CN109949365B - Vehicle designated position parking method and system based on road surface feature points - Google Patents
Vehicle designated position parking method and system based on road surface feature points Download PDFInfo
- Publication number
- CN109949365B CN109949365B CN201910154967.8A CN201910154967A CN109949365B CN 109949365 B CN109949365 B CN 109949365B CN 201910154967 A CN201910154967 A CN 201910154967A CN 109949365 B CN109949365 B CN 109949365B
- Authority
- CN
- China
- Prior art keywords
- point
- vehicle
- points
- parking
- camera
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a vehicle designated position parking method and system based on road surface characteristic points, which calculate a homography matrix from a ground plane to a camera pixel plane of each parking position according to a chessboard diagram, collect image reference characteristic points containing road signs near the corresponding vehicle positions, and then make and construct a sample library by combining various information of the parking positions; screening and matching current characteristic points of the real-time image with corresponding characteristic points in a sample library; the method comprises the steps of calculating the relative position relation between the current position of a camera and an image reference point by combining the current feature points obtained by screening and matching with a homography matrix from a ground plane to an image plane, and finally adding the obtained relative position relation to a known GPS landmark position point in a sample library to obtain an absolute position point of a current vehicle, so that the absolute position of the current vehicle is obtained by calculating the relative relation with the landmark by using vision, accurate parking is realized, the process of constructing the sample library is greatly simplified during image matching, and the operability is improved.
Description
Technical Field
The invention relates to a vehicle designated position parking method and system based on road surface feature points, and relates to the technical field of intersection of image recognition and machine vision.
Background
In intelligent driving, vehicle position information represents the position state of the current vehicle on the road or in the environment, and is an important reference for answering "where the vehicle is". The high-precision position information is divided into absolute position information and relative position information. The absolute Position information is generally obtained by a Global Positioning System (GPS), and is expressed by using longitude and latitude elevation information with a certain point as a reference origin. The method for acquiring the absolute position information through the GPS is an important way for intelligent driving, and generally, the higher the position accuracy of the GPS is, the higher the selling price is. The major disadvantage of the GPS is that the condition of star loss exists in scenes such as long tunnels, large overpasses, urban forests and the like, and the measurement error jumps sharply and can reach several meters or even dozens of meters generally. The relative position information calculates the relative relation between the common road attribute and a certain point of the vehicle, thereby obtaining the relative position of the vehicle and the road sign. A common approach uses a sensor (e.g., a millimeter wave radar, a camera, etc.) to detect information on the relative distance between a road sign (e.g., a lane line, a vehicle, a street lamp, a guideboard, etc.) and a vehicle, thereby estimating the positional relationship between the vehicle and the road sign. Absolute and relative position complementation has been a hotspot of research in recent years. Some road signs fixed on the road surface can utilize high-precision GPS equipment to collect the absolute positions of the road signs in advance, and the real-time relative relationship between the vehicle and the road signs can be calculated according to the absolute position relationship of the road signs, for example, a camera collects images in real time to detect the road signs and then performs fusion matching with a high-precision map, and a high-precision Inertial Measurement Unit (IMU) and the camera are fused, etc.
When a camera is used to designate a position on a road surface, ground landmarks or features are usually detected, and template matching or deep learning is mainly used. The template matching mainly utilizes a small sample library (relative to deep learning), common features in the sample library are extracted to form a template, then target classification is carried out, and the speed is generally high; deep learning mainly utilizes a large sample library, sets a multilayer neural network, directly outputs classification results, and is generally slow. Both methods involve the preparation of a sample library. At present, a lot of free high-quality data sets even labeled exist on public materials, but the final effect is greatly reduced in actual projects according to different types of selection of customers and cameras, and therefore, a project sample library is indispensable to construct. When a sample library is manufactured, a required target object is found by manually acquiring data (videos or images) in advance according to each frame or each image, and the target object is manually marked or cut to be in a proper size and stored. The mode of relying on manual production has the disadvantages of complicated operation, low efficiency, no guarantee of quality and very high labor cost.
Disclosure of Invention
In view of this, the invention provides a method and a system for parking at a specified position of a vehicle based on road surface feature points, which can simplify the process of constructing a sample library and improve the operability.
A vehicle specified-position parking method based on road surface feature points comprises the following steps:
s1, collecting images containing nearby road signs at each parking position, extracting reference feature points, calculating a homography matrix from a ground plane of each parking position to a pixel plane of a vehicle-mounted camera according to a chessboard diagram, and combining various information of each parking position to make and construct a sample library;
s2, the vehicle-mounted system receives the input preset serial number of the parking position and acquires corresponding parking position information from a sample library according to the preset serial number of the parking position;
s3, collecting an image acquired by a vehicle-mounted camera in real time, detecting and extracting a current feature point in the real-time image, calculating a homography matrix of a real-time frame and a sample frame, and screening and matching the current feature point of the real-time image and a reference feature point corresponding to the parking position in a sample library;
s4, calculating the relative position relation between the current position of the camera and the image reference point by combining the current feature points obtained by screening and matching with a homography matrix from the ground plane to the image plane;
and S5, adding the obtained relative position relation and the relative distance from the reference origin of the chessboard pattern to the known GPS landmark position point and the known GPS landmark position point in the sample library to obtain the absolute position point of the current vehicle, wherein the vehicle-mounted system realizes accurate parking according to the absolute position point of the current vehicle.
A vehicle specified position parking system based on road surface characteristic points comprises the following functional modules:
the system comprises a sample library establishing module, a characteristic analysis module and a characteristic analysis module, wherein the sample library establishing module is used for acquiring images containing nearby road signs at each parking position, extracting reference characteristic points, calculating a homography matrix from a ground plane of each parking position to a pixel plane of a vehicle-mounted camera according to a chessboard diagram, and combining various information of each parking position to make and establish a sample library;
the information acquisition module is used for receiving the input preset serial number of the parking position by the vehicle-mounted system and acquiring corresponding parking position information from the sample library according to the preset serial number of the parking position;
the characteristic point screening and matching module is used for acquiring images acquired by the vehicle-mounted camera in real time, detecting and extracting current characteristic points in the real-time images, calculating homography matrixes of the real-time frames and the sample frames, and screening and matching the current characteristic points of the real-time images and reference characteristic points corresponding to the parking positions in the sample library;
the relative position calculation module is used for calculating the relative position relation between the current position of the camera and the image reference point by combining the current feature points obtained by screening and matching with a homography matrix from the ground plane to the image plane;
and the absolute position parking module is used for adding the obtained relative position relation and the relative distance from the chessboard pattern reference origin to the known GPS landmark position point in the sample library to obtain the absolute position point of the current vehicle, and the vehicle-mounted system realizes accurate parking according to the absolute position point of the current vehicle.
The invention relates to a vehicle designated position parking method and system based on road surface characteristic points, which is characterized in that a homography matrix from a ground plane to a camera pixel plane of each parking position is calculated according to a chessboard diagram, an image containing nearby road signs is collected at each parking position, reference characteristic points are extracted, and then a sample library is manufactured and constructed by combining various information of each parking position; screening and matching the current characteristic point of the real-time image with a reference characteristic point corresponding to the parking position in a sample library; and combining the current feature points obtained by screening and matching with a homography matrix from a ground plane to an image plane to calculate the relative position relationship between the current position of the camera and the image reference point, and finally adding the obtained relative position relationship to the known GPS landmark position points in the sample library to obtain the absolute position points of the current vehicle, wherein the absolute position points of the current vehicle are accurately parked by the vehicle-mounted system. By adopting the vehicle specified position parking method based on the road surface characteristic points, only a plurality of corresponding images are required to be collected at the parking position, then the absolute position of the current vehicle is obtained by calculating the relative relation between the current vehicle and the road sign by using vision, the accurate parking is realized, the process of constructing a sample library is greatly simplified during image matching, and the operability is improved.
Drawings
FIG. 1 is a flow chart of a method for parking a vehicle at a specified position based on road surface feature points according to the present invention;
FIG. 2 is a schematic diagram of the steps of a vehicle parking method based on road surface feature points according to the present invention;
FIG. 3 is a block diagram of the step S1 in FIG. 1;
FIG. 4 is a block diagram of step S3 of FIG. 1;
FIG. 5 is a block diagram of step S4 of FIG. 1;
FIG. 6 is a specific application diagram of functional modules of the vehicle parking system at a specified position based on road surface feature points;
FIG. 7 is a block diagram of the elements of the sample library building block of the present invention;
FIG. 8 is a block diagram of the elements of the feature point filtering and matching module according to the present invention;
FIG. 9 is a block diagram of the elements of the relative position calculation module of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to the accompanying drawings and embodiments, it being understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention.
As shown in fig. 1 and fig. 2, an embodiment of the present invention provides a method for parking a vehicle at a specified position based on road surface feature points, where the method for parking a vehicle at a specified position based on road surface feature points includes the following steps:
s1, collecting images containing nearby road signs at each parking position, extracting reference feature points, calculating a homography matrix from a ground plane of each parking position to a pixel plane of a vehicle-mounted camera according to a chessboard diagram, and making and constructing a sample library by combining various information of each parking position.
As shown in fig. 3, the step S1 includes the following sub-steps:
s11, calibrating an internal reference matrix of the vehicle-mounted camera;
s12, shooting an image containing a road sign and an image of a chessboard pattern placed on the ground at each parking position by using a vehicle-mounted camera;
s13, extracting pixel angular points between the physical angular points of the ground chessboard pattern and the cameras from the images of the chessboard pattern placed on the ground, and calculating a homography matrix from the ground plane of each parking position to the pixel plane of the vehicle-mounted camera according to the chessboard pattern by utilizing the pixel angular points between the physical angular points of the ground chessboard pattern and the cameras;
s14, extracting reference feature points and feature descriptors from road sign-containing images shot by a vehicle-mounted camera;
and S15, taking the preset number, the image, the reference feature point, the feature descriptor, the known actual GPS road sign position point and the deviation record of the road sign position point and the chessboard pattern reference origin of each parking position as the map attributes of the sample library, and finishing the map construction of the sample library.
The landmarks comprise L-shaped marks, box numbers, special marked lines and the like, stop points at different positions are distinguished according to ground marks, and the marks are provided with independent marks. When a vehicle-mounted camera is used for shooting the ground on a vehicle, the closer the vehicle-mounted camera to the ground is, the better the effect of detecting ground characteristic points is, and the calculation accuracy of the vertical shooting position of the camera is higher, but in practice, the camera is installed too low, the camera is shaken violently due to shaking of a vehicle body and potholes on a road surface, meanwhile, when the vehicle is used for shooting the ground vertically, the camera is too low, the available field range used by the camera is too small, when the vehicle runs, the repeated area of two continuous frames is too small, the matching difficulty is increased, and the algorithm processing speed is required to be increased. Therefore, the size of θ may be slightly inclined so as to cover the ground completely with the main field of view of the camera in practice.
Specifically, the parking points at different parking positions are marked with different numbers, denoted as Pi (i is 1,2,3,4 \8230);, and the attributes of the parking positions are mainly represented by the parking position numbers, the Landmark GPS coordinates, the physical offsets (x, y) from the Landmark, the image block Patch feature points, and the image block Patch feature descriptors, denoted as P (position numbers, landmark GPS, patch feature points, patch descriptors). The sequence of parking positions for the sample library is denoted by S, and the entire sample library is composed of a plurality of P points with sequences, denoted by S { P1, \8230;, pi, \8230; }.
For example, in actual operation, the vehicle stops at a position stop point of position number 013, and the Landmark GPS point at this time is noted. Then shooting two images in the image scene, wherein one image is an image simultaneously containing a road sign and a chessboard pattern and an image with the same angle but only containing the road sign, setting a first angular point at the position of the upper left corner of the chessboard pattern as a chessboard pattern reference original point, and calculating to obtain a homography matrix from the ground plane to the camera pixel plane according to the one-to-one mapping relation between the angular point of the camera pixel plane and the physical angular points of the chessboard pattern in the actual physical plane; and extracting and storing the characteristic points and the characteristic descriptors in the images containing the checkerboard images. After the homography matrix from the ground plane to the camera pixel plane is obtained through calculation and the feature points and the feature descriptors are extracted, the image containing the chessboard pattern can be deleted, only the image containing the road signs is stored in the sample library and is recorded as a training frame, and therefore the information production of the parking space in the sample library is completed. And respectively collecting the relevant information of each parking point, and manufacturing a finished sample library. Preferably, the XY coordinate system of the chessboard pattern and the coordinate system of the road sign are arranged in parallel, the chessboard pattern is printed by a high-precision laser printer, and the physical size of the small grid is measured in advance.
And S2, the vehicle-mounted system receives the input preset serial number of the parking position and acquires corresponding parking position information from the sample library according to the preset serial number of the parking position.
That is, when the user needs to park the vehicle at the parking spot numbered 013, the number of the parking spot is first input into the in-vehicle system: 013, the vehicle system then extracts the parking position information relating to the parking spot number 013 from the sample library.
And S3, acquiring an image acquired by the vehicle-mounted camera in real time, detecting and extracting a current feature point in the real-time image, calculating a homography matrix of the real-time frame and the sample frame, and screening and matching the current feature point of the real-time image and a reference feature point corresponding to the parking position in the sample library.
As shown in fig. 4, the step S3 includes the following sub-steps:
s31, acquiring an image acquired by a vehicle-mounted camera in real time, detecting and extracting a current feature point in the real-time image, and calculating a homography matrix of a real-time frame and a sample frame;
s32, screening the current characteristic point of the real-time image and a reference characteristic point corresponding to the parking position in a sample library by using a Hessian threshold value and a characteristic point distance;
and S33, performing consistency matching on the interior points of the feature points by using a RANSAC algorithm, and filtering irrelevant exterior points.
In the process of backing up, the vehicle-mounted camera starts to acquire real-time images, then a common image feature detection algorithm is adopted to extract feature points and feature descriptors in the real-time acquired images, and if the feature points are not detected, the next frame of image is continuously detected.
The Scale-invariant feature transform (SIFT) feature detection, speedup feature detection, object FAST and related BRIEF feature detection, FAST From Acquired Segment Test (FAST) feature detection and HARRIS feature detection are supported at present.
After the feature points and the feature descriptors are detected, screening the current feature points of the real-time image and the reference feature points corresponding to the parking position in the sample library through a preset Hessian threshold value and a feature point distance, and then storing the interior points of the feature points by utilizing a RANSAC algorithm provided by OpenCV; and using RANSAC algorithm to carry out consistency matching on the interior points of the feature points.
And S4, calculating the relative position relation between the current position of the camera and the image reference point by combining the current feature point obtained by screening and matching with the homography matrix from the ground plane to the image plane.
As shown in fig. 5, the step S4 includes the following sub-steps:
s41, calculating a rotation matrix and a translation matrix of the camera position to an image reference origin point with a checkerboard diagram in a sample library by combining the current feature points obtained by screening and matching with a camera internal reference matrix and a homography matrix from a ground plane to an image plane;
s42, solving the XYZ relative relation between the camera center point and the chessboard pattern reference origin point by utilizing the geometric antipodal relation of the rotation matrix and the translation matrix, and calculating the distance between the camera center point and the chessboard pattern reference origin point according to the XYZ relative relation between the camera center point and the chessboard pattern reference origin point;
s43, adding the distance from the central point of the camera to the reference origin of the chessboard pattern and the offset from the camera to the vehicle reference point to obtain the position distance from the reference origin of the chessboard pattern to the vehicle reference point.
Specifically, the homography matrix and the camera intrinsic parameter matrix are decomposed into R and T matrices by using a decomplexie homographiymat function, and the transposed left-hand multiplication translation matrix of the matrices is XYZ coordinates, and the specific formula is as follows:
in the above formula, λ is a scale factor, fx is a focal length of the camera in the x direction, fy is a focal length of the camera in the y direction, dx is a pixel size in the x direction, dy is a pixel size in the y direction, u0 is an x coordinate of an image principal point, and v0 is a y coordinate of the image principal point; r is a 3x3 rotation matrix from the physical plane to the pixel plane, T is a 3x1 translation matrix from the physical plane to the pixel plane, and 1 is a 3x1 unit matrix; u and v are respectively an x coordinate and a y coordinate of a pixel plane, and Xw, yw and Zw are physical coordinates of a target from the center of the camera under a world coordinate system; the matrix composed of fx, dx, fy, dy, u0, and v0 is called a camera internal reference matrix.
And S5, adding the obtained relative position relation and the relative distance from the reference origin of the chessboard pattern to the known GPS landmark position point and the known GPS landmark position point in the sample library to obtain the absolute position point of the current vehicle, wherein the vehicle-mounted system realizes accurate parking according to the absolute position point of the current vehicle.
And inputting the position distance from the reference origin of the chessboard diagram to the vehicle reference point into a vehicle-mounted system for parking test, and applying the distance offset to correct in real time according to the test result, wherein the distance offset is calibrated if the parking precision meets the requirement. And finally, calculating to obtain the absolute position coordinates of the vehicle reference point position according to the distance calculated by the camera, the position distance from the chessboard pattern reference origin to the vehicle reference point and the physical offset of the Landmark in the sample library.
The invention relates to a vehicle designated position parking method based on road surface characteristic points, which is characterized in that a homography matrix from a ground plane to a camera pixel plane of each parking position is calculated according to a chessboard diagram, images containing nearby road signs are collected at each parking position, reference characteristic points are extracted, and then a sample library is manufactured and constructed by combining various information of each parking position; screening and matching the current characteristic point of the real-time image with a reference characteristic point corresponding to the parking position in a sample library; and combining the current feature points obtained by screening and matching with a homography matrix from a ground plane to an image plane to calculate the relative position relationship between the current position of the camera and the image reference point, and finally adding the obtained relative position relationship to the known GPS landmark position points in the sample library to obtain the absolute position points of the current vehicle, wherein the absolute position points of the current vehicle are accurately parked by the vehicle-mounted system. By adopting the vehicle specified position parking method based on the road surface characteristic points, only a plurality of corresponding images are required to be collected at the parking position, then the absolute position of the current vehicle is obtained by calculating the relative relation between the current vehicle and the road sign by using vision, the accurate parking is realized, the process of constructing a sample library is greatly simplified during image matching, and the operability is improved.
Based on the vehicle designated position parking method based on the road surface characteristic points, the invention also provides a vehicle designated position parking system based on the road surface characteristic points, as shown in fig. 6, the vehicle designated position parking system based on the road surface characteristic points comprises the following functional modules:
the sample library establishing module 10 is used for collecting images containing nearby road signs at each parking position, extracting reference feature points, calculating a homography matrix from a ground plane of each parking position to a pixel plane of a vehicle-mounted camera according to a chessboard diagram, and combining various information of each parking position to make and establish a sample library;
the information acquisition module 20 is used for the vehicle-mounted system to receive the input preset serial number of the parking position and acquire corresponding parking position information from the sample library according to the preset serial number of the parking position;
the feature point screening and matching module 30 is used for acquiring images acquired by the vehicle-mounted camera in real time, detecting and extracting current feature points in the real-time images, calculating homography matrixes of the real-time frames and the sample frames, and screening and matching the current feature points of the real-time images and reference feature points corresponding to the parking positions in the sample library;
the relative position calculating module 40 is used for calculating the relative position relationship between the current position of the camera and the image reference point by combining the current feature points obtained by screening and matching with a homography matrix from the ground plane to the image plane;
and the absolute position parking module 50 is used for adding the obtained relative position relationship and the relative distance from the chessboard pattern reference origin to the known GPS landmark position point in the sample library, namely the absolute position point of the current vehicle, and the vehicle-mounted system realizes accurate parking according to the absolute position point of the current vehicle.
As shown in fig. 7, the sample library establishing module 10 includes the following functional units:
the internal reference calibration unit 11 is used for calibrating an internal reference matrix of the vehicle-mounted camera;
the image acquisition unit 12 is used for shooting an image containing a road sign and an image of a chessboard pattern placed on the ground by using a vehicle-mounted camera at each parking position;
a homography matrix calculation unit 13, configured to extract pixel angular points between the physical angular points of the ground chessboard pattern and the cameras from the images of the chessboard pattern placed on the ground, and calculate a homography matrix from the ground plane of each parking position to the pixel plane of the vehicle-mounted camera according to the chessboard pattern by using the pixel angular points between the physical angular points of the ground chessboard pattern and the cameras; specifically, the matrix calculation unit is specifically configured to extract pixel angular points between the physical angular points of the ground chessboard pattern and the camera from the image in which the chessboard pattern is placed on the ground, and calculate a homography matrix from the ground plane to the camera pixel plane according to a one-to-one mapping relationship between the angular points of the camera pixel plane and the physical angular points of the chessboard pattern in the actual physical plane;
a reference feature extraction unit 14, configured to extract feature points and feature descriptors from road sign-containing images captured by a vehicle-mounted camera;
and the information normalization unit 15 is configured to use the preset number, the image, the reference feature point, the feature descriptor, the known actual GPS landmark position point, and the offset record of the landmark position point and the chessboard pattern reference origin of each parking position as the map attributes of the sample library, so as to complete the map construction of the sample library.
As shown in fig. 8, the feature point screening and matching module 30 includes the following functional units:
a current feature point extracting unit 31, configured to collect an image obtained by the vehicle-mounted camera in real time, detect and extract a current feature point in the real-time image, and calculate a homography matrix of the real-time frame and the sample frame;
a feature point screening unit 32, configured to screen a current feature point of the real-time image and a reference feature point corresponding to the parking position in the sample library by using a Hessian threshold and a feature point distance, and store an interior point of the feature point by using an RANSAC algorithm;
and a feature point matching unit 33, configured to perform consistency matching on the interior points of the feature points by using the RANSAC algorithm.
As shown in fig. 9, the relative position calculation module 40 includes the following functional units:
the RT matrix resolving unit 41 is used for combining the current feature points obtained by screening and matching with the camera internal reference matrix and the homography matrix from the ground plane to the image plane to calculate a rotation matrix and a translation matrix from the camera position to the image reference origin with the checkerboard graph in the sample library;
the coordinate calculating unit 42 is used for calculating the XYZ relative relationship between the camera center point and the chessboard pattern reference origin by using the geometric antipodal relationship between the rotation matrix and the translation matrix, and calculating the distance between the camera center point and the chessboard pattern reference origin according to the XYZ relative relationship between the camera center point and the chessboard pattern reference origin;
and a position distance calculation unit 43, configured to add the distance from the center point of the camera to the reference origin of the checkerboard pattern and the offset from the camera to the vehicle reference point, so as to obtain the position distance from the reference origin of the checkerboard pattern to the vehicle reference point.
The above apparatus embodiments and method embodiments are in one-to-one correspondence, and for the sake of simplicity, reference may be made to the method embodiments.
In the present specification, the embodiments are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other.
Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative components and steps have been described above generally in terms of their functionality in order to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module may reside in random access memory, read only memory, electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable magnetic disk, a CD-ROM, or any other form of storage medium known in the art.
While the present invention has been described with reference to the embodiments shown in the drawings, the present invention is not limited to the embodiments, which are illustrative and not restrictive, and it will be apparent to those skilled in the art that various changes and modifications can be made therein without departing from the spirit and scope of the invention as defined in the appended claims.
Claims (10)
1. The vehicle specified-position parking method based on the road surface characteristic points is characterized by comprising the following steps of:
s1, collecting images containing nearby road signs at each parking position, extracting reference feature points, calculating a homography matrix from a ground plane of each parking position to a pixel plane of a vehicle-mounted camera according to a chessboard diagram, and combining various information of each parking position to make and construct a sample library;
s2, the vehicle-mounted system receives the input preset serial number of the parking position and acquires corresponding parking position information from a sample library according to the preset serial number of the parking position;
s3, collecting an image acquired by a vehicle-mounted camera in real time, detecting and extracting a current feature point in the real-time image, calculating a homography matrix of a real-time frame and a sample frame, and screening and matching the current feature point of the real-time image and a reference feature point corresponding to the parking position in a sample library;
s4, calculating the relative position relation between the current position of the camera and the image reference point by combining the current feature points obtained by screening and matching with a homography matrix from the ground plane to the image plane;
and S5, adding the obtained relative position relation and the relative distance from the reference origin of the chessboard pattern to the known GPS landmark position point and the known GPS landmark position point in the sample library to obtain the absolute position point of the current vehicle, wherein the vehicle-mounted system realizes accurate parking according to the absolute position point of the current vehicle.
2. The method for parking a vehicle at a specified position based on road surface feature points according to claim 1, wherein the step S1 comprises the sub-steps of:
s11, calibrating an internal reference matrix of the vehicle-mounted camera;
s12, shooting an image containing a road sign and an image of a chessboard pattern placed on the ground at each parking position by using a vehicle-mounted camera;
s13, extracting pixel angular points between the physical angular points of the ground chessboard pattern and the cameras from the images of the chessboard pattern placed on the ground, and calculating a homography matrix from the ground plane of each parking position to the pixel plane of the vehicle-mounted camera according to the chessboard pattern by utilizing the pixel angular points between the physical angular points of the ground chessboard pattern and the cameras;
s14, extracting reference feature points and feature descriptors from road sign-containing images shot by a vehicle-mounted camera;
and S15, taking the preset number, the image, the reference feature point, the feature descriptor, the known actual GPS road sign position point and the deviation record of the road sign position point and the chessboard pattern reference origin of each parking position as the map attributes of the sample library, and finishing the map construction of the sample library.
3. A method for parking a vehicle at a specified position based on road surface characteristic points according to claim 1, wherein said step S3 comprises the sub-steps of:
s31, acquiring an image acquired by a vehicle-mounted camera in real time, detecting and extracting a current feature point in the real-time image, and calculating a homography matrix of a real-time frame and a sample frame;
s32, screening the current characteristic point of the real-time image and a reference characteristic point corresponding to the parking position in a sample library by using a Hessian threshold value and a characteristic point distance;
and S33, performing consistency matching on the interior points of the current feature points by using a RANSAC algorithm, and filtering irrelevant exterior points.
4. The method for parking a vehicle at a specified position based on road surface feature points as claimed in claim 1, wherein said step S4 comprises the sub-steps of:
s41, calculating a rotation matrix and a translation matrix of the camera position to an image reference origin point with a checkerboard pattern in a sample library by combining the current feature points obtained by screening and matching with a camera internal reference matrix, a homography matrix of a real-time frame and a sample frame and a homography matrix from a ground plane to an image plane;
s42, solving the XYZ relative relation between the camera center point and the chessboard pattern reference origin point by utilizing the geometric antipodal relation of the rotation matrix and the translation matrix, and calculating the distance between the camera center point and the chessboard pattern reference origin point according to the XYZ relative relation between the camera center point and the chessboard pattern reference origin point;
s43, adding the distance between the central point of the camera and the reference origin of the checkerboard pattern and the offset between the camera and the vehicle reference point to obtain the position distance between the reference origin of the checkerboard pattern and the vehicle reference point.
5. The method for parking vehicles at specified positions based on road surface feature points as claimed in claim 1, wherein the specific method for calculating the homography matrix from the ground plane of each parking position to the pixel plane of the vehicle-mounted camera according to the chessboard diagram by using the physical corner points of the chessboard diagram on the ground and the pixel corner points between the cameras is as follows:
and calculating to obtain a homography matrix from the ground plane to the camera pixel plane according to the one-to-one mapping relation between the corner points of the camera pixel plane and the physical corner points of the chessboard pattern in the actual physical plane.
6. The vehicle specified-position parking system based on the road surface characteristic points is characterized by comprising the following functional modules:
the sample library establishing module is used for collecting images containing nearby road signs at each parking position, extracting reference characteristic points, calculating a homography matrix from a ground plane of each parking position to a pixel plane of the vehicle-mounted camera according to the chessboard diagram, and combining various information of each parking position to make and establish a sample library;
the information acquisition module is used for receiving the input preset serial number of the parking position by the vehicle-mounted system and acquiring corresponding parking position information from the sample library according to the preset serial number of the parking position;
the characteristic point screening and matching module is used for acquiring images acquired by the vehicle-mounted camera in real time, detecting and extracting current characteristic points in the real-time images, calculating homography matrixes of the real-time frames and the sample frames, and screening and matching the current characteristic points of the real-time images and reference characteristic points corresponding to the parking position in the sample library;
the relative position calculation module is used for calculating the relative position relation between the current position of the camera and the image reference point by combining the current feature points obtained by screening and matching with a homography matrix from the ground plane to the image plane;
and the absolute position parking module is used for adding the obtained relative position relationship and the relative distance from the chessboard pattern reference origin to the known GPS landmark position point in the sample library to obtain the absolute position point of the current vehicle, and the vehicle-mounted system realizes accurate parking according to the absolute position point of the current vehicle.
7. The system for parking a vehicle at a specified position based on the road surface feature points as claimed in claim 6, wherein the sample library establishing module comprises the following functional units:
the internal reference calibration unit is used for calibrating an internal reference matrix of the vehicle-mounted camera;
the image acquisition unit is used for shooting an image containing a road sign and an image of a chessboard pattern placed on the ground by using a vehicle-mounted camera at each parking position;
the homography matrix calculation unit is used for extracting pixel angular points between the physical angular points of the ground chessboard pattern and the cameras from the images of the chessboard pattern placed on the ground, and calculating homography matrixes from the ground plane of each parking position to the pixel plane of the vehicle-mounted camera according to the chessboard pattern by utilizing the pixel angular points between the physical angular points of the ground chessboard pattern and the cameras;
the reference feature extraction unit is used for extracting reference feature points and feature descriptors from road sign-containing images shot by the vehicle-mounted camera;
and the information normalizing unit is used for taking the preset number, the image, the reference feature point, the feature descriptor, the known actual GPS landmark position point and the deviation record of the landmark position point and the chessboard pattern reference origin of each parking position as the map attribute of the sample library, namely completing the map construction of the sample library.
8. The system for parking a vehicle at a specified position based on the road surface feature points as claimed in claim 6, wherein the feature point screening and matching module comprises the following functional units:
the current feature point extraction unit is used for acquiring images acquired by the vehicle-mounted camera in real time, detecting and extracting current feature points in the real-time images, and calculating a homography matrix of the real-time frames and the sample frames;
the characteristic point screening unit is used for screening the current characteristic point of the real-time image and the reference characteristic point corresponding to the parking position in the sample library by utilizing a Hessian threshold value and a characteristic point distance;
and the characteristic point matching unit is used for performing consistency matching on the interior points of the characteristic points by using a RANSAC algorithm and filtering irrelevant exterior points.
9. The system according to claim 6, wherein the relative position calculating module includes the following functional units:
the RT matrix resolving unit is used for calculating a rotation matrix and a translation matrix of the camera position to an image reference origin point with a chessboard diagram in a sample library by combining the current feature points obtained by screening and matching with a camera internal parameter matrix, a homography matrix of a real-time frame and a sample frame and a homography matrix from a ground plane to an image plane;
the coordinate calculation unit is used for calculating the XYZ relative relation between the camera center point and the chessboard diagram reference origin point by using the geometric antipodal relation of the rotation matrix and the translation matrix, and calculating the distance between the camera center point and the chessboard diagram reference origin point according to the XYZ relative relation between the camera center point and the chessboard diagram reference origin point;
and the position distance calculation unit is used for adding the distance from the center point of the camera to the reference origin of the checkerboard pattern and the offset from the camera to the vehicle reference point to obtain the position distance from the reference origin of the checkerboard pattern to the vehicle reference point.
10. The system of claim 6, wherein the homography matrix calculation unit is specifically configured to extract pixel corner points between the physical corner points of the ground chessboard pattern and the cameras from the images of the chessboard pattern placed on the ground, and calculate a homography matrix from the ground plane to the camera pixel plane according to a one-to-one mapping relationship between the corner points of the camera pixel plane and the physical corner points of the chessboard pattern in the actual physical plane.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910154967.8A CN109949365B (en) | 2019-03-01 | 2019-03-01 | Vehicle designated position parking method and system based on road surface feature points |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910154967.8A CN109949365B (en) | 2019-03-01 | 2019-03-01 | Vehicle designated position parking method and system based on road surface feature points |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109949365A CN109949365A (en) | 2019-06-28 |
| CN109949365B true CN109949365B (en) | 2022-12-02 |
Family
ID=67008061
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910154967.8A Active CN109949365B (en) | 2019-03-01 | 2019-03-01 | Vehicle designated position parking method and system based on road surface feature points |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109949365B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112669377B (en) * | 2019-10-15 | 2023-09-05 | 杭州海康威视数字技术股份有限公司 | Parking space detection method and device, electronic equipment and storage medium |
| EP3809313A1 (en) * | 2019-10-16 | 2021-04-21 | Ningbo Geely Automobile Research & Development Co. Ltd. | A vehicle parking finder support system, method and computer program product for determining if a vehicle is at a reference parking location |
| CN110969655B (en) * | 2019-10-24 | 2023-08-18 | 百度在线网络技术（北京）有限公司 | Method, device, equipment, storage medium and vehicle for detecting parking space |
| CN110907955B (en) * | 2019-12-02 | 2021-02-09 | 荣讯塑胶电子制品(深圳)有限公司 | Positioning instrument fault identification system |
| CN111047904B (en) * | 2019-12-17 | 2020-11-20 | 北京科技大学 | Vehicle position information detection system and method based on tic-tac-toe calibration line |
| CN111540023B (en) * | 2020-05-15 | 2023-03-21 | 阿波罗智联(北京)科技有限公司 | Monitoring method and device of image acquisition equipment, electronic equipment and storage medium |
| CN112184818B (en) * | 2020-10-09 | 2022-06-10 | 重庆邮电大学 | Vision-based vehicle positioning method and parking lot management system applying same |
| CN113129376A (en) * | 2021-04-22 | 2021-07-16 | 青岛联合创智科技有限公司 | Checkerboard-based camera real-time positioning method |
| CN114485682B (en) * | 2021-12-30 | 2023-06-27 | 武汉光庭信息技术股份有限公司 | Positioning method based on SLAM technology |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101183427A (en) * | 2007-12-05 | 2008-05-21 | 浙江工业大学 | Computer vision based peccancy parking detector |
| CN106529587A (en) * | 2016-08-26 | 2017-03-22 | 北京航空航天大学 | Visual course identification method based on target point identification |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100982192B1 (en) * | 2008-12-09 | 2010-09-14 | 한국과학기술연구원 | A Method for Geo-tagging of Pictures and Apparatus thereof |
-
2019
- 2019-03-01 CN CN201910154967.8A patent/CN109949365B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101183427A (en) * | 2007-12-05 | 2008-05-21 | 浙江工业大学 | Computer vision based peccancy parking detector |
| CN106529587A (en) * | 2016-08-26 | 2017-03-22 | 北京航空航天大学 | Visual course identification method based on target point identification |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109949365A (en) | 2019-06-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109949365B (en) | Vehicle designated position parking method and system based on road surface feature points | |
| CN110285793B (en) | Intelligent vehicle track measuring method based on binocular stereo vision system | |
| CN110322702B (en) | Intelligent vehicle speed measuring method based on binocular stereo vision system | |
| CN111179358B (en) | Calibration method, device, equipment and storage medium | |
| CN109059954B (en) | Method and system for supporting high-precision map lane line real-time fusion update | |
| CN110598743A (en) | Target object labeling method and device | |
| CN110490936B (en) | Calibration method, device and equipment of vehicle camera and readable storage medium | |
| CN111261016B (en) | Road map construction method and device and electronic equipment | |
| KR100915600B1 (en) | Method for measuring 3-dimensinal coordinates of images using a target for ground control point | |
| CN109146958B (en) | Traffic sign space position measuring method based on two-dimensional image | |
| CN113139031B (en) | Method and related device for generating traffic sign for automatic driving | |
| CN107589069B (en) | Non-contact type measuring method for object collision recovery coefficient | |
| CN111932627A (en) | Marker drawing method and system | |
| CN115205382A (en) | Target positioning method and device | |
| CN112424568A (en) | System and method for constructing high-definition map | |
| CN111783597B (en) | Method and device for calibrating driving trajectory, computer equipment and storage medium | |
| CN111191596B (en) | Closed area drawing method, device and storage medium | |
| CN112488022A (en) | Panoramic monitoring method, device and system | |
| CN111598956A (en) | Calibration method, device and system | |
| CN116645404A (en) | Photo GPS position information extraction matching method and extraction matching device | |
| CN110148205B (en) | Three-dimensional reconstruction method and device based on crowdsourcing image | |
| CN111243021A (en) | Vehicle-mounted visual positioning method and system based on multiple combined cameras and storage medium | |
| CN113674361B (en) | Vehicle-mounted all-round-looking calibration implementation method and system | |
| CN115249345A (en) | Traffic jam detection method based on oblique photography three-dimensional live-action map | |
| CN117152361B (en) | Remote sensing image visibility estimation method based on attention mechanism |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |