target line and the static point cloud set, and the point cloud distribution characteristics of the target line are calculated, including the length of the line segment corresponding to the target line.

[0079] Optionally, the Euclidean distance between each point cloud in the static point cloud set and the target line is calculated sequentially; all point clouds whose Euclidean distance to the target line is less than or equal to a fourth preset value are obtained to form a feature point cloud set. In the feature point cloud set, the point clouds whose Euclidean distance to the target line is greater than zero and less than or equal to the fourth preset value are the point clouds located near the target line; based on the feature point cloud set, the point cloud distribution characteristics of the target line are calculated.

[0080] Given the equation of the straight line y = a q x+b q The Euclidean distance between a point cloud (x0, y0) in space and the line is calculated using Formula 2, which is:

[0081]

[0082] Based on this, in this embodiment of the invention, the static point cloud set is represented by Q. S It means that Q S It includes S point clouds, Q S =[(R x R y )1, (R x R y )2,…,(R x R y ) s ], traverse the set of static point clouds, for any point (R) in the set of static point clouds x R y ) i Calculate the distance between this point and the target line:

[0083]

[0084] if d i If the distance is less than or equal to ε, it indicates that the point cloud has an Euclidean distance to the target line that is less than or equal to the fourth preset value, where ε is the preset minimum distance threshold, i.e., the fourth preset value in this step. Assuming ε = 0.5, all point clouds with an Euclidean distance to the target line that is less than or equal to the fourth preset value are obtained to obtain the feature point cloud set. In this embodiment of the invention, the feature point cloud set can be Q. H This is represented by the feature point cloud set. Since the feature point cloud set is used to represent the point cloud distribution characteristics of the target line, the point cloud distribution characteristics of the target line, such as the length of the line segment corresponding to the target line, can be analyzed based on the feature point cloud set.

[0085] Optionally, the difference between the maximum and minimum y-coordinate values ​​in the feature point cloud set can be calculated to obtain the length of the line segment corresponding to the target line; or, the difference between the maximum and minimum x-coordinate values ​​in the feature point cloud set can be calculated to obtain the length of the line segment corresponding to the target line.

[0086] For example, the maximum value of the y-coordinate in the feature point cloud set is Y. max The minimum value is Y min Then the length of the line segment corresponding to the target line is LH = Y. max -Y min .

[0087] In step 105, the presence of buildings is determined based on the point cloud distribution characteristics of the target line.

[0088] In real-world parking environments, buildings exhibit relatively obvious straight-line characteristics. Based on this principle, the length of a straight line can be used to determine whether a target is a building. Different buildings exhibit different line segment length characteristics. Based on this principle, the length of the line segment corresponding to the target's straight line can be analyzed to distinguish buildings from other targets, as well as to distinguish the type of building, such as a column or a wall.

[0089] Optionally, the point cloud distribution features of the target line also include the density features of the line segments corresponding to the target line. Calculating the point cloud distribution features of the target line based on the feature point cloud set further includes: dividing the line segments corresponding to the target line into a predetermined number of continuous but non-overlapping sub-segments; and determining the number of point clouds contained in each sub-segment based on the feature point cloud set. The number of point clouds contained in each sub-segment is used to represent the density features of the line segments corresponding to the target line. For any sub-segment, the number of point clouds contained in the sub-segment includes the point clouds located in the sub-segment and the point clouds located near the sub-segment.

[0090] For example, the line segment corresponding to the target line is divided into three continuous but non-overlapping segments. Based on the point cloud in the feature point cloud set, the line segment corresponding to the target line is divided into three segments using the following formula 3: Formula 3 is:

[0091]

[0092] The density characteristics of the line segment corresponding to the target line are obtained by counting the number of point clouds included in each sub-line segment.

[0093] Optionally, if the number of point clouds on the target line is greater than or equal to the fifth preset value, and the length of the line segment corresponding to the target line is greater than or equal to the preset length, and the number of point clouds contained in each sub-line segment is greater than or equal to the sixth preset value, then it is determined that a building exists.

[0094] For example, taking a building as a wall, if the fifth preset value is 40, the preset length is 3 meters, and the sixth preset value is 10, then if the above conditions are met simultaneously, it is determined that a building exists and the building is a wall.

[0095] In one possible implementation, if the process of determining the target line in step 103 is carried out in the following manner: y subsets with a point cloud quantity greater than or equal to a third preset value are determined from multiple subsets, and linear fitting is performed sequentially on the point clouds in the y subsets to obtain the target line, wherein the target line comprises y lines. For example, at least one subset in step 102 comprises 5 subsets: subset 1, comprising 48 point clouds; subset 2, comprising 46 point clouds; subset 3, comprising 35 point clouds; subset 4, comprising 12 point clouds; and subset 5, comprising 4 point clouds. In this embodiment of the invention, subsets with a point cloud quantity greater than or equal to 40, namely subset 1 and subset 2, are selected, and linear fitting is performed on the point clouds in subset 1 and subset 2 respectively to obtain two lines. Since the number of point clouds in the subset is already limited to 40 or more in this step, the judgment rule in determining the wall can also be set as follows: if the number of point clouds in the feature point cloud set corresponding to the target straight line is greater than or equal to a seventh preset value, and the length of the line segment corresponding to the target straight line is greater than or equal to a preset length, and the number of point clouds contained in each sub-line segment is greater than or equal to a sixth preset value, then it is determined that a building exists. The seventh preset value is greater than the fifth preset value. For example, when the number of point clouds located on the straight line is used as the judgment criterion, the fifth preset value is set to 40. Therefore, when the number of point clouds in the feature point cloud set corresponding to the target straight line is used as the judgment criterion, the seventh preset value can be set to a value greater than 40, such as 50. The specific numerical settings can be modified according to the actual application scenario, and this embodiment of the invention does not limit this.

[0096] In step 106, if a building exists, targets located outside the building are removed based on the building's location.

[0097] The feature point cloud set is also used to represent the location of buildings.

[0098] If a building is determined to exist through step 105, the feature point cloud set can be used to represent the location of the building. By removing targets located outside the building through the location of the building, false targets in the identification results can be effectively removed, thereby improving the radar's identification accuracy.

[0099] This invention provides a target recognition method. By processing point cloud images, a static point cloud set is obtained. Since the spatial point cloud distribution of targets such as buildings, walls, and pillars is closest to a straight line in a parking environment, a straight line is obtained by data analysis of the static point cloud set. By analyzing the characteristics of the straight line, it is determined whether a building, such as a wall, exists. If it does, targets located outside the building, i.e., false targets, are removed from all identified targets, thereby improving the radar recognition accuracy in a parking environment.

[0100] Figure 3A flowchart illustrating the implementation of the target recognition method provided in an embodiment of the present invention is shown below, and is described in detail below:

[0101] In step 301, the point cloud image is processed to obtain a static point cloud set.

[0102] The stationary point cloud set includes all point clouds in a stationary state in the point cloud image, which is obtained by processing the echo signals received by the radar installed on the vehicle.

[0103] For details on how to implement this step, please refer to [link / reference]. Figure 1 The steps 101 in the corresponding embodiments will not be repeated in this embodiment of the invention.

[0104] In step 302, for each stationary point cloud P in the set of stationary point clouds... j Based on n preset different angle values, construct n straight lines passing through the stationary point cloud in the xy coordinate system, and determine the perpendicular distance of each line to the origin. Obtain the perpendicular distance corresponding to each angle value, and construct P. j n sets of polar coordinate parameters (θ) in the polar coordinate system based on Hough transform i , ρ i ).

[0105] For any given angle value, the angle value is greater than or equal to 0 degrees and less than 180 degrees, where 1≤i≤n and 1≤j≤s.

[0106] The Hough transform is a feature extraction method used in digital image processing. It utilizes the transformation between two coordinate spaces to map curves or lines of the same shape in one coordinate space to a point in the other space in the form of peaks. The main idea of ​​the Hough transform is to exchange the parameters and variables of an equation. A straight line in a Cartesian coordinate system can be represented by y = kx + b, where (x, y) are variables and (k, b) are parameters. By exchanging the parameter space and the variable space, the equation b = y - kx is represented using the parameter space (k, b). Assuming a point (x0, y0) is known in the variable space, it is represented in the parameter space as a straight line b = -x0k + y0. Thus, the problem of "collinear points" in the (x, y) space is transformed into the problem of "intersecting lines" in the (k, b) parameter space, and the number of intersecting lines equals the number of collinear points.

[0107] The core idea of ​​the Hough transform of a straight line is the mapping relationship between points and lines. Figure 4 This is a schematic diagram illustrating the mapping relationship between points and lines in two coordinate spaces based on the Hough transform. (See attached diagram) Figure 4 The kb coordinate system and the xy coordinate system have the following... Figure 4The three dashed lines in the diagram represent the corresponding relationships. Point P1 in the xy-coordinate system corresponds to line L1 in the kb-coordinate system; similarly, point P2 corresponds to line L2. The intersection point P3(k0, b0) in the kb-coordinate system corresponds to the slope and intercept of line L0 in the xy-coordinate system. In other words, if several lines in the kb-coordinate system intersect at a single point, then several points on that common line in the xy-coordinate system will correspond to it. Based on this property, given some discrete points on the common line, the equation of the line connecting these discrete points can be found using the Hough transform.

[0108] However, there are cases where a straight line in the xy coordinate system is perpendicular or nearly perpendicular to the x-axis. In such cases, the slope k of this line is infinite, making it impossible to represent in the kb coordinate system. To solve this problem, it can be represented in the polar coordinate system. The parametric equation expression in the polar coordinate system is as follows:

[0109] ρ = x cos(θ) + y sin(θ)

[0110] Where ρ represents the perpendicular distance from the line to the origin, and θ represents the angle between the perpendicular line from the origin and the positive x-axis. Figure 5 This diagram illustrates the parameter transformation correspondence between a straight line in rectangular and polar coordinate systems based on the Hough transform. (See attached diagram.) Figure 5 In the polar coordinate system, the intersection point P3(ρ0, θ0) of curves S1 and S2 is the distance ρ0 from the origin of the straight line L0 in the xy coordinate system and the angle θ0 between the straight line L0 and the positive x-axis.

[0111] In this embodiment of the invention, a polar coordinate system is used for the Hough transformation. The n different angle values ​​are θ1, θ2, ... θ i 、…θ n And 0°≤θ i <180°, 1≤i≤n; where, the method for determining n different angle values ​​can be: let θ=[0°, 180°] and take 180 angle parameters at intervals of Δθ=1°, namely 0°, 1°, 2°……179°, for a point cloud (R) in a static point cloud set. x R y )1 can yield 180 sets of (θ, ρ) line parameters [(θ1, ρ1), (θ2, ρ2), ..., (θ...] 180 , ρ 180 [ ], meaning a point cloud may consist of 180 straight lines. For point cloud P j The set of 180 line parameters corresponding to this point cloud can be denoted as w. j =[(θ1, ρ1) j ,(θ2,ρ2) j , …, (θ 180, ρ 180 ) j ], where 1≤j≤s, then the total set of polar coordinate parameters formed by all point clouds is W=[w1,w2,…w j , ...w s ].

[0112] Given that the number of stationary point clouds is s, the set of stationary point clouds is represented by Q. S It means that Q S It includes S point clouds, Q S =[(R x R y )1, (R x R y )2,…,(R x R y ) s Substituting these values ​​into the formula ρ=x cos(θ)+ysin(θ), we obtain the (θ, ρ) corresponding to s×180 lines. s×180 parameter.

[0113] In step 303, for the angle θ among n different angle values i Iterate through the polar coordinate parameters of each point cloud at the given angle value in the s point clouds, and obtain all point clouds that intersect at the same point in the polar coordinate system at that angle value, forming a subset.

[0114] Optionally, in polar coordinates, when the angle value is θ i If the absolute value of the difference between the two ρ values ​​of the corresponding curves of two point clouds in the polar coordinate system is less than or equal to the second preset value, then it is determined that the curves corresponding to the two point clouds intersect at the same point in the polar coordinate system at that angle value.

[0115] For example, if the same angle θ i Iterate through all ρ values ​​corresponding to the given angle in the total set of polar coordinate parameters W. If the ρ values ​​of two point clouds at that angle are... To detect threshold, For the second preset value in this step, it is assumed that the curves corresponding to the two point clouds intersect in the polar coordinate system at that angle value. When θ1, iterate through the number of curves that intersect with (θ1, ρ1)1. If the number of curves is greater than 2, the point cloud intersecting at that point is constructed into a subset. By analogy, iterate through each angle value to obtain all the intersection points, that is, obtain all the subsets.

[0116] In step 304, each of the n different angle values ​​is traversed to obtain at least one subset.

[0117] Combining step 303, iterate through each of the n different angle values ​​to obtain at least one subset.

[0118] In step 305, the target line is determined based on at least one subset.

[0119] In step 306, a feature point cloud set is obtained to represent the point cloud distribution characteristics of the target line based on the target line and the static point cloud set, and the point cloud distribution characteristics of the target line are calculated, including the length of the line segment corresponding to the target line.

[0120] In step 307, the presence of buildings is determined based on the point cloud distribution characteristics of the target straight line.

[0121] In step 308, if a building exists, targets located outside the building are removed based on the building's location.

[0122] The feature point cloud set is also used to represent the location of buildings.

[0123] For details on the implementation of steps 305 to 308, please refer to [link / reference]. Figure 1 Steps 103 to 106 of the corresponding embodiments will not be repeated in this embodiment of the invention.

[0124] This invention provides a target recognition method. By processing point cloud images, a static point cloud set is obtained. Since the spatial point cloud distribution of targets such as buildings, walls, and pillars is closest to a straight line in a parking environment, the static point cloud set is analyzed and a straight line is obtained through Hough transform. By analyzing the characteristics of the straight line, it is determined whether a building, such as a wall, exists. If it does, targets located outside the building, i.e., false targets, are removed from all identified targets, thereby improving the radar recognition accuracy in a parking environment.

[0125] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0126] The following are device embodiments of the present invention. For details not described in detail, please refer to the corresponding method embodiments described above.

[0127] Figure 6 A schematic diagram of the target recognition device provided in an embodiment of the present invention is shown. For ease of explanation, only the parts related to the embodiment of the present invention are shown, and are described in detail below:

[0128] like Figure 6As shown, the target recognition device 6 includes: a static point cloud set acquisition module 61, a subset acquisition module 62, a target line determination module 63, a point cloud distribution feature calculation module 64, a building judgment module 65, and a target recognition module 66;

[0129] The stationary point cloud set acquisition module 61 is used to process the point cloud image to obtain a stationary point cloud set, wherein the stationary point cloud set includes all point clouds in the point cloud image that are in a stationary state. The point cloud image is obtained by processing the echo signal received by the radar installed on the vehicle.

[0130] The subset acquisition module 62 is used to process the point clouds in the static point cloud set to obtain at least one subset. For any subset, all point clouds in the subset are located on the same straight line, and the subset includes all point clouds in the static point cloud set that are located on the straight line. The number of point clouds in each subset of at least one subset is greater than 2.

[0131] The target line determination module 63 is used to determine the target line based on at least one subset.

[0132] The point cloud distribution feature calculation module 64 is used to obtain a feature point cloud set representing the point cloud distribution features of the target line based on the target line and the static point cloud set, and to calculate the point cloud distribution features of the target line, including the length of the line segment corresponding to the target line.

[0133] The building detection module 65 is used to determine whether a building exists based on the point cloud distribution characteristics of the target straight line;

[0134] The target recognition module 66 is used to remove targets located outside of buildings based on the location of the buildings if buildings exist.

[0135] This invention provides a target recognition device that processes point cloud images to obtain a static point cloud set. Since the spatial point cloud distribution of targets such as buildings, walls, and pillars is closest to a straight line in a parking environment, the device analyzes the static point cloud set and obtains a straight line through Hough transform. By analyzing the characteristics of the straight line, it determines whether a building, such as a wall, exists. If so, targets located outside the building, i.e., false targets, are removed from all identified targets, thereby improving the radar recognition accuracy in a parking environment.

[0136] In one possible implementation, the stationary point cloud set acquisition module 61 is used to determine that any point cloud in the point cloud image is a stationary point cloud if it satisfies a first preset formula. The first preset formula is:

[0137] |V car ×cos(θ p)+V p |≤ε

[0138] Among them, V car θ is used to represent the speed of a vehicle. p V is used to represent the angle of the point cloud. p The velocity of the point cloud is used to represent the speed of the point cloud, and ε is used to represent the first preset value;

[0139] Obtain all point clouds in the point cloud image that satisfy the first preset formula to obtain a static point cloud set.

[0140] In one possible implementation, the subset acquisition module 62 is used for each stationary point cloud P in the set of stationary point clouds. j Based on n preset different angle values, construct n straight lines passing through the stationary point cloud in the xy coordinate system, and determine the perpendicular distance of each line to the origin. Obtain the perpendicular distance corresponding to each angle value, and construct P. j n sets of polar coordinate parameters (θ) in the polar coordinate system based on Hough transform i , ρ i ), where for any angle value, the angle value is greater than or equal to 0 degrees and the angle value is less than 180 degrees, where 1≤i≤n, 1≤j≤s;

[0141] For the angle θ among n different angle values i Traverse the polar coordinate parameters of each point cloud at the given angle value in the s point clouds, and obtain all point clouds that intersect at the same point in the polar coordinate system at the given angle value, forming a subset;

[0142] By iterating through each of the n different angle values, at least one subset is obtained.

[0143] In one possible implementation, the subset acquisition module 62 is used to obtain the subset in polar coordinates when the angle value is θ. i If the absolute value of the difference between the two ρ values ​​of the corresponding curves of two point clouds in the polar coordinate system is less than or equal to the second preset value, then it is determined that the curves corresponding to the two point clouds intersect at the same point in the polar coordinate system at that angle value.

[0144] In one possible implementation, the target line determination module 63 is used for:

[0145] If at least one subset contains only one subset, perform linear fitting on the point cloud in that subset to obtain the target straight line;

[0146] If at least one subset comprises multiple subsets, the subset with the largest number of point clouds is determined among the multiple subsets, and the point clouds in the subset with the largest number of point clouds are linearly fitted to obtain the target straight line; or, the multiple subsets are sorted in descending order of the number of point clouds, the first m subsets are selected, and the point clouds in the m subsets are linearly fitted sequentially to obtain the target straight line, wherein the target straight line comprises m straight lines; or, y subsets with a number of point clouds greater than or equal to a third preset value are determined among the multiple subsets, and the point clouds in the y subsets are linearly fitted sequentially to obtain the target straight line, wherein the target straight line comprises y straight lines.

[0147] In one possible implementation, the point cloud distribution feature calculation module 64 is used for:

[0148] Calculate the Euclidean distance between each point in the static point cloud set and the target line in turn;

[0149] Obtain all point clouds whose Euclidean distance to the target line is less than or equal to the fourth preset value, and obtain a feature point cloud set. In the feature point cloud set, the point clouds whose Euclidean distance to the target line is greater than zero and less than or equal to the fourth preset value are the point clouds located near the target line.

[0150] Calculate the point cloud distribution characteristics of the target line based on the feature point cloud set.

[0151] In one possible implementation, the point cloud distribution feature calculation module 64 is used for:

[0152] Calculate the difference between the maximum and minimum y-coordinate values ​​in the feature point cloud set to obtain the length of the line segment corresponding to the target line;

[0153] Alternatively, the difference between the maximum and minimum x-coordinate values ​​in the feature point cloud set can be calculated to obtain the length of the line segment corresponding to the target line.

[0154] In one possible implementation, the point cloud distribution feature calculation module 64 is also used for:

[0155] Divide the line segment corresponding to the target line into a predetermined number of continuous but non-overlapping sub-segments;

[0156] Based on the feature point cloud set, determine the number of point clouds contained in each sub-line segment. The number of point clouds contained in each sub-line segment is used to represent the density feature of the line segment corresponding to the target line. For any sub-line segment, the number of point clouds contained in the sub-line segment includes the point cloud located in the line segment and the point cloud located near the sub-line segment.

[0157] In one possible implementation, the building determination module 65 is used for:

[0158] If the number of point clouds on the target line is greater than or equal to the fifth preset value, and the length of the line segment corresponding to the target line is greater than or equal to the preset length, and the number of point clouds contained in each sub-line segment is greater than or equal to the sixth preset value, then it is determined that a building exists.

[0159] The target recognition device provided in this embodiment can be used to execute the target recognition method embodiment described above. Its implementation principle and technical effect are similar, and will not be described again here.

[0160] Figure 7 This is a schematic diagram of a target recognition device provided in an embodiment of the present invention. Figure 7 As shown, the target recognition device 7 of this embodiment includes: a processor 70, a memory 71, and a computer program 72 stored in the memory 71 and executable on the processor 70. When the processor 70 executes the computer program 72, it implements the steps in the various target recognition method embodiments described above, for example... Figure 1 Steps 101 to 106 are shown. Alternatively, when the processor 70 executes the computer program 72, it implements the functions of each module / unit in the above-described device embodiments, for example... Figure 6 The functions of modules 61 to 66 are shown.

[0161] For example, the computer program 72 can be divided into one or more modules / units, which are stored in the memory 71 and executed by the processor 70 to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program 72 in the target recognition device 7.

[0162] The target identification device 7 may be a target identification device, equipment, or module installed on a vehicle. The target identification device 7 may include, but is not limited to, a processor 70 and a memory 71. Those skilled in the art will understand that... Figure 7 This is merely an example of the target identification device 7 and does not constitute a limitation on the target identification device 7. It may include more or fewer components than shown, or combine certain components, or different components. For example, the target identification device may also include input / output devices, network access devices, buses, etc.

[0163] The processor 70 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0164] The memory 71 can be an internal storage unit of the target identification device 7, such as a hard disk or memory of the target identification device 7. The memory 71 can also be an external storage device of the target identification device 7, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the target identification device 7. Furthermore, the memory 71 can include both internal storage units and external storage devices of the target identification device 7. The memory 71 is used to store the computer program and other programs and data required by the target identification device. The memory 71 can also be used to temporarily store data that has been output or will be output.

[0165] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0166] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0167] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0168] In the embodiments provided by this invention, it should be understood that the disclosed apparatus / target identification device and method can be implemented in other ways. For example, the apparatus / target identification device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0169] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0170] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0171] If the integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above-described embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various target identification method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0172] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A target recognition method, characterized in that, include: The point cloud image is processed to obtain a static point cloud set, wherein the static point cloud set includes all point clouds in the point cloud image that are in a static state, and the point cloud image is obtained by processing the echo signal received by the radar installed on the vehicle. The point clouds in the static point cloud set are processed to obtain at least one subset. For any subset, all point clouds in the subset are located on the same straight line, and the subset includes all point clouds in the static point cloud set that are located on the straight line. The number of point clouds in each subset of the at least one subset is greater than 2. Determine the target line based on at least one subset; Based on the target straight line and the static point cloud set, a feature point cloud set is obtained to represent the point cloud distribution characteristics of the target straight line, and the point cloud distribution characteristics of the target straight line are calculated. The point cloud distribution characteristics include the length of the line segment corresponding to the target straight line and the density characteristics of the line segment corresponding to the target straight line. Based on the point cloud distribution characteristics of the target straight line, determine whether a building exists; If a building exists, remove targets located outside the building based on the building's location; The method for determining the density characteristics of the line segment corresponding to the target straight line includes: The line segment corresponding to the target straight line is divided into a predetermined number of continuous but non-overlapping sub-segments. Based on the feature point cloud set, the number of point clouds contained in each sub-segment is determined. The number of point clouds contained in each sub-segment is used to represent the density characteristics of the line segment corresponding to the target line. For any sub-segment, the number of point clouds contained in the sub-segment includes the point cloud located in the line segment and the point cloud located near the sub-segment.

2. The target recognition method according to claim 1, characterized in that, The process of processing the point cloud image to obtain a static point cloud set includes: For any point cloud in the point cloud image, if the point cloud satisfies a first preset formula, then the point cloud is determined to be a stationary point cloud. The first preset formula is: in, Used to indicate the vehicle's speed. Used to represent the angle of the point cloud. Used to represent the velocity of the point cloud. Used to represent the first preset value; Obtain all point clouds in the point cloud image that satisfy the first preset formula to obtain the static point cloud set.

3. The target recognition method according to claim 1, characterized in that, The static point cloud set includes s point clouds, and the process of processing the point clouds in the static point cloud set to obtain at least one subset includes: For each stationary point cloud in the set of stationary point clouds Based on n preset different angle values, construct n straight lines passing through the stationary point cloud in the xy coordinate system, and determine the perpendicular distance of each line to the origin. Obtain the perpendicular distance corresponding to each angle value, and construct... n sets of polar coordinate parameters in the polar coordinate system based on Hough transform ( ), where for any angle value, the angle value is greater than or equal to 0 degrees and the angle value is less than 180 degrees, where 1 ≤ ≤n, 1≤ j ≤s; For the angle among the n different angle values Traverse the polar coordinate parameters of each point cloud in the s point clouds corresponding to the angle value, and obtain all point clouds that intersect at the same point in the polar coordinate system at the angle value, forming a subset; By iterating through each of the n different angle values, at least one subset is obtained.

4. The target recognition method according to claim 3, characterized in that, The angle value among the n different angle values By iterating through the s point clouds, the polar coordinate parameters corresponding to each point cloud at that angle value are obtained, resulting in all point clouds intersecting at the same point in the polar coordinate system at that angle value, including: In the polar coordinate system, when the angle value is When two point clouds correspond to two curves in the polar coordinate system, If the absolute value of the difference is less than or equal to the second preset value, then it is determined that the curves corresponding to the two point clouds intersect at the same point in the polar coordinate system at that angle value.

5. The target recognition method according to claim 1, characterized in that, If the at least one subset includes only one subset, then determining the target line based on the at least one subset includes: The target straight line is obtained by performing linear fitting on the point cloud in the subset; If the at least one subset includes multiple subsets, then determining the target line based on the at least one subset includes: Among the plurality of subsets, determine the subset with the largest number of point clouds, and perform linear fitting on the point clouds in the subset with the largest number of point clouds to obtain the target straight line; Alternatively, the subsets can be sorted in descending order of the number of point clouds, the first m subsets can be selected, and the point clouds in the m subsets can be linearly fitted sequentially to obtain the target line, wherein the target line includes m lines. Alternatively, among the plurality of subsets, determine y subsets with a point cloud quantity greater than or equal to a third preset value, and sequentially perform linear fitting on the point clouds in the y subsets to obtain the target straight line, wherein the target straight line includes y straight lines.

6. The target recognition method according to any one of claims 1 to 5, characterized in that, The step of obtaining a feature point cloud set representing the point cloud distribution characteristics of the target line based on the target line and the static point cloud set, and calculating the point cloud distribution characteristics of the target line includes: Calculate the Euclidean distance between each point cloud in the static point cloud set and the target line in sequence; Obtain all point clouds whose Euclidean distance to the target line is less than or equal to a fourth preset value to obtain the feature point cloud set. In the feature point cloud set, the point clouds whose Euclidean distance to the target line is greater than zero and less than or equal to the fourth preset value are the point clouds located near the target line. Based on the feature point cloud set, calculate the point cloud distribution characteristics of the target straight line.

7. The target recognition method according to claim 6, characterized in that, The step of calculating the point cloud distribution features of the target straight line based on the feature point cloud set includes: Calculate the difference between the maximum and minimum y-coordinate values ​​in the feature point cloud set to obtain the length of the line segment corresponding to the target line; Alternatively, the length of the line segment corresponding to the target line can be obtained by calculating the difference between the maximum and minimum x-coordinate values ​​in the feature point cloud set.

8. The target recognition method according to claim 1, characterized in that, The step of determining whether a building exists based on the point cloud distribution characteristics of the target straight line includes: If the number of point clouds on the target straight line is greater than or equal to a fifth preset value, and the length of the line segment corresponding to the target straight line is greater than or equal to a preset length, and the number of point clouds contained in each sub-line segment is greater than or equal to a sixth preset value, then it is determined that a building exists.

9. A target identification device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 8 above.

Images