Parking space identification method, device, computer equipment and storage medium
By adding path planning and data generation methods to the aerial view of the vehicle, obstructions are eliminated and missing information are filled, the problem of inaccurate positioning of parking spaces in automatic parking in vehicles is solved, and higher accuracy and cost reduction are achieved.
Patent Information
- Application Number
- CN202411002894.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-25
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2044-07-25
AI Technical Summary
During the automatic parking process of the vehicle, the parking space is not positioned accurately enough. It may be that the vehicle blocks the parking space lines or the camera is blocked by obstacles, resulting in the parking space detection being inaccurate enough, affecting the accuracy of parking space.
By adding path planning to a single vehicle aerial view and introducing data generation methods into the vehicle surround view camera system, a target aerial view is generated, parking space perception detection positioning is optimized, obstacle image range is eliminated, missing ground information is filled, and target parking is determined. Path to control vehicle parking.
It improves the accuracy of parking space positioning, ensures that vehicles can accurately park in the parking space, avoids inaccurate detection problems caused by obstacles, and reduces the cost of manual collection and labeling.
Smart Images

Figure CN118521996B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of autonomous driving technology, and in particular to a parking space recognition method, device, computer equipment and storage medium. Background Art
[0002] For the automatic parking function, in addition to identifying parking spaces during the patrol phase, the car also needs to continuously accurately locate the parking space during the automatic parking process. If the perception is stopped later, the initial parking space location information and wheel stopper information may not be accurate enough. In the surround vision perception solution, we will map the parking space to a bird's-eye view, locate the parking space by detecting the parking space on the bird's-eye view, and may also add other modal perception content such as radar to determine whether it is a parking space.
[0003] At present, during the automatic parking process, the vehicle may block the parking space lines on the ground, which will affect the positioning of the parking space. The camera may also be blocked by other obstacles, which will make the parking space detection inaccurate and affect the positioning of the parking space. Summary of the invention
[0004] In view of this, the present invention provides a parking space recognition method, device, computer equipment and storage medium to solve the problem of inaccurate positioning of parking spaces when vehicles are automatically parked.
[0005] In a first aspect, the present invention provides a method for identifying a parking space, comprising: obtaining an initial bird's-eye view corresponding to a target vehicle; the initial bird's-eye view includes an initial position of the target vehicle, a target parking space position corresponding to the target vehicle, and an initial parking path, wherein the initial parking path is a parking path determined based on the initial position and the target parking space position; based on each parking path point of the initial parking path, generating a target bird's-eye view corresponding to each parking path point; based on each target bird's-eye view, determining a target parking path between a current position of the target vehicle and the target parking space position; and controlling the target vehicle to enter the target parking space according to the target parking path.
[0006] The parking space recognition method provided by the embodiment of the present invention, after acquiring the initial bird's-eye view corresponding to the target vehicle, generates the target bird's-eye view corresponding to each parking path point according to each parking path point of the initial parking path, and determines the target parking path between the current position of the target vehicle and the position of the target parking space according to each target bird's-eye view, and controls the target vehicle to enter the target parking space according to the target parking path. Therefore, it is realized to add path planning to a single vehicle bird's-eye view map, and introduce a data generation method in the vehicle surround camera system, so as to optimize the accuracy of parking space perception detection and positioning during the vehicle parking process.
[0007] In an optional implementation, based on each parking path point of the initial parking path, a target bird's-eye view corresponding to each parking path point is generated, including: acquiring a vehicle coordinate system corresponding to the initial position and a pixel coordinate system corresponding to the initial bird's-eye view; converting the vehicle coordinate system to the pixel coordinate system to obtain a first position of the initial position in the pixel coordinate system; based on the first position, determining a second position of each parking path point in the pixel coordinate system; and based on the first position and each second position, generating a target bird's-eye view corresponding to each parking path point.
[0008] The parking space recognition method provided by the embodiment of the present invention converts the vehicle coordinate system to the pixel coordinate system to obtain the first position of the initial position in the pixel coordinate system after obtaining the vehicle coordinate system corresponding to the initial position and the pixel coordinate system corresponding to the initial bird's-eye view, and determines the second position of each parking path point in the pixel coordinate system according to the first position. A target bird's-eye view corresponding to each parking path point is generated according to the first position and each second position, so as to perform data enhancement on each parking path point to improve the accuracy of parking space positioning.
[0009] In an optional embodiment, a target bird's-eye view corresponding to each parking path point is generated based on the first position and each second position, including: obtaining a first camera extrinsic parameter corresponding to the first position and a previous position point corresponding to each second position; determining a target camera extrinsic parameter corresponding to each parking path point based on the first position, the second position, the previous position point and the first camera extrinsic parameter; and generating a target bird's-eye view corresponding to each parking path point based on each target camera extrinsic parameter.
[0010] The parking space recognition method provided by the embodiment of the present invention, after obtaining the first camera extrinsic parameter corresponding to the first position and the previous position point corresponding to each second position, determines the target camera extrinsic parameter corresponding to each parking path point according to the obtained first position, second position, previous position point and first camera extrinsic parameter information, and generates a target bird's-eye view corresponding to each parking path point according to each target camera extrinsic parameter, thereby performing data enhancement on each parking path point and generating a target bird's-eye view corresponding to each parking path point to optimize the positioning of the parking space.
[0011] In an optional implementation, it is detected whether there are obstacles in the corresponding surround camera image for each parking path point; when it is determined that there are obstacles in the surround camera image, the image range corresponding to the obstacle is eliminated to generate a target image range; the missing ground information in the target image range is filled to obtain a target surround camera image corresponding to each parking path point; based on the target surround camera image, a target bird's-eye view corresponding to each parking path point is generated.
[0012] The parking space recognition method provided by the embodiment of the present invention detects whether there are obstacles in the corresponding surround camera image of each parking path point. When it is determined that there are obstacles in the surround camera image, the image range corresponding to the obstacle is eliminated to obtain the target image range. The missing ground information in the target image range is filled to obtain the target surround camera image corresponding to each parking path point, and then the target bird's-eye view corresponding to each parking path point is generated according to the target surround camera image. Therefore, the problem of inaccurate parking space detection and parking space positioning is avoided when the camera carried by the vehicle is blocked by other obstacles, thereby improving the accuracy of parking space detection and ensuring the accuracy of parking space positioning.
[0013] In an optional implementation, eliminating an image range corresponding to an obstacle to generate a target image range includes: identifying the type of obstacle; determining whether the obstacle needs to be eliminated based on the type; when it is determined that the obstacle needs to be eliminated, identifying a elimination position of the obstacle; and eliminating the obstacle at the elimination position to obtain a target image range.
[0014] The parking space recognition method provided by the embodiment of the present invention identifies the type of obstacle, determines whether the obstacle needs to be removed according to the type information, and when it is determined that the obstacle needs to be removed, identifies the removal position of the obstacle, and removes the obstacle at the removal position to obtain a target image range, thereby avoiding the problem of inaccurate parking space detection caused by the camera carried by the vehicle being blocked by other obstacles, and further improving the accuracy of parking space detection.
[0015] In an optional implementation, the missing ground information in the target image range is filled to obtain a target surround camera image corresponding to each parking path point, including: identifying a rejection image range corresponding to a rejection position, and determining environmental information corresponding to the rejection image range; predicting the missing ground information of the rejection image range based on the environmental information; and filling the predicted ground information into the missing ground corresponding to the rejection image range to obtain a target surround camera image.
[0016] The parking space recognition method provided by the embodiment of the present invention, after identifying the rejection image range corresponding to the rejection position and determining the environmental information corresponding to the rejection image range, predicts the missing ground information of the rejection image range according to the determined environmental information, and fills the predicted ground information into the missing ground corresponding to the rejection image range to obtain a target surround camera image, so that the target surround camera image can display complete parking space information, avoids the parking space being blocked by other obstacles, and maximizes the positioning accuracy of the parking space.
[0017] In an optional embodiment, a target parking path between a current position of a target vehicle and a target parking space is determined based on each target bird's-eye view, including: determining a target parking path point corresponding to the current position based on each target bird's-eye view; determining a parking direction based on the target parking path point and the target parking space; and generating a target parking path based on the parking direction.
[0018] The parking space recognition method provided by the embodiment of the present invention determines the target parking path point corresponding to the current position according to the target bird's-eye views generated above, determines the parking direction according to the target parking path point and the target parking space position, and generates the target parking path, thereby optimizing the positioning information of the parking space, so as to control the vehicle to be parked smoothly in the parking space according to the target parking path.
[0019] In an optional implementation, an offline database is constructed based on the target bird's-eye view, and the offline database is deployed in the target vehicle; when the target vehicle is parking, the parking path of the target vehicle is optimized based on the offline database.
[0020] The parking space recognition method provided by the embodiment of the present invention constructs an offline database through a target bird's-eye view, and deploys the offline database in the target vehicle. When the target vehicle parks, the parking path of the target vehicle is optimized according to the offline database, thereby eliminating the need for manual collection and labeling of parking space and vehicle environment information, thereby reducing costs.
[0021] In a second aspect, the present invention provides a parking space identification device, comprising: an acquisition module, used to acquire an initial bird's-eye view corresponding to a target vehicle; the initial bird's-eye view includes an initial position of the target vehicle, a target parking space position corresponding to the target vehicle, and an initial parking path, wherein the initial parking path is a parking path determined based on the initial position and the target parking space position; a generation module, used to generate a target bird's-eye view corresponding to each parking path point based on each parking path point of the initial parking path; a determination module, used to determine a target parking path between a current position of the target vehicle and the target parking space position based on each target bird's-eye view; and a control module, used to control the target vehicle to enter the target parking space according to the target parking path.
[0022] In a third aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the parking space identification method of the first aspect or any corresponding embodiment thereof by executing the computer instructions.
[0023] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the parking space identification method of the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0025] Figure 1 is a schematic flow chart of a method for identifying a parking space according to an embodiment of the present invention;
[0026] Figure 2 is a flow chart of another parking space identification method according to an embodiment of the present invention;
[0027] Figure 3 is a schematic diagram of a pixel coordinate system according to an embodiment of the present invention;
[0028] Figure 4 is a flow chart of another parking space identification method according to an embodiment of the present invention;
[0029] Figure 5 is a flow chart of another parking space identification method according to an embodiment of the present invention;
[0030] Figure 6 is a schematic diagram of parking directions according to an embodiment of the present invention;
[0031] Figure 7 is a structural block diagram of a vehicle identification device according to an embodiment of the present invention;
[0032] Figure 8 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION
[0033] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.
[0034] For the automatic parking function, in addition to identifying parking spaces during the patrol phase, the car also needs to continuously accurately locate the parking space during the automatic parking process. If the perception is stopped later, the initial parking space location information and wheel stopper information may not be accurate enough. In the surround vision perception solution, we will map the parking space to a bird's-eye view, locate the parking space by detecting the parking space on the bird's-eye view, and may also add other modal perception content such as radar to determine whether it is a parking space.
[0035] At present, during the automatic parking process, the vehicle may block the parking space lines on the ground. Since the parking space is blocked in the image, the positioning of the parking space is affected. The camera may also be blocked by other obstacles, which may result in inaccurate detection of the parking space and affect the positioning of the parking space. If the parking space is not in the right position relative to the vehicle, the vehicle may not be parked correctly, causing scratches with the vehicles next to it or compressing the parking space of other vehicles.
[0036] In view of this, the technical solution of the present invention proposes an offline data enhancement solution, which optimizes the accuracy of parking space perception, detection and positioning during vehicle parking by adding path planning to a single vehicle bird's-eye view map and introducing a data generation method in the vehicle surround camera system.
[0037] According to an embodiment of the present invention, an embodiment of a method for identifying a parking space is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.
[0038] In this embodiment, a parking space recognition method is provided, which can be used for computer devices, such as tablet computers, desktop computers, etc. Figure 1 is a flow chart of a method for identifying a parking space according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:
[0039] Step S101, obtaining an initial bird's-eye view corresponding to the target vehicle.
[0040] The initial bird's-eye view is an image obtained by the computer device through data processing based on the initial surround camera image sent by the target vehicle when the target vehicle is at the initial position. The initial surround camera image is an image taken by multiple cameras carried by the target vehicle and located in different directions. The initial surround camera image includes the target parking space information and surrounding environment information corresponding to the target vehicle. The computer device obtains the initial position information obtained by the positioning device carried by the target vehicle and extracts the target parking space information and surrounding environment information in the initial surround camera image, and obtains the initial parking path using the path planning algorithm.
[0041] Specifically, the computer device can receive the initial positioning information and initial surround camera image sent by the target vehicle in real time, and generate the corresponding initial bird's-eye view after data processing the initial positioning information and the initial surround camera image. The computer device can also select a bird's-eye view of the patrol phase with pre-marked parking spaces from the previously stored bird's-eye view as the initial bird's-eye view, randomly select a target parking space in the initial bird's-eye view, and use the parking path planning algorithm to obtain several parking paths between the target vehicle position and the target parking space position, and randomly select one of them as the initial parking path, that is, the initial bird's-eye view at this time carries the initial position of the target vehicle, the target parking space position corresponding to the target vehicle, and the initial parking path.
[0042] Step S102: generating a target bird's-eye view corresponding to each parking path point based on each parking path point of the initial parking path.
[0043] The parking path points are the actual travel points of the target vehicle during the parking process. Specifically, the computer device controls the target vehicle to travel according to the initial parking path. During this process, the computer device can receive the surround camera images corresponding to the parking path points sent by the target vehicle in real time, and after processing the positioning information of each parking path point and each surround camera image, generate the target bird's-eye view corresponding to each parking path point. The computer device can also select a bird's-eye view of the patrol phase with pre-marked parking spaces from the previously stored bird's-eye view as the initial bird's-eye view. At this time, the initial bird's-eye view carries the positioning information and surround camera images corresponding to each parking path point. After processing the positioning information of each parking path point and each surround camera image, generate the target bird's-eye view corresponding to each parking path point.
[0044] Step S103, based on each target bird's-eye view, determine a target parking path between the current position of the target vehicle and the target parking space position.
[0045] The computer device receives the current location information sent by the target vehicle. It is understandable that the initial location information and the current location information of the target vehicle can be determined based on the positioning device carried by the target vehicle, such as GPS. The target parking path is the actual driving path between the current location of the target vehicle and the target parking space. Specifically, after receiving the current location information sent by the target vehicle, the computer device can query the target parking space information corresponding to the current location of the target vehicle based on the target vehicle information and target parking space information carried by each target bird's-eye view, and then determine the target parking path after performing path planning based on the current location information and the target parking space information.
[0046] Step S104: Control the target vehicle to drive into the target parking space according to the target parking path.
[0047] The computer device sends the determined target parking path to the controller of the target vehicle, and the controller parses the target parking path to obtain the position information and direction information included in the target parking path, so that the target vehicle can automatically drive into the target parking space according to the position information and direction information.
[0048] The parking space recognition method provided by the embodiment of the present invention, after acquiring the initial bird's-eye view corresponding to the target vehicle, generates the target bird's-eye view corresponding to each parking path point according to each parking path point of the initial parking path, and determines the target parking path between the current position of the target vehicle and the position of the target parking space according to each target bird's-eye view, and controls the target vehicle to enter the target parking space according to the target parking path. Therefore, it is realized to add path planning to a single vehicle bird's-eye view map, and introduce a data generation method in the vehicle surround camera system, so as to optimize the accuracy of parking space perception detection and positioning during the vehicle parking process.
[0049] In this embodiment, a parking space recognition method is provided, which can be used in the above-mentioned computer devices, such as tablet computers, desktop computers, etc. Figure 2 is a flow chart of a method for identifying a parking space according to an embodiment of the present invention. Figure 2 As shown, the process includes the following steps:
[0050] Step S201, obtain the initial bird's-eye view corresponding to the target vehicle. Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.
[0051] Step S202: generating a target bird's-eye view corresponding to each parking path point based on each parking path point of the initial parking path.
[0052] Specifically, the above step S202 includes:
[0053] Step S2021, obtaining the vehicle coordinate system corresponding to the initial position and the pixel coordinate system corresponding to the initial bird's-eye view.
[0054] The vehicle coordinate system is a coordinate system used to represent the actual position of the vehicle, which may be a world coordinate system or a camera coordinate system, which is not limited here.
[0055] Specifically, the computer device receives the initial position information sent by the target vehicle, such as (x1, y1, z1), and establishes a corresponding vehicle coordinate system according to the initial position information.
[0056] Specifically, the computer device calculates the pixel center point corresponding to the initial bird's-eye view according to the image parameters of the initial bird's-eye view, such as length information and width information, and establishes a pixel coordinate system with the pixel center point as the origin. Figure 3 As shown, the image length and image width corresponding to the initial bird's-eye view are both 100 pixels (px), that is, the size corresponding to the initial bird's-eye view is 100px×100px. The computer device divides 100px by 2 to determine the pixel center point position corresponding to the initial bird's-eye view ( , ),by( , ) is used as the origin to establish a pixel coordinate system.
[0057] Step S2022, converting the vehicle coordinate system to the pixel coordinate system to obtain a first position of the initial position in the pixel coordinate system.
[0058] The first position is the pixel coordinate position corresponding to the initial position in the pixel coordinate system. Specifically, the computer device determines the relevant parameters of the camera, such as the camera intrinsic parameters and the camera extrinsic parameters, according to the type of camera carried by the pre-queried target vehicle, and calculates the first position of the initial position in the pixel coordinate system in combination with the vehicle coordinate information of the initial position. For example, after obtaining the camera intrinsic parameters such as the focal length, imaging origin, and distortion coefficient, and the camera extrinsic parameters such as the rotation vector and the translation vector, the computer device generates a camera intrinsic parameter matrix according to the focal length, imaging origin, and distortion coefficient, and generates a camera extrinsic parameter matrix according to the rotation vector and the translation vector, and multiplies the initial position (x1, y1, z1), the camera intrinsic parameter matrix, and the camera extrinsic parameter matrix to obtain the first position ( , ).
[0059] Step S2023: Determine a second position of each parking path point in the pixel coordinate system based on the first position.
[0060] The second position is the pixel coordinate position of the parking path point in the pixel coordinate system. Specifically, the computer device determines the first position ( , ), the initial bird's-eye view image is parsed to find the pixel coordinates of the target parking space ( , ) and the surrounding environment information between the initial position and the target parking space position, and using a path planning algorithm, such as the Dubins Algorithm, to obtain a target pixel path corresponding to the initial parking path on the initial bird's-eye view. For each pixel parking path point of the target pixel path, the computer device determines a second position of each parking path point in the pixel coordinate system according to the position of each pixel parking path point on the initial bird's-eye view and the pixel coordinate system.
[0061] Step S2024: Generate a target bird's-eye view corresponding to each parking path point based on the first position and each second position.
[0062] Specifically, the computer device calculates the target camera extrinsic parameters corresponding to each parking path point based on the relevant parameters corresponding to the first position and the relevant parameters of each second position, generates a surround camera image corresponding to each parking path point based on the acquired camera intrinsic parameters and camera extrinsic parameters, and performs data processing on the surround camera image to obtain a target bird's-eye view corresponding to each parking path point.
[0063] In some optional implementations, the above step S2024 includes:
[0064] Step a1, obtaining the first camera external parameter corresponding to the first position and the previous position point corresponding to each second position.
[0065] The first camera extrinsic parameter is the initial camera extrinsic parameter corresponding to the target vehicle at the initial position. Specifically, the computer device can receive the initial camera extrinsic parameter sent by the target vehicle and use the initial camera extrinsic parameter as the first camera extrinsic parameter corresponding to the first position. The camera type corresponding to the target vehicle can also be queried in advance, and the corresponding first camera extrinsic parameter is determined according to the camera type, which is not limited here.
[0066] Specifically, the computer device directly determines the position information of the previous path point corresponding to each pixel parking path point according to the relevant position information of the target pixel path determined above, that is, determines the previous position point corresponding to each second position.
[0067] Step a2: determining the target camera extrinsic parameters corresponding to each parking path point based on the first position, the second position, the previous position point and the first camera extrinsic parameters.
[0068] The target camera external parameters are the actual camera external parameters corresponding to each parking path point. Specifically, the computer device calculates the target camera external parameters according to the first position ( ) and each second position ( ) to obtain the translation matrix Tt corresponding to each parking path point. The computer device calculates the position of the second position and the previous position point corresponding to each second position ( ) to obtain the translation matrix Tr corresponding to each parking path point. ,in , multiply the translation matrix and the translation matrix to obtain the external parameter transformation matrix corresponding to each parking path point.
[0069] The computer device multiplies the first camera extrinsic parameter by each extrinsic parameter conversion matrix to obtain the target camera extrinsic parameter corresponding to each parking path point.
[0070] Step a3: Generate a target bird's-eye view corresponding to each parking path point based on the external parameters of each target camera.
[0071] The computer device generates a surround camera image corresponding to each parking path point according to the acquired camera intrinsic parameters and target camera extrinsic parameters, and performs data processing on the surround camera image to obtain a target bird's-eye view image corresponding to each parking path point. The data processing method may be inverse perspective mapping, but this is not limited here.
[0072] In the above embodiment, after obtaining the first camera extrinsic parameter corresponding to the first position and the previous position point corresponding to each second position, the target camera extrinsic parameter corresponding to each parking path point is determined according to the obtained first position, second position, previous position point and first camera extrinsic parameter information, and a target bird's-eye view corresponding to each parking path point is generated according to each target camera extrinsic parameter, so as to perform data enhancement on each parking path point and generate a target bird's-eye view corresponding to each parking path point to optimize the positioning of the parking space.
[0073] Step S203: Based on each target bird's-eye view, determine the target parking path between the current position of the target vehicle and the target parking space. Figure 1 Step S103 of the illustrated embodiment will not be described in detail here.
[0074] Step S204: Control the target vehicle to enter the target parking space according to the target parking path. Figure 1 Step S104 of the illustrated embodiment will not be described in detail here.
[0075] The parking space recognition method provided by the embodiment of the present invention converts the vehicle coordinate system to the pixel coordinate system to obtain the first position of the initial position in the pixel coordinate system after obtaining the vehicle coordinate system corresponding to the initial position and the pixel coordinate system corresponding to the initial bird's-eye view, and determines the second position of each parking path point in the pixel coordinate system according to the first position. A target bird's-eye view corresponding to each parking path point is generated according to the first position and each second position, so as to perform data enhancement on each parking path point to improve the accuracy of parking space positioning.
[0076] In this embodiment, a parking space recognition method is provided, which can be used in the above-mentioned computer devices, such as tablet computers, desktop computers, etc. Figure 4 is a flow chart of a method for identifying a parking space according to an embodiment of the present invention. Figure 4 As shown, the process includes the following steps:
[0077] Step S301, obtain the initial bird's-eye view corresponding to the target vehicle. Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.
[0078] Step S302: Generate a target bird's-eye view corresponding to each parking path point based on each parking path point of the initial parking path. Figure 1 Step S102 of the illustrated embodiment will not be described in detail here.
[0079] Step S303: Based on the bird's-eye view of each target, determine the target parking path between the current position of the target vehicle and the target parking space. Figure 1 Step S103 of the illustrated embodiment will not be described in detail here.
[0080] Step S304: Control the target vehicle to drive into the target parking space according to the target parking path. Figure 1 Step S104 of the illustrated embodiment will not be described in detail here.
[0081] Step S305 , detecting whether there are obstacles in the corresponding surround camera image at each parking path point.
[0082] Obstacles are objects that block the camera's field of view, such as the vehicle itself, pillars, other vehicles, road cones, pedestrians, vehicles, and other non-ground objects. Specifically, the computer device can detect whether there are obstacles in the corresponding surround camera image at each parking path point based on the carried target detection algorithm.
[0083] Step S306: when it is determined that there is an obstacle in the surround camera image, the image range corresponding to the obstacle is eliminated to generate a target image range.
[0084] The target image range is the image range remaining after the image range corresponding to the obstacle is removed from the surround camera image. Specifically, when the computer device determines that there is an obstacle in the surround camera image according to the target detection algorithm, it identifies each obstacle type according to the type recognition algorithm, determines the image range corresponding to the obstacle according to the obstacle type, and removes the image range to obtain the target image range.
[0085] Specifically, the above step S306 includes:
[0086] Step S3061, identifying the type of obstacle.
[0087] The types of obstacles may include the vehicle itself, pillars, other vehicles, road cones, pedestrians and vehicles, etc. Specifically, the computer device may identify the types of each obstacle and the range outline of each obstacle according to a semantic segmentation network algorithm.
[0088] Step S3062: Determine whether the obstacle needs to be removed based on the type.
[0089] The computer device determines the type of obstacle to be removed according to the user's pre-selected operation. When the type of obstacle to be removed is identified, the current obstacle is removed; when the type of obstacle not to be removed is identified, the current obstacle is skipped.
[0090] Step S3063: When it is determined that an obstacle needs to be removed, the removal position of the obstacle is identified.
[0091] When the computer device determines that an obstacle needs to be removed, the position of the obstacle in the pixel coordinate system is determined according to the pixel coordinate system, and the position of the obstacle in the pixel coordinate system is determined as the removal position of the obstacle.
[0092] Step S3064, removing obstacles at the removal position to obtain the target image range.
[0093] The computer device determines the range outline of the obstacle according to the determined removal position, removes the obstacles in the surround camera image according to the range outline of the obstacle, and generates an image range after the obstacle is removed, and the image range is the target image range.
[0094] In the above implementation, by identifying the type of obstacle and determining whether the obstacle needs to be removed based on the type information, when it is determined that the obstacle needs to be removed, the removal position of the obstacle is identified, and the obstacle at the removal position is removed to obtain the target image range, thereby avoiding the problem of inaccurate parking space detection due to the camera carried by the vehicle being blocked by other obstacles, and further improving the accuracy of parking space detection.
[0095] Step S307, filling in the missing ground information in the target image range to obtain the target surround camera image corresponding to each parking path point.
[0096] The computer device fills in the ground information in the missing range in the target image range according to the carried ground generation model, updates the surround camera image corresponding to each parking path point, and obtains the target surround camera image corresponding to each parking path point.
[0097] The ground generation model is a training model for ground generation. A large amount of ground sample information is collected and input into the model for training. When the ground information around the missing range is identified, it is matched with the ground sample information in the model, and the matching result is output as the ground information corresponding to the missing range.
[0098] Specifically, the above step S307 includes:
[0099] Step S3071, identifying the removal image range corresponding to the removal position, and determining the environmental information corresponding to the removal image range.
[0100] The excluded image range is an image range that is cropped according to the obstacle outline. Specifically, the computer device can find out the missing image range in the target image range, that is, the excluded image range, and determine the ground information around the excluded image range according to the surround camera image, and determine the ground information around the excluded image range as the environmental information corresponding to the excluded image range.
[0101] Step S3072: Eliminate missing ground information in the image range based on environmental information prediction.
[0102] The missing ground information is used to characterize the missing ground information in the removed image range. Specifically, the computer device predicts the ground information corresponding to the removed image range based on the ground information around the removed image range to obtain the missing ground information of the removed image range.
[0103] Step S3073, fill the predicted ground information into the missing ground corresponding to the removed image range to obtain the target surround camera image.
[0104] The target surround camera image is a surround camera image updated according to the presence of obstacles. Specifically, the computer device can fill the predicted ground information into the missing ground corresponding to the removed image range according to a filling model, such as an inpaint-anything model, and determine the filled surround camera image as the target surround camera image.
[0105] In the above implementation, after identifying the rejection image range corresponding to the rejection position and determining the environmental information corresponding to the rejection image range, the missing ground information of the rejection image range is predicted according to the determined environmental information, and the predicted ground information is filled into the missing ground corresponding to the rejection image range to obtain the target surround camera image, so that the target surround camera image can display the complete parking space information, avoid the parking space being blocked by other obstacles, and maximize the positioning accuracy of the parking space.
[0106] Step S308: Generate a target bird's-eye view corresponding to each parking path point based on the target surround camera image.
[0107] The computer device determines the relevant parameters of the inverse perspective mapping pre-defined by the user, including the viewing angle height, viewing angle direction, focal length, the distance between the top of the top view and the original image, etc., and calculates the perspective transformation matrix based on the above parameters to perform inverse perspective mapping on the target surround camera image to obtain the target bird's-eye view.
[0108] The parking space recognition method provided by the embodiment of the present invention detects whether there are obstacles in the corresponding surround camera image of each parking path point. When it is determined that there are obstacles in the surround camera image, the image range corresponding to the obstacle is eliminated to obtain the target image range. The missing ground information in the target image range is filled to obtain the target surround camera image corresponding to each parking path point, and then the target bird's-eye view corresponding to each parking path point is generated according to the target surround camera image. Therefore, the problem of inaccurate parking space detection and parking space positioning is avoided when the camera carried by the vehicle is blocked by other obstacles, thereby improving the accuracy of parking space detection and ensuring the accuracy of parking space positioning.
[0109] In this embodiment, a parking space recognition method is provided, which can be used in the above-mentioned computer devices, such as tablet computers, desktop computers, etc. Figure 5 is a flow chart of a method for identifying a parking space according to an embodiment of the present invention. Figure 5 As shown, the process includes the following steps:
[0110] Step S401, obtain the initial bird's-eye view corresponding to the target vehicle. Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.
[0111] Step S402: Generate a target bird's-eye view corresponding to each parking path point based on each parking path point of the initial parking path. Figure 1 Step S102 of the illustrated embodiment will not be described in detail here.
[0112] Step S403, based on each target bird's-eye view, determine a target parking path between the current position of the target vehicle and the target parking space position.
[0113] Specifically, the above step S403 includes:
[0114] Step S4031, based on each target bird's-eye view, determine the target parking path point corresponding to the current position.
[0115] The target parking path point is a path point compared with the current position of the target vehicle. Specifically, the computer device determines the initial position, the target parking space position and the position of each parking path point corresponding to the parking path carried by each target bird's-eye view from the tag information corresponding to each target bird's-eye view, compares the current position sent by the target vehicle with each parking path point, and obtains the target parking path point and the target parking space position corresponding to the current position.
[0116] Step S4032, determining the parking direction based on the target parking path point and the target parking space position.
[0117] The computer device obtains the direction angle between the target parking path point and the target parking space position according to the above determined target parking path point position and target parking space position, such as Figure 6 As shown, the direction angle Determine the parking direction.
[0118] Step S4033: generating a target parking path based on the parking direction.
[0119] The computer device generates a target parking path between the target parking path point and the target parking space position according to the parking direction and in combination with the route planning algorithm. The route planning algorithm includes RRT algorithm, PRM algorithm, etc. Taking the RRT algorithm as an example, the target parking path point is set as the starting point, the target parking space position is set as the end point to establish a tree, and the target parking path point is used as the root node. The environmental information between the target parking path point and the target parking space position is read, and a random point is randomly selected between the target parking path point and the target parking space position, and the current entire tree is traversed to find the point closest to the random point, and a step length is extended from the nearest point to the random point to obtain a new point A, and whether point A is on the obstacle is detected. If it is on the obstacle, the current cycle is jumped out and a new random point is selected again. If it is not on the obstacle, point A is inserted into the entire tree. When the distance between point A and the target parking space position is less than a certain threshold range, the search is exited, the entire parking path is generated, and the parking path is determined as the target parking path.
[0120] The parking space recognition method provided by the embodiment of the present invention determines the target parking path point corresponding to the current position according to the target bird's-eye views generated above, determines the parking direction according to the target parking path point and the target parking space position, and generates the target parking path, thereby optimizing the positioning information of the parking space, so as to control the vehicle to be parked smoothly in the parking space according to the target parking path.
[0121] Step S404: Control the target vehicle to enter the target parking space according to the target parking path. Figure 1 Step S104 of the illustrated embodiment will not be described in detail here.
[0122] Step S405: construct an offline database based on the target bird's-eye view.
[0123] The off-line database is a database including each target bird's-eye view, which is deployed in the target vehicle. Specifically, the computer device can transmit the database to the parking system of the target vehicle so that the target vehicle can park according to the relevant parameters of the target bird's-eye view.
[0124] Step S406: When the target vehicle is parking, the parking path of the target vehicle is optimized based on the offline database.
[0125] When the target vehicle is parking, the target vehicle can retrieve the bird's-eye view of each target in the offline database, extract the initial position, target parking space position and position information of each parking path point corresponding to the parking path included in each target bird's-eye view, determine the current parking path and current parking space position corresponding to the target vehicle, so as to optimize the parking path.
[0126] The parking space recognition method provided by the embodiment of the present invention constructs an offline database through a target bird's-eye view, and deploys the offline database in the target vehicle. When the target vehicle parks, the parking path of the target vehicle is optimized according to the offline database, thereby eliminating the need for manual collection and labeling of parking space and vehicle environment information, thereby reducing costs.
[0127] In this embodiment, a parking space identification device is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.
[0128] This embodiment provides a parking space recognition device, such as Figure 7 As shown, including:
[0129] The acquisition module 501 is used to acquire an initial bird's-eye view corresponding to the target vehicle. The initial bird's-eye view includes an initial position of the target vehicle, a target parking space position corresponding to the target vehicle, and an initial parking path, wherein the initial parking path is a parking path determined based on the initial position and the target parking space position.
[0130] The first generating module 502 is configured to generate a target bird's-eye view corresponding to each parking path point based on each parking path point of the initial parking path.
[0131] The determination module 503 is used to determine a target parking path between the current position of the target vehicle and the target parking space position based on each target bird's-eye view.
[0132] The control module 504 is used to control the target vehicle to enter the target parking space according to the target parking path.
[0133] In some optional implementations, the first generating module 502 includes:
[0134] The first acquisition submodule is used to acquire the vehicle coordinate system corresponding to the initial position and the pixel coordinate system corresponding to the initial bird's-eye view.
[0135] The conversion submodule is used to convert the vehicle coordinate system into the pixel coordinate system to obtain a first position of the initial position in the pixel coordinate system.
[0136] A second acquisition submodule, configured to determine a second position of each parking path point in a pixel coordinate system based on the first position;
[0137] The first generating submodule is used to generate a target bird's-eye view corresponding to each parking path point based on the second position and each first position.
[0138] In some optional implementations, the first generation submodule includes:
[0139] The acquisition unit is used to acquire the first camera external parameter corresponding to the first position and the previous position point corresponding to each second position.
[0140] The determination unit is used to determine the target camera extrinsic parameters corresponding to each parking path point based on the first position, the second position, the previous position point and the first camera extrinsic parameters.
[0141] The generating unit is used to generate a target bird's-eye view corresponding to each parking path point based on each target camera external parameter.
[0142] In some optional embodiments, the above device further includes:
[0143] The detection module is used to detect whether there are obstacles at each parking path point in the corresponding surround camera image.
[0144] The elimination module is used to eliminate the image range corresponding to the obstacle and generate the target image range when it is determined that there is an obstacle in the surround camera image.
[0145] The filling module is used to fill in the missing ground information in the target image range to obtain the target surround camera image corresponding to each parking path point.
[0146] The second generating module is used to generate a target bird's-eye view corresponding to each parking path point based on the target surround camera image.
[0147] In some optional implementations, the above-mentioned rejection module includes:
[0148] The first identification submodule is used to identify the type of obstacle.
[0149] The first determination submodule is used to determine whether an obstacle needs to be removed based on its type.
[0150] The second identification submodule is used to identify the removal position of the obstacle when it is determined that the obstacle needs to be removed.
[0151] The elimination submodule is used to eliminate obstacles at the elimination position to obtain the target image range.
[0152] In some optional embodiments, the filling module includes:
[0153] The third identification submodule is used to identify the rejection image range corresponding to the rejection position and determine the environmental information corresponding to the rejection image range.
[0154] The prediction submodule is used to predict the missing ground information of the culled image range based on the environmental information.
[0155] The filling submodule is used to fill the predicted ground information into the missing ground corresponding to the removed image range to obtain the target surround camera image.
[0156] In some optional implementations, the determination module 503 includes:
[0157] The second determination submodule is used to determine the target parking path point corresponding to the current position based on each target bird's-eye view.
[0158] The third determination submodule is used to determine the parking direction based on the target parking path point and the target parking space position.
[0159] The second generating submodule is used to generate a target parking path based on the parking direction.
[0160] In some optional embodiments, the above device further includes:
[0161] A building module is used to build an offline database based on the target bird's-eye view, and the offline database is deployed in the target vehicle.
[0162] The optimization module is used to optimize the parking path of the target vehicle based on the offline database when the target vehicle is parking.
[0163] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.
[0164] The vehicle identification device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.
[0165] The vehicle identification device provided by the embodiment of the present invention, after acquiring the initial bird's-eye view corresponding to the target vehicle, generates a target bird's-eye view corresponding to each parking path point according to each parking path point of the initial parking path, and determines the target parking path between the current position of the target vehicle and the target parking space according to each target bird's-eye view, and controls the target vehicle to enter the target parking space according to the target parking path. Therefore, it is achieved to add path planning to a single vehicle bird's-eye view map, and introduce a data generation method in the vehicle surround camera system, so as to optimize the accuracy of parking space perception detection and positioning during the vehicle parking process.
[0166] The embodiment of the present invention also provides a computer device having the above Figure 7 An identification device for the vehicle shown.
[0167] See also Figure 8 , Figure 8 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 8 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 8 A processor 10 is taken as an example.
[0168] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.
[0169] The memory 20 stores instructions executable by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.
[0170] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.
[0171] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.
[0172] The computer device also includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Figure 8 The example of connecting through bus is taken in the following.
[0173] The input device 30 can receive input digital or character information, and generate key signal input related to the user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator bar, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 may include a display device, an auxiliary lighting device (e.g., an LED) and a tactile feedback device (e.g., a vibration motor), etc. The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display and a plasma display. In some optional embodiments, the display device may be a touch screen.
[0174] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.
[0175] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.
Claims
1. A method for identifying a parking space, characterized in that: The method comprises: Acquire an initial bird's-eye view corresponding to the target vehicle; the initial bird's-eye view includes an initial position of the target vehicle, a target parking space position corresponding to the target vehicle, and an initial parking path, wherein the initial parking path is a parking path determined based on the initial position and the target parking space position; Based on each parking path point of the initial parking path, generating a target bird's-eye view corresponding to each parking path point; Based on each of the target bird's-eye views, determining a target parking path between the current position of the target vehicle and the target parking space position; Controlling the target vehicle to drive into the target parking space according to the target parking path; The step of generating a target bird's-eye view corresponding to each parking path point based on each parking path point of the initial parking path includes: Acquire a vehicle coordinate system corresponding to the initial position and a pixel coordinate system corresponding to the initial bird's-eye view; Converting the vehicle coordinate system to the pixel coordinate system to obtain a first position of the initial position in the pixel coordinate system; Based on the first position, determining a second position of each of the parking path points in the pixel coordinate system; Generating a target bird's-eye view corresponding to each parking path point based on the first position and each of the second positions includes: Obtaining a first camera extrinsic parameter corresponding to the first position and a previous position point corresponding to each of the second positions; Determine a target camera extrinsic parameter corresponding to each of the parking path points based on the first position, the second position, the previous position point, and the first camera extrinsic parameter; Based on the external parameters of each target camera, a target bird's-eye view corresponding to each parking path point is generated.
2. The method according to claim 1, characterized in that The method further comprises: Detecting whether there is an obstacle at each of the parking path points in the corresponding surround camera image; When it is determined that the obstacle exists in the surround camera image, the image range corresponding to the obstacle is eliminated to generate a target image range; Filling the missing ground information in the target image range to obtain a target surround camera image corresponding to each parking path point; Based on the target surround camera image, a target bird's-eye view image corresponding to each of the parking path points is generated.
3. The method according to claim 2, characterized in that The step of removing the image range corresponding to the obstacle to generate a target image range includes: identifying the type of the obstacle; Determine whether the obstacle needs to be removed based on the type; When it is determined that the obstacle needs to be removed, identifying the removal position of the obstacle; Obstacles located at the removal position are removed to obtain the target image range.
4. The method according to claim 2, characterized in that: The step of filling the missing ground information in the target image range to obtain the target surround camera image corresponding to each parking path point includes: Identify a rejection image range corresponding to a rejection position, and determine environmental information corresponding to the rejection image range; Predicting missing ground information of the removed image range based on the environmental information; The predicted ground information is filled into the missing ground corresponding to the removed image range to obtain the target surround camera image.
5. The method according to claim 1, characterized in that The step of determining a target parking path between the current position of the target vehicle and the target parking space position based on each of the target bird's-eye views includes: Determining a target parking path point corresponding to the current position based on each of the target bird's-eye views; Determining a parking direction based on the target parking path point and the target parking space position; The target parking path is generated based on the parking direction.
6. The method according to claim 1, characterized in that Also includes: Building an offline database based on the target bird's-eye view, wherein the offline database is deployed in the target vehicle; When the target vehicle is parking, a parking path of the target vehicle is optimized based on the offline database.
7. A parking space identification device, characterized in that: The device comprises: an acquisition module, configured to acquire an initial bird's-eye view corresponding to a target vehicle; the initial bird's-eye view includes an initial position of the target vehicle, a target parking space position corresponding to the target vehicle, and an initial parking path, wherein the initial parking path is a parking path determined based on the initial position and the target parking space position; A first generating module, for generating a target bird's-eye view corresponding to each parking path point based on each parking path point of the initial parking path, comprising: acquiring a vehicle coordinate system corresponding to the initial position and a pixel coordinate system corresponding to the initial bird's-eye view; converting the vehicle coordinate system to the pixel coordinate system to obtain a first position of the initial position in the pixel coordinate system; determining a second position of each parking path point in the pixel coordinate system based on the first position; generating a target bird's-eye view corresponding to each parking path point based on the first position and each second position, comprising: acquiring a first camera extrinsic parameter corresponding to the first position and a previous position point corresponding to each second position; determining a target camera extrinsic parameter corresponding to each parking path point based on the first position, the second position, the previous position point and the first camera extrinsic parameter; generating a target bird's-eye view corresponding to each parking path point based on each target camera extrinsic parameter; A determination module, configured to determine a target parking path between the current position of the target vehicle and the target parking space position based on each of the target bird's-eye views; A control module is used to control the target vehicle to enter the target parking space according to the target parking path.
8. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the parking space identification method according to any one of claims 1 to 6 by executing the computer instructions.
9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the parking space identification method according to any one of claims 1 to 6.
