Lane line detection method, device, electronic device and storage medium

Abstract

The present application provides a lane line detection method, device, electronic device and computer-readable storage medium. The lane line detection method includes: acquiring a state image containing a lane line to be detected; performing prediction processing based on the state image to obtain regional pixel point information of each connected area of ​​the lane line to be detected, and the center axis distance information between each pixel point in each connected area and the lane line to be detected; determining the center axis pixel point set of each connected area based on the regional pixel point information and the center axis distance information; performing clustering processing on the center axis pixel point set to obtain a target center axis pixel point set for each lane line to be detected; performing fitting processing based on the target center axis pixel point set to obtain the curve data of each lane line to be detected. The present application can avoid the problem of being unable to accurately detect lane lines for lower quality images, thereby improving the detection accuracy of lane lines.

Description

Technical Field

[0001] The present application relates to the field of intelligent transportation technology, and in particular to a lane line detection method, device, electronic device and computer-readable storage medium. Background Art

[0002] In recent years, artificial intelligence (AI) technologies have developed rapidly. As a major application scenario in the field of AI, safe driving has received increasing attention and attention from industry professionals. Accurate lane detection plays a crucial role in ensuring safe driving.

[0003] In the existing technology, lane line edge information is mainly extracted based on the image through edge detection technology, and then the lane lines are fitted through processing such as Hough transform.

[0004] However, existing lane line detection methods have high requirements on image quality. In scenarios with poor image quality (such as too bright, too dark, blurred lane lines, etc.), the edge information of the lane lines is destroyed, resulting in the inability to accurately detect the lane lines. Summary of the Invention

[0005] The present application provides a lane line detection method, device, electronic device and computer-readable storage medium, aiming to solve the problem in the prior art that lane lines cannot be accurately detected for low-quality images.

[0006] In a first aspect, the present application provides a lane line detection method, the method comprising:

[0007] Obtain a state image containing lane lines to be detected;

[0008] Performing prediction processing on the state image to obtain regional pixel information of each connected region of the lane line to be detected, and distance information between each pixel in each connected region and the center axis of the lane line to be detected;

[0009] Determining a set of central axis pixels of each connected region according to the region pixel information and the central axis distance information;

[0010] Clustering the central axis pixel point set to obtain a target central axis pixel point set for each lane line to be detected;

[0011] Fitting processing is performed on the target central axis pixel point set to obtain curve data of each lane line to be detected, wherein the curve data is used to indicate the lane line to be detected.

[0012] In a second aspect, the present application provides a lane line detection device, the lane line detection device comprising:

[0013] An acquisition unit, configured to acquire a state image including lane lines to be detected;

[0014] a prediction unit, configured to perform prediction processing based on the state image to obtain regional pixel information of each connected region of the lane line to be detected, and distance information between each pixel in each connected region and the center axis of the lane line to be detected;

[0015] The prediction unit is further configured to determine a set of central axis pixels of each connected region based on the region pixel information and the central axis distance information;

[0016] A clustering unit is used to perform clustering processing on the central axis pixel point set to obtain a target central axis pixel point set for each lane line to be detected;

[0017] A fitting unit is used to perform fitting processing based on the target central axis pixel point set to obtain curve data of each lane line to be detected, wherein the curve data is used to indicate the lane line to be detected.

[0018] In a possible implementation of the present application, the clustering unit is further configured to:

[0019] performing fitting processing on each point in the set of central axis pixels of each connected region according to a preset first polynomial to obtain a first fitting curve for each connected region;

[0020] Acquiring curve characteristic information of the first fitting curve;

[0021] According to the curve feature information, the central axis pixel point set is clustered to obtain a target central axis pixel point set for each lane line to be detected.

[0022] In a possible implementation of the present application, the prediction unit is further configured to:

[0023] Calling the segmentation subnetwork in the trained segmentation network to perform segmentation prediction processing based on the state image to obtain regional pixel point information of each connected area of ​​the lane line to be detected;

[0024] The regression subnetwork in the trained segmentation network is called to perform regression prediction processing according to the state image to obtain the distance information between each pixel point in each connected area and the center axis of the lane line to be detected.

[0025] In a possible implementation of the present application, the lane detection device further includes a training unit, which is specifically configured to:

[0026] Acquire a target image, wherein the target image has been annotated with actual regional pixel point information and actual central axis distance information;

[0027] Calling a segmentation subnetwork in a preset segmentation network to perform segmentation processing according to the target image, determine the connected area information of the lane lines in the target image, and obtain a connected area prediction result of the target image;

[0028] Calling the regression subnetwork in the preset segmentation network to perform regression processing based on the target image, determine the distance information between each pixel in the target image and the center axis of the lane line in the target image, and obtain a prediction result of the center axis distance of the target image;

[0029] The preset segmentation network is trained according to the connected region prediction result and the median distance prediction result to obtain a trained segmentation network.

[0030] In one possible implementation of the present application, the regional pixel point information includes a first horizontal coordinate value and a first vertical coordinate value of each pixel point in each connected area, and the center axis distance information includes a center axis horizontal coordinate difference and a center axis vertical coordinate difference between each pixel point in each connected area and the lane line to be detected, and the prediction unit is further configured to:

[0031] Determine, based on the first abscissa value and the median abscissa difference, a second abscissa value of each median pixel point of each connected region; and determine, based on the first ordinate value and the median ordinate difference, a second ordinate value of each median pixel point of each connected region;

[0032] The central axis pixel point set of each connected area is determined according to the second horizontal coordinate value and the second vertical coordinate value.

[0033] In a possible implementation of the present application, the fitting unit is further configured to:

[0034] According to a preset second polynomial, each point in the target central axis pixel point set is fitted to obtain curve data of each lane line to be detected.

[0035] In a possible implementation of the present application, before the step of performing prediction processing according to the state image, the acquiring unit is further configured to:

[0036] Performing inverse perspective transformation on the state image to obtain a transformed state image;

[0037] In a possible implementation of the present application, the prediction unit is further configured to:

[0038] Prediction processing is performed according to the transformed state image.

[0039] In a third aspect, the present application also provides an electronic device, which includes a processor and a memory, wherein a computer program is stored in the memory, and when the processor calls the computer program in the memory, it executes the steps of any lane line detection method provided in the present application.

[0040] In a fourth aspect, the present application also provides a computer-readable storage medium on which a computer program is stored, and the computer program is loaded by a processor to execute the steps in the lane line detection method.

[0041] This application performs predictive processing on a state image containing lane lines to be detected to obtain regional pixel point information for each connected region of the lane lines to be detected, as well as information on the center axis distance between each pixel point in each connected region and the lane lines to be detected; based on the regional pixel point information and the center axis distance information, the center axis pixel point set of each connected region is determined; clustering processing is performed on the center axis pixel point set to obtain a target center axis pixel point set for each lane line to be detected; and fitting processing is performed on the target center axis pixel point set to obtain curve data for each lane line to be detected. On the one hand, the detection of lane lines of arbitrary shapes can be achieved.

[0042] On the other hand, since the center axis of the lane line to be detected most directly reflects the position of the lane line, this application takes the center axis of the lane line to be detected as the starting point. The center axis pixel point set of each connected area is determined by prediction and based on the regional pixel point information and the center axis distance information. The center axis pixel point set is clustered to obtain the target center axis pixel point set of each lane line to be detected to detect the lane line to be detected. In addition, there is no need to rely on edge detection technology to extract lane line edge information, thereby avoiding the problem that the existing lane line detection method has high requirements on image quality and cannot accurately detect lane lines for lower quality images, thereby improving the detection accuracy of lane lines. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0044] Figure 1 This is a schematic diagram of a lane line scenario provided by an embodiment of the present application;

[0045] Figure 2 This is a flow chart of a lane line detection method provided in an embodiment of the present application;

[0046] Figure 3This is a schematic diagram of a connected area provided in an embodiment of the present application;

[0047] Figure 4 This is a schematic diagram of the center axis of the lane line to be detected provided in an embodiment of the present application;

[0048] Figure 5 2 is a schematic diagram comparing the state image provided in the embodiment of the present application before and after the inverse perspective transformation process;

[0049] Figure 6 This is a schematic diagram of the central axis pixel points of the connected area provided in the embodiment of the present application;

[0050] Figure 7 This is a schematic diagram of the overall framework of the trained segmentation network provided in an embodiment of the present application;

[0051] Figure 8 This is a schematic structural diagram of an embodiment of a lane detection device provided in an embodiment of the present application;

[0052] Figure 9 It is a schematic diagram of the structure of an embodiment of an electronic device provided in the embodiments of the present application. DETAILED DESCRIPTION

[0053] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0054] In the description of the embodiments of the present application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of the present application, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0055] In order to enable any person skilled in the art to implement and use the present application, the following description is provided. In the following description, details are listed for the purpose of explanation. It should be understood that one of ordinary skill in the art will recognize that the present application can be implemented without using these specific details. In other examples, well-known processes will not be elaborated in detail to avoid obscuring the description of the embodiments of the present application with unnecessary details. Therefore, the present application is not intended to be limited to the embodiments shown, but is consistent with the widest range of principles and features disclosed in accordance with the embodiments of the present application.

[0056] The embodiments of the present application provide a lane detection method, device, electronic device, and computer-readable storage medium. The lane detection device can be integrated into an electronic device, which can be a server, a terminal, or other device.

[0057] First, before introducing the embodiments of the present application, the following is an introduction to the application background of the embodiments of the present application.

[0058] In recent years, artificial intelligence (AI) technologies have developed rapidly. As a major application scenario in AI, safe driving has garnered increasing attention and focus within the industry. Lane detection, a key component of safe driving, aims to locate lane markings in images. However, due to the complexity of road scenes, lane detection is one of the more challenging tasks in computer vision.

[0059] like Figure 1 As shown, Figure 1 This is a schematic diagram of a lane line scenario provided in an embodiment of the present application.

[0060] The existing lane line detection solution still has the following problems:

[0061] 1. Current lane detection solutions based on traditional image methods primarily use edge detection techniques to extract lane edge information and then perform Hough transforms to fit lane lines. These algorithms suffer from poor robustness and are sensitive to image quality requirements. They are not well suited for road scenarios such as overly bright or dark images, blurry images, and blurred lane lines.

[0062] 2. The current lane line detection method based on the drivable area is too dependent on large-scale road scene datasets, and the ability to describe the spatial position distribution of lane lines is not robust.

[0063] Based on the above-mentioned defects of the existing related technologies, the embodiments of the present application provide a lane line detection method, which at least to a certain extent overcomes the defects of the existing related technologies.

[0064] The executor of the lane line detection method of the embodiment of the present application can be the lane line detection device provided in the embodiment of the present application, or different types of electronic devices such as a server device, a physical host or a user equipment (UE) that integrates the lane line detection device. Among them, the lane line detection device can be implemented in hardware or software, and the UE can specifically be a terminal device such as a smart phone, a tablet computer, a laptop computer, a PDA, a desktop computer or a personal digital assistant (PDA).

[0065] The electronic device can operate in a standalone mode or in a device cluster mode. By applying the lane line detection method provided in the embodiment of the present application, the problem of being unable to accurately detect lane lines for lower-quality images can be avoided, thereby improving the lane line detection accuracy.

[0066] Next, we will begin to introduce the lane line detection method provided in the embodiment of the present application. In the embodiment of the present application, an electronic device is used as the execution subject. For the sake of simplicity and ease of description, the execution subject will be omitted in the subsequent method embodiments. The lane line detection method includes: acquiring a state image containing the lane line to be detected; performing prediction processing based on the state image to obtain regional pixel point information of each connected area of ​​the lane line to be detected, and the center axis distance information between each pixel point in each connected area and the lane line to be detected; determining the center axis pixel point set of each connected area based on the regional pixel point information and the center axis distance information; performing clustering processing on the center axis pixel point set to obtain the target center axis pixel point set of each lane line to be detected; performing fitting processing based on the target center axis pixel point set to obtain the curve data of each lane line to be detected.

[0067] Reference Figure 2 , Figure 2 A flowchart of a lane detection method provided in an embodiment of the present application. It should be noted that although a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than shown here. The lane detection method includes steps S10 to S50, wherein:

[0068] S10: Acquire a state image containing lane lines to be detected.

[0069] The state image refers to an image containing lane lines. The lane lines contained in the state image can be one or more. Unless otherwise specified, the state image referred to below refers to the image containing the lane lines to be detected.

[0070] There are many ways to obtain a state image containing the lane line to be detected. For example, a lane image taken from a driving angle can be obtained as the state image containing the lane line to be detected.

[0071] Specifically, in actual applications, the electronic device that applies the lane line detection method provided in the embodiments of the present application may directly include the vehicle's camera in hardware, and locally store the images captured by the camera, which can be directly read inside the electronic device; alternatively, the electronic device may establish a network connection with the vehicle's camera, and obtain the images captured by the camera online from the camera based on the network connection; alternatively, the electronic device may read the images captured by the camera from a related storage medium that stores the images captured by the vehicle's camera. The specific acquisition method is not limited here.

[0072] The camera can capture images according to a preset shooting method, such as setting the shooting height, shooting direction, or shooting distance. The specific shooting method can be adjusted according to the camera itself and is not limited here. The images captured by the camera are composed of multiple frames, which can be combined into a video through a timeline.

[0073] S20 , performing prediction processing based on the state image to obtain regional pixel point information of each connected area of ​​the lane line to be detected, and distance information between each pixel point in each connected area and the center axis of the lane line to be detected.

[0074] The following details the pixel information of each connected region of the lane line to be detected and the distance between each pixel in each connected region and the center axis of the lane line to be detected.

[0075] In the first aspect, the connected region refers to a mask of the lane line region to be detected in the state image. In the embodiment of the present application, the mask of the lane line to be detected (i.e., the connected region) is obtained by prediction. Each connected region can be approximately regarded as a segment of each lane line to be detected, that is, each lane line to be detected can be approximately regarded as consisting of one or more connected regions.

[0076] The regional pixel information of each connected region refers to the position information of each pixel in the connected region. The position information of each pixel can be represented by coordinates. For example, each pixel in the connected region (denoted as point M) can be represented by coordinates. i ) corresponds to the camera coordinates (denoted as (x i ,y i )), as the position information of each pixel.

[0077] When acquiring the state image, the camera coordinates of each pixel in the state image can also be acquired simultaneously, that is, each pixel in the state image can be represented by the camera coordinates. In the following embodiments of the present application, the coordinates mentioned are taken as an example of the camera coordinates of the state image. It is understood that the camera coordinates of the pixels in the state image can also be further converted.

[0078] Please refer to Figure 3 , Figure 3 This is a schematic diagram of a connected area provided in an embodiment of the present application. Figure 3 (a) is the state image containing the lane line to be detected; Figure 3 (b) shows the regional pixel point information of each connected area of ​​the lane line to be detected after predicting the state image in (a).

[0079] Due to factors such as the instability of connected region prediction, the fact that each lane line to be detected is not continuously connected, and the quality of the state image, the mask map of each lane line area may not be a complete connected region, but rather multiple connected regions in a "disconnected" state. To reduce data interference and improve data processing efficiency and quality, connected regions with an area smaller than a preset area threshold can be filtered out before subsequent data processing. The preset area threshold can be set according to specific needs and is not limited here.

[0080] like Figure 3 As shown in , for example, the areas of connected regions 1, 2, 3, 4, 5, 6, and 7 are 3 cm 2 , 1.2cm 2 , 1.5cm 2 , 2cm 2 , 1cm 2 , 0.2cm 2 , 0.1cm 2 , the preset area threshold is 1cm 2 , then the connected areas 6 and 7 can be filtered out, and the connected areas 1, 2, 3, 4, and 5 can be retained, as shown in Figure 3 As shown in (c).

[0081] Secondly, since the actual lane line to be detected is not a single line segment, but a line segment with a certain width, in order to facilitate data processing, the concept of "central axis" is proposed in the embodiment of this application, which mainly refers to a single line segment formed by the center point position of the lane line to be detected with a certain width. Figure 4 As shown, Figure 4 This is a schematic diagram of the center axis of the lane line to be detected provided in an embodiment of the present application. Figure 4 (a) shows a state image containing two lane lines to be detected. Figure 4 Each dashed line segment in (b) represents the center axis of the corresponding lane line to be detected.

[0082] After predicting the state image, each lane line to be detected can be considered to be composed of one or more connected regions. Therefore, the median axis of each lane line to be detected will be considered the median axis of each connected region. In other words, the curve formed by the median pixels of the lane line to be detected within each connected region will be considered the median axis of that connected region.

[0083] The median distance information refers to the distance information between each pixel point in each connected region and the median axis of the connected region (hereinafter referred to as "median distance information of the connected region"). In some embodiments, the median distance information specifically refers to the distance between each pixel point in each connected region (denoted as point M i , the corresponding coordinates are expressed as (x i ,y i )), the distance difference information between the x-axis and y-axis of the connected area (denoted as (dx i ,dy i )).

[0084] For better explanation, in the embodiment of the present application, the distance difference information between each pixel point in each connected area and the central axis of the connected area on the x-axis and y-axis is taken as the central axis distance information. It is understandable that in some cases, the distance difference information between each pixel point in each connected area and the central axis of the connected area on the x-axis can also be used as the central axis distance information. Or the distance difference information between each pixel point in each connected area and the central axis of the connected area on the y-axis can be used as the central axis distance information.

[0085] Furthermore, in order to better utilize the spatial position information of the lane line, the state image can be subjected to inverse perspective transformation and then prediction processing to obtain the regional pixel point information of each connected area of ​​the lane line to be detected, as well as the distance information between each pixel point in each connected area and the center axis of the lane line to be detected.

[0086] That is, before the step of "performing prediction processing based on the state image", the method further includes: performing inverse perspective transformation processing on the state image to obtain a transformed state image. The step of "performing prediction processing based on the state image" specifically includes: performing prediction processing based on the transformed state image.

[0087] For details, please refer to Figure 5 , Figure 5 3 is a schematic diagram comparing the state image provided in the embodiment of the present application before and after the inverse perspective transformation processing. Figure 5 (a) shows the state image before the inverse perspective transformation process. Figure 5 (b) in the middle shows the state image after the inverse perspective transformation processing.

[0088] It can be seen that by performing inverse perspective transformation on the original captured state image, the original state image is converted into a bird's-eye view, so as to better utilize the spatial position information of the lane line, providing more favorable conditions for the next step (step S40) of clustering the central axis pixel point set, making the algorithm more robust.

[0089] S30. Determine a central axis pixel point set of each connected area according to the area pixel point information and the central axis distance information.

[0090] The central axis pixel set of each connected region refers to the set of central axis pixel points contained in the connected region (and the lane line to be detected corresponding to the connected region). To simplify the description, in the following text, each central axis pixel point contained in the connected region (and the lane line to be detected corresponding to the connected region) is referred to as the central axis pixel point (or the central axis pixel point of the connected region).

[0091] For easier understanding, please refer to Figure 6 , Figure 6 This is a schematic diagram of the central axis pixel points of the connected area provided in an embodiment of the present application. Figure 6 (a) represents the state image, Figure 6 (b) is an enlarged schematic diagram of the connected area C. Figure 6 (b) shows that each pixel point in the connected area C (denoted as point M i )’s position information (e.g., expressed in coordinates, denoted as (x i ,y i )), and each pixel point in each connected area (M i ) and the center axis distance information of the lane line to be detected (denoted as (dx i ,dy i )), each central axis pixel point in the connected area is determined respectively.

[0092] In some embodiments, each central axis pixel point in the connected region can be represented by coordinates. Specifically, each central axis pixel point in the connected region (denoted as point N) can be determined by the following formula (1): i ) coordinates.

[0093] (x i ',y i ')=(x i +dx i ,y i +dy i ) Formula (1)

[0094] Among them, (x i ',y i ') represents the central axis pixel N in the connected areai The coordinate value of (x i ,y i ) represents the pixel point M in the connected area i The coordinate value, dx i Represents the pixel point M in the connected area i The difference between the horizontal coordinate of the median axis of the connected area, dy i Represents the pixel point M in the connected area i The difference between the vertical coordinate and the median axis of the connected region.

[0095] In some embodiments, the regional pixel point information includes a first horizontal coordinate value and a first vertical coordinate value of each pixel point in each connected area, and the center axis distance information includes a center axis horizontal coordinate difference and a center axis vertical coordinate difference between each pixel point in each connected area and the lane line to be detected. Step S30 may specifically include: determining a second horizontal coordinate value of each center axis pixel point in each connected area based on the first horizontal coordinate value and the center axis horizontal coordinate difference; determining a second vertical coordinate value of each center axis pixel point in each connected area based on the first vertical coordinate value and the center axis vertical coordinate difference; and determining a center axis pixel point set of each connected area based on the second horizontal coordinate value and the second vertical coordinate value.

[0096] The first horizontal coordinate value refers to the horizontal coordinate value corresponding to each pixel point in the connected area. The first vertical coordinate value refers to the vertical coordinate value corresponding to each pixel point in the connected area. The first horizontal coordinate value and the first vertical coordinate value are used to indicate the value of each pixel point (denoted as point M) in each connected area. i )'s coordinate position (denoted as (x i ,y i )).

[0097] The second horizontal coordinate value refers to the horizontal coordinate value corresponding to each central axis pixel point in the connected area. The second vertical coordinate value refers to the vertical coordinate value corresponding to each central axis pixel point in the connected area. That is, the second horizontal coordinate value and the second vertical coordinate value are used to indicate each central axis pixel point in the connected area (denoted as point N i )'s coordinate position (denoted as (x i ',y i '))

[0098] Specifically, for easier understanding, please continue to refer to Figure 6For example, the coordinate positions of the pixel points M1, M2, M3, and M4 in the connected area C are (x1, y1), (x2, y2), (x3, y3), and (x4, y4), and the distance information between M1, M2, M3, and M4 and the central axis of the connected area C are (dx1, dy1), (dx2, dy2), (dx3, dy3), and (dx4, dy4), respectively. Then, the coordinate positions of the central axis pixel points N1, N2, N3, and N4 in the connected area C can be determined as (x1', y1'), (x2', y2'), (x3', y3'), and (x4', y4'). Similarly, the set of central axis pixel points of each connected area can be determined.

[0099] From the above content, it can be seen that since each pixel point in the state image has a unique corresponding coordinate value, by using the coordinate value to represent each pixel point in each connected area, the central axis pixel point set of each connected area can be accurately and quickly determined.

[0100] S40 , clustering the central axis pixel point set to obtain a target central axis pixel point set for each lane line to be detected.

[0101] At this time, in some embodiments, according to a preset first polynomial, each point in the central axis pixel point set of each connected area is fitted to obtain a first fitting curve of each connected area; curve feature information of the first fitting curve is obtained; according to the curve feature information, the central axis pixel point set is clustered to obtain a target central axis pixel point set for each lane line to be detected.

[0102] The preset first polynomial can be any polynomial of multiple orders, such as a first-order polynomial, a second-order polynomial, or a third-order polynomial, and can be preset according to actual needs. For example, when the lane line to be detected is a straight line, the preset first polynomial can be a first-order polynomial, as shown in the following formula (2). The preset first polynomial can be expressed as:

[0103] f j (x) = b j x+c j Formula (2)

[0104] Among them, f j (x) represents the vertical coordinate value of each point in the set of central axis pixels of connected region j; x represents the horizontal coordinate value of each point in the set of central axis pixels of connected region j; b j 、c j are constants respectively, and are determined by the curve characteristics of the curve formed by each point in the set of median pixel points of the connected region j. Each parameter in the preset first polynomial is an unknown number.

[0105] For another example, when the lane line to be detected is a straight line or a curve, the preset first polynomial can also be a second-order polynomial, as shown in the following formula (3). The preset first polynomial can be expressed as:

[0106] f j (x) = a j x 2 +b j x+c j Formula (3)

[0107] Among them, f j (x) represents the ordinate value of each point in the set of central axis pixels of connected region j; x represents the abscissa value of each point in the set of central axis pixels of connected region j; a j 、b j 、c j are constants respectively, and are determined by the curve characteristics of the curve formed by each point in the set of median pixel points of the connected region j. Each parameter in the preset first polynomial is an unknown number.

[0108] The first fitting curve is a curve obtained by fitting each point in the set of central axis pixels of each connected region. A first fitting curve can be determined for each connected region.

[0109] Specifically, first, according to the preset first polynomial, each point in the set of median pixel points of each connected region is fitted to obtain the first fitting curve of each connected region. For example, according to the above formula (3), each point in the set of median pixel points of connected region 1 is fitted to obtain the first fitting curve 1 of connected region 1 (expressed as f1(x)=a1x 2 +b1x+c1, where a1, b1, and c1 are constants respectively); each point in the set of central axis pixels of the connected region 2 is fitted to obtain the first fitting curve 2 of the connected region 2 (expressed as f2(x)=a2x 2 +b2x+c2, where a2, b2, and c2 are constants respectively).

[0110] Then, the curve characteristic information of the first fitting curve is obtained, for example, the opening information of the first fitting curve (using a j Indicates), location information (using (represented), the vertical coordinate value range of the first fitting curve in the state image, the distance between any two closest end points of the first fitting curve, and other information.

[0111] Finally, based on the curve feature information of the first fitting curve, the central axis pixel point set is clustered to obtain the target central axis pixel point set for each lane line to be detected. For example, in some embodiments, when any two first fitting curves (such as curve 1 and curve 2) are detected to meet the following five conditions, the central axis pixel point sets corresponding to the two first fitting curves are clustered into the same cluster set. Each central axis pixel point set contained in each cluster set is used as the target central axis pixel point set for each lane line to be detected, for subsequent use in fitting the curve data of the lane line to be detected.

[0112] Condition 1: Parameters a1 and a2 of curve 1 and curve 2 satisfy the following formula (4).

[0113] a1*a2>=0 Formula (4)

[0114] Among them, the parameters of curve 1 and curve 2 satisfy formula (4) and formula (5), which means that the opening directions of curve 1 and curve 2 are the same.

[0115] Condition 2: Parameters c1 and c2 of curve 1 and curve 2 satisfy the following formula (5).

[0116] c1*c2>=0 formula (5)

[0117] Among them, the parameters of curve 1 and curve 2 satisfy formula (5), which means that the positions of curve 1 and curve 2 in the y-axis direction are the same or close.

[0118] Condition 3: The parameters a1, a2, b1, and b2 of Curve 1 and Curve 2 satisfy the following formula (6).

[0119]

[0120] Wherein, T1 is a preset empirical value that can be set according to specific needs. The specific value of T1 is not limited here. The parameters of curve 1 and curve 2 satisfy formula (6), indicating that the central axis positions of curve 1 and curve 2 are the same or close.

[0121] Condition 4: The vertical coordinate value range of curve 1 in the state image does not overlap with the vertical coordinate value range of curve 2 in the state image.

[0122] Condition 5: The distance between the closest endpoints of curve 1 and curve 2 is less than a preset threshold value T2. T2 is a preset empirical value that can be set according to specific needs. There is no restriction on the specific value of T2.

[0123] By limiting the above five conditions, the central axis pixels of multiple connected areas belonging to the same lane line to be detected can be grouped, providing an accurate data basis for subsequent fitting to determine the curve data of the lane line to be detected, thereby improving the accuracy of the curve data of the lane line to be detected, that is, improving the detection accuracy of the lane line to be detected.

[0124] From the above content, it can be seen that by fitting each point in the central axis pixel point set of each connected area to obtain the first fitting curve of each connected area, and clustering the central axis pixel point set according to the curve feature information of the first fitting curve, the central axis pixel points of multiple connected areas belonging to the same lane line to be detected can be grouped, which provides an accurate data basis for the subsequent fitting to determine the curve data of the lane line to be detected, thereby improving the accuracy of the curve data of the lane line to be detected, that is, improving the detection accuracy of the lane line to be detected.

[0125] S50 , performing fitting processing according to the target central axis pixel point set to obtain curve data of each lane line to be detected.

[0126] The curve data of each lane line to be detected is used to indicate the position of the lane line to be detected in the status image.

[0127] After determining the target central axis pixel point set for each lane line to be detected, fitting can be performed based on each point in the target central axis pixel point set to determine the curve data of each lane line to be detected, thereby realizing the detection of the lane line to be detected in the status image.

[0128] In some embodiments, each point in the target mid-axis pixel point set may be fitted using the least squares method to obtain curve data for each lane line to be detected.

[0129] As can be seen from the above, by performing prediction processing on the state image containing the lane line to be detected, regional pixel information for each connected area of ​​the lane line to be detected and the distance information between each pixel in each connected area and the center axis of the lane line to be detected are obtained. Based on the regional pixel information and center axis distance information, the center axis pixel set of each connected area is determined. Clustering processing is performed on the center axis pixel set to obtain the target center axis pixel set for each lane line to be detected. Fitting processing is performed on the target center axis pixel set to obtain the curve data for each lane line to be detected. On the one hand, lane lines of arbitrary shapes can be detected.

[0130] On the other hand, since the center axis of the lane line to be detected most directly reflects the position of the lane line, the embodiment of the present application takes the center axis of the lane line to be detected as the starting point. The center axis pixel point set of each connected area is determined by prediction and based on the regional pixel point information and the center axis distance information. The center axis pixel point set is clustered to obtain the target center axis pixel point set of each lane line to be detected to detect the lane line to be detected. In addition, there is no need to rely on edge detection technology to extract lane line edge information, thereby avoiding the problem that the existing lane line detection method has high requirements on image quality and cannot accurately detect lane lines for lower quality images, thereby improving the detection accuracy of lane lines.

[0131] In some embodiments, step S50 may specifically include: performing fitting processing on each point in the target central axis pixel point set according to a preset second polynomial to obtain curve data of each lane line to be detected.

[0132] The preset second polynomial can be any polynomial of multiple orders, such as a first-order polynomial, a second-order polynomial, or a third-order polynomial, and can be preset according to actual needs. For example, when the lane line to be detected is a straight line, the preset second polynomial can be a first-order polynomial, as shown in the following formula (7). The preset second polynomial can be expressed as:

[0133] f k (x) = b k x+c k Formula (7)

[0134] Among them, f k (x) represents the ordinate value of each point in the target mid-axis pixel point set for each lane line k to be detected; x represents the abscissa value of each point in the target mid-axis pixel point set for each lane line k to be detected; b k 、c k are constants respectively, and are determined by the curve characteristics of the curve formed by each point in the set of median pixel points of the connected region j. Each parameter in the preset second polynomial is an unknown number.

[0135] For another example, when the lane line to be detected is a straight line or a curve, the preset second polynomial can also be a second-order polynomial, as shown in the following formula (8). The preset second polynomial can be expressed as:

[0136] F k (x) = a k x 2 +b k x+c k Formula (8)

[0137] Among them, f k(x) represents the ordinate value of each point in the target mid-axis pixel point set for each lane line k to be detected; x represents the abscissa value of each point in the target mid-axis pixel point set for each lane line k to be detected; a k 、b k 、c k are constants, respectively, and are determined by the curve characteristics of the curve formed by each point in the target center axis pixel set of each lane line k to be detected. Each parameter in the preset second polynomial is an unknown number.

[0138] Among them, the specific implementation of "fitting each point in the target central axis pixel point set according to a preset second polynomial to obtain the curve data of each lane line to be detected" is similar to the implementation of "fitting each point in the central axis pixel point set of each connected area according to a preset first polynomial to obtain the first fitting curve of each connected area". For details, please refer to the above step S40 and will not be repeated here.

[0139] In some embodiments, prediction processing can be performed using the trained segmentation network state image to obtain regional pixel information for each connected region of the lane line to be detected, as well as distance information between each pixel in each connected region and the center axis of the lane line to be detected. Specifically, step S20 may include: calling a segmentation subnetwork within the trained segmentation network to perform segmentation prediction processing based on the state image to obtain regional pixel information for each connected region of the lane line to be detected; and calling a regression subnetwork within the trained segmentation network to perform regression prediction processing based on the state image to obtain distance information between each pixel in each connected region and the center axis of the lane line to be detected.

[0140] Please refer to Figure 7 , Figure 7 This is a schematic diagram of the overall framework of the trained segmentation network provided in an embodiment of the present application.

[0141] like Figure 7 As shown, the trained segmentation network may include a backbone network, a segmentation sub-network, and a regression sub-network. The specific training process of the trained segmentation network may refer to the steps A1 to A4 below, which will not be described in detail here.

[0142] The backbone network, which can be a convolutional neural network, is used to output the image features of the state image, providing image feature data for the subsequent segmentation subnetwork and regression subnetwork to ensure the normal prediction of regional pixel information and median distance information. The backbone network takes the state image as input and outputs the image features of the state image.

[0143] The segmentation subnetwork is used to output regional pixel information. The segmentation subnetwork takes the image features of the state image as input and outputs regional pixel information.

[0144] The regression subnetwork is used to output the median distance information. The regression subnetwork takes the image features of the state image as input and outputs the median distance information.

[0145] From the above content, it can be seen that since the trained segmentation network can fully learn and explore better expressions for connected region segmentation and median distance information regression, by calling the trained segmentation network to predict regional pixel information and median distance information based on the state image, the regional pixel information and median distance information can be made more accurate, thereby improving the detection accuracy of lane lines.

[0146] In some embodiments, the trained segmentation network is obtained by training through the following steps A1 to A4, wherein:

[0147] A1. Acquire the target image.

[0148] The target image is an image containing lane lines used to train the segmentation network. The target image is annotated with the actual region pixel information and the actual center axis distance information.

[0149] The actual area pixel point information marked in the target image refers to the actual position information of the connected area of ​​the lane lines contained in the target image.

[0150] The actual median axis distance information marked in the target image refers to the actual distance information between each point in each connected area in the target image and the median axis of the connected area.

[0151] A2. Calling a segmentation subnetwork in a preset segmentation network, performing segmentation processing according to the target image, determining the connected area information of the lane lines in the target image, and obtaining a connected area prediction result of the target image.

[0152] The connected region prediction result refers to the information of the connected region of the lane line in the target image obtained through prediction.

[0153] In some embodiments, the pre-set segmentation network includes a backbone network, a segmentation sub-network, and a regression sub-network, corresponding to the post-training segmentation network. The roles of the backbone network, segmentation sub-network, and regression sub-network, as well as their input and output, can be found in the above description of the "post-training segmentation network," and will not be further elaborated here.

[0154] A3. Call the regression subnetwork in the preset segmentation network, perform regression processing based on the target image, determine the distance information between each pixel point in the target image and the center axis of the lane line in the target image, and obtain the center axis distance prediction result of the target image.

[0155] The center axis distance prediction result refers to the distance information between each pixel point in each connected area (of the lane line contained in the target image) and the center axis of the connected area obtained through prediction.

[0156] Specifically, the regression subnetwork in the preset segmentation network is called. First, regression processing is performed on the target image to determine the distance between each pixel in the target image and the center axis of the lane line in the target image. Then, based on the distance between each pixel in the target image and the center axis of the lane line in the target image, the distance between each pixel in the connected area (the predicted lane line in the target image) and the center axis of the corresponding lane line in the target image is obtained, thereby obtaining the center axis distance prediction result of the target image.

[0157] A4. Training the preset segmentation network according to the connected region prediction result and the median distance prediction result to obtain a trained segmentation network.

[0158] Specifically, on the one hand, the segmentation loss value of the segmentation sub-network is determined based on the connected region prediction results and the actual region pixel information that has been annotated in the target image. For example, a cross entropy loss function can be used to determine the segmentation loss value of the segmentation sub-network based on the connected region prediction results and the actual region pixel information.

[0159] On the other hand, the segmentation loss value of the regression sub-network is determined based on the medial axis distance prediction result and the actual medial axis distance information annotated in the target image. For example, a cross-entropy loss function can be used to determine the regression loss value of the regression sub-network based on the medial axis distance prediction result and the actual medial axis distance information.

[0160] Then, the sum of the segmentation loss value of the segmentation subnetwork and the regression loss value of the regression subnetwork is used as the total training loss value of the segmentation network, and the preset segmentation network is back-propagated and trained until the preset segmentation network converges (that is, meets the preset stop training conditions), and the preset segmentation network is used as the trained segmentation network.

[0161] At this point, the trained segmentation network can be used to perform prediction processing based on the image to obtain regional pixel information of each connected area of ​​the lane line to be detected, as well as the distance information between each pixel in each connected area and the center axis of the lane line to be detected.

[0162] As can be seen from the above, by training the preset segmentation network based on a target image annotated with actual regional pixel information and actual center-axis distance information, a trained segmentation network is obtained. This allows the trained segmentation network to perform prediction processing based on the image, obtaining regional pixel information for each connected region of the lane line to be detected, as well as the center-axis distance information between each pixel in each connected region and the lane line to be detected. Due to the strong self-learning ability of the neural network, regional pixel information and center-axis distance information can be determined more accurately, thereby improving lane line detection accuracy.

[0163] In order to better implement the lane line detection method in the embodiment of the present application, based on the lane line detection method, the embodiment of the present application also provides a lane line detection device, such as Figure 8 FIG. 8 is a schematic structural diagram of an embodiment of a lane detection device according to an embodiment of the present application. The lane detection device 800 includes:

[0164] An acquisition unit 801 is configured to acquire a state image including lane lines to be detected;

[0165] A prediction unit 802 is configured to perform prediction processing based on the state image to obtain pixel information of each connected region of the lane line to be detected, and distance information between each pixel in each connected region and the center axis of the lane line to be detected;

[0166] The prediction unit 802 is further configured to determine a set of central axis pixels of each connected region based on the region pixel information and the central axis distance information;

[0167] A clustering unit 803 is configured to perform clustering processing on the central axis pixel point set to obtain a target central axis pixel point set for each lane line to be detected;

[0168] The fitting unit 804 is used to perform fitting processing based on the target central axis pixel point set to obtain curve data of each lane line to be detected, wherein the curve data is used to indicate the lane line to be detected.

[0169] In some embodiments, the clustering unit 803 is further configured to:

[0170] performing fitting processing on each point in the set of central axis pixels of each connected region according to a preset first polynomial to obtain a first fitting curve for each connected region;

[0171] Acquiring curve characteristic information of the first fitting curve;

[0172] According to the curve feature information, the central axis pixel point set is clustered to obtain a target central axis pixel point set for each lane line to be detected.

[0173] In some embodiments, the prediction unit 802 is further configured to:

[0174] Calling the segmentation subnetwork in the trained segmentation network to perform segmentation prediction processing based on the state image to obtain regional pixel point information of each connected area of ​​the lane line to be detected;

[0175] The regression subnetwork in the trained segmentation network is called to perform regression prediction processing according to the state image to obtain the distance information between each pixel point in each connected area and the center axis of the lane line to be detected.

[0176] In some embodiments, the lane detection device further includes a training unit (not shown in the figure), which is specifically configured to:

[0177] Acquire a target image, wherein the target image has been annotated with actual regional pixel point information and actual central axis distance information;

[0178] Calling a segmentation subnetwork in a preset segmentation network to perform segmentation processing according to the target image, determine the connected area information of the lane lines in the target image, and obtain a connected area prediction result of the target image;

[0179] Calling the regression subnetwork in the preset segmentation network to perform regression processing based on the target image, determine the distance information between each pixel in the target image and the center axis of the lane line in the target image, and obtain a prediction result of the center axis distance of the target image;

[0180] The preset segmentation network is trained according to the connected region prediction result and the median distance prediction result to obtain a trained segmentation network.

[0181] In some embodiments, the regional pixel point information includes a first horizontal coordinate value and a first vertical coordinate value of each pixel point in each connected area, and the center axis distance information includes a horizontal coordinate difference and a vertical coordinate difference between each pixel point in each connected area and the lane line to be detected. The prediction unit 802 is further configured to:

[0182] Determine, based on the first abscissa value and the median abscissa difference, a second abscissa value of each median pixel point of each connected region; and determine, based on the first ordinate value and the median ordinate difference, a second ordinate value of each median pixel point of each connected region;

[0183] The central axis pixel point set of each connected area is determined according to the second horizontal coordinate value and the second vertical coordinate value.

[0184] In some embodiments, the fitting unit 804 is further configured to:

[0185] According to a preset second polynomial, each point in the target central axis pixel point set is fitted to obtain curve data of each lane line to be detected.

[0186] In some embodiments, before the step of performing prediction processing according to the state image, the acquiring unit 801 is further configured to:

[0187] Performing inverse perspective transformation on the state image to obtain a transformed state image;

[0188] In some embodiments, the prediction unit 802 is further configured to:

[0189] Prediction processing is performed according to the transformed state image.

[0190] In specific implementation, the above units can be implemented as independent entities, or can be arbitrarily combined to be implemented as the same or several entities. The specific implementation of the above units can be found in the previous method embodiments and will not be repeated here.

[0191] Since the lane line detection device can perform the following Figures 1 to 7 Corresponding to the steps in the lane line detection method in any embodiment, the present application can be implemented as follows Figures 1 to 7 The beneficial effects that can be achieved by the lane line detection method in any embodiment are detailed in the previous description and will not be repeated here.

[0192] In addition, in order to better implement the lane line detection method in the embodiment of the present application, based on the lane line detection method, the embodiment of the present application also provides an electronic device, see Figure 9 , Figure 9 The electronic device provided in the embodiment of the present application includes a processor 901, which is used to execute the computer program stored in the memory 902 to implement the following Figures 1 to 7 Corresponding to each step of the lane line detection method in any embodiment; or, when the processor 901 is used to execute the computer program stored in the memory 902, the following is implemented Figure 8 The functions of each unit in the corresponding embodiment.

[0193] For example, the computer program may be divided into one or more modules / units, one or more of which are stored in the memory 902 and executed by the processor 901 to implement the embodiments of the present application. One or more modules / units may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program in the computer device.

[0194] The electronic device may include, but is not limited to, a processor 901 and a memory 902. Those skilled in the art will appreciate that the illustrations are merely examples of electronic devices and do not limit the electronic device. The electronic device may include more or fewer components than shown, or a combination of certain components, or different components. For example, the electronic device may also include input and output devices, network access devices, and buses, and the processor 901, memory 902, input and output devices, and network access devices are connected via a bus.

[0195] The processor 901 may be a central processing unit (CPU), or other general-purpose processors, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor. The processor is the control center of the electronic device and connects various parts of the entire electronic device using various interfaces and lines.

[0196] The memory 902 can be used to store computer programs and / or modules. The processor 901 implements various functions of the computer device by running or executing the computer programs and / or modules stored in the memory 902 and accessing the data stored in the memory 902. The memory 902 may mainly include a program storage area and a data storage area. The program storage area may store an operating system and at least one application required for a function (such as a sound playback function, an image playback function, etc.); the data storage area may store data generated based on the use of the electronic device (such as audio data, video data, etc.). In addition, the memory may include high-speed random access memory and non-volatile memory, such as a hard disk, internal memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0197] Those skilled in the art will clearly understand that for the convenience and brevity of description, the specific working process of the lane line detection device, electronic device and its corresponding units described above can refer to the following. Figures 1 to 7The description of the lane line detection method in any embodiment will not be repeated here.

[0198] Those skilled in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be accomplished by instructions, or by controlling related hardware through instructions. The instructions may be stored in a computer-readable storage medium and loaded and executed by a processor.

[0199] To this end, an embodiment of the present application provides a computer-readable storage medium, which stores multiple instructions, which can be loaded by a processor to execute the present application as follows: Figures 1 to 7 For the steps in the lane line detection method in any embodiment, the specific operations can be referred to as follows: Figures 1 to 7 The description of the lane line detection method in any embodiment will not be repeated here.

[0200] The computer-readable storage medium may include a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0201] Due to the instructions stored in the computer readable storage medium, the present application can be executed as follows: Figures 1 to 7 Corresponding to the steps in the lane line detection method in any embodiment, the present application can be implemented as follows Figures 1 to 7 The beneficial effects that can be achieved by the lane line detection method in any embodiment are detailed in the previous description and will not be repeated here.

[0202] The above is a detailed introduction to a lane line detection method, device, electronic device and computer-readable storage medium provided in the embodiments of the present application. Specific examples are used herein to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for technical personnel in this field, based on the ideas of the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A lane line detection method, characterized in that: The method comprises: Obtain a state image containing lane lines to be detected; Calling the segmentation subnetwork in the trained segmentation network to perform segmentation prediction processing based on the state image to obtain regional pixel point information of each connected area of ​​the lane line to be detected; Calling the regression subnetwork in the trained segmentation network to perform regression prediction processing based on the state image to obtain distance information between each pixel point in each connected area and the center axis of the lane line to be detected; Determining a set of central axis pixels of each connected region according to the region pixel information and the central axis distance information; performing fitting processing on each point in the set of central axis pixels of each connected region according to a preset first polynomial to obtain a first fitting curve for each connected region; Acquiring curve characteristic information of the first fitting curve; Clustering the central axis pixel point set according to the curve feature information to obtain a target central axis pixel point set for each lane line to be detected; Fitting processing is performed on the target central axis pixel point set to obtain curve data of each lane line to be detected, wherein the curve data is used to indicate the lane line to be detected.

2. The lane line detection method according to claim 1, characterized in that: The method further comprises: Acquire a target image, wherein the target image has been annotated with actual area pixel point information and actual central axis distance information; Calling a segmentation subnetwork in a preset segmentation network to perform segmentation processing according to the target image, determine the connected area information of the lane lines in the target image, and obtain a connected area prediction result of the target image; Calling the regression subnetwork in the preset segmentation network to perform regression processing based on the target image, determine the distance information between each pixel in the target image and the center axis of the lane line in the target image, and obtain a prediction result of the center axis distance of the target image; The preset segmentation network is trained according to the connected region prediction result and the median distance prediction result to obtain a trained segmentation network.

3. The lane line detection method according to claim 1, characterized in that: The regional pixel point information includes a first horizontal coordinate value and a first vertical coordinate value of each pixel point in each connected area, and the central axis distance information includes a central axis horizontal coordinate difference and a central axis vertical coordinate difference between each pixel point in each connected area and the lane line to be detected. Determining the central axis pixel point set of each connected area based on the regional pixel point information and the central axis distance information includes: Determine, based on the first abscissa value and the median abscissa difference, a second abscissa value of each median pixel point of each connected region; and determine, based on the first ordinate value and the median ordinate difference, a second ordinate value of each median pixel point of each connected region; The central axis pixel point set of each connected area is determined according to the second horizontal coordinate value and the second vertical coordinate value.

4. The lane line detection method according to claim 1, characterized in that: The fitting process is performed based on the target central axis pixel point set to obtain the curve data of each lane line to be detected, including: According to a preset second polynomial, each point in the target central axis pixel point set is fitted to obtain curve data of each lane line to be detected.

5. The lane line detection method according to any one of claims 1 to 4, characterized in that: The method further comprises: Performing inverse perspective transformation on the state image to obtain a transformed state image; The performing prediction processing according to the state image includes: Prediction processing is performed according to the transformed state image.

6. A lane line detection device, characterized in that: The lane line detection device includes: An acquisition unit, configured to: acquire a state image including a lane line to be detected; The prediction unit is configured to: call the segmentation subnetwork in the trained segmentation network to perform segmentation prediction processing based on the state image to obtain regional pixel information of each connected region of the lane line to be detected; call the regression subnetwork in the trained segmentation network to perform regression prediction processing based on the state image to obtain distance information between each pixel in each connected region and the center axis of the lane line to be detected; The prediction unit is further configured to determine a set of central axis pixels of each connected region based on the region pixel information and the central axis distance information; The clustering unit is configured to: perform fitting processing on each point in the set of central axis pixels of each connected region according to a preset first polynomial to obtain a first fitting curve for each connected region; obtain curve feature information of the first fitting curve; and perform clustering processing on the set of central axis pixels according to the curve feature information to obtain a target set of central axis pixels for each lane line to be detected; A fitting unit is used to perform fitting processing based on the target central axis pixel point set to obtain curve data of each lane line to be detected, wherein the curve data is used to indicate the lane line to be detected.

7. An electronic device, characterized in that: The method comprises a processor and a memory, wherein a computer program is stored in the memory, and when the processor calls the computer program in the memory, the lane line detection method according to any one of claims 1 to 5 is executed.

8. A computer-readable storage medium, characterized in that A computer program is stored thereon, and the computer program is loaded by a processor to execute the steps in the lane line detection method according to any one of claims 1 to 5.

Images