Parking space positioning method and device for parking scene, equipment and storage medium

Abstract

The invention relates to a parking space positioning method and device for a parking scene, equipment and a storage medium. The method comprises the following steps: acquiring a first image around a vehicle; performing feature extraction on the first image to obtain a first image feature of the first image; determining a first feature layer of the first image feature under the first resolution and a second feature layer of the first image feature under the second resolution; the first resolution is lower than the second resolution; using the first feature layer to predict first parking space coordinates of all parking spaces around the vehicle and first wheel stopper coordinates of wheel stoppers in each parking space; predicting a second parking space coordinate of at least one parking space in a preset range around the vehicle and a second wheel stopper coordinate of a wheel stopper in each parking space by using the second feature layer; and performing parking operation on the vehicle based on the first parking space coordinate, the first wheel stopper coordinate, the second parking space coordinate and the second wheel stopper coordinate. According to the invention, the accuracy of automatic parking can be improved.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and in particular to a parking space positioning method, device, equipment and storage medium for a parking scene. Background Art

[0002] With the rapid development of autonomous driving technology, parking, as one of the important application scenarios in autonomous driving, has received widespread attention. In the task of autonomous driving parking, accurate positioning of parking spaces and wheel chocks is the core link to achieve safe parking.

[0003] Currently, parking spaces and wheel stoppers are usually located using visual parking space detection and scene perception based on neural networks. This method has poor recognition effect, which affects the user experience of using automatic parking. Summary of the invention

[0004] One of the purposes of the present invention is to provide a parking space positioning method, device, equipment and storage medium for a parking scene.

[0005] In a first aspect, the present invention provides a parking space positioning method for a parking scene, the method comprising:

[0006] Acquire a first image around the vehicle; perform feature extraction on the first image to obtain a first image feature of the first image;

[0007] Determining a first feature layer of a first image feature at a first resolution and a second feature layer at a second resolution; the first resolution is lower than the second resolution;

[0008] Predicting first parking space coordinates of all parking spaces around the vehicle and first wheel stopper coordinates of wheel stoppers in each parking space by using the first feature layer; predicting second parking space coordinates of at least one parking space within a preset range around the vehicle and second wheel stopper coordinates of wheel stoppers in each parking space by using the second feature layer;

[0009] A parking operation is performed on the vehicle based on the first parking space coordinate, the first wheel chock coordinate, the second parking space coordinate, and the second wheel chock coordinate.

[0010] Based on the above scheme, the present invention can effectively improve the vehicle's perception and positioning accuracy of parking spaces and wheel stoppers within a short range while satisfying long-distance and large-range positioning by combining prediction results with different resolutions, thereby achieving more efficient and accurate parking operations.

[0011] In the parking space positioning method for the above parking scene, the first image includes the second image and the third image, and the first image around the vehicle is acquired; and feature extraction is performed on the first image to obtain the first image feature of the first image, including:

[0012] acquiring a second image around the vehicle using a forward-looking wide-angle camera;

[0013] Acquire a third image around the vehicle using a fisheye camera;

[0014] Performing feature extraction on the second image and the third image respectively to obtain a second image feature of the second image and a third image feature of the third image;

[0015] The second image feature and the third image feature are fused to obtain the first image feature.

[0016] In the parking space positioning method for the above parking scene, the fisheye camera includes multiple fisheye cameras arranged on the vehicle body at multiple different positions; the method further includes:

[0017] Using multiple fisheye cameras to obtain multiple images of the vehicle's surroundings at multiple different orientations;

[0018] Feature extraction is performed on the multiple images to obtain third image features of the multiple images.

[0019] In the parking space positioning method for the above-mentioned parking scenario, the multiple fisheye cameras include at least a left front fisheye camera arranged on the left front side of the vehicle body, a right front fisheye camera arranged on the right front side of the vehicle body, a left rear fisheye camera arranged on the left rear side of the vehicle body, and a right rear fisheye camera arranged on the right rear side of the vehicle body.

[0020] In the parking space positioning method for the above parking scene, the second image feature and the third image feature are fused to obtain the first image feature, including:

[0021] The second image feature and the third image feature are fused by inverse perspective transformation, and the coordinate systems of the second image feature and the third image feature are unified to obtain the first image feature.

[0022] In the parking space positioning method for the above parking scene, the parking operation of the vehicle is performed based on the first parking space coordinate, the first wheel stopper coordinate, the second parking space coordinate and the second wheel stopper coordinate, including:

[0023] determining a target parking space coordinate of at least one parking space in the first parking space coordinate, and determining a target wheel chock coordinate of a wheel chock in each parking space of the at least one parking space;

[0024] Using the second parking space coordinates of at least one parking space to update the target parking space coordinates in the first parking space coordinates, to obtain updated first parking space coordinates;

[0025] updating the target wheel chock coordinates in the first wheel chock coordinates using the second wheel chock coordinates of the wheel chock in each parking space of the at least one parking space to obtain updated first wheel chock coordinates;

[0026] A parking operation is performed on the vehicle based on the updated first parking space coordinates and the updated first wheel stopper coordinates.

[0027] In the parking space positioning method for the above parking scenario, the parking operation of the vehicle is performed based on the updated first parking space coordinates and the updated first wheel stopper coordinates, including:

[0028] Determine the current position coordinates of the vehicle;

[0029] determining the parking space coordinates to be parked based on the updated first parking space coordinates, and determining the wheel chock coordinates in the parking space coordinates to be parked based on the updated first wheel chock coordinates;

[0030] The vehicle is parked according to the current position coordinates, the parking space coordinates and the wheel stopper coordinates.

[0031] In the parking space positioning method for the above parking scene, the parking space coordinates include the corner point coordinates of the four corner points of the parking space, and the wheel chock coordinates include the center point coordinates of the center point of the wheel chock.

[0032] In a second aspect, the present invention provides a parking space positioning device for a parking scene, and the parking space positioning device for a parking scene includes:

[0033] An acquisition unit is used to acquire a first image around the vehicle; perform feature extraction on the first image to obtain a first image feature of the first image;

[0034] A determining unit, configured to determine a first feature layer of a first image feature at a first resolution and a second feature layer at a second resolution; the first resolution is lower than the second resolution;

[0035] A prediction unit is used to predict first parking space coordinates of all parking spaces around the vehicle and first wheel stopper coordinates of wheel stoppers in each parking space by using the first feature layer; and predict second parking space coordinates of at least one parking space within a preset range around the vehicle and second wheel stopper coordinates of wheel stoppers in each parking space by using the second feature layer;

[0036] The parking unit is used for parking the vehicle based on the first parking space coordinate, the first wheel stopper coordinate, the second parking space coordinate and the second wheel stopper coordinate.

[0037] In a third aspect, the present invention provides a parking space positioning device for a parking scenario, the device comprising: a processor, a memory and a communication bus; when the processor executes a running program stored in the memory, the method of the parking space positioning device applied to the parking scenario is implemented.

[0038] In a fourth aspect, the present invention provides a computer-readable storage medium, characterized in that a computer program is stored thereon, and when the computer program is executed by a processor, the method of the parking space positioning device applied to the parking scene is implemented.

[0039] In a fifth aspect, the present invention provides a computer program product, including a computer program or instructions, characterized in that when the computer program or instructions are executed by a processor, the method of the parking space positioning device applied to the parking scene is implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 A schematic diagram of a flow chart of a parking space positioning method for a parking scenario provided by an embodiment of the present invention;

[0041] Figure 2 A schematic flow chart of an exemplary method for determining a first image feature provided by an embodiment of the present invention;

[0042] Figure 3 A schematic diagram of an exemplary parking method provided in an embodiment of the present invention Figure 1 ;

[0043] Figure 4 A schematic diagram of an exemplary parking method provided in an embodiment of the present invention Figure 2 ;

[0044] Figure 5 A schematic diagram of a process flow of locating a parking space in an exemplary parking scenario provided by an embodiment of the present invention;

[0045] Figure 6 A schematic diagram of the structure of a parking space positioning device for a parking scene provided by an embodiment of the present invention;

[0046] Figure 7 A schematic diagram of the composition structure of a parking space positioning device for a parking scenario provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0047] The following will describe the embodiments of the present invention with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention, not for limiting the scope of protection of the present invention.

[0048] With the rapid development of autonomous driving technology, parking, as one of the important application scenarios in autonomous driving, has received widespread attention. In the task of autonomous parking, accurate positioning of parking spaces and wheel chocks are the core links to achieve safe parking: accurate positioning of parking spaces is related to whether the vehicle can be parked safely in a limited space, while accurate detection of wheel chocks can effectively prevent the vehicle from colliding or exceeding the predetermined parking area during parking.

[0049] Traditional parking detection solutions usually use a single feature layer to predict the corner positions of parking spaces and wheel stoppers respectively. However, single-resolution feature detection solutions face the following challenges in complex parking scenarios:

[0050] (1) A feature layer with a single resolution may not be able to take into account both the requirements of long-distance and large-scale perception and short-distance and high-precision detection, thus affecting the overall accuracy and safety of parking.

[0051] (2) The single-resolution detection scheme cannot integrate the information of the two key elements, parking space and wheel stopper, which leads to network structure redundancy and waste of storage and computing resources.

[0052] In view of the above technical problems, the present invention provides a parking space positioning method for a parking scene, which is implemented by a parking space positioning device for a parking scene. Figure 1 A flow chart of a parking space positioning method for a parking scene provided by an embodiment of the present invention is shown as follows: Figure 1 As shown, the steps S100 to S103 are included:

[0053] Step S100: Acquire a first image around the vehicle; perform feature extraction on the first image to obtain first image features of the first image.

[0054] In an embodiment of the present invention, a parking space positioning device for a parking scene acquires a first image around a vehicle; and performs feature extraction on the first image to obtain a first image feature of the first image.

[0055] In some embodiments, the first image may be an environmental image around the vehicle. When a parking operation is triggered, the vehicle may obtain an environmental image around the vehicle, and compress the environmental image through a feature extraction network to generate first image features with higher information density and smaller storage space. The feature extraction network may be a feature extraction network based on a convolutional neural network (CNN) or a feature extraction network based on a Transformer (a neural network based on a self-attention mechanism). The specific feature extraction method may be selected according to actual conditions, and the present invention does not make any specific limitation here.

[0056] Step S101: determining a first feature layer of a first image feature at a first resolution and a second feature layer at a second resolution; the first resolution is lower than the second resolution.

[0057] In an embodiment of the present invention, after obtaining the first image feature of the first image, the parking space positioning device for the parking scene determines a first feature layer of the first image feature at a first resolution and a second feature layer at a second resolution; the first resolution is lower than the second resolution.

[0058] In some embodiments, after obtaining the first image feature of the first image, the parking space positioning device of the parking scene needs to use a convolutional neural network to further extract feature layers at different resolutions, namely, the first feature layer at low resolution and the second feature layer at high resolution.

[0059] It should be noted that since the low-resolution first feature layer can cover a larger range of image information, although the resolution is low, it can perceive a larger range; while the high-resolution second feature layer focuses on a smaller image area with a higher resolution and can accurately locate the target position, it is necessary to extract feature layers at different resolutions at the same time.

[0060] Step S102, using the first feature layer to predict the first parking space coordinates of all parking spaces around the vehicle and the first wheel stopper coordinates of the wheel stopper in each parking space; using the second feature layer to predict the second parking space coordinates of at least one parking space within a preset range around the vehicle and the second wheel stopper coordinates of the wheel stopper in each parking space.

[0061] In an embodiment of the present invention, after determining the first feature layer of the first image feature at the first resolution and the second feature layer at the second resolution, the parking space positioning device of the parking scene uses the first feature layer to predict the first parking space coordinates of all parking spaces around the vehicle and the first wheel stopper coordinates of the wheel stopper in each parking space; and uses the second feature layer to predict the second parking space coordinates of at least one parking space within a preset range around the vehicle and the second wheel stopper coordinates of the wheel stopper in each parking space.

[0062] In some embodiments, after extracting the first feature layer, the parking space positioning device of the parking scene can predict the parking spaces within the entire range around the vehicle through the first feature layer, because the low-resolution first feature layer can cover a larger range of image information, and obtain the first parking space coordinates of each parking space roughly positioned within the entire range, as well as the first wheel stopper coordinates of the wheel stopper in each parking space.

[0063] In some embodiments, after extracting the second feature layer, the parking space positioning device of the parking scene can predict the parking space within a close distance of the vehicle through the second feature layer, because the high-resolution second feature layer focuses on a smaller image area, and thus obtain the second parking space coordinates of each parking space accurately positioned within a close distance, as well as the second wheel stopper coordinates of the wheel stopper in each parking space.

[0064] Step S103: parking the vehicle based on the first parking space coordinates, the first wheel stopper coordinates, the second parking space coordinates, and the second wheel stopper coordinates.

[0065] In an embodiment of the present invention, the parking space positioning device of the parking scene performs a parking operation on the vehicle based on the first parking space coordinates, the first wheel stopper coordinates, the second parking space coordinates and the second wheel stopper coordinates after predicting the first parking space coordinates and the first wheel stopper coordinates, the second parking space coordinates and the second wheel stopper coordinates.

[0066] In some embodiments, the parking space positioning device of the parking scene can select an optimal parking space for the vehicle based on the above coordinates and the current coordinates of the vehicle, and perform parking operations after obtaining the first parking space coordinates of each parking space under rough positioning in the full range and the first wheel stopper coordinates of the wheel stopper in each parking space, and obtaining the second parking space coordinates of each parking space under precise positioning within a close distance and the second wheel stopper coordinates of the wheel stopper in each parking space.

[0067] In an optional embodiment of the present invention, the first image includes the second image and the third image, and the first image around the vehicle is acquired; Figure 2 The parking space positioning device of the parking scene extracts features from the first image to obtain first image features of the first image, including the following steps S200 to S203:

[0068] Step S200: Acquire a second image around the vehicle using a forward-looking wide-angle camera.

[0069] In an embodiment of the present invention, the parking space positioning device for a parking scene uses a forward-looking wide-angle camera to acquire a second image around the vehicle.

[0070] In some embodiments, the forward-looking wide-angle camera is a camera equipped with a wide-angle lens, which is generally used to capture images in a larger range; the specific forward-looking wide-angle camera can be selected according to actual conditions, and the present invention does not make specific limitations here.

[0071] Step S201: Acquire a third image around the vehicle using a fisheye camera.

[0072] In the embodiment of the present invention, the parking space positioning device for the parking scene uses a fisheye camera to obtain a third image around the vehicle.

[0073] In some embodiments, a fisheye camera refers to a camera with a fisheye lens, which can capture a wider picture than an ordinary lens; the specific fisheye camera can be selected according to actual conditions, and the present invention does not make any specific limitation here.

[0074] Step S202: extract features from the second image and the third image respectively to obtain second image features of the second image and third image features of the third image.

[0075] In the embodiment of the present invention, after acquiring the second image and the third image, the parking space positioning device for the parking scene extracts features from the second image and the third image respectively to obtain the second image feature of the second image and the third image feature of the third image.

[0076] In some embodiments, the parking space positioning device of the parking scene extracts features from the second image and the third image through a feature extraction network to obtain second image features and third image features with higher information density and smaller occupied storage space.

[0077] Step S203: fuse the second image feature and the third image feature to obtain the first image feature.

[0078] In the embodiment of the present invention, after obtaining the second image feature of the second image and the third image feature of the third image, the parking space positioning device for the parking scene fuses the second image feature and the third image feature to obtain the first image feature.

[0079] In some embodiments, after obtaining the second image feature of the forward-looking wide-angle camera and the third image feature of the fisheye camera, the parking space positioning device of the parking scene needs to perform feature fusion on the second image feature and the third image feature, so as to obtain a first image feature that fuses the second image feature and the third image feature. Thus, the first image feature includes all features of the environmental image around the vehicle acquired by the forward-looking wide-angle camera and the fisheye camera.

[0080] In an optional embodiment of the present invention, the fisheye camera includes multiple fisheye cameras arranged on the vehicle body at multiple different positions, and the parking space positioning device for the parking scene can also perform the following steps: using multiple fisheye cameras to obtain multiple images around the vehicle at multiple different positions; performing feature extraction on the multiple images to obtain third image features of the multiple images.

[0081] In an optional embodiment of the present invention, the multiple fisheye cameras include at least a left front fisheye camera arranged on the left front side of the vehicle body, a right front fisheye camera arranged on the right front side of the vehicle body, a left rear fisheye camera arranged on the left rear side of the vehicle body, and a right rear fisheye camera arranged on the right rear side of the vehicle body.

[0082] In some embodiments, the vehicle is provided with a left front fisheye camera on the left front side of the vehicle body, a right front fisheye camera on the right front side of the vehicle body, a left rear fisheye camera on the left rear side of the vehicle body, and a right rear fisheye camera on the right rear side of the vehicle body according to different viewing angles. Thus, the environmental image of the vehicle at each viewing angle can be acquired through these fisheye cameras. Therefore, the second image at least includes the left front image of the vehicle surrounding captured by the left front fisheye camera, the right front image of the vehicle surrounding captured by the right front fisheye camera, the left rear image of the vehicle surrounding captured by the left rear fisheye camera, and the right rear image of the vehicle surrounding captured by the right rear fisheye camera. The specific distribution of the fisheye cameras on the vehicle and the number of the fisheye cameras can be set according to actual conditions, and the present invention does not make specific limitations here.

[0083] In some embodiments, after acquiring the left front image, right front image, left rear image and right rear image, the parking space positioning device of the parking scene can extract the features of the left front image, right front image, left rear image and right rear image through a feature extraction network, thereby obtaining the third image features corresponding to the left front image, right front image, left rear image and right rear image.

[0084] In an optional embodiment of the present invention, a parking space positioning device for a parking scene fuses the second image feature and the third image feature to obtain a first image feature, including the following steps: performing inverse perspective transformation on the second image feature and the third image feature, and aligning the coordinate systems of the second image feature and the third image feature to obtain the first image feature.

[0085] In some embodiments, the parking space positioning device of the parking scene uses inverse perspective mapping (IPM) when fusing the second image feature and the third image feature to obtain the first image feature. The IPM technology can convert images of different perspectives into a unified perspective, thereby achieving effective alignment and fusion of the second image feature of the forward wide-angle camera and the third image feature of the fisheye camera at different perspectives on the same plane.

[0086] In an optional embodiment of the present invention, reference Figure 3 The parking space positioning device of the parking scene performs a parking operation on the vehicle based on the first parking space coordinate, the first wheel stopper coordinate, the second parking space coordinate and the second wheel stopper coordinate, including the following steps S300 to S303:

[0087] Step S300: determining a target parking space coordinate of at least one parking space in the first parking space coordinate, and determining a target wheel chock coordinate of a wheel chock in each parking space of the at least one parking space.

[0088] In an embodiment of the present invention, when a parking space positioning device of a parking scene performs a parking operation on a vehicle based on a first parking space coordinate, a first wheel stopper coordinate, a second parking space coordinate and a second wheel stopper coordinate, it is first necessary to determine a target parking space coordinate of at least one parking space in the first parking space coordinate, and determine a target wheel stopper coordinate of the wheel stopper in each parking space of at least one parking space.

[0089] In some embodiments, during the parking process of a vehicle, it is necessary not only to know the rough situation of parking spaces within the entire range, but also to know the precise positioning of parking spaces in close proximity in order to facilitate precise parking. Therefore, the parking process needs to be based on the first parking space coordinates and the first wheel stopper coordinates under the rough positioning within the entire range, and the second parking space coordinates and the second wheel stopper coordinates under the precise positioning within close proximity, that is, the first parking space coordinates, the first wheel stopper coordinates, the second parking space coordinates and the second wheel stopper coordinates are combined.

[0090] In some embodiments, when merging the first parking space coordinates, the first wheel stopper coordinates, the second parking space coordinates and the second wheel stopper coordinates, the target parking space coordinates of at least one parking space accurately positioned within a close distance in the first parking space coordinates can be determined first, and the target wheel stopper coordinates of the wheel stopper in each parking space of at least one parking space can be determined. At this time, the target parking space coordinates and the target wheel stopper coordinates are coordinates under rough positioning.

[0091] Step S301: Use the second parking space coordinates of at least one parking space to update the target parking space coordinates in the first parking space coordinates to obtain updated first parking space coordinates.

[0092] In an embodiment of the present invention, after determining the target parking space coordinates of at least one parking space in the first parking space coordinates, the parking space positioning device of the parking scene updates the target parking space coordinates in the first parking space coordinates using the second parking space coordinates of at least one parking space to obtain updated first parking space coordinates.

[0093] In some embodiments, the second parking space coordinates of at least one parking space precisely positioned at a close distance may be used to replace the target parking space coordinates of at least one same parking space roughly positioned within the full range. Thus, the updated first parking space coordinates are composed of the parking space coordinates of the long-distance parking space roughly positioned and the parking space coordinates of the close-range parking space precisely positioned.

[0094] Step S302: updating the target wheel chock coordinates in the first wheel chock coordinates by using the second wheel chock coordinates of the wheel chock in each parking space of at least one parking space to obtain updated first wheel chock coordinates.

[0095] In an embodiment of the present invention, after determining the target wheel chock coordinates of the wheel chock in each parking space of at least one parking space, the parking space positioning device of the parking scene updates the target wheel chock coordinates in the first wheel chock coordinates using the second wheel chock coordinates of the wheel chock in each parking space of at least one parking space to obtain the updated first wheel chock coordinates.

[0096] In some embodiments, the target wheel chock coordinates of the wheel chocks in each parking space of the same at least one parking space under the rough positioning within the full range may be replaced by the second wheel chock coordinates of the wheel chocks in each parking space of the at least one parking space under the precise positioning within the close range, whereby the updated first wheel chock coordinates are composed of the wheel chock coordinates in the long-distance parking space under the rough positioning and the wheel chock coordinates in the close-range parking space under the precise positioning.

[0097] Step S303: parking the vehicle based on the updated first parking space coordinates and the updated first wheel stopper coordinates.

[0098] In the embodiment of the present invention, after obtaining the updated first parking space coordinates and the updated first parking space coordinates, the parking space positioning device of the parking scene performs a parking operation on the vehicle based on the updated first parking space coordinates and the updated first wheel stopper coordinates.

[0099] In an optional embodiment of the present invention, reference Figure 4 The parking space positioning device of the parking scene performs a parking operation on the vehicle based on the updated first parking space coordinates and the updated first wheel stopper coordinates, including the following steps S400 to S402:

[0100] Step S400: Determine the current position coordinates of the vehicle.

[0101] In the embodiment of the present invention, when the parking space positioning device of the parking scene performs a parking operation on the vehicle based on the updated first parking space coordinates and the updated first wheel stopper coordinates, it is necessary to first determine the current position coordinates of the vehicle.

[0102] In some embodiments, the current position coordinates of the vehicle may include the position coordinates of the center of the current front wheel or rear wheel rotation axis of the vehicle, and may also include the current position coordinates of each wheel on the vehicle; the specific current position coordinates can be determined based on actual conditions, and the embodiments of the present invention do not make specific limitations here.

[0103] Step S401: determining the parking space coordinates to be parked based on the updated first parking space coordinates, and determining the wheel chock coordinates in the parking space coordinates to be parked based on the updated first wheel chock coordinates.

[0104] In an embodiment of the present invention, after determining the current position coordinates of the vehicle, the parking space positioning device of the parking scene determines the parking space coordinates to be parked based on the updated first parking space coordinates, and determines the wheel stopper coordinates in the parking space coordinates to be parked based on the updated first wheel stopper coordinates.

[0105] In some embodiments, a parking space positioning device of a parking scene can select a target parking space to be parked from the surrounding parking spaces. Generally speaking, a parking space closer to the vehicle is selected. Therefore, the parking space coordinates corresponding to the target parking space in the updated first parking space coordinates are the parking space coordinates precisely positioned at a close distance, and the wheel stopper coordinates corresponding to the target parking space in the updated first wheel stopper coordinates are the wheel stopper coordinates precisely positioned at a close distance. The parking operation of the vehicle can be achieved based on the precisely positioned parking space coordinates to be parked and the wheel stopper coordinates.

[0106] Step S402: Parking the vehicle according to the current position coordinates, the parking space coordinates and the wheel stopper coordinates.

[0107] In an embodiment of the present invention, after determining the parking space coordinates and the wheel stopper coordinates in the parking space coordinates, the parking space positioning device of the parking scene performs parking operations on the vehicle according to the current position coordinates, the parking space coordinates and the wheel stopper coordinates.

[0108] In an optional implementation manner of the present invention, the parking space coordinates include corner point coordinates of four corner points of the parking space, and the wheel chock coordinates include center point coordinates of a center point of the wheel chock.

[0109] In some embodiments, since the coordinates of the parking space to be parked include the corner point coordinates of the four corner points of the parking space to be parked, and the wheel stopper coordinates of the parking space to be parked include the center point coordinates of the center point of the wheel stopper, the parking route of the vehicle can be planned according to the corner point coordinates of the four corner points of the parking space to be parked and the distance between the center point coordinates of the center point of the wheel stopper and the current position coordinates, and the control strategy of the steering wheel and the control strategy of the brake can be further determined, so as to successfully park the vehicle in the parking space to be parked.

[0110] An embodiment of the present invention provides a parking space positioning method for a parking scene, the method comprising: obtaining a first problem; obtaining a first image around a vehicle; performing feature extraction on the first image to obtain a first image feature of the first image; determining a first feature layer of the first image feature at a first resolution and a second feature layer at a second resolution; the first resolution is lower than the second resolution; using the first feature layer to predict first parking space coordinates of all parking spaces around the vehicle and first wheel stopper coordinates of a wheel stopper in each parking space; using the second feature layer to predict second parking space coordinates of at least one parking space within a preset range around the vehicle and second wheel stopper coordinates of a wheel stopper in each parking space; performing a parking operation on the vehicle based on the first parking space coordinates, the first wheel stopper coordinates, the second parking space coordinates and the second wheel stopper coordinates; adopting the above implementation scheme, the present invention can effectively improve the perception and positioning accuracy of the vehicle to the parking space and the wheel stopper within a short range on the basis of satisfying long-distance and large-range positioning by combining prediction results of different resolutions, thereby achieving more efficient and accurate parking operations.

[0111] Based on the above embodiments, the present invention provides a schematic diagram of a flow chart of parking space positioning in an exemplary parking scenario, referring to Figure 5 , the specific steps are as follows:

[0112] Step 1: Use a forward-looking wide-angle camera 501 and four fisheye cameras 502 to perceive the vehicle's surrounding environment 500.

[0113] Specifically, the acquisition of images of the vehicle surroundings 500 is the basis of parking perception technology. The present invention uses five cameras, namely, a front wide-angle camera, a left front fisheye camera, a right front fisheye camera, a left rear fisheye camera, and a right rear fisheye camera, to collect data in real time, ensuring that synchronous image information is acquired from different perspectives at the same time, thereby improving the accuracy of subsequent image fusion processing.

[0114] Step 2: extract the wide-angle camera features 503 of the front-view wide-angle camera 501 and the fisheye camera features 504 of the four fisheye cameras 502 .

[0115] Specifically, after acquiring images around the vehicle, a feature extraction network is used to compress these images, thereby generating image features with higher information density and smaller storage space. Common feature extraction methods include feature extraction networks based on convolutional neural networks (CNN) and feature extraction networks based on Transformers. The technical solution of the present invention can adapt to these two feature extraction networks, thereby flexibly responding to different types of image feature extraction requirements.

[0116] Step 3: The wide-angle camera feature 503 and the fisheye camera feature 504 are fused through the IPM graph.

[0117] Specifically, after obtaining the wide-angle camera feature 503 (the second image feature in the above embodiment) and the fisheye camera feature 504 (the third image feature in the above embodiment), the present invention fuses the wide-angle camera feature 503 and the fisheye camera feature 504 through the IPM graph to obtain the fused feature 505 (the first image feature in the above embodiment). The IPM technology can convert images of different perspectives into a unified perspective, thereby achieving effective alignment and fusion of the image features of the forward-looking wide-angle camera 501 and the four fisheye cameras 502 on the same plane.

[0118] Step 4: Use a convolutional neural network to extract the low-resolution feature layer 506 and the high-resolution feature layer 507.

[0119] Specifically, this solution further extracts low-resolution feature layer 506 and high-resolution feature layer 507 through convolutional neural network. Low-resolution feature layer 506 covers a larger range of image information, and although the resolution is lower, it can perceive a larger range; while high-resolution feature layer 507 focuses on a smaller range of image areas, has a higher resolution, and can accurately locate the target position.

[0120] Step 5: Use the low-resolution feature layer 506 and the high-resolution feature layer 507 to predict the full range, close-range parking spaces and wheel stopper corner positions respectively.

[0121] Specifically, based on the characteristics of the large perception range of the low-resolution feature layer 506 and the high positioning accuracy of the high-resolution feature layer 507, the present invention roughly locates the full-range parking spaces and wheel stopper coordinates 508 on the low-resolution feature layer 506, and accurately locates the close-range parking spaces and wheel stopper coordinates 509 on the high-resolution feature layer 507.

[0122] Step 6: Replace the close-range corner points generated by the low-resolution feature layer with the fine close-range corner point positions predicted in the high-resolution feature layer.

[0123] Specifically, in the final processing stage, the present invention will use the more detailed close-range parking space and wheel chock coordinates 509 predicted in the high-resolution feature layer 507 to replace the relatively rough close-range parking space and wheel chock coordinates predicted by the low-resolution feature layer 506, and obtain the final full-range parking space and wheel chock coordinates 510. Through this process, the positioning accuracy of the model for parking spaces and wheel chocks in a close range can be significantly improved, ensuring that the vehicle can be accurately parked even in complex environments.

[0124] Based on the above embodiment, in another embodiment of the present invention, a parking space positioning device for a parking scene is provided. Figure 6 A schematic diagram of the structure of a parking space positioning device for a parking scene provided by an embodiment of the present invention is shown in FIG. Figure 6As shown, the parking space positioning device 6 of the parking scene includes:

[0125] The acquisition unit 600 is used to acquire a first image around the vehicle; perform feature extraction on the first image to obtain a first image feature of the first image;

[0126] A determining unit 601 is configured to determine a first feature layer of a first image feature at a first resolution and a second feature layer at a second resolution; the first resolution is lower than the second resolution;

[0127] The prediction unit 602 is used to predict the first parking space coordinates of all parking spaces around the vehicle and the first wheel stopper coordinates of the wheel stopper in each parking space by using the first feature layer; predict the second parking space coordinates of at least one parking space within a preset range around the vehicle and the second wheel stopper coordinates of the wheel stopper in each parking space by using the second feature layer;

[0128] The parking unit 603 is used to perform a parking operation on the vehicle based on the first parking space coordinate, the first wheel stopper coordinate, the second parking space coordinate and the second wheel stopper coordinate.

[0129] Optionally, the parking space positioning device 6 for the parking scene further includes: a data processing unit;

[0130] The acquisition unit 600 is further configured to acquire a second image around the vehicle using the forward-looking wide-angle camera; and acquire a third image around the vehicle using the fisheye camera;

[0131] The data processing unit is used to extract features from the second image and the third image respectively to obtain second image features of the second image and third image features of the third image; and to fuse the second image features and the third image features to obtain first image features.

[0132] Optionally, the acquisition unit 600 is further configured to acquire multiple images of the surroundings of the vehicle at multiple different orientations using multiple fisheye cameras;

[0133] The data processing unit is also used to extract features from the multiple images to obtain third image features of the multiple images.

[0134] Optionally, the data processing unit is further used to perform inverse perspective transformation and fusion on the second image feature and the third image feature, and unify the coordinate systems of the second image feature and the third image feature to obtain the first image feature.

[0135] Optionally, the determining unit 601 is further configured to determine a target parking space coordinate of at least one parking space in the first parking space coordinate, and determine a target wheel chock coordinate of a wheel chock in each parking space of the at least one parking space;

[0136] The data processing unit is further used to update the target parking space coordinates in the first parking space coordinates using the second parking space coordinates of at least one parking space to obtain updated first parking space coordinates; update the target wheel stopper coordinates in the first wheel stopper coordinates using the second wheel stopper coordinates of the wheel stopper in each parking space of at least one parking space to obtain updated first wheel stopper coordinates;

[0137] The parking unit 603 is further configured to perform a parking operation on the vehicle based on the updated first parking space coordinates and the updated first wheel stopper coordinates.

[0138] Optionally, the determination unit 601 is further configured to determine the current position coordinates of the vehicle; determine the parking space coordinates to be parked based on the updated first parking space coordinates, and determine the wheel stopper coordinates in the parking space coordinates to be parked based on the updated first wheel stopper coordinates;

[0139] The parking unit 603 is also used to park the vehicle according to the current position coordinates, the parking space coordinates and the wheel stopper coordinates.

[0140] An embodiment of the present invention provides a parking space positioning device for a parking scene, the device comprising: an acquisition unit, used to acquire a first image around a vehicle; extract features from the first image to obtain first image features of the first image; a determination unit, used to determine a first feature layer of the first image feature at a first resolution and a second feature layer at a second resolution; the first resolution is lower than the second resolution; a prediction unit, used to predict first parking space coordinates of all parking spaces around the vehicle and first wheel stopper coordinates of wheel stoppers in each parking space by using the first feature layer; predict second parking space coordinates of at least one parking space within a preset range around the vehicle and second wheel stopper coordinates of wheel stoppers in each parking space by using the second feature layer; a parking unit, used to perform a parking operation on the vehicle based on the first parking space coordinates, the first wheel stopper coordinates, the second parking space coordinates and the second wheel stopper coordinates; by adopting the above implementation scheme, the present invention can effectively improve the perception and positioning accuracy of the vehicle to the parking space and the wheel stopper within a short range on the basis of satisfying long-distance and large-range positioning by combining prediction results of different resolutions, thereby achieving more efficient and accurate parking operations.

[0141] Figure 7 A schematic diagram of the composition structure of a parking space positioning device for a parking scene provided by an embodiment of the present invention. In practical applications, based on the same disclosed concept of the above embodiments, Figure 7 As shown, the parking space positioning device 7 for the parking scene of this embodiment includes: a processor 700 , a memory 701 and a communication bus 702 .

[0142] In the specific embodiment, the acquisition unit 600, the determination unit 601, the prediction unit 602, the parking unit 603, and the data processing unit can be implemented by a processor 700 located on the parking space positioning device 7 of the parking scene, and the processor 700 can be at least one of an application-specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing image processing device (DSPD), a programmable logic image processing device (PLD), a field programmable gate array (FPGA), a CPU, a controller, a microcontroller, and a microprocessor. It can be understood that for the parking space positioning device 7 of different parking scenes, the electronic device used to implement the above-mentioned processor function can also be other, which is not specifically limited in this embodiment.

[0143] In the embodiment of the present invention, the communication bus 702 is used to realize the connection and communication between the processor 700 and the memory 701; when the processor 700 executes the running program stored in the memory 701, the following parking space positioning method for the parking scene is realized:

[0144] Acquire a first image around the vehicle; perform feature extraction on the first image to obtain a first image feature of the first image;

[0145] Determining a first feature layer of a first image feature at a first resolution and a second feature layer at a second resolution; the first resolution is lower than the second resolution;

[0146] Predicting first parking space coordinates of all parking spaces around the vehicle and first wheel stopper coordinates of wheel stoppers in each parking space by using the first feature layer; predicting second parking space coordinates of at least one parking space within a preset range around the vehicle and second wheel stopper coordinates of wheel stoppers in each parking space by using the second feature layer;

[0147] A parking operation is performed on the vehicle based on the first parking space coordinate, the first wheel chock coordinate, the second parking space coordinate, and the second wheel chock coordinate.

[0148] Optionally, the processor 700 is also used to obtain a second image of the surroundings of the vehicle using a forward wide-angle camera; obtain a third image of the surroundings of the vehicle using a fisheye camera; perform feature extraction on the second image and the third image respectively to obtain second image features of the second image and third image features of the third image; and fuse the second image features and the third image features to obtain first image features.

[0149] Optionally, the processor 700 is further used to obtain multiple images of the surroundings of the vehicle at multiple different directions using multiple fisheye cameras; and perform feature extraction on the multiple images to obtain third image features of the multiple images.

[0150] Optionally, the processor 700 is further configured to perform inverse perspective transformation and fusion on the second image feature and the third image feature, and unify the coordinate systems of the second image feature and the third image feature to obtain the first image feature.

[0151] Optionally, the processor 700 is further used to determine the target parking space coordinates of at least one parking space in the first parking space coordinates, determine the target wheel stopper coordinates of the wheel stopper in each parking space of the at least one parking space; update the target parking space coordinates in the first parking space coordinates using the second parking space coordinates of the at least one parking space to obtain updated first parking space coordinates; update the target wheel stopper coordinates in the first wheel stopper coordinates using the second wheel stopper coordinates of the wheel stopper in each parking space of the at least one parking space to obtain updated first wheel stopper coordinates; and perform parking operations on the vehicle based on the updated first parking space coordinates and the updated first wheel stopper coordinates.

[0152] Optionally, the processor 700 is further used to determine the current position coordinates of the vehicle; determine the coordinates of the parking space to be parked based on the updated first parking space coordinates, and determine the wheel stopper coordinates in the parking space coordinates to be parked based on the updated first wheel stopper coordinates; and perform parking operations on the vehicle according to the current position coordinates, the parking space coordinates to be parked, and the wheel stopper coordinates.

[0153] An embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon. The computer-readable storage medium stores one or more programs. The one or more programs can be executed by one or more processors and are applied to a parking space positioning device for a parking scene. The computer program implements a parking space positioning method for the parking scene as described above.

[0154] An embodiment of the present invention provides a computer program product, including a computer program or instructions. When the computer program or instructions are executed by a processor, the parking space positioning method for the parking scenario as described above is implemented.

[0155] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of hardware embodiments, software embodiments, or embodiments combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) containing computer-usable program code.

[0156] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0157] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0158] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0159] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A parking space positioning method for a parking scene, characterized in that: The method comprises: Acquire a first image around the vehicle; perform feature extraction on the first image to obtain first image features of the first image; Determining a first feature layer of the first image feature at a first resolution and a second feature layer at a second resolution; the first resolution is lower than the second resolution; Predicting first parking space coordinates of all parking spaces around the vehicle and first wheel stopper coordinates of wheel stoppers in each parking space by using the first feature layer; predicting second parking space coordinates of at least one parking space within a preset range around the vehicle and second wheel stopper coordinates of wheel stoppers in each parking space by using the second feature layer; A parking operation is performed on the vehicle based on the first parking space coordinates, the first wheel chock coordinates, the second parking space coordinates, and the second wheel chock coordinates.

2. The parking space positioning method for parking scenes according to claim 1, characterized in that: The first image includes a second image and a third image, and the acquiring of the first image around the vehicle; and extracting features from the first image to obtain first image features of the first image include: Acquire a second image around the vehicle using a forward-looking wide-angle camera; Acquire a third image around the vehicle using a fisheye camera; Performing feature extraction on the second image and the third image respectively to obtain a second image feature of the second image and a third image feature of the third image; The second image feature and the third image feature are fused to obtain the first image feature.

3. The parking space positioning method for parking scenes according to claim 2, characterized in that: The fisheye camera includes a plurality of fisheye cameras arranged on a plurality of different positions of the vehicle body; the method further includes: Using the multiple fisheye cameras to acquire multiple images of the surroundings of the vehicle at the multiple different positions; Perform feature extraction on the multiple images to obtain the third image features of the multiple images.

4. The parking space positioning method for parking scenes according to claim 3, characterized in that: The multiple fisheye cameras include at least a left front fisheye camera arranged on the left front side of the vehicle body, a right front fisheye camera arranged on the right front side of the vehicle body, a left rear fisheye camera arranged on the left rear side of the vehicle body, and a right rear fisheye camera arranged on the right rear side of the vehicle body.

5. The parking space positioning method for parking scenes according to claim 2, characterized in that: The fusing the second image feature and the third image feature to obtain the first image feature includes: The second image feature and the third image feature are fused by inverse perspective transformation, and the coordinate systems of the second image feature and the third image feature are unified to obtain the first image feature.

6. The parking space positioning method for parking scenes according to claim 1, characterized in that: The performing a parking operation on the vehicle based on the first parking space coordinate, the first wheel stopper coordinate, the second parking space coordinate, and the second wheel stopper coordinate comprises: determining a target parking space coordinate of the at least one parking space in the first parking space coordinate, and determining a target wheel chock coordinate of a wheel chock in each parking space of the at least one parking space; Using the second parking space coordinates of the at least one parking space, the target parking space coordinates in the first parking space coordinates are updated to obtain updated first parking space coordinates; updating the target wheel chock coordinates in the first wheel chock coordinates by using the second wheel chock coordinates of the wheel chock in each parking space of the at least one parking space to obtain updated first wheel chock coordinates; A parking operation is performed on the vehicle based on the updated first parking space coordinates and the updated first wheel stopper coordinates.

7. The parking space positioning method for parking scenes according to claim 6, characterized in that: The performing a parking operation on the vehicle based on the updated first parking space coordinates and the updated first wheel stopper coordinates comprises: Determining the current position coordinates of the vehicle; determining the parking space coordinates to be parked based on the updated first parking space coordinates, and determining the wheel chock coordinates in the parking space coordinates to be parked based on the updated first wheel chock coordinates; The vehicle is parked according to the current position coordinates, the parking space coordinates and the wheel stopper coordinates.

8. The parking space positioning method for a parking scene according to any one of claims 1 to 7, characterized in that: The parking space coordinates include corner point coordinates of four corner points of the parking space, and the wheel chock coordinates include center point coordinates of a center point of the wheel chock.

9. A parking space positioning device for a parking scene, characterized in that: The device comprises: An acquisition unit, configured to acquire a first image around the vehicle; perform feature extraction on the first image to obtain a first image feature of the first image; a determining unit, configured to determine a first feature layer of the first image feature at a first resolution and a second feature layer at a second resolution; the first resolution is lower than the second resolution; A prediction unit, configured to predict first parking space coordinates of all parking spaces around the vehicle and first wheel stopper coordinates of wheel stoppers in each parking space by using the first feature layer; and predict second parking space coordinates of at least one parking space within a preset range around the vehicle and second wheel stopper coordinates of wheel stoppers in each parking space by using the second feature layer; A parking unit is used to perform a parking operation on the vehicle based on the first parking space coordinate, the first wheel chock coordinate, the second parking space coordinate, and the second wheel chock coordinate.

10. A parking space positioning device for a parking scene, characterized in that: The device comprises: a processor, a memory and a communication bus; when the processor executes the running program stored in the memory, the method according to any one of claims 1 to 8 is implemented.

11. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

12. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images