CN111508260A - Vehicle parking space detection method, device and system - Google Patents
Vehicle parking space detection method, device and system Download PDFInfo
- Publication number
- CN111508260A CN111508260A CN201910089956.6A CN201910089956A CN111508260A CN 111508260 A CN111508260 A CN 111508260A CN 201910089956 A CN201910089956 A CN 201910089956A CN 111508260 A CN111508260 A CN 111508260A
- Authority
- CN
- China
- Prior art keywords
- parking space
- parking
- vehicle
- detection
- information
- 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.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/14—Traffic control systems for road vehicles indicating individual free spaces in parking areas
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
- G06V20/586—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of parking space
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
Abstract
The application relates to a method, a device and a system for detecting a vehicle parking space, wherein the method comprises the following steps: receiving images collected by camera devices arranged around a vehicle and carrying out panoramic stitching to obtain a bird-eye view; carrying out linear detection on the aerial view, and combining and screening according to the detected straight lines to obtain preliminary parking space area information; inputting the preliminary parking space region information into a preset parking space model to obtain a parking space detection result and sending the parking space detection result to an automatic parking system of the vehicle; tracking the detected parking space, correcting the position of the parking space in real time, and sending the corrected parking space information to a vehicle guidance system; the parking space model is obtained by training according to historical parking space picture samples. The parking space model obtained by training through analyzing historical parking space picture samples is used for parking space identification, accurate detection under a special environment can be achieved, the problems of fuzzy parking space detection and detection accuracy are solved, and compared with a traditional visual parking space detection method, the parking space detection reliability is improved.
Description
Technical Field
The application relates to the technical field of intelligent driving of vehicles, in particular to a method, a device and a system for detecting a parking space of a vehicle.
Background
With the improvement of living standard of people, automobiles gradually enter thousands of households to become the most important transportation means. With the increasing holding quantity of automobiles, the requirements for automation and intellectualization of the automobiles are also higher, and one of the requirements is automatic parking. Automatic parking is mainly divided into two types in the market at present, wherein the first type is based on ultrasonic radar, and the second type is based on visual automatic parking.
Traditional vision parking stall detection method is through the mark parking stall characteristics such as the line that contains that acquires, angle point, parking stall line colour, detects the parking stall position, nevertheless is unsatisfactory to the parking stall detection effect under the special environment such as grass planting brick, brings very big puzzlement, and traditional vision parking stall detection method still exists the shortcoming that detection reliability is low.
Disclosure of Invention
In view of the above, it is desirable to provide a method, an apparatus and a system for detecting a parking space of a vehicle, which can improve reliability of detecting the parking space.
A vehicle parking space detection method, comprising: receiving images collected by camera devices arranged around a vehicle and carrying out panoramic stitching to obtain a bird-eye view; carrying out linear detection on the aerial view, and combining and screening according to the detected straight lines to obtain preliminary parking space area information; inputting the preliminary parking space region information into a preset parking space model to obtain a parking space detection result and sending the parking space detection result to an automatic parking system of the vehicle; the parking space model is obtained by training according to historical parking space picture samples.
According to the vehicle parking space detection method, after the acquired images are subjected to panoramic stitching to obtain the aerial view, the aerial view is subjected to linear detection and screening to obtain preliminary parking space area information, then the preliminary parking space area information is identified and confirmed by combining a parking space model obtained according to historical parking space picture sample training, and finally determined detection results are sent to an automatic parking system of a vehicle so as to be used for the automatic parking system to perform automatic parking control. The parking space model obtained by training through analyzing historical parking space picture samples is used for parking space identification, accurate detection under a special environment can be achieved, the problems of fuzzy parking space detection and detection accuracy are solved, and compared with a traditional visual parking space detection method, the parking space detection reliability is improved.
In one embodiment, the historical parking stall picture samples comprise various types of fuzzy parking stall picture samples collected and sorted in day and night. Model training is carried out by combining various fuzzy parking space picture samples in daytime and at night, so that the recognition accuracy of the parking space model can be ensured, and the accuracy of parking space detection is improved.
In one embodiment, the performing line detection on the aerial view, and combining and screening according to the detected lines to obtain preliminary parking space area information includes: preprocessing the aerial view to obtain a preprocessed graph; and detecting straight lines in the preprocessed image, and combining and screening according to the detected straight lines to obtain the information of the primary parking space area.
The synthesized aerial view can be preprocessed to more clearly highlight the vehicle position line, so that the accuracy is improved for subsequent detection of straight lines, and the accuracy of vehicle parking space detection is improved.
In one embodiment, the pre-processing includes illumination compensation, horizontal and vertical gradient processing, and contrast enhancement.
Illumination compensation, horizontal gradient and vertical gradient processing and contrast enhancement are carried out to the aerial view in proper order, restrain the highlight and show the car position line to strengthen the high bright car position line that forms to outstanding parking stall line, make the parking stall line in the image show by force, improve follow-up parking stall accuracy that detects.
In one embodiment, the detecting a straight line in the preprocessed image, and combining and screening according to the detected straight line to obtain the preliminary parking space region information includes: performing linear detection on the preprocessed images, and combining the preprocessed images according to the detected linear to obtain a quadrilateral area; and screening the quadrilateral area according to preset parking place attribute data to obtain preliminary parking place area information.
And carrying out linear detection on the preprocessed image to obtain a possible parking space line, and then screening the obtained parking space line according to the parking space attribute data to obtain a preliminary parking space region so as to be used as a subsequent input parking space model for parking space detection. The parking space detection is carried out by combining the preliminary screening and the model detection, and the detection efficiency and the accuracy are improved.
In one embodiment, the slot attribute data includes a horizontal slot width and a vertical slot width.
Meanwhile, the detected parking space lines are screened by utilizing the horizontal parking space width and the vertical parking space width to obtain possible parking space areas, so that the accuracy of parking space screening is ensured, and meanwhile, the detection efficiency and the reliability of a follow-up parking space model can be improved.
In one embodiment, after the information of the preliminary parking space area is input into a preset parking space model, a parking space detection result is obtained and sent to an automatic parking system of a vehicle, the method further includes the following steps: and tracking the detected parking space in the parking process of the vehicle, correcting the position of the parking space in real time, and sending the corrected parking space information to a vehicle guidance system.
When the parking space detection result is sent to an automatic parking system of the vehicle for automatic parking operation, the detected parking space is corrected in real time in the parking process, and the corrected parking space information is sent to a vehicle guidance system to provide parking guidance, so that the accuracy of parking space detection is further improved.
In one embodiment, the tracking and real-time correcting the parking space position of the detected parking space during the parking process of the vehicle, and sending the corrected parking space information to the vehicle guidance system includes: carrying out parking space feature extraction on a bird view obtained by panoramic stitching in the vehicle parking process to obtain parking space features; carrying out dead reckoning according to preset vehicle motion parameters to obtain parking space position information; and correcting the parking space according to the parking space position information and the parking space characteristics, and sending the corrected parking space information to a vehicle guidance system.
And the parking space is corrected in real time by combining the extracted parking space characteristics and the parking space position information deduced by the navigation position, so that the accuracy and the reliability of parking space detection are ensured.
A vehicle parking space detection apparatus comprising: the image splicing module is used for receiving images collected by the camera devices arranged around the vehicle and carrying out panoramic splicing to obtain a bird's-eye view; the area screening module is used for carrying out linear detection on the aerial view, and combining and screening according to the detected straight lines to obtain preliminary parking place area information; the parking space detection module is used for inputting the preliminary parking space region information into a preset parking space model, obtaining a parking space detection result and sending the parking space detection result to an automatic parking system of the vehicle; the parking space model is obtained by training according to historical parking space picture samples.
Above-mentioned vehicle parking stall detection device trains the parking stall model that obtains through analysis historical parking stall picture sample and carries out the parking stall discernment, can realize having solved fuzzy parking stall detection problem and detection precision problem to the accurate detection under the special environment, compares in traditional vision parking stall detection method, has improved parking stall detection reliability.
In one embodiment, the apparatus further comprises: and the parking space tracking module is used for inputting the preliminary parking space region information into a preset parking space model by the parking space detection module, obtaining a parking space detection result and sending the parking space detection result to an automatic parking system of the vehicle, tracking the detected parking space in the parking process of the vehicle, correcting the position of the parking space in real time, and sending the corrected parking space information to a vehicle guidance system.
When the parking space detection result is sent to an automatic parking system of the vehicle for automatic parking operation, the detected parking space is corrected in real time in the parking process, and the corrected parking space information is sent to a vehicle guidance system to provide parking guidance, so that the accuracy of parking space detection is further improved.
The utility model provides a vehicle parking stall detecting system, includes camera device and on-vehicle treater, camera device sets up around the vehicle, on-vehicle treater is connected camera device, camera device is used for gathering vehicle environment image all around to with the image transmission who gathers to on-vehicle treater, on-vehicle treater is used for above-mentioned mode to carry out vehicle parking stall and tracks, and will obtain the parking stall testing result and send the automatic parking system to the vehicle, with the parking stall information transmission after the correction to vehicle bootstrap system.
Above-mentioned vehicle parking stall detecting system trains the parking stall model that obtains through analysis historical parking stall picture sample and carries out the parking stall discernment, can realize having solved fuzzy parking stall detection problem and detection precision problem to the accurate detection under the special environment, compares in traditional vision parking stall detection method, has improved parking stall detection reliability.
In one embodiment, the camera device is a fisheye camera. The panoramic camera capable of monitoring the parking space of the vehicle can independently realize large-range panoramic shooting without dead angle monitoring by using the fisheye camera, so that the detection range is expanded, and the reliability of vehicle parking space detection is further improved.
Drawings
FIG. 1 is a flow chart of a method for detecting a parking space of a vehicle according to an embodiment;
FIG. 2 is a flowchart illustrating a process of performing a line detection on the aerial view, and performing a combination and a screening according to the detected line to obtain information of a preliminary parking space area according to an embodiment of the present invention;
FIG. 3 is a flowchart of a method for detecting a parking space of a vehicle according to another embodiment;
fig. 4 is a flowchart illustrating that, in an embodiment, the detected parking space is tracked and the parking space position is corrected in real time during the parking process of the vehicle, and the corrected parking space information is sent to the vehicle guidance system;
FIG. 5 is a schematic view of a scene of a grass planting brick stall in one embodiment;
FIG. 6 is a schematic view of a red brick parking space in an embodiment;
FIG. 7 is a bird's eye view of the panoramic stitching in one embodiment;
FIG. 8 is an image of the bird's eye view of FIG. 7 after preprocessing;
FIG. 9 is a line graph illustrating the detection of a pre-processed image according to an embodiment;
FIG. 10 is a line graph of the pre-processed image detected in another embodiment;
FIG. 11 is a line graph of a pre-processed image detected in a further embodiment;
FIG. 12 is a line graph of a pre-processed image detected in a further embodiment;
FIG. 13 is a diagram of a vehicle position after a lane line is tracked and corrected according to an embodiment;
fig. 14 is a block diagram showing the construction of a parking space detecting device for a vehicle according to an embodiment;
fig. 15 is a block diagram showing the construction of a parking space detection apparatus for a vehicle according to another embodiment;
fig. 16 is a block diagram of a parking space system of a vehicle according to an embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
In one embodiment, there is provided a vehicle parking space detection method, as shown in fig. 1, including:
step S110: and receiving images collected by the camera devices arranged around the vehicle for panoramic stitching to obtain a bird's-eye view.
Specifically, images captured by the camera device may be received by the onboard processor. The specific number and type of the camera devices are not unique, for example, 4 fisheye cameras are adopted, the fisheye cameras can be respectively arranged in the front direction, the rear direction, the left direction and the right direction of the vehicle, and the vehicle-mounted processor acquires images collected by all the fisheye cameras arranged around the vehicle and carries out panoramic stitching to obtain a bird's-eye view image so as to be used for detecting parking spaces around the vehicle.
In addition, before step S110, the method may further include the step of calibrating the camera devices disposed around the vehicle. Similarly, taking 4 fisheye cameras as an example, the fisheye cameras are installed at four positions fixed at the front, the back, the left and the right of the vehicle and are calibrated to obtain internal and external parameters of the four cameras, distortion correction is performed on the cameras, and the visible area of the cameras is set to achieve good bird's-eye view splicing. After the driver starts the automatic parking system, the bird's-eye view can be generated through images collected by the calibrated four cameras and used for subsequent detection.
Step S120: and carrying out linear detection on the aerial view, and combining and screening according to the detected straight lines to obtain preliminary parking space area information.
After the vehicle-mounted processor splices the acquired images to obtain a bird's-eye view, straight lines in the bird's-eye view are detected, the detected straight lines are combined to obtain possible parking space areas, and then the parking space areas obtained through combination are screened to obtain preliminary parking space area information. Specifically, in one embodiment, as shown in fig. 2, step S120 includes step S122 and step S124.
Step S122: and preprocessing the aerial view to obtain a preprocessed graph. The method for preprocessing the bird's-eye view is not exclusive, and the preprocessing can be performed by a traditional segmentation algorithm. For a traditional segmentation algorithm, the aim of clearly highlighting the parking space lines of various fuzzy parking spaces in the day and at night is finally achieved, the parts which are not the parking space lines are basically filtered or become very dark, and the accuracy is improved for detecting straight lines later.
Step S124: and detecting straight lines in the preprocessed image, and combining and screening according to the detected straight lines to obtain the information of the primary parking space area. Specifically, the straight lines in the preprocessed image can be detected and adjusted by hough transform detection straight line mode to reduce the interference lines. And after the possible parking space lines are obtained, the parking space lines are combined and screened, and possible parking space areas are extracted.
In this embodiment, the vehicle-parking space line can be more clearly highlighted by preprocessing the aerial view obtained by synthesis, so as to improve the precision for subsequent detection of the straight line and improve the accuracy of vehicle parking space detection.
Further, in one embodiment, the pre-processing includes illumination compensation, horizontal and vertical gradient processing, and contrast enhancement. Specifically, illumination compensation is performed through gamma conversion, so that the effect of inhibiting strong light and balancing light is achieved; and secondly, the parking space line is protruded through the horizontal gradient and the vertical gradient, and the protruded parking space line is enhanced through contrast enhancement to form a high-brightness parking space line. Illumination compensation, horizontal gradient and vertical gradient processing and contrast enhancement are carried out to the aerial view in proper order, restrain the highlight and show the car position line to strengthen the high bright car position line that forms to outstanding parking stall line, make the parking stall line in the image show by force, improve follow-up parking stall accuracy that detects.
In one embodiment, step S124 includes: performing linear detection on the preprocessed images, and combining the preprocessed images according to the detected straight lines to obtain a quadrilateral area; and screening the quadrilateral area according to preset parking place attribute data to obtain preliminary parking place area information.
Specifically, after straight lines in the image are detected, intersection points of the straight lines are calculated, the intersection points are placed into a linked list and combined pairwise to obtain possible parking space quadrangles, and then the possible parking space quadrangles are screened according to parking space attribute data to obtain preliminary parking space region information. And carrying out linear detection on the preprocessed image to obtain a possible parking space line, and then screening the obtained parking space line according to the parking space attribute data to obtain a preliminary parking space region so as to be used as a subsequent input parking space model for parking space detection. The parking space detection is carried out by combining the preliminary screening and the model detection, and the detection efficiency and the accuracy are improved.
It is to be understood that the specific content of the slot attribute data is not unique, and in one embodiment, the slot attribute data includes a horizontal slot width and a vertical slot width. Meanwhile, the detected parking space lines are screened by utilizing the horizontal parking space width and the vertical parking space width to obtain possible parking space areas, so that the accuracy of parking space screening is ensured, and meanwhile, the detection efficiency and the reliability of a follow-up parking space model can be improved.
Step S130: and inputting the preliminary parking space region information into a preset parking space model to obtain a parking space detection result and sending the parking space detection result to an automatic parking system of the vehicle.
The parking space model is obtained by training according to historical parking space picture samples. In one embodiment, the historical parking spot picture samples comprise various types of fuzzy parking spot picture samples collected and sorted in the day and at night. Model training is carried out by combining various fuzzy parking space picture samples in daytime and at night, so that the recognition accuracy of the parking space model can be ensured, and the accuracy of parking space detection is improved. Specifically, various fuzzy parking stall picture samples including daytime and evening parking stall pictures can be collected and sorted in advance, and the collected historical parking stall picture samples are trained to obtain a parking stall model capable of identifying the parking stall region. The model training mode is not unique, in this embodiment, a parking space image can be subjected to a parking space training model through a lenet5 or cifar-10 network model of a coffee (Convolutional Architecture for Fast FeatureEmbed, Convolutional neural network framework), a parking space line part is accurately identified and judged, whether the parking space line part is a parking space or not is judged, and if the parking space line part is a parking space, a parking space line is finally determined and drawn and a parking space position is obtained; if the parking space is not the parking space, the next possible parking space area is selected for judgment. After the parking space model is used for detecting the initial parking space region information, the parking space detection result is sent to an automatic parking system of the vehicle, so that the automatic parking system can execute automatic parking operation.
According to the vehicle parking space detection method, the parking space identification is carried out through the parking space model obtained by analyzing the historical parking space picture sample for training, the accurate detection under the special environment can be realized, the problems of fuzzy parking space detection and detection accuracy are solved, and compared with the traditional visual parking space detection method, the parking space detection reliability is improved.
In one embodiment, as shown in fig. 3, after step S130, the method may further include step S140.
Step S140: and tracking the detected parking space in the parking process of the vehicle, correcting the position of the parking space in real time, and sending the corrected parking space information to a vehicle guidance system.
After receiving the parking space detection result, the automatic parking system reminds a driver whether to carry out a parking task of the parking space if the parking space is detected. And carrying out automatic parking operation after receiving a confirmation instruction of the driver. And the vehicle-mounted processor continues to perform fuzzy parking space tracking detection in the parking process of the vehicle. In the process of parking tracking, the vehicle-mounted processor highlights a highlight vehicle position line on the look-around aerial view by utilizing a traditional segmentation algorithm, corrects the position of the parking space in real time and transmits parking space information to the vehicle guiding system.
In this embodiment, when the parking space detection result is sent to the automatic parking system of the vehicle to perform the automatic parking operation, the detected parking space is corrected in real time in the parking process, and the corrected parking space information is sent to the vehicle guidance system to provide parking guidance, so that the accuracy of parking space detection is further improved.
In one embodiment, as shown in fig. 4, step S140 includes steps S142 to S146.
Step S142: and carrying out parking space feature extraction on the aerial view obtained by panoramic splicing in the parking process of the vehicle to obtain parking space features.
Step S144: and carrying out dead reckoning according to preset vehicle motion parameters to obtain the parking space position information.
Step S146: and correcting the parking space according to the parking space position information and the parking space characteristics, and sending the corrected parking space information to a vehicle guidance system.
Specifically, the fuzzy parking space tracking correction comprises characteristic detection, a parking space state tracker and vehicle motion model tracking. Firstly, a vehicle motion model tracks the approximate position of a parking space obtained through dead reckoning, then feature extraction is carried out through feature detection (feature extraction of a fuzzy parking space line is carried out on an interested area of the parking space in a high-brightness parking space line image segmented by a traditional segmentation algorithm), and errors generated are corrected in real time. The parking space state tracker tracks the parking space information given by the characteristic detection and the vehicle motion model to obtain the current accurate parking space information and transmits the current accurate parking space information to the vehicle guiding system. And the parking space is corrected in real time by combining the extracted parking space characteristics and the parking space position information deduced by the navigation position, so that the accuracy and the reliability of parking space detection are ensured.
In order to better understand the above-mentioned vehicle parking space detection method, the following detailed explanation is made with reference to specific embodiments.
Traditional vision parking stall detection is not ideal to the parking stall detection effect on the grass planting brick, and what we meet in the life at ordinary times often is these parking stalls. The parking spaces marked by the paints with the colors of red, yellow, white and the like on the grass planting bricks or the parking space lines formed by piling up colored bricks and stones are basically blurred, and the corner points of the parking spaces are not clear. Some parking space angle points are made of bricks and stones, most of the parking space angle points and parking space lines are worn, and even no parking space angle point exists, such as the parking space angle points shown in fig. 5 and 6. The fuzzy parking spaces on the grass bricks usually account for a large proportion, the detection rate is low (less than 10%) at present, a high target detection rate (more than 90%) is achieved, and great trouble is brought to parking space detection, but the problems are not well solved by the existing method.
The vehicle parking space detection method is based on panoramic looking around to perform fuzzy parking space detection and tracking, a bird's-eye view is mainly looked around as an input image, the part of a brightened parking space line is highlighted through a traditional segmentation algorithm, a parking space line is detected through a detection line mode, the parking space line is preliminarily confirmed and screened according to the width range of the parking space line (the width of a vertical parking space is 2.1-4.0 m, the width of a horizontal parking space is 5.0-6.5 m), and finally, a possible parking space area is placed into a parking space model trained through lenet5 or cifar-10 (the parking space model is a mask model trained through a gray scale map of the parking space and comprises three convolution layers, two pooling layers and two full connection layers) to perform fine identification confirmation, and the position of the parking space line is finally determined.
Detection of fuzzy parking spaces:
1) the aerial view is preprocessed through a traditional segmentation algorithm, the algorithms comprise illumination compensation, horizontal gradient and vertical gradient processing, contrast enhancement and the like, the parking space line part is highlighted, various fuzzy parking spaces can be clearly highlighted in the daytime and at night, other parking space lines are basically filtered or are very dark, and the accuracy is improved for detecting straight lines later. The method comprises the following specific steps: firstly, illumination compensation is carried out through gamma conversion, and the effects of inhibiting strong light and balancing light are achieved; secondly, protruding the car position line through a horizontal gradient and a vertical gradient; and then the outstanding parking space lines are enhanced through contrast enhancement to form high-brightness parking space lines. Fig. 7 and 8 are the original image and the conventional divided parking space line image, respectively, and fig. 8 shows the effect of the plant brick that the parking space protrudes from the parking space line in a blurred manner, so that the parking space line is highlighted.
2) The straight line is detected by a hough conversion detection straight line mode and then adjusted, so that interference lines are reduced. Fig. 9 to 12 are each a line graph detected for each image. And calculating to obtain the intersection point of the straight lines through the obtained possible vehicle position lines. And putting the intersection points into a linked list, combining every two intersection points to obtain possible parking space quadrangles, and screening the possible parking space quadrangles according to the parking space attributes.
3) Collecting and sorting various fuzzy parking stall picture samples, including the parking stall pictures in the daytime and at night. And (3) training a parking space model for the parking space picture through a lens 5 or cifar-10 network model of the cafe, accurately identifying and judging the part of the parking space line, and judging whether the parking space line is a parking space. If the parking space is the parking space, finally determining and drawing a parking space line, and obtaining the position of the parking space; if the parking space is not the parking space, the next possible parking space area is selected for judgment.
To there being mark parking stall detection part in this application, can adopt following mode:
1) the four fisheye cameras are arranged at the front, the back, the left and the right of the vehicle, each camera is accurately calibrated, and the calibrated four cameras can generate a bird's-eye view.
2) And starting the automatic parking system, and enabling the vehicle to advance along the lane direction, wherein the parking space is in the front right direction of the vehicle.
3) And at the moment, carrying out parking space detection, carrying out a highlighted car position line on the aerial view through a traditional segmentation algorithm, carrying out parking space line identification and judgment, finding a parking space mode through characteristic detection (the parking space width is accurately identified by a parking space model) when the car is close to the parking space for a certain distance, judging the current parking space type, and simultaneously transmitting a parking space detection result to an automatic parking system.
4) The automatic parking system reports to the vehicle owner whether the parking task of the parking space is needed.
Fuzzy parking space tracking: the method is used for tracking the detected parking places in the parking process, and in the parking tracking process, the highlight parking place lines are highlighted on the panoramic bird's eye view by utilizing the traditional segmentation algorithm, such as the image 13, the parking place positions are corrected in real time, and the parking place information is transmitted to a vehicle guiding system. The fuzzy parking space tracking comprises characteristic detection, parking space state tracker and vehicle motion model tracking. Firstly, a vehicle motion model tracks and obtains the approximate position of a parking space through dead reckoning, then feature extraction is carried out through feature detection (feature extraction of a fuzzy parking space line is carried out on an interested area of the parking space in a high-brightness parking space line image segmented by a traditional segmentation algorithm), and an error generated by the vehicle motion model is corrected in real time by a parking space state tracker.
For the tracking part with marked parking spaces in the parking process, the following modes can be adopted:
1) after the car owner selects the parking space to park, the car guiding system works, at the moment, the parking space detection is stopped, and the parking space tracking starts to work.
2) The car owner operates according to the instruction given by the automatic parking system and formally carries out the parking task.
3) In the parking process, extracting the characteristics of the parking space line in the region of interest of the parking space in the high-brightness parking space line image segmented by the traditional segmentation algorithm, and positioning the parking space position according to the characteristics; and according to the vehicle motion parameters, giving the position of the current parking space by using a dead reckoning motion model.
4) The parking space state tracker tracks the parking space information given by the characteristic detection and the vehicle motion model to obtain the current accurate parking space information and transmits the current accurate parking space information to the vehicle guiding system.
By the vehicle parking space detection method, the problem of fuzzy parking space detection is solved, the problem of fuzzy parking space detection accuracy is solved, and the problem of fuzzy parking space tracking in the parking process is also solved. The fuzzy parking space detection on the grass planting bricks is supported, the detection rate is improved by 90%, and the detection speed is improved. The precision of the fuzzy parking space detection on the grass planting bricks is improved, and the current accurate parking space information can be given in real time no matter whether the parking space is blocked or not in the parking process.
In one embodiment, a parking space detection apparatus for a vehicle, as shown in fig. 14, includes an image stitching module 110, an area filtering module 120, and a parking space detection module 130.
The image stitching module 110 is configured to receive images collected by the camera devices disposed around the vehicle and perform panoramic stitching to obtain an aerial view. In addition, before receiving the image, the image stitching module 110 may further perform calibration on the cameras disposed around the vehicle. After the driver starts the automatic parking system, the bird's-eye view can be generated through images collected by the calibrated four cameras and used for subsequent detection.
The area screening module 120 is configured to perform line detection on the aerial view, and combine and screen the line according to the detected line to obtain preliminary parking space area information. Specifically, the region screening module 120 preprocesses the bird's-eye view to obtain a preprocessed graph; and detecting straight lines in the preprocessed image, and combining and screening according to the detected straight lines to obtain the information of the primary parking space area.
Further, in one embodiment, the pre-processing includes illumination compensation, horizontal and vertical gradient processing, and contrast enhancement. In one embodiment, the region screening module 120 performs line detection on the preprocessed images, and combines the preprocessed images according to the detected lines to obtain a quadrilateral region; and screening the quadrilateral area according to preset parking place attribute data to obtain preliminary parking place area information. It is to be understood that the specific content of the slot attribute data is not unique, and in one embodiment, the slot attribute data includes a horizontal slot width and a vertical slot width.
The parking space detection module 130 is configured to input the preliminary parking space region information into a preset parking space model, obtain a parking space detection result, and send the parking space detection result to an automatic parking system of the vehicle. The parking space model is obtained by training according to historical parking space picture samples.
In one embodiment, as shown in FIG. 15, the apparatus further includes a space tracking module 140.
The parking space tracking module 140 is configured to enable the parking space detection module 130 to input the preliminary parking space region information into a preset parking space model, obtain a parking space detection result, send the parking space detection result to an automatic parking system of the vehicle, track the detected parking space in the parking process of the vehicle, correct the position of the parking space in real time, and send the corrected parking space information to a vehicle guidance system.
In one embodiment, the parking space tracking module 140 performs parking space feature extraction on a bird's eye view obtained by panoramic stitching in the parking process of the vehicle to obtain parking space features; carrying out dead reckoning according to preset vehicle motion parameters to obtain parking space position information; and correcting the parking space according to the parking space position information and the parking space characteristics, and sending the corrected parking space information to a vehicle guidance system.
For specific definition of the vehicle parking space detection device, reference may be made to the above definition of the vehicle parking space detection method, which is not described herein again. The modules in the vehicle parking space detection device can be wholly or partially realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
Above-mentioned vehicle parking stall detection device trains the parking stall model that obtains through analysis historical parking stall picture sample and carries out the parking stall discernment, can realize having solved fuzzy parking stall detection problem and detection precision problem to the accurate detection under the special environment, compares in traditional vision parking stall detection method, has improved parking stall detection reliability.
In an embodiment, as shown in fig. 16, a vehicle parking space detection system is provided, and includes a camera device 210 and an onboard processor 220, where the camera device 210 is disposed around a vehicle, the onboard processor 220 is connected to the camera device 210, the camera device 210 is configured to collect an environment image around the vehicle and send the collected image to the onboard processor 220, and the onboard processor 220 is configured to perform vehicle parking space detection and tracking according to the above manner, send an obtained parking space detection result to an automatic parking system of the vehicle, and send corrected parking space information to a vehicle guidance system.
The specific number and type of the image capturing devices 210 are not exclusive, and the image capturing devices 210 may be fish-eye cameras. The panoramic camera capable of monitoring the parking space of the vehicle can independently realize large-range panoramic shooting without dead angle monitoring by using the fisheye camera, so that the detection range is expanded, and the reliability of vehicle parking space detection is further improved. Taking 4 fisheye cameras as an example, the fisheye cameras can be respectively arranged in the front direction, the rear direction, the left direction and the right direction of the vehicle. The onboard processor 220 may specifically employ an onboard embedded processor.
Above-mentioned vehicle parking stall detecting system trains the parking stall model that obtains through analysis historical parking stall picture sample and carries out the parking stall discernment, can realize having solved fuzzy parking stall detection problem and detection precision problem to the accurate detection under the special environment, compares in traditional vision parking stall detection method, has improved parking stall detection reliability.
The computer instructions may be stored in or transmitted from a website, computer, server, or data center via a wired (e.g., coaxial cable, fiber optic, digital subscriber line (DS L)) or wireless (e.g., infrared, wireless, microwave, etc.) manner to another website, computer, server, or data center.
One of ordinary skill in the art will appreciate that all or part of the processes in the methods of the above embodiments may be implemented by hardware related to instructions of a computer program, which may be stored in a computer-readable storage medium, and when executed, may include the processes of the above method embodiments. And the aforementioned storage medium includes: various media that can store program codes, such as a read-only memory (ROM) or a Random Access Memory (RAM), a magnetic disk, or an optical disk.
The technical features of the embodiments described above may be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the embodiments described above are not described, but should be considered as being within the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (12)
1. A vehicle parking space detection method, characterized by comprising:
receiving images collected by camera devices arranged around a vehicle and carrying out panoramic stitching to obtain a bird-eye view;
carrying out linear detection on the aerial view, and combining and screening according to the detected straight lines to obtain preliminary parking space area information;
inputting the preliminary parking space region information into a preset parking space model to obtain a parking space detection result and sending the parking space detection result to an automatic parking system of the vehicle; the parking space model is obtained by training according to historical parking space picture samples.
2. The method of claim 1, wherein the historical parking picture samples comprise various types of blurred parking picture samples collected during the day and at night.
3. The method of claim 1, wherein the detecting the line of the bird's eye view image, and combining and screening the line of the bird's eye view image to obtain the preliminary parking space area information comprises:
preprocessing the aerial view to obtain a preprocessed graph;
and detecting straight lines in the preprocessed image, and combining and screening according to the detected straight lines to obtain the information of the primary parking space area.
4. The method of claim 3, wherein the pre-processing comprises illumination compensation, horizontal and vertical gradient processing, and contrast enhancement.
5. The method according to claim 3, wherein the detecting the straight lines in the preprocessed image, and the combining and screening according to the detected straight lines to obtain the preliminary parking space region information comprises:
performing linear detection on the preprocessed images, and combining the preprocessed images according to the detected linear to obtain a quadrilateral area;
and screening the quadrilateral area according to preset parking place attribute data to obtain preliminary parking place area information.
6. The method of claim 5, wherein the slot attribute data includes a horizontal slot width and a vertical slot width.
7. The method according to any one of claims 1 to 6, wherein after the preliminary parking space region information is input into a preset parking space model, a parking space detection result is obtained and sent to an automatic parking system of a vehicle, the method further comprises the following steps:
and tracking the detected parking space in the parking process of the vehicle, correcting the position of the parking space in real time, and sending the corrected parking space information to a vehicle guidance system.
8. The method of claim 7, wherein the tracking the detected parking space and correcting the parking space position in real time during the parking process of the vehicle, and sending the corrected parking space information to a vehicle guidance system comprises:
carrying out parking space feature extraction on a bird view obtained by panoramic stitching in the vehicle parking process to obtain parking space features;
carrying out dead reckoning according to preset vehicle motion parameters to obtain parking space position information;
and correcting the parking space according to the parking space position information and the parking space characteristics, and sending the corrected parking space information to a vehicle guidance system.
9. A vehicle parking space detection apparatus, characterized by comprising:
the image splicing module is used for receiving images collected by the camera devices arranged around the vehicle and carrying out panoramic splicing to obtain a bird's-eye view;
the area screening module is used for carrying out linear detection on the aerial view, and combining and screening according to the detected straight lines to obtain preliminary parking place area information;
the parking space detection module is used for inputting the preliminary parking space region information into a preset parking space model, obtaining a parking space detection result and sending the parking space detection result to an automatic parking system of the vehicle; the parking space model is obtained by training according to historical parking space picture samples.
10. The apparatus of claim 9, further comprising:
and the parking space tracking module is used for inputting the preliminary parking space region information into a preset parking space model by the parking space detection module, obtaining a parking space detection result and sending the parking space detection result to an automatic parking system of the vehicle, tracking the detected parking space in the parking process of the vehicle, correcting the position of the parking space in real time, and sending the corrected parking space information to a vehicle guidance system.
11. The vehicle parking space detection system is characterized by comprising a camera device and an on-board processor, wherein the camera device is arranged around a vehicle, the on-board processor is connected with the camera device, the camera device is used for collecting images of the environment around the vehicle and sending the collected images to the on-board processor, the on-board processor is used for detecting and tracking a vehicle parking space according to the mode of any one of claims 1 to 8, sending the obtained parking space detection result to an automatic parking system of the vehicle, and sending the corrected parking space information to a vehicle guide system.
12. The system of claim 11, wherein the camera device is a fisheye camera.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910089956.6A CN111508260A (en) | 2019-01-30 | 2019-01-30 | Vehicle parking space detection method, device and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910089956.6A CN111508260A (en) | 2019-01-30 | 2019-01-30 | Vehicle parking space detection method, device and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111508260A true CN111508260A (en) | 2020-08-07 |
Family
ID=71877291
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910089956.6A Pending CN111508260A (en) | 2019-01-30 | 2019-01-30 | Vehicle parking space detection method, device and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111508260A (en) |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112036385A (en) * | 2020-11-04 | 2020-12-04 | 天津天瞳威势电子科技有限公司 | Library position correction method and device, electronic equipment and readable storage medium |
| CN112183413A (en) * | 2020-09-30 | 2021-01-05 | 东风柳州汽车有限公司 | Parking space detection method and device, storage medium and vehicle |
| CN112289070A (en) * | 2020-10-20 | 2021-01-29 | 广州小鹏自动驾驶科技有限公司 | Parking space detection method and device, vehicle and readable medium |
| CN112330977A (en) * | 2020-10-28 | 2021-02-05 | 广州小鹏自动驾驶科技有限公司 | Automatic parking method and device |
| CN112562391A (en) * | 2020-11-30 | 2021-03-26 | 广州小鹏自动驾驶科技有限公司 | Parking space updating method and device |
| CN113313964A (en) * | 2021-05-08 | 2021-08-27 | 深圳市大道至简信息技术有限公司 | Roadside parking management method and system thereof |
| CN113327241A (en) * | 2021-06-16 | 2021-08-31 | 南昌航空大学 | Visual detection method and system for surface defects of bearing end face |
| CN113538523A (en) * | 2021-09-17 | 2021-10-22 | 魔视智能科技(上海)有限公司 | Parking space detection tracking method, electronic equipment and vehicle |
| CN113705474A (en) * | 2021-08-30 | 2021-11-26 | 北京易航远智科技有限公司 | Parking space detection method and device |
| CN114141055A (en) * | 2020-08-13 | 2022-03-04 | 纵目科技（上海）股份有限公司 | Parking space detection device and detection method of intelligent parking system |
| WO2022096027A1 (en) * | 2020-11-04 | 2022-05-12 | 天津天瞳威势电子科技有限公司 | Garage space tracking method and apparatus |
| US11351932B1 (en) | 2021-01-22 | 2022-06-07 | Toyota Motor Engineering & Manufacturing North America, Inc. | Vehicles and methods for generating and displaying composite images |
| CN114882727A (en) * | 2022-03-15 | 2022-08-09 | 深圳市德驰微视技术有限公司 | Parking space detection method based on domain controller, electronic device and storage medium |
| CN115083199A (en) * | 2021-03-12 | 2022-09-20 | 上海汽车集团股份有限公司 | Parking space information determination method and related equipment thereof |
| CN115565158A (en) * | 2022-11-17 | 2023-01-03 | 广汽埃安新能源汽车股份有限公司 | Parking space detection method and device, electronic equipment and computer readable medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130162825A1 (en) * | 2011-12-23 | 2013-06-27 | Hyundai Motor Company | Avm top view based parking support system |
| CN107424116A (en) * | 2017-07-03 | 2017-12-01 | 浙江零跑科技有限公司 | Position detecting method of parking based on side ring depending on camera |
| CN107719361A (en) * | 2017-10-10 | 2018-02-23 | 深圳市豪恩汽车电子装备股份有限公司 | Automatic parking assisting method and system based on image vision |
| CN107886080A (en) * | 2017-11-23 | 2018-04-06 | 同济大学 | One kind is parked position detecting method |
| CN108140312A (en) * | 2015-10-22 | 2018-06-08 | 日产自动车株式会社 | Parking assistance method and parking aid |
-
2019
- 2019-01-30 CN CN201910089956.6A patent/CN111508260A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130162825A1 (en) * | 2011-12-23 | 2013-06-27 | Hyundai Motor Company | Avm top view based parking support system |
| CN108140312A (en) * | 2015-10-22 | 2018-06-08 | 日产自动车株式会社 | Parking assistance method and parking aid |
| CN107424116A (en) * | 2017-07-03 | 2017-12-01 | 浙江零跑科技有限公司 | Position detecting method of parking based on side ring depending on camera |
| CN107719361A (en) * | 2017-10-10 | 2018-02-23 | 深圳市豪恩汽车电子装备股份有限公司 | Automatic parking assisting method and system based on image vision |
| CN107886080A (en) * | 2017-11-23 | 2018-04-06 | 同济大学 | One kind is parked position detecting method |
Non-Patent Citations (1)
| Title |
|---|
| 朱道明 等: "《建筑安防技术》", 31 January 2013, 东华大学出版社 * |
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114141055A (en) * | 2020-08-13 | 2022-03-04 | 纵目科技（上海）股份有限公司 | Parking space detection device and detection method of intelligent parking system |
| CN114141055B (en) * | 2020-08-13 | 2024-04-16 | 纵目科技（上海）股份有限公司 | Parking space detection device and method of intelligent parking system |
| CN112183413A (en) * | 2020-09-30 | 2021-01-05 | 东风柳州汽车有限公司 | Parking space detection method and device, storage medium and vehicle |
| CN112183413B (en) * | 2020-09-30 | 2024-05-14 | 东风柳州汽车有限公司 | Parking space detection method and device, storage medium and vehicle |
| CN112289070A (en) * | 2020-10-20 | 2021-01-29 | 广州小鹏自动驾驶科技有限公司 | Parking space detection method and device, vehicle and readable medium |
| CN112330977A (en) * | 2020-10-28 | 2021-02-05 | 广州小鹏自动驾驶科技有限公司 | Automatic parking method and device |
| CN112330977B (en) * | 2020-10-28 | 2022-05-13 | 广州小鹏自动驾驶科技有限公司 | Automatic parking method and device |
| CN112036385B (en) * | 2020-11-04 | 2021-02-02 | 天津天瞳威势电子科技有限公司 | Library position correction method and device, electronic equipment and readable storage medium |
| CN112036385A (en) * | 2020-11-04 | 2020-12-04 | 天津天瞳威势电子科技有限公司 | Library position correction method and device, electronic equipment and readable storage medium |
| WO2022096027A1 (en) * | 2020-11-04 | 2022-05-12 | 天津天瞳威势电子科技有限公司 | Garage space tracking method and apparatus |
| CN112562391A (en) * | 2020-11-30 | 2021-03-26 | 广州小鹏自动驾驶科技有限公司 | Parking space updating method and device |
| US11351932B1 (en) | 2021-01-22 | 2022-06-07 | Toyota Motor Engineering & Manufacturing North America, Inc. | Vehicles and methods for generating and displaying composite images |
| CN115083199A (en) * | 2021-03-12 | 2022-09-20 | 上海汽车集团股份有限公司 | Parking space information determination method and related equipment thereof |
| CN115083199B (en) * | 2021-03-12 | 2024-02-27 | 上海汽车集团股份有限公司 | Parking space information determining method and related equipment thereof |
| CN113313964A (en) * | 2021-05-08 | 2021-08-27 | 深圳市大道至简信息技术有限公司 | Roadside parking management method and system thereof |
| CN113327241A (en) * | 2021-06-16 | 2021-08-31 | 南昌航空大学 | Visual detection method and system for surface defects of bearing end face |
| CN113705474B (en) * | 2021-08-30 | 2022-04-15 | 北京易航远智科技有限公司 | Parking space detection method and device |
| CN113705474A (en) * | 2021-08-30 | 2021-11-26 | 北京易航远智科技有限公司 | Parking space detection method and device |
| CN113538523B (en) * | 2021-09-17 | 2021-12-31 | 魔视智能科技(上海)有限公司 | Parking space detection tracking method, electronic equipment and vehicle |
| CN113538523A (en) * | 2021-09-17 | 2021-10-22 | 魔视智能科技(上海)有限公司 | Parking space detection tracking method, electronic equipment and vehicle |
| CN114882727A (en) * | 2022-03-15 | 2022-08-09 | 深圳市德驰微视技术有限公司 | Parking space detection method based on domain controller, electronic device and storage medium |
| CN114882727B (en) * | 2022-03-15 | 2023-09-05 | 深圳市德驰微视技术有限公司 | Parking space detection method based on domain controller, electronic equipment and storage medium |
| CN115565158A (en) * | 2022-11-17 | 2023-01-03 | 广汽埃安新能源汽车股份有限公司 | Parking space detection method and device, electronic equipment and computer readable medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111508260A (en) | Vehicle parking space detection method, device and system | |
| CN109949594B (en) | Real-time traffic light identification method | |
| US8995723B2 (en) | Detecting and recognizing traffic signs | |
| US8452053B2 (en) | Pixel-based texture-rich clear path detection | |
| US11371851B2 (en) | Method and system for determining landmarks in an environment of a vehicle | |
| CN111209780A (en) | Lane line attribute detection method and device, electronic device and readable storage medium | |
| WO2020248910A1 (en) | Target detection method and device | |
| CN111523368B (en) | Information processing device, server, and traffic management system | |
| CN110658539B (en) | Vehicle positioning method, device, vehicle and computer readable storage medium | |
| CN110046584B (en) | Road crack detection device and detection method based on unmanned aerial vehicle inspection | |
| CN111967396A (en) | Processing method, device and equipment for obstacle detection and storage medium | |
| CN111507145A (en) | Method, system and device for detecting barrier at storage position of embedded vehicle-mounted all-round looking system | |
| CN111127520B (en) | Vehicle tracking method and system based on video analysis | |
| CN114663852A (en) | Method and device for constructing lane line graph, electronic equipment and readable storage medium | |
| CN116413725A (en) | Barrier detection method based on camera and millimeter wave radar data fusion | |
| CN112241004B (en) | Object recognition device | |
| CN113071500A (en) | Method and device for acquiring lane line, computer equipment and storage medium | |
| CN113538967B (en) | Vehicle-road cooperation device and method under crossroad scene | |
| CN114379544A (en) | Automatic parking system, method and device based on multi-sensor pre-fusion | |
| WO2008046458A1 (en) | Method and device for capturing the surroundings in motor vehicles with the aid of aerial images | |
| CN110826364A (en) | Stock position identification method and device | |
| CN112509321A (en) | Unmanned aerial vehicle-based driving control method and system for urban complex traffic situation and readable storage medium | |
| JP7407427B2 (en) | Information processing method and information processing system | |
| CN106991415A (en) | Image processing method and device for vehicle-mounted fisheye camera | |
| CN113378719B (en) | Lane line identification method, lane line identification device, computer equipment and storage medium |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200807 |
|
| RJ01 | Rejection of invention patent application after publication |