Intersection-oriented high-precision map construction method, device and equipment

Abstract

The embodiment of the invention provides an intersection-oriented high-precision map construction method, device and equipment. The method comprises the following steps: acquiring initial point cloud data and an initial image acquired by roadside equipment; wherein the initial point cloud data and the initial image comprise a plurality of road surface elements of a target intersection; performing compensation processing on the initial point cloud data to obtain target point cloud data; determining first point cloud data of each pavement element according to the initial image and the target point cloud data; obtaining fusion point cloud data of each pavement element according to the target point cloud data and the first point cloud data; and constructing a high-precision map of the target intersection according to the fused point cloud data of each pavement element. The method is used for achieving the effects of improving the accuracy of the map data and reducing the map making cost at the same time.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence, and in particular to a method, device and equipment for constructing a high-precision map for intersections. Background Art

[0002] The development of high-precision map technology provides a solid foundation for the development of intelligent driving technology.

[0003] The production of high-precision maps at intersections usually requires data collection vehicles to collect data from the same intersection multiple times, and then build a high-precision map based on the data collected multiple times.

[0004] However, in this method, the collection equipment installed on the collection vehicle is limited by the field of view, resulting in the obstruction of the collection data at the intersection, and multiple collections are required, resulting in high map production costs. Summary of the invention

[0005] The embodiments of the present application provide a method, device and equipment for constructing a high-precision map for intersections, so as to improve the accuracy of map data while reducing the cost of map production.

[0006] In a first aspect, an embodiment of the present application provides a method for constructing a high-precision map for an intersection, comprising:

[0007] Acquire initial point cloud data and initial image collected by roadside equipment; wherein the initial point cloud data and the initial image include multiple road surface elements of the target intersection;

[0008] Performing compensation processing on the initial point cloud data to obtain target point cloud data;

[0009] Determine the first point cloud data of each road surface element according to the initial image and the target point cloud data; and obtain the fused point cloud data of each road surface element according to the target point cloud data and the first point cloud data;

[0010] A high-precision map of the target intersection is constructed based on the fused point cloud data of each road surface element.

[0011] In a possible implementation manner, the compensating the initial point cloud data to obtain target point cloud data includes:

[0012] Determine a point set according to a first distance of each data point in the initial point cloud data; wherein the first distance represents the distance from the data point to the roadside equipment; and the data points in the point set represent data points whose first distance is greater than or equal to a preset threshold;

[0013] For each data point in the point set, K neighboring points of the data point are determined, wherein K is a positive integer; and the target point cloud data is obtained through interpolation processing based on the neighboring points and the data point.

[0014] In a possible implementation manner, obtaining the target point cloud data through interpolation processing based on the neighboring points and the data point includes:

[0015] Performing interpolation processing on the data point and the neighboring point to obtain an interpolation data point;

[0016] The intensity value of the interpolated data point is determined according to the intensity value of the data point and the intensity value of the neighboring point to obtain the target point cloud data.

[0017] In a possible implementation manner, determining the first point cloud data of each road surface element according to the initial image and the target point cloud data includes:

[0018] Acquire a road surface intensity image of the target point cloud data; and perform correction processing on the initial image to obtain a target image; wherein the pixel points of the road surface intensity image correspond to a plurality of data points in the target point cloud data;

[0019] The first point cloud data of each road surface element is determined according to the target image and the road surface intensity image.

[0020] In a possible implementation manner, determining the first point cloud data of each road surface element according to the target image and the road surface intensity image includes:

[0021] Performing a second segmentation process on the target image to obtain a plurality of road surface elements in the target image;

[0022] Determine the projection relationship of the pixel points in the road surface intensity image through projection processing; wherein the projection relationship represents the corresponding relationship between the pixel points in the road surface intensity image and the road surface elements in the target image;

[0023] The first point cloud data of each road surface element is determined according to the projection relationship of the pixel points.

[0024] In a possible implementation manner, obtaining fused point cloud data of each road surface element according to the target point cloud data and the first point cloud data includes:

[0025] Acquire a road surface intensity image of the target point cloud data; wherein the pixel points of the road surface intensity image correspond to a plurality of data points in the target point cloud data;

[0026] Performing a first segmentation process on the road surface intensity image to obtain second point cloud data of each road surface element;

[0027] According to the first point cloud data and the second point cloud data, fused point cloud data of each road surface element is obtained.

[0028] In a possible implementation manner, obtaining fused point cloud data of each road surface element according to the first point cloud data and the second point cloud data includes:

[0029] The first point cloud data of each road surface element is added to the second point cloud data of the corresponding road surface element to obtain fused point cloud data of each road surface element.

[0030] In a second aspect, an embodiment of the present application provides a device for constructing a high-precision map for an intersection, comprising:

[0031] An acquisition module, used to acquire initial point cloud data and initial image collected by roadside equipment; wherein the initial point cloud data and the initial image include multiple road surface elements of the target intersection;

[0032] A compensation module, used for performing compensation processing on the initial point cloud data to obtain target point cloud data;

[0033] A determination module, configured to determine first point cloud data of each road surface element according to the initial image and the target point cloud data; and to obtain fused point cloud data of each road surface element according to the target point cloud data and the first point cloud data;

[0034] A construction module is used to construct a high-precision map of the target intersection based on the fused point cloud data of each road surface element.

[0035] In a third aspect, an embodiment of the present application provides an electronic device, including: a memory, a processor;

[0036] The memory stores computer-executable instructions;

[0037] The processor executes the computer-executable instructions stored in the memory, so that the processor executes the above first aspect and / or various possible implementations of the first aspect.

[0038] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the first aspect above and / or various possible implementations of the first aspect.

[0039] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the above first aspect and / or various possible implementation methods of the first aspect.

[0040] The method, device and equipment for constructing high-precision maps for intersections provided in the embodiments of the present application first compensate the initial point cloud data and the initial image collected by the roadside equipment to obtain the target point cloud data, and then obtain the first point cloud data of each road element based on the target point cloud data and the initial image, and then obtain the fused point cloud data of each road element, and use the fused point cloud data of each road element to construct a high-precision map of the target intersection. In this way, since the roadside equipment can collect more comprehensive road elements at the intersection, the electronic equipment can reduce the occlusion probability of the road elements and reduce the number of collections by constructing the map based on the point cloud data and images collected by the roadside equipment, that is, reduce the cost of map production. At the same time, the electronic equipment can compensate the point cloud data collected by the roadside equipment, and further supplement the collected point cloud data based on the two-dimensional image data, thereby improving the accuracy and completeness of the point cloud data, and thus improving the accuracy of the high-precision map. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0042] Figure 1 A schematic diagram of an application scenario provided for this application;

[0043] Figure 2 Schematic diagram of the process of constructing a high-precision map for intersections provided in this application Figure 1 ;

[0044] Figure 3 A schematic diagram of point cloud data collected by a roadside device provided in this application;

[0045] Figure 4 A schematic diagram of circle center calculation provided for this application;

[0046] Figure 5 A schematic diagram of contour point extraction provided in this application;

[0047] Figure 6 Schematic diagram of the process of constructing a high-precision map for intersections provided in this application Figure 2 ;

[0048] Figure 7 A schematic diagram of an interpolation process provided for this application;

[0049] Figure 8 A schematic diagram of the structure of a device for constructing a high-precision map for intersections provided in this application;

[0050] Fig. 9A schematic diagram of the structure of the electronic device provided in this application.

[0051] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0052] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0053] The road surface elements at intersections are complex and diverse and the traffic volume is large. Therefore, in order to ensure that the complete road surface elements at the intersection are collected, it is usually necessary for the collection vehicle to collect data from the same intersection multiple times, and then build a high-precision map of the intersection based on the data collected multiple times.

[0054] However, in this method, the collection equipment installed on the collection vehicle is limited by the field of view, resulting in the obstruction of the collected data at the intersection. In other words, the integrity and accuracy of the collected data are low, and the data must be collected multiple times, resulting in high map production costs.

[0055] In view of this, the present application provides a method for constructing a high-precision map for intersections, which uses point cloud data collected by roadside equipment to construct a high-precision map for intersections. Since roadside equipment is usually set at a high place at the intersection, the occlusion of road surface elements in the collected point cloud data can be reduced to ensure the integrity of the data. In addition, one collection can cover the road surface elements within the intersection range, reducing the collection cost. At the same time, compensation processing is performed on the point cloud data collected by the roadside equipment to ensure the accuracy of the data, thereby ensuring the accuracy of the generated high-precision map.

[0056] The execution subject of this application can be an electronic device with processing capabilities, such as a computer, a server, etc., and this application does not limit this.

[0057] Figure 1 A schematic diagram of an application scenario provided for this application, such as Figure 1As shown, traffic lights are usually installed on traffic poles at intersections, and roadside equipment can also be installed on the traffic poles. The roadside equipment can include sensors for taking two-dimensional images (such as cameras) and sensors for collecting point clouds (such as laser radars) to collect road surface elements in different data forms at the intersection. By installing roadside equipment at a high position, it is possible to cover road surface elements within the intersection in one acquisition, reduce acquisition costs, and reduce the probability of road surface occlusion in the collected data.

[0058] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0059] Figure 2 Schematic diagram of the process of constructing a high-precision map for intersections provided in this application Figure 1 ,like Figure 2 As shown, the method includes:

[0060] S201. Acquire initial point cloud data and initial images collected by roadside equipment.

[0061] Exemplarily, the above-mentioned initial point cloud data represents point cloud data (three-dimensional data) collected by roadside equipment, and the initial point cloud data includes multiple road surface elements of the target intersection. It should be understood that the initial point cloud data includes a large number of data points, each of which has a three-dimensional coordinate and an intensity value. The above-mentioned initial image represents two-dimensional image data collected by roadside equipment, and the initial image includes multiple road surface elements of the target intersection. Among them, the above-mentioned road surface elements represent elements on the road surface, such as lane lines, zebra crossings, lane markings (such as left turn lanes, through lanes), etc.

[0062] In some possible implementations, the electronic device may periodically extract the collected initial point cloud data and initial image from the roadside device; alternatively, the roadside device stores the collected initial point cloud data and initial image in a database, and the electronic device obtains the initial point cloud data and initial image from the database; alternatively, the roadside device reports the initial point cloud data and initial image to the electronic device in real time; alternatively, the electronic device receives the initial point cloud data and initial image imported by an external device, wherein the external device may be, for example, a USB flash drive; or alternatively, the electronic device communicates with other electronic devices to obtain the initial point cloud data and initial image.

[0063] Optionally, after the electronic device acquires the initial point cloud data and the initial image, the initial point cloud data and the initial image are aligned based on the time attribute to ensure that the subsequent processing is the initial point cloud data and the initial image within the same time, thereby ensuring the accuracy of the data.

[0064] S202: Perform compensation processing on the initial point cloud data to obtain target point cloud data.

[0065] Exemplarily, the target point cloud data refers to point cloud data obtained by performing compensation processing on the initial point cloud data.

[0066] The inventor of the present application has found that since the roadside equipment is usually installed on one side of the intersection, the point cloud data collected by the roadside equipment is Figure 3 As shown, Figure 3 This is a schematic diagram of point cloud data collected by a roadside device provided in this application. As the distance from the roadside device increases, the density of the point cloud on the opposite side of the intersection gradually decreases ( Figure 3 The data points in the intersection are increasingly sparse from bottom to top), which results in the point cloud data on the opposite side of the intersection not accurately reflecting the road surface information on the opposite side of the intersection. To address this problem, the present application performs compensation processing on the initial point cloud data to improve the accuracy of the point cloud data.

[0067] In some possible implementations, the electronic device can first determine the distance between each data point in the initial point cloud data and the sensor based on the coordinate position of the sensor in the roadside device, add the data points whose distance is greater than or equal to the preset threshold to the point set, and then perform interpolation processing on each data point in the point set to obtain multiple interpolated data points, and perform weighted processing based on the distance, assign the interpolated data point an intensity value, and add the interpolated data point to the initial point cloud data to obtain the target point cloud data. In this way, the problem of sparse point cloud data points caused by the distance of the sensor can be compensated, and the accuracy of the point cloud data can be improved.

[0068] In some possible implementations, a movable acquisition platform may be used to collect sample point cloud data, such as a drone equipped with a laser radar, and then an initial deep learning model may be trained based on the collected sample point cloud data to obtain a trained deep learning model, which can learn the distribution characteristics of the point cloud data and thus achieve compensation processing of the point cloud data. The deep learning model may be, for example, a Generative Adversarial Network (GAN) model, a Variational Autoencoder (VAE) model, etc., which is not limited in the embodiments of the present application.

[0069] Taking the generative adversarial network model as an example, the electronic device can input the initial point cloud data into the trained generative adversarial network model and output the target point cloud data. This method can make up for the sparse point cloud data density problem caused by the sensor distance and improve the integrity of the point cloud data. In addition, this method can be used to compensate for the initial point cloud data in combination with the actual distribution characteristics of the real point cloud data, so that the target point cloud data after compensation is more accurate.

[0070] S203: Determine first point cloud data of each road surface element according to the initial image and the target point cloud data.

[0071] Exemplarily, the first point cloud data of the road surface element represents data points belonging to the road surface element obtained based on the target point cloud data and the initial image. In other words, the first point cloud data of the road surface element includes multiple data points.

[0072] As mentioned above, the density of the point cloud data on the opposite side of the intersection gradually decreases as the distance from the roadside equipment increases. In order to further ensure the accuracy of the point cloud data, this application further compensates the point cloud data based on the initial image on the basis of the aforementioned target point cloud data to improve the accuracy and completeness of the point cloud data.

[0073] In some possible implementations, the electronic device may first perform image instance segmentation processing on the initial image to obtain pixel points of multiple road surface elements, and then project the pixel points in the road surface intensity image obtained based on the target point cloud data, whose pixel values ​​are greater than a preset threshold, into the initial image to obtain the correspondence between the pixel points in the road surface intensity image and the road surface elements in the initial image, and the pixel points in the road surface intensity image correspond to multiple data points. Therefore, the data points corresponding to the road surface elements in the initial image can be obtained, that is, the first point cloud data of the road surface elements.

[0074] S204: Obtain fused point cloud data of each road surface element according to the target point cloud data and the first point cloud data.

[0075] Exemplarily, as mentioned above, the first point cloud data is the data points belonging to the road surface elements obtained based on the initial image and the target point cloud data. Therefore, the electronic device can first perform image instance segmentation processing according to the target point cloud data to obtain the second point cloud data of each road surface element, that is, the second point cloud data is the data points belonging to the road surface elements obtained based on the target point cloud data, and then fuse the two to obtain more accurate data points belonging to the road surface elements, that is, the fused point cloud data of the road surface elements. For example, the electronic device can add the data points in the first point cloud data of the road surface element to the second point cloud data of the road surface element to obtain the fused point cloud data of the road surface element.

[0076] Through the above steps, the three-dimensional point cloud data is supplemented with the two-dimensional image, further improving the accuracy and completeness of the point cloud data.

[0077] S205: Construct a high-precision map of the target intersection based on the fused point cloud data of each road surface element.

[0078] Exemplarily, the fused point cloud data of the road surface elements includes multiple data points. The electronic device can perform vectorization processing based on the fused point cloud data of the road surface elements to obtain vector data, such as polygons or line segments, and build a high-precision map based on the vector data.

[0079] In some possible implementations, the electronic device can perform denoising and clustering processing in sequence based on the fused point cloud data of each road surface element to obtain the fused point cloud data of the processed road surface element; then, the electronic device can determine the boundary of the road surface element based on a preset boundary extraction algorithm for the fused point cloud data of each processed road surface element, generate vector data of the road surface element based on the boundary of the road surface element and the fused point cloud data of the processed road surface element, and then construct a high-precision map based on the vector data of each road surface element. It should be noted that this application does not limit the type of boundary extraction algorithm, and does not limit how to construct a high-precision map based on vector data.

[0080] In one example, the boundary extraction algorithm may be, for example, an Alpha-shapes algorithm. Based on the Alpha-shapes algorithm, the electronic device may first perform projection processing on the fused point cloud data of the processed road surface elements, convert the three-dimensional fused point cloud data into two-dimensional fused point cloud data, and perform boundary extraction on the two-dimensional fused point cloud data. Specifically, the algorithm may include the following steps:

[0081] (1) For any point p(x, y) to be judged in the two-dimensional fused point cloud data, according to the preset rolling circle radius α, search for all points within a distance of less than 2α from the point p in the two-dimensional fused point cloud data, which are recorded as the point set Q;

[0082] (2) Select any point p1(x1,y1) in the point set Q, and calculate the coordinates of the two circle centers o1(x o1 ,y o1 ) and o2(x o2 ,y o2 ),refer to Figure 4 As shown, Figure 4 A schematic diagram of circle center calculation provided in this application, the calculation formulas of circle centers o1 and o2 are as follows:

[0083]

[0084]

[0085]

[0086]

[0087] in, S 2 =(x-x1) 2 +(y-y1) 2 , S 2 It can be regarded as the square of the distance between p and p1, and then H can be regarded as a proportional coefficient based on the rolling circle radius α and the distance S between p and p1. By controlling the proportional coefficient H, the position of the center coordinate can be controlled.

[0088] (3) For all points in the point set Q except point p1, calculate the distance to the center o1 and o2 respectively. If the distance of all points to o1 or o2 is greater than α, then point p is marked as a contour point and the judgment is terminated.

[0089] (4) If the distances from all points except point p1 to the centers o1 and o2 are not all greater than α, refer to Figure 5 As shown, Figure 5 A schematic diagram of contour point extraction provided in this application is shown in FIG. 1 , all points in the point set Q are rotated as point p1, the coordinates of the circle center are recalculated, and point p is judged according to step (3). If there is a point that marks point p as a contour point, point p is determined to be a contour point, otherwise point p is determined to be a non-contour point.

[0090] Through the above steps, a set of all the points marked as contour points can be obtained, and these contour points can constitute the boundary of the road surface element.

[0091] The method for constructing a high-precision map for intersections provided in the embodiment of the present application is based on the initial point cloud data and initial image collected by the roadside equipment, and firstly compensates the initial point cloud data to obtain the target point cloud data, and then obtains the first point cloud data of each road surface element based on the target point cloud data and the initial image, and then obtains the fused point cloud data of each road surface element, and uses the fused point cloud data of each road surface element to construct a high-precision map of the target intersection. In this way, since the roadside equipment can collect more comprehensive road surface elements at the intersection, the electronic equipment constructs the map based on the point cloud data and images collected by the roadside equipment, which can reduce the occlusion probability of the road surface elements and reduce the number of collections, that is, reduce the cost of map production. At the same time, the electronic equipment can compensate the point cloud data collected by the roadside equipment, and further supplement the collected point cloud data based on the two-dimensional image data, thereby improving the accuracy and completeness of the point cloud data, and thus improving the accuracy of the high-precision map.

[0092] Figure 6 Schematic diagram of the process of constructing a high-precision map for intersections provided in this application Figure 2 ,like Figure 6 As shown, in this embodiment Figure 2 Based on the embodiment, a method for constructing a high-precision map for an intersection is described in detail, and the method includes:

[0093] S301. Acquire initial point cloud data and initial images collected by roadside equipment.

[0094] It should be noted that this step is similar to the aforementioned step S201 and will not be repeated here.

[0095] S302: Determine a point set according to a first distance of each data point in the initial point cloud data.

[0096] Exemplarily, the first distance represents the distance from the data point to the roadside equipment, or the distance between the data point and the sensor. The data points in the point set represent data points whose first distance is greater than or equal to a preset threshold.

[0097] For example, the electronic device can determine the first distance of the data point based on the coordinates of the data point and the coordinates of the sensor based on a preset distance calculation formula. If the first distance is greater than or equal to a preset threshold, the data point is added to the point set to obtain the point set.

[0098] S303: For each data point in the point set, determine K neighboring points of the data point.

[0099] Exemplarily, the K neighboring points represent the K data points closest to the data point, and the K is a positive integer. The present application does not limit the value of K. For example, K can be 5, that is, 5 neighboring points of the data point are determined. For example, the electronic device can calculate the distance between the data point and other data points within a preset distance around the data point, and sort them in ascending order, and use the other data points corresponding to the first K distances as the K neighboring points.

[0100] S304: Obtain target point cloud data through interpolation processing based on neighboring points and the data point.

[0101] Exemplarily, the electronic device may perform interpolation processing between the neighboring point and the data point to obtain a plurality of interpolated data points, and then add the interpolated data points to the initial point cloud data to obtain the target point cloud data.

[0102] Specifically, the electronic device may perform interpolation processing based on the data point and the neighboring point to obtain the interpolated data point; determine the intensity value of the interpolated data point based on the intensity value of the data point and the intensity value of the neighboring point to obtain the target point cloud data. Figure 7 A schematic diagram of an interpolation process provided by this application, refer to Figure 7 As shown, the electronic device can perform interpolation processing on the line between the data point and the neighboring point, wherein the interpolation interval is a preset value; and then the number of interpolation data points can be determined based on the distance between the data point and the neighboring point and the interpolation interval to obtain multiple interpolation data points, and the intensity value of the data point and the intensity value of the neighboring point can be inversely distance weighted to obtain the intensity value of each interpolation data point.

[0103] For example, the distance between the interpolated data point and the data point is distance 1, and the distance between the interpolated data point and the neighboring point is distance 2. The intensity value of the interpolated data point can be expressed as distance 2*weight coefficient / distance 1.

[0104] In this way, the intensity value of the interpolated data points at different positions can be determined using the weight coefficient. The weight coefficient can be calculated based on the distance between the interpolated data point and the data point. The smaller the distance, the larger the weight coefficient. This application does not limit the calculation method of the weight coefficient.

[0105] Through this step, the density of the target point cloud data can be improved by using interpolation processing, and the characteristics of the road surface elements can be maintained.

[0106] S305: Obtain a road surface strength image of the target point cloud data.

[0107] Exemplarily, the road surface intensity image can reflect the intensity characteristics of different areas of the road surface, and the pixel points of the road surface intensity image correspond to multiple data points in the target point cloud data.

[0108] In some possible implementations, the electronic device may first extract the target road point cloud data based on a preset road point cloud extraction algorithm to obtain the target road point cloud data, and then grid the target road point cloud data based on a preset spatial resolution (e.g., one grid per meter) to obtain multiple grids. In other words, the target road point cloud data is divided into multiple grids, each grid including multiple data points, and then the average intensity value of the data points in each grid can be calculated as the pixel value of the grid to generate a road intensity image. It should be noted that the present application does not limit the type of road point cloud extraction algorithm.

[0109] For example, the road surface point cloud extraction algorithm may be a Random Sample Consensus (RanSAC) algorithm, and the electronic device may extract road surface point cloud data from the target point cloud data based on the RanSAC algorithm. Specifically, the RanSAC algorithm may include the following steps:

[0110] (1) Assume that the road surface is a plane or a structure close to a plane, and define a plane model to describe the road surface. The model parameters of the plane model are three non-collinear points. It should be understood that a plane can be determined based on three non-collinear points.

[0111] (2) Randomly select three non-collinear data points from the target point cloud data as the minimum sample set (MSS), which is used to initialize the model parameters of the plane model.

[0112] (3) Fit a plane model based on the least squares method and minimum sample set.

[0113] (4) Substitute the remaining data points in the target point cloud data into the fitted plane model and calculate the distance from the data point to the plane. If the distance is less than a certain set threshold, the data point is determined to be an interior point (i.e., a road surface point), otherwise it is an exterior point (i.e., a non-road surface point).

[0114] (5) Count the number of inliers. If it is determined that the number of inliers does not reach a preset threshold, repeat steps (2) to (5) until a preset number of iterations is reached or a model with the largest number of inliers is found, thereby obtaining a final road surface model.

[0115] Through the above steps, the internal point set corresponding to the final road surface model, that is, the road surface point cloud data, can be obtained.

[0116] In this way, the electronic device can perform subsequent processing only on the road surface point cloud data to obtain the point cloud data of the road surface elements, thereby reducing the amount of calculation, avoiding interference from other point cloud data, and improving the accuracy of the point cloud data of the road surface elements.

[0117] S306: Perform a first segmentation process on the road surface intensity image to obtain second point cloud data of each road surface element.

[0118] Exemplarily, the second point cloud data is data points belonging to road surface elements obtained based on the target point cloud data. The electronic device can perform a first segmentation process on the road surface intensity image based on the preset first image instance segmentation model to obtain pixel points corresponding to each road surface element, and the pixel points in the road surface intensity image correspond to multiple data points in the target point cloud data, thereby obtaining the second point cloud data corresponding to the road surface element. It should be noted that the present application does not limit the type of the first image instance segmentation model.

[0119] S307: Correct the initial image to obtain a target image.

[0120] Exemplarily, the electronic device may sequentially perform denoising, distortion correction, image enhancement, high-pass filtering and other correction processes on the initial image to obtain the target image. For example, the initial image is denoised to reduce the impact of sensor noise on subsequent analysis. Then, lens distortion correction is performed to correct the geometric distortion caused by the camera lens to ensure the geometric accuracy of the image. Next, image enhancement techniques, such as histogram equalization or adaptive contrast adjustment, are applied to improve the clarity and detail of the image.

[0121] S308: Determine first point cloud data of each road surface element according to the target image and the road surface intensity image.

[0122] Exemplarily, the electronic device may first perform image instance segmentation processing on the target image to obtain pixel points of the road surface elements in the target image, and then project the pixel points in the road surface intensity image whose pixel values ​​are greater than a preset threshold value onto the target image to obtain the correspondence between the pixel points in the road surface intensity image and the road surface elements in the target image. The pixel points in the road surface intensity image correspond to multiple data points, and therefore, the data points corresponding to the road surface elements in the target image can be obtained, that is, the first point cloud data of the road surface elements.

[0123] Specifically, the electronic device can first perform a second segmentation process on the target image based on a preset second image instance segmentation model to obtain multiple road surface elements in the target image; determine the projection relationship of the pixel points in the road surface intensity image through projection processing; wherein the projection relationship represents the corresponding relationship between the pixel points in the road surface intensity image and the road surface elements in the target image; and determine the first point cloud data of each road surface element according to the projection relationship of the pixel points. For example, if the pixel points in the road surface intensity image whose pixel values ​​are greater than a preset threshold are projected into the target image, then the projection point corresponding to the pixel point will belong to the road surface element identified in the target image, and then the projection relationship can be obtained.

[0124] It should be noted that this application does not limit the type of the second image instance segmentation model.

[0125] S309: Add the first point cloud data of each road surface element to the second point cloud data of the corresponding road surface element to obtain fused point cloud data of each road surface element.

[0126] Exemplarily, through the above steps, the road surface element can obtain the data point set 2 of the first point cloud data and the data point set 1 of the second point cloud data, and the data points in the data point set 2 obtained based on the two-dimensional image are added to the data point set 1 obtained based on the three-dimensional point cloud data to obtain the fused point cloud data of the road surface element. In this way, the three-dimensional point cloud data is supplemented based on the two-dimensional image, which can avoid the deviation caused by directly converting the two-dimensional data into three-dimensional data and improve the accuracy of the point cloud data.

[0127] S310: construct a high-precision map of the target intersection based on the fused point cloud data of each road surface element.

[0128] It should be noted that this step is similar to the aforementioned step S206 and will not be repeated here.

[0129] In the method for constructing a high-precision map for intersections provided in an embodiment of the present application, the electronic device can use interpolation to perform compensation processing based on the initial point cloud data collected by the roadside equipment to obtain target point cloud data, and obtain road surface point cloud data based on the road surface point cloud extraction algorithm, and then obtain a road surface intensity image, and obtain second point cloud data of road surface elements based on the road surface intensity image; and use the road surface intensity image and the initial image collected by the roadside equipment to perform projection processing to obtain first point cloud data of road surface elements, add the first point cloud data to the second point cloud data of the corresponding road surface elements, obtain fused point cloud data of road surface elements, and then construct a high-precision map based on the fused point cloud data of each road surface element.

[0130] In this way, the electronic device constructs a map based on the point cloud data and images collected by the roadside equipment, which can reduce the probability of occlusion of road elements and the number of collection times, that is, reduce the cost of map production. At the same time, the electronic device can compensate for the point cloud data collected by the roadside equipment in the form of interpolation to make up for the lack of point cloud data caused by the collection distance of the sensor, initially improve the accuracy and completeness of the point cloud data, and further supplement the collected point cloud data based on the two-dimensional image data, further improve the accuracy and completeness of the point cloud data, and thus improve the accuracy of the high-precision map.

[0131] Figure 8 A schematic diagram of the structure of a device for constructing a high-precision map for intersections provided in this application, such as Figure 8 As shown, the construction device 400 for the high-precision map for intersections provided in this embodiment includes:

[0132] The acquisition module 401 is used to acquire the initial point cloud data and the initial image collected by the roadside equipment; wherein the initial point cloud data and the initial image include a plurality of road surface elements of the target intersection;

[0133] A compensation module 402 is used to perform compensation processing on the initial point cloud data to obtain target point cloud data;

[0134] The determination module 403 is used to determine the first point cloud data of each road surface element according to the initial image and the target point cloud data; and obtain the fused point cloud data of each road surface element according to the target point cloud data and the first point cloud data;

[0135] The construction module 404 is used to construct a high-precision map of the target intersection based on the fused point cloud data of each road surface element.

[0136] In a possible implementation, the compensation module 402 is specifically configured to:

[0137] Determine a point set according to a first distance of each data point in the initial point cloud data; wherein the first distance represents the distance from the data point to the roadside equipment; and the data points in the point set represent data points whose first distance is greater than or equal to a preset threshold;

[0138] For each data point in the point set, K neighboring points of the data point are determined, wherein K is a positive integer; and the target point cloud data is obtained through interpolation processing based on the neighboring points and the data point.

[0139] In a possible implementation, the compensation module 402 is specifically configured to:

[0140] Performing interpolation processing on the data point and the neighboring point to obtain an interpolation data point;

[0141] The intensity value of the interpolated data point is determined according to the intensity value of the data point and the intensity value of the neighboring point to obtain the target point cloud data.

[0142] In a possible implementation, the determination module 403 is specifically configured to:

[0143] Acquire a road surface intensity image of the target point cloud data; and perform correction processing on the initial image to obtain a target image; wherein the pixel points of the road surface intensity image correspond to a plurality of data points in the target point cloud data;

[0144] The first point cloud data of each road surface element is determined according to the target image and the road surface intensity image.

[0145] In a possible implementation, the determination module 403 is specifically configured to:

[0146] Performing a second segmentation process on the target image to obtain a plurality of road surface elements in the target image;

[0147] Determine the projection relationship of the pixel points in the road surface intensity image through projection processing; wherein the projection relationship represents the corresponding relationship between the pixel points in the road surface intensity image and the road surface elements in the target image;

[0148] The first point cloud data of each road surface element is determined according to the projection relationship of the pixel points.

[0149] In a possible implementation, the compensation module 402 is specifically configured to:

[0150] Acquire a road surface intensity image of the target point cloud data; wherein the pixel points of the road surface intensity image correspond to a plurality of data points in the target point cloud data;

[0151] Performing a first segmentation process on the road surface intensity image to obtain second point cloud data of each road surface element;

[0152] According to the first point cloud data and the second point cloud data, fused point cloud data of each road surface element is obtained.

[0153] In a possible implementation, the determination module 403 is specifically configured to:

[0154] The first point cloud data of each road surface element is added to the second point cloud data of the corresponding road surface element to obtain fused point cloud data of each road surface element.

[0155] The device for constructing a high-precision map for intersections provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effects are similar, and are not described in detail in this embodiment.

[0156] Fig. 9 This is a schematic diagram of the structure of the electronic device provided in this application. Fig. 9 As shown, the electronic device 500 provided in this embodiment includes: at least one processor 501 and a memory 502. Optionally, the device 500 also includes a communication component 503. The processor 501, the memory 502 and the communication component 503 are connected via a bus 504.

[0157] In a specific implementation process, at least one processor 501 executes the computer-executable instructions stored in the memory 502, so that at least one processor 501 executes the above method.

[0158] The specific implementation process of the processor 501 can be found in the above method embodiment, and its implementation principle and technical effect are similar, so this embodiment will not be repeated here.

[0159] In the above embodiments, it should be understood that the processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the invention may be directly implemented as being executed by a hardware processor, or may be executed by a combination of hardware and software modules in the processor.

[0160] The memory may include a high-speed memory (Random Access Memory, RAM), and may also include a non-volatile memory (Non-volatile Memory, NVM), such as at least one disk memory.

[0161] The bus may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, the bus in the drawings of the present application is not limited to only one bus or one type of bus.

[0162] The present application also provides a computer program product, including a computer program, which implements the above method when executed by a processor.

[0163] The present application also provides a computer-readable storage medium, in which computer-executable instructions are stored. When a processor executes the computer-executable instructions, the above method is implemented.

[0164] The above-mentioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The readable storage medium can be any available medium that can be accessed by a general or special-purpose computer.

[0165] An exemplary readable storage medium is coupled to a processor so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can be located in an application specific integrated circuit (Application Specific Integrated Circuits, referred to as: ASIC). Of course, the processor and the readable storage medium can also exist in the device as discrete components.

[0166] The division of units is only a logical function division, and there may be other divisions in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interface, device or unit, which can be electrical, mechanical or other forms.

[0167] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0168] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0169] If the function is implemented in the form of 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, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0170] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, disk or optical disk and other media that can store program codes.

[0171] Finally, it should be noted that those skilled in the art will readily conceive of other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. The present invention is intended to cover any variations, uses or adaptations of the present invention, which follow the general principles of the present invention and include common knowledge or customary technical means in the art not disclosed by the present invention, are not limited to the precise structure described above and shown in the drawings, and may be modified and changed in various ways without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.

Claims

1. A method for constructing a high-precision map for intersections, characterized in that: include: Acquire initial point cloud data and initial image collected by roadside equipment; wherein the initial point cloud data and the initial image include multiple road surface elements of the target intersection; Performing compensation processing on the initial point cloud data to obtain target point cloud data; Determine the first point cloud data of each road surface element according to the initial image and the target point cloud data; and obtain the fused point cloud data of each road surface element according to the target point cloud data and the first point cloud data; A high-precision map of the target intersection is constructed based on the fused point cloud data of each road surface element.

2. The method according to claim 1, characterized in that The compensating the initial point cloud data to obtain target point cloud data includes: Determine a point set according to a first distance of each data point in the initial point cloud data; wherein the first distance represents the distance from the data point to the roadside equipment; and the data points in the point set represent data points whose first distance is greater than or equal to a preset threshold; For each data point in the point set, K neighboring points of the data point are determined, wherein K is a positive integer; and the target point cloud data is obtained through interpolation processing based on the neighboring points and the data point.

3. The method according to claim 2, characterized in that The step of obtaining the target point cloud data by interpolation processing based on the neighboring points and the data point includes: Performing interpolation processing on the data point and the neighboring point to obtain an interpolation data point; The intensity value of the interpolated data point is determined according to the intensity value of the data point and the intensity value of the neighboring point to obtain the target point cloud data.

4. The method according to claim 1, characterized in that: Determining first point cloud data of each road surface element according to the initial image and the target point cloud data includes: Acquire a road surface intensity image of the target point cloud data; and perform correction processing on the initial image to obtain a target image; wherein the pixel points of the road surface intensity image correspond to a plurality of data points in the target point cloud data; The first point cloud data of each road surface element is determined according to the target image and the road surface intensity image.

5. The method according to claim 4, characterized in that The step of determining first point cloud data of each road surface element according to the target image and the road surface intensity image comprises: Performing a second segmentation process on the target image to obtain a plurality of road surface elements in the target image; Determine the projection relationship of the pixel points in the road surface intensity image through projection processing; wherein the projection relationship represents the corresponding relationship between the pixel points in the road surface intensity image and the road surface elements in the target image; The first point cloud data of each road surface element is determined according to the projection relationship of the pixel points.

6. The method according to any one of claims 1 to 5, characterized in that: The step of obtaining fused point cloud data of each road surface element according to the target point cloud data and the first point cloud data includes: Acquire a road surface intensity image of the target point cloud data; wherein the pixel points of the road surface intensity image correspond to a plurality of data points in the target point cloud data; Performing a first segmentation process on the road surface intensity image to obtain second point cloud data of each road surface element; According to the first point cloud data and the second point cloud data, fused point cloud data of each road surface element is obtained.

7. The method according to claim 6, characterized in that The step of obtaining fused point cloud data of each road surface element according to the first point cloud data and the second point cloud data includes: The first point cloud data of each road surface element is added to the second point cloud data of the corresponding road surface element to obtain fused point cloud data of each road surface element.

8. A device for constructing a high-precision map for intersections, characterized in that: include: An acquisition module, used to acquire initial point cloud data and initial image collected by roadside equipment; wherein the initial point cloud data and the initial image include multiple road surface elements of the target intersection; A compensation module, used for performing compensation processing on the initial point cloud data to obtain target point cloud data; A determination module, configured to determine first point cloud data of each road surface element according to the initial image and the target point cloud data; and to obtain fused point cloud data of each road surface element according to the target point cloud data and the first point cloud data; A construction module is used to construct a high-precision map of the target intersection based on the fused point cloud data of each road surface element.

9. An electronic device, characterized in that: include: Memory, processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory, so that the processor performs the method according to any one of claims 1 to 7.

10. A computer-readable storage medium / computer program product, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 7 when executed by a processor; and / or, The computer program product comprises a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images