Parking space reconstruction method, automatic parking method, system and mobile communication equipment

Abstract

The invention provides a parking space reconstruction method, an automatic parking method, an automatic parking system and mobile communication equipment. A mobile communication device of a target vehicle user carries an application program used for parking space reconstruction. The parking space reconstruction method specifically comprises the steps of collecting parking space image information of a target parking space based on the application program; performing feature matching according to the parking space image information to obtain a matching feature pair, and performing triangulation based on the matching feature pair to generate a dense point cloud according to the matching feature pair after triangulation; and generating a target parking space structure model according to the dense point cloud. According to the method and the device, the limitation of singly depending on a sensor of a vehicle under the scenes of narrow parking space, prominent objects and the like is overcome, more comprehensive parking space information is provided, an automatic parking system can be ensured to plan a parking path more accurately, and subsequent problems caused by inaccurate parking are reduced.

Description

Technical Field

[0001] The present application relates to the field of assisted driving technology, and in particular to a parking space reconstruction method, an automatic parking method, a system, and a mobile communication device. Background Art

[0002] With the advancement of intelligent vehicles, automatic parking systems have become a crucial feature for enhancing driving convenience. Currently, most existing automatic parking systems rely on the vehicle's built-in sensors, including four fisheye cameras (front, rear, left, and right), as well as several millimeter-wave radars. These sensors are effective in typical parking scenarios, enabling vehicles to park automatically.

[0003] However, existing automated parking systems exhibit significant limitations in complex scenarios. Visual and radar sensors lack detection range and algorithms. For one thing, the system often fails to accurately estimate the position of objects within the parking space, leading to errors. This can cause collisions with surrounding objects during parking, increasing safety risks. Furthermore, these sensors lack the ability to effectively detect suspended objects above a certain height. For example, pipes and hanging signs in parking lots may be unrecognizable, creating potential dangers during parking.

[0004] Furthermore, current automated parking systems have a low recognition rate for non-standard scenarios, such as odd-shaped and multi-story parking spaces. The irregular shapes of odd-shaped spaces and the unique spatial structures of multi-story parking spaces make it difficult for existing sensors and algorithms to accurately identify and analyze them. This makes it difficult for the system to effectively plan parking paths in these non-standard scenarios, hindering the subsequent automated parking maneuver.

[0005] To sum up, although the existing automatic parking system provides convenience to a certain extent, there are still many problems in handling complex scenarios and non-standard parking spaces. Summary of the Invention

[0006] In order to solve the above technical problems, the present application provides a parking space reconstruction method, an automatic parking method, a system and a mobile communication device that can be used in various parking scenarios and can improve parking accuracy.

[0007] Specifically, the present application provides a parking space reconstruction method. The target vehicle user's mobile communication device is equipped with an application for parking space reconstruction. The parking space reconstruction method specifically includes:

[0008] Based on the application, parking space image information of the target parking space is collected; feature matching is performed based on the parking space image information to obtain matching feature pairs, and triangulation is performed based on the matching feature pairs to generate a dense point cloud based on the matching feature pairs after triangulation; and a spatial structure model of the target parking space is generated based on the dense point cloud.

[0009] In the above technical solution, parking space images are collected through mobile communication devices and the spatial structure of the parking space is generated. This overcomes the limitations of relying solely on the vehicle's own sensors in scenarios such as narrow parking spaces and protruding objects, and provides more comprehensive parking space information, which helps ensure that the automatic parking system plans the parking path more accurately and reduces subsequent problems caused by inaccurate parking.

[0010] Furthermore, the parking space image information includes at least parking space boundary information, parking space ground information and parking space surrounding environment information; the collection of parking space image information of the target parking space includes: calling the camera of the mobile communication device through the application to collect parking space image information through the camera; wherein, in the process of collecting parking space image information, the application outputs a collection position prompt in real time according to the current collection scene, so that the target vehicle user moves the collection direction of the camera according to the collection position prompt.

[0011] In the above technical solution, multi-directional image information is collected by the camera of the mobile communication device, which can intuitively display the actual situation of the parking space and help the automatic parking system to accurately plan the parking path; during the collection process, the application outputs the collection position prompt, so that the collected image can cover all key positions of the parking space, avoiding information loss and further enriching the integrity of the parking space information; in addition, the user does not need to learn additional complex operating skills, but only needs to move the camera according to the prompts of the application. The operation is simple and intuitive, which lowers the user threshold; at the same time, the mobile communication device is portable, and the user can collect parking space images at any time and place required, without being restricted by the vehicle location and sensor installation situation, making parking space reconstruction more flexible and convenient, and able to adapt to various different parking scenarios.

[0012] Furthermore, after completing the acquisition of parking space image information, it also includes: dedistorting the parking space image information based on the camera intrinsic parameter matrix, and performing feature extraction based on the dedistorted parking space image information through a preset feature extraction algorithm to generate a feature descriptor set.

[0013] In the above technical solution, camera lenses inevitably have certain optical errors during the manufacturing process. By dedistorting the parking space image, these distortion effects can be eliminated, allowing the image to more accurately reflect the true shape and size of the parking space, providing a more reliable data basis for subsequent feature matching and parking space reconstruction; through feature extraction, key information in the image can be extracted, reducing data dimensions, reducing the amount of calculation, and improving the efficiency of subsequent processing.

[0014] Furthermore, the generating of the dense point cloud includes: screening matching feature pairs based on the feature descriptor set, and calculating the basic matrix based on the matching feature pairs; calculating the essential matrix based on the basic matrix and the camera intrinsic parameter matrix, and decomposing the essential matrix into a rotation matrix and a translation vector.

[0015] In the above technical solution, by screening matching feature pairs based on a feature descriptor set, truly corresponding feature point pairs can be selected from numerous features, and erroneous or inaccurate matches can be removed, thereby improving the accuracy and reliability of feature matching; by calculating the basic matrix, the matching feature pairs can be further verified and optimized, making subsequent three-dimensional reconstruction more accurate; and by decomposing the essential matrix, the relative position and posture changes of the parking space image can be clarified.

[0016] Furthermore, the generating of the dense point cloud also includes: calculating the three-dimensional point coordinates based on the triangulation principle according to the rotation matrix, the translation vector and the matching feature pair, so as to generate an initial sparse point cloud based on the three-dimensional point coordinates.

[0017] In the above technical solution, the principle of triangulation is to use different perspectives and matching feature pairs to calculate the coordinates of three-dimensional points through geometric relationships; the initial sparse point cloud is composed of these three-dimensional points, which is a preliminary three-dimensional representation of the spatial structure of the parking space and provides a basic framework for the subsequent generation of more detailed dense point clouds; when generating the initial sparse point cloud, only the three-dimensional points corresponding to the matching feature pairs are considered, and the amount of data is relatively small, which to a certain extent reduces the computational complexity and improves the computational efficiency, while also being able to quickly obtain the approximate three-dimensional structure of the parking space.

[0018] Furthermore, the generating of the dense point cloud further includes: generating a dense point cloud according to the basic matrix, the essential matrix and the initial sparse point cloud.

[0019] In the above technical solution, the dense point cloud contains more three-dimensional points than the initial sparse point cloud, and can describe the spatial structure of the parking space in more detail; by combining the basic matrix, the essential matrix and the initial sparse point cloud, the initial sparse point cloud can be interpolated and refined using more information in the image, thereby generating a denser point cloud and providing richer three-dimensional information about the parking space; the dense point cloud can more accurately reflect the true shape and details of the parking space, such as the boundaries of the parking space and the flatness of the ground, which helps the automatic parking system to plan the parking path more accurately and improve the accuracy and safety of parking.

[0020] Furthermore, after generating the target parking space spatial structure model, the method further includes: transmitting the target parking space spatial structure model to the automatic parking system of the target vehicle through the application.

[0021] In the above technical solution, the target parking space spatial structure model includes detailed information such as parking space boundaries, ground conditions, and surrounding environment. After receiving this complete data, the automatic parking system can plan the parking path more accurately. In real parking lots, there are various complex parking scenarios, such as narrow parking spaces, irregular parking spaces, and surrounding obstacles. Detailed parking space spatial structure information can help the automatic parking system better deal with these complex situations.

[0022] Furthermore, based on the same concept, the present application also provides an automatic parking method, comprising:

[0023] A target parking space spatial structure model is obtained, and path planning is performed according to the target parking space spatial structure model, so as to execute automatic parking based on the path planning result; wherein the target parking space spatial structure model is obtained by the parking space reconstruction method.

[0024] In the above-mentioned technical solution, the spatial structure model of the target parking space is obtained through a parking space reconstruction method, which includes detailed information such as parking space boundaries, ground conditions, and surrounding environment. The automatic parking system performs path planning based on such comprehensive and accurate information, and can fully consider various actual situations, such as narrow parking spaces, irregular parking space shapes, and the location of surrounding obstacles. Compared with traditional path planning methods that only rely on the limited information of the vehicle's own sensors, it can plan a more reasonable and accurate parking path, greatly improving the success rate and accuracy of parking.

[0025] Furthermore, based on the same concept, the present application also provides an automatic parking system, comprising a memory and a processor; the memory stores computer instructions for the automatic parking method, and the processor is used to execute the computer instructions stored in the memory.

[0026] In the above technical solution, the automatic parking system has good compatibility and can be adapted to vehicles of different models and brands. As long as the vehicle has the corresponding hardware interface and communication capabilities, the system can be installed and used, providing automakers with a universal automatic parking solution and reducing R&D costs and cycles.

[0027] Furthermore, based on the same concept, the present application also provides a mobile communication device equipped with an application for parking space reconstruction, wherein the application adopts the parking space reconstruction method to obtain a target parking space spatial structure model.

[0028] Compared with the prior art, the present invention has the following advantages:

[0029] This application innovatively uses mobile communication devices to collect parking space image information, and then constructs a precise parking space structure model, breaking the traditional reliance on the vehicle's own sensors. It demonstrates strong advantages in extremely challenging scenarios such as narrow parking spaces and protruding objects. The comprehensive and detailed parking space information is like injecting intelligent eyes into the automatic parking system, enabling it to plan parking paths with unprecedented accuracy, fundamentally reducing a series of subsequent problems caused by inaccurate parking. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a flow chart of the parking space reconstruction method described in this application.

[0031] Figure 2 This is a schematic diagram of the parking space collection path described in this application.

[0032] Figure 3 This is a flow chart of the automatic parking method described in this application.

[0033] Figure 4 This is a framework diagram of the automatic parking system described in this application.

[0034] Figure 5 This is a framework diagram of the mobile communication device described in this application. DETAILED DESCRIPTION

[0035] The following is a further detailed description of a parking space reconstruction method, automatic parking method, system and mobile communication device of the present application in conjunction with specific embodiments and drawings.

[0036] See Figure 1 The present application provides a parking space reconstruction method. The target vehicle user's mobile communication device is equipped with an application for parking space reconstruction. The parking space reconstruction method specifically includes the following steps S100-S300.

[0037] Step S100: collecting parking space image information of a target parking space based on the application.

[0038] Step S200: performing feature matching according to the parking space image information to obtain matching feature pairs, and performing triangulation based on the matching feature pairs to generate a dense point cloud according to the matching feature pairs after triangulation.

[0039] Step S300: generating a target parking space spatial structure model according to the dense point cloud.

[0040] In some embodiments, the user opens the parking space reconstruction APP in the mobile phone, and collects parking space image information of the target parking space through the collection path prompt output by the APP. The parking space image information covers the parking space boundary, ground features and obstacles; the APP is also equipped with an SfM (Structure from Motion) algorithm, which uses the SfM algorithm to perform feature extraction, feature matching and triangulation on the collected parking space image information, generating an initial sparse point cloud, and then further generating a dense point cloud through a multi-view stereo (MVS) algorithm, and then outputting the target parking space spatial structure model to the vehicle end based on the generated dense point cloud.

[0041] Next, steps S100-S300 are described in detail.

[0042] The parking space image information includes at least parking space boundary information, parking space ground information and parking space surrounding environment information; the parking space image information of the target parking space in the step S100 includes: calling the camera of the mobile communication device through the application to collect parking space image information through the camera; wherein, in the process of collecting parking space image information, the application outputs a collection position prompt in real time according to the current collection scene, so that the target vehicle user moves the collection direction of the camera according to the collection position prompt.

[0043] In some embodiments, the user opens a parking space reconstruction APP in a mobile communication device (such as a mobile phone) to call the camera for continuous shooting, and follows the collection position prompt output by the APP, such as Figure 2 The dotted path shown starts from the parking space entrance to capture the corner points of the entrance, then enters the parking space, and captures objects around the parking space along the parking space line until the last corner point of the parking space entrance is captured; the APP requires at least a preset frame (such as 100 frames) of parking space image information to be captured, and the shooting range covers the obstacle surface up to a preset meter (such as 2 meters) near the four corner points of the parking space; the user uses a mobile communication device to capture a multi-view image sequence of the parking space, covering the parking space boundary (i.e., parking space boundary information), ground features (i.e., parking space ground information) and surrounding obstacles (i.e., parking space surrounding environment information).

[0044] It should be noted that the shooting frame number and shooting range can be set by those skilled in the art and are not limited thereto.

[0045] In the above technical solution, multi-directional image information is collected by the camera of the mobile communication device, which can intuitively display the actual situation of the parking space and help the automatic parking system to accurately plan the parking path; during the collection process, the application outputs the collection position prompt, so that the collected image can cover all key positions of the parking space, avoiding information loss and further enriching the integrity of the parking space information; in addition, the user does not need to learn additional complex operating skills, but only needs to move the camera according to the prompts of the application. The operation is simple and intuitive, which lowers the user threshold; at the same time, the mobile communication device is portable, and the user can collect parking space images at any time and place required, without being restricted by the vehicle location and sensor installation situation, making parking space reconstruction more flexible and convenient, and able to adapt to various different parking scenarios.

[0046] Furthermore, after completing the acquisition of parking space image information, step S100 also includes: dedistorting the parking space image information based on the camera intrinsic parameter matrix, and performing feature extraction based on the dedistorted parking space image information through a preset feature extraction algorithm to generate a feature descriptor set.

[0047] In some embodiments, the camera intrinsic parameter matrix is ​​as follows: The preset feature extraction algorithm is, for example, SIFT (Scale-Invariant Feature Transform) or ORB (Oriented FAST and Rotated BRIEF).

[0048] Among them, SIFT mainly constructs the scale space through the Gaussian difference function, and searches for extreme points in the image at different scales. These extreme points may be corner points, edge points and other points with significant features in the image; accurately determines the position and scale of key points by fitting a three-dimensional quadratic function, and removes key points with low contrast and edge response to improve the stability of the features; assigns one or more directions to each key point, calculates the gradient direction histogram in the neighborhood of the key point, and takes the direction corresponding to the peak in the histogram as the main direction of the key point, so that the feature has rotation invariance; in the neighborhood around the key point, rotates the sampling area according to the main direction and divides it into multiple sub-areas, calculates the gradient direction histogram of each sub-area, and combines these histograms into a 128-dimensional feature vector as the descriptor of the key point, and then obtains the feature descriptor set F = {f i |f i ∈R d}.

[0049] ORB uses the FAST (Features from Accelerated Segment Test) algorithm to detect corners in an image. The FAST algorithm quickly determines whether a pixel is a corner by comparing the brightness of the pixel with its neighboring pixels. It performs non-maximum suppression on the detected key points to remove redundant key points, and uses the Harris corner response function to sort the key points and select key points with higher response values. It calculates the main direction of each key point, and determines the center of mass position by calculating the image moment in the neighborhood of the key point, thereby obtaining the direction of the key point. It uses the BRIEF (Binary Robust Independent Elementary Features) algorithm to generate feature descriptors. BRIEF is a binary descriptor that randomly selects pixel pairs in the neighborhood of the key point for comparison to generate a binary string as a feature descriptor, and then obtains the feature descriptor subset F = {f i |f i ∈R d}.

[0050] In the above technical solution, camera lenses inevitably have certain optical errors during the manufacturing process. By dedistorting the parking space image, these distortion effects can be eliminated, allowing the image to more accurately reflect the true shape and size of the parking space, providing a more reliable data basis for subsequent feature matching and parking space reconstruction; through feature extraction, key information in the image can be extracted, reducing data dimensions, reducing the amount of calculation, and improving the efficiency of subsequent processing.

[0051] Furthermore, the step S200 includes: screening matching feature pairs based on the feature descriptor set, and calculating a basic matrix based on the matching feature pairs; calculating an essential matrix based on the basic matrix and the camera intrinsic parameter matrix, and decomposing the essential matrix into a rotation matrix and a translation vector.

[0052] In some embodiments, the app is equipped with a SfM algorithm that extracts, matches, and analyzes feature points in an image to reconstruct the 3D spatial structure of the parking space. The algorithm can identify features such as the parking space's boundaries and corners in the image, and calculate the position of these features in three-dimensional space through triangulation, thereby constructing a 3D model of the parking space. Specifically:

[0053] First, the RANSAC algorithm is used to filter matching feature pairs based on the feature descriptor set, and then the basic matrix F and the essential matrix E are calculated to decompose the camera relative pose (R, t); where E = K -T FK -1 , E=[t]×R, [t]× is the skew-symmetric matrix of the translation vector t, and R is the rotation matrix.

[0054] In the above technical solution, by screening matching feature pairs based on a feature descriptor set, truly corresponding feature point pairs can be selected from numerous features, and erroneous or inaccurate matches can be removed, thereby improving the accuracy and reliability of feature matching; by calculating the basic matrix, the matching feature pairs can be further verified and optimized, making subsequent three-dimensional reconstruction more accurate; and by decomposing the essential matrix, the relative position and posture changes of the parking space image can be clarified.

[0055] Furthermore, the step S200 further includes: calculating three-dimensional point coordinates based on the principle of triangulation according to the rotation matrix, the translation vector and the matching feature pair, so as to generate an initial sparse point cloud based on the three-dimensional point coordinates.

[0056] In some embodiments, the matching feature points are triangulated to generate an initial sparse point cloud P sparse ={P j ∣P j =(X j ,Y j ,Z j )}; where (X j ,Y j ,Z j ) are the three-dimensional point coordinates.

[0057] In the above technical solution, the principle of triangulation is to use different perspectives and matching feature pairs to calculate the coordinates of three-dimensional points through geometric relationships; the initial sparse point cloud is composed of these three-dimensional points, which is a preliminary three-dimensional representation of the spatial structure of the parking space and provides a basic framework for the subsequent generation of more detailed dense point clouds; when generating the initial sparse point cloud, only the three-dimensional points corresponding to the matching feature pairs are considered, and the amount of data is relatively small, which to a certain extent reduces the computational complexity and improves the computational efficiency, while also being able to quickly obtain the approximate three-dimensional structure of the parking space.

[0058] Furthermore, the step S200 further includes: generating a dense point cloud according to the basic matrix, the essential matrix and the initial sparse point cloud.

[0059] In some embodiments, multi-view stereo (MVS) is used to generate a dense point cloud. The MSV algorithm uses multi-view image information to perform feature matching, depth estimation and other operations between different images to add more three-dimensional points on the basis of the initial sparse point cloud, thereby obtaining a dense point cloud.

[0060] In the above technical solution, the dense point cloud contains more three-dimensional points than the initial sparse point cloud, and can describe the spatial structure of the parking space in more detail; by combining the basic matrix, the essential matrix and the initial sparse point cloud, the initial sparse point cloud can be interpolated and refined using more information in the image, thereby generating a denser point cloud and providing richer three-dimensional information about the parking space; the dense point cloud can more accurately reflect the true shape and details of the parking space, such as the boundaries of the parking space and the flatness of the ground, which helps the automatic parking system to plan the parking path more accurately and improve the accuracy and safety of parking.

[0061] In summary, this application uses mobile communication devices to collect parking space images and generate the spatial structure of the parking space, overcoming the limitations of relying solely on the vehicle's own sensors in scenarios such as narrow parking spaces and protruding objects. It also provides more comprehensive parking space information, which helps ensure that the automatic parking system plans the parking path more accurately and reduces subsequent problems caused by inaccurate parking.

[0062] Furthermore, after generating the target parking space spatial structure model, step S300 further includes: transmitting the target parking space spatial structure model to the automatic parking system of the target vehicle through the application.

[0063] In some embodiments, the mobile communication device and the vehicle terminal can realize the transmission of the target parking space spatial structure model through Bluetooth, WIFI, NFC, etc.

[0064] In the above technical solution, the target parking space spatial structure model includes detailed information such as parking space boundaries, ground conditions, and surrounding environment. After receiving this complete data, the automatic parking system can plan the parking path more accurately. In real parking lots, there are various complex parking scenarios, such as narrow parking spaces, irregular parking spaces, and surrounding obstacles. Detailed parking space spatial structure information can help the automatic parking system better deal with these complex situations.

[0065] Further, based on the same concept, see Figure 3 , the present application also provides an automatic parking method, comprising:

[0066] A target parking space spatial structure model is obtained, and path planning is performed according to the target parking space spatial structure model, so as to execute automatic parking based on the path planning result; wherein the target parking space spatial structure model is obtained by the parking space reconstruction method.

[0067] In some embodiments, based on the spatial structure model of the target parking space, the angle and distance required for the vehicle to enter the space are accurately calculated to avoid collisions with surrounding obstacles. If there are protruding objects, the location of the objects is accurately identified and the parking path is adjusted. Path planning can be implemented using any existing path planning algorithm and is not limited here.

[0068] In the above-mentioned technical solution, the spatial structure model of the target parking space is obtained through a parking space reconstruction method, which includes detailed information such as parking space boundaries, ground conditions, and surrounding environment. The automatic parking system performs path planning based on such comprehensive and accurate information, and can fully consider various actual situations, such as narrow parking spaces, irregular parking space shapes, and the location of surrounding obstacles. Compared with traditional path planning methods that only rely on the limited information of the vehicle's own sensors, it can plan a more reasonable and accurate parking path, greatly improving the success rate and accuracy of parking.

[0069] Further, based on the same concept, see Figure 4 The present application also provides an automatic parking system, comprising a memory and a processor; the memory stores computer instructions for the automatic parking method, and the processor is used to execute the computer instructions stored in the memory.

[0070] In some embodiments, the processor executes computer instructions corresponding to the automatic parking method in the memory to obtain an optimal parking path, and then controls the vehicle's steering actuator, power actuator and / or braking actuator so that the vehicle completes parking based on the optimal parking path.

[0071] In the above technical solution, the automatic parking system has good compatibility and can be adapted to vehicles of different models and brands. As long as the vehicle has the corresponding hardware interface and communication capabilities, the system can be installed and used, providing automakers with a universal automatic parking solution and reducing R&D costs and cycles.

[0072] Further, based on the same concept, see Figure 5 The present application also provides a mobile communication device equipped with an application for parking space reconstruction, wherein the application adopts the parking space reconstruction method to obtain a target parking space spatial structure model.

[0073] Although example embodiments have been described herein with reference to the accompanying drawings, it should be understood that the above example embodiments are merely illustrative and are not intended to limit the scope of the present application. Various changes and modifications may be made therein by those skilled in the art without departing from the scope and spirit of the present application. All such changes and modifications are intended to be included within the scope of the present application as required by the appended claims.

[0074] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0075] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units described is merely a logical function division. In actual implementation, other division methods may be used, such as combining or integrating multiple units or components into another device, or ignoring or not performing some features.

[0076] The various component embodiments of the present application can be implemented in hardware, or in a software module running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that a microprocessor or digital signal processor (DSP) can be used in practice to implement some or all of the functions of some modules according to the embodiments of the present application. The application can also be implemented as a part or all of a device program (e.g., a computer program and a computer program product) for performing the method described herein. Such a program implementing the present application can be stored on a computer-readable medium, or can have the form of one or more signals. Such a signal can be downloaded from an Internet website, or provided on a carrier signal, or provided in any other form.

[0077] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply the existence of any such actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element.

[0078] Although the present application is described in conjunction with the above specific embodiments, it is obvious that those skilled in the art can make many substitutions, modifications and variations based on the above content. Therefore, all such substitutions, improvements and variations are included in the spirit and scope of the appended claims.

Claims

1. A parking space reconstruction method, characterized in that: The target vehicle user's mobile communication device is equipped with an application for parking space reconstruction. The parking space reconstruction method specifically includes: collecting parking space image information of the target parking space based on the application; Performing feature matching based on the parking space image information to obtain matching feature pairs, and performing triangulation based on the matching feature pairs to generate a dense point cloud based on the triangulated matching feature pairs; And, generating a target parking space spatial structure model according to the dense point cloud.

2. The parking space reconstruction method according to claim 1, characterized in that: The parking space image information at least includes parking space boundary information, parking space ground information and parking space surrounding environment information; The collecting of parking space image information of the target parking space includes: calling the camera of the mobile communication device through the application to collect parking space image information through the camera; In the process of collecting parking space image information, the application program outputs a collection position prompt in real time according to the current collection scene, so that the target vehicle user moves the collection direction of the camera according to the collection position prompt.

3. The parking space reconstruction method according to claim 2, characterized in that: After completing the acquisition of parking space image information, it also includes: dedistorting the parking space image information based on the camera intrinsic parameter matrix, and extracting features based on the dedistorted parking space image information through a preset feature extraction algorithm to generate a feature descriptor set.

4. The parking space reconstruction method according to claim 3, characterized in that: Generating a dense point cloud includes: screening matching feature pairs based on the feature descriptor set, and calculating a basic matrix according to the matching feature pairs; The intrinsic matrix is ​​calculated according to the fundamental matrix and the camera intrinsic parameter matrix, and the intrinsic matrix is ​​decomposed into a rotation matrix and a translation vector.

5. The parking space reconstruction method according to claim 4, characterized in that: Generating a dense point cloud further includes: According to the rotation matrix, the translation vector and the matching feature pair, three-dimensional point coordinates are calculated based on the triangulation principle to generate an initial sparse point cloud based on the three-dimensional point coordinates.

6. The parking space reconstruction method according to claim 5, characterized in that: Generating a dense point cloud further includes: A dense point cloud is generated according to the fundamental matrix, the essential matrix and the initial sparse point cloud.

7. The parking space reconstruction method according to claim 1, characterized in that: After generating the target parking space structure model, it also includes: The target parking space spatial structure model is transmitted to the automatic parking system of the target vehicle through the application.

8. An automatic parking method, characterized in that: include: Obtaining a target parking space spatial structure model, and performing path planning according to the target parking space spatial structure model, so as to execute automatic parking based on the path planning result; Wherein, the target parking space spatial structure model is obtained by the parking space reconstruction method according to any one of claims 1-6.

9. An automatic parking system, characterized in that: The method comprises a memory and a processor; the memory stores computer instructions of the automatic parking method according to claim 8, and the processor is used to execute the computer instructions stored in the memory.

10. A mobile communication device, characterized in that: The utility model is equipped with an application for parking space reconstruction, wherein the application adopts the parking space reconstruction method according to any one of claims 1 to 6 to obtain a target parking space spatial structure model.

Images