Parking space retrieval system based on machine vision

Abstract

The invention provides a parking space retrieval system based on machine vision. The parking space retrieval system comprises a parking lot, the parking lot is connected with a supporting column, the supporting column is connected with a rotating mechanism, the rotating mechanism is connected with a C-shaped plate, the C-shaped plate is connected with a camera, the camera is connected with the rotating mechanism, and the camera is connected with the parking space retrieval system. The parking space retrieval system comprises a data acquisition module, a model establishment module, a background screening module, a vehicle distribution calculation module, a parking space judgment module, a route planning module, a signal sending module and a client module. Model training is carried out through customized image recognition, information such as behavior characteristics of a vehicle collected by a camera is processed and analyzed, whether the vehicle is parked in a parking space at present is judged, data is transmitted to a client module, the position of an empty parking space can be mastered at the first time when the vehicle enters a parking lot, and the parking space is more accurate. A large amount of time is saved to a certain extent, and congestion caused by the fact that vehicles find parking spaces on the road surface is avoided.

Description

Technical Field

[0001] The present invention mainly relates to the field of parking space retrieval, and in particular to a parking space retrieval system based on machine vision. Background Art

[0002] With the improvement of living standards, almost every family in the country owns a private car, but they face the pressure of parking areas. Every time they drive to shopping malls and other places, they need to spend a lot of time looking for parking spaces, which brings certain troubles to people's pace of life. It is difficult to give accurate parking routes, resulting in a lot of time wasted. Summary of the invention

[0003] The present invention mainly provides a parking space retrieval system based on machine vision to solve the technical problems raised in the above background technology.

[0004] The technical solution adopted by the present invention to solve the above technical problems is: a parking space retrieval system based on machine vision, comprising a parking lot, one side of the parking lot is connected to a pillar, the top of the pillar is connected to a rotating mechanism, the top of the rotating mechanism is connected to a C-shaped plate, the top of the C-shaped plate is rotatably connected to a camera, one side of the camera is connected to the rotating mechanism, and one side of the top of the camera is connected to the parking space retrieval system;

[0005] The parking space retrieval system comprises:

[0006] A data acquisition module, used to acquire images of parking spaces and vehicles in the parking lot;

[0007] Model building module, used for image classification, image detection and background modeling to obtain the candidate parking space area;

[0008] A background screening module, used to find vehicles around parking spaces and filter artificial landscape images around parking spaces;

[0009] The vehicle distribution calculation module is used to calculate the distribution of vehicles on the time axis and form a dynamic vehicle motion trajectory image;

[0010] Parking space judgment module, used to screen and verify the correctness of calculations and select parking spaces;

[0011] A route planning module, used to provide a route for the vehicle to enter the parking space;

[0012] A signal sending module, used for sending the signal provided by the route planning module;

[0013] The client module is used to receive the signal sent by the signal sending module and perform navigation.

[0014] In the present invention, the camera is convenient for photographing the parking lot and providing an image of the parking lot.

[0015] Furthermore, the parking space retrieval system also includes a mobile phone terminal for receiving a vehicle positioning signal sent by the client module.

[0016] Furthermore, the data acquisition module includes:

[0017] A car entry and exit collection unit, used to obtain the number of cars in the parking lot, and the car entry and exit collection unit sends a signal to the model building module;

[0018] The visual monitoring unit is used to obtain the picture of the parking lot in real time. The visual monitoring unit sends the image to the model building module. The visual monitoring unit is electrically connected to the camera.

[0019] Furthermore, the model building module includes:

[0020] An image collating unit, used to sort the images collected by the visual monitoring unit and detect the images;

[0021] The background modeling unit is used to perform parking lot background modeling according to the image sorted by the image sorting unit. The background modeling unit sends data to the background screening module, and the background screening module sends data to the vehicle distribution calculation module.

[0022] Furthermore, the parking space determination module includes:

[0023] A screening and verification unit, used to perform binarization and clustering operations on the spatial axis according to the motion trajectory image, screen out clusters with a recognition accuracy greater than 50% and perform verification, and the screening and verification unit receives the motion trajectory image sent by the vehicle distribution calculation module;

[0024] The parking space selection unit is used to receive the data sent by the screening and verification unit, select the parking space closest to the vehicle, and send a parking space signal to the route planning module.

[0025] Furthermore, the receiving client module includes:

[0026] An on-vehicle signal receiving unit, used to receive the signal sent by the signal sending module;

[0027] The route navigation unit is used to receive the signal sent by the vehicle-mounted signal receiving unit to perform navigation operations, and the route navigation unit sends a positioning signal to the mobile phone terminal.

[0028] Furthermore, the background screening module uses a Selective Search algorithm to perform logical screening on the candidate parking space areas.

[0029] Furthermore, the rotating mechanism includes a rotating motor connected to the pillar and one side of the C-shaped plate, a worm connected to the driving end of the rotating motor and a worm wheel meshingly connected to one side of the worm, one bottom of the worm wheel is rotated and connected to the pillar, and one side of the other worm wheel is connected to a camera.

[0030] In the present invention, the worm wheel and the worm screw cooperate to realize a self-locking function, and the rotation of the worm wheel facilitates the adjustment of the angle of the camera.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] The present invention collects images in a parking lot through a camera, and transmits the images to an image sorting unit for sorting and detection, performs parking lot background modeling through a background modeling unit to obtain an alternative parking space area, filters out artificial landscape images around the parking space through a background screening module, forms a dynamic vehicle motion trajectory image through a vehicle distribution calculation module, performs binarization and clustering operations on the obtained motion trajectory image on the spatial axis through a screening verification unit, filters out clusters with a recognition accuracy rate of less than 50%, extracts features from each of the remaining clusters, and performs accuracy verification on the time axis to determine whether a vehicle is parked in a parking space, selects a parking space closest to the vehicle through a parking space selection unit, and performs navigation operations on the parking space through a route navigation unit, so that the driver can quickly find a parking space. The system performs model training through customized image recognition, processes and analyzes information such as the behavior characteristics of the vehicle collected by the camera, determines whether a car is currently parked in the parking space, and then transmits the data to a client module, so that the position of the empty parking space can be grasped at the first time when entering the parking lot, which saves a lot of time to a certain extent and avoids the congestion caused by vehicles looking for parking spaces on the road.

[0033] The present invention will be explained in detail below with reference to the accompanying drawings and specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 is a flow chart of the parking space retrieval system of the present invention;

[0035] Figure 2 It is a schematic diagram of the overall structure of the present invention;

[0036] Figure 3 for Figure 2 Enlarged view of point A in the middle.

[0037] In the figure: 1. parking lot; 10. data acquisition module; 11. car entry and exit acquisition unit; 12. visual monitoring unit; 2. pillar; 20. model building module; 21. image sorting unit; 22. background modeling unit; 3. rotating mechanism; 30. background screening module; 31. rotating motor; 32. worm; 33. worm gear; 4. C-type plate; 40. vehicle distribution calculation module; 5. camera; 50. parking space judgment module; 51. screening and verification unit; 52. parking space selection unit; 6. parking space retrieval system; 60. route planning module; 70. signal sending module; 80. client module; 81. vehicle signal receiving unit; 82. route navigation unit; 90. mobile terminal. DETAILED DESCRIPTION

[0038] To facilitate understanding of the present invention, the present invention will be described more comprehensively below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings, but the present invention can be implemented in different forms and is not limited to the embodiments described in the text. On the contrary, these embodiments are provided to make the content disclosed in the present invention more thorough and comprehensive.

[0039] It should be noted that when an element is referred to as being "fixed on" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used in this article are for illustrative purposes only.

[0040] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by technicians in the technical field to which the present invention belongs. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0041] For examples, please refer to the attached Figure 1-3 , a parking space retrieval system based on machine vision, comprising a parking lot 1, one side of the parking lot 1 is connected to a pillar 2, the top of the pillar 2 is connected to a rotating mechanism 3, the top of the rotating mechanism 3 is connected to a C-shaped plate 4, the top of the C-shaped plate 4 is rotatably connected to a camera 5, one side of the camera 5 is connected to the rotating mechanism 3, and one side of the top of the camera 5 is connected to a parking space retrieval system 6;

[0042] The parking space retrieval system 6 comprises:

[0043] A data acquisition module 10 is used to acquire images of parking spaces and vehicles in the parking lot 1;

[0044] A model building module 20 is used for image classification, image detection and background modeling to obtain a candidate parking space area;

[0045] A background screening module 30, for finding vehicles around the parking space and filtering artificial landscape images around the parking space;

[0046] The vehicle distribution calculation module 40 is used to calculate the distribution of vehicles on the time axis and form a dynamic vehicle motion trajectory image;

[0047] The parking space determination module 50 is used to screen and verify the correctness of the calculation and select a parking space;

[0048] A route planning module 60, for providing a route for a vehicle to enter a parking space;

[0049] A signal sending module 70, used to send the signal provided by the route planning module 60;

[0050] The client module 80 is used to receive the signal sent by the signal sending module 70 and perform navigation.

[0051] It should be noted that, in the embodiment, the scene images in the parking lot 1 are collected by the camera 5, and the model training is performed by performing customized image recognition on the collected scene images in the parking lot, and the model is uploaded to the model building module 20 to generate an API, and the vehicle distribution calculation module 40 processes and analyzes information such as the behavior characteristics of the vehicle, and on the timeline, the distribution of the vehicle is calculated using the monitoring results of several consecutive frames to form a dynamic vehicle motion trajectory image.

[0052] For examples, please refer to the attached Figure 1 The parking space retrieval system 6 also includes a mobile terminal 90 for receiving a vehicle positioning signal sent by the client module 80 .

[0053] It should be noted that, in the embodiment, the mobile terminal 90 obtains the vehicle positioning information to facilitate the user to find his vehicle and prevent the user from forgetting the location of the vehicle.

[0054] For examples, please refer to the attached Figure 1 , the data acquisition module 10 comprises:

[0055] The car entry and exit collection unit 11 is used to obtain the number of cars in the parking lot 1, and the car entry and exit collection unit 11 sends a signal to the model building module 20;

[0056] The visual monitoring unit 12 is used to obtain the image of the parking lot 1 in real time. The visual monitoring unit 12 sends the image to the model building module 20 . The visual monitoring unit 12 is electrically connected to the camera 5 .

[0057] It should be noted that, in the embodiment, the vehicle entry and exit collection unit 11 is convenient for monitoring the number of vehicles entering and exiting the parking lot 1, so as to facilitate calculation of the number of idle parking spaces.

[0058] For examples, please refer to the attached Figure 1 , the model building module 20 includes:

[0059] An image collating unit 21 is used to sort the images collected by the visual monitoring unit 12 and detect the images;

[0060] The background modeling unit 22 is used to perform background modeling of the parking lot 1 according to the image sorted by the image sorting unit 21 . The background modeling unit 22 sends data to the background screening module 30 , and the background screening module 30 sends data to the vehicle distribution calculation module 40 .

[0061] For examples, please refer to the attached Figure 1 , the parking space determination module 50 includes:

[0062] A screening and verification unit 51 is used to perform binarization and clustering operations on the spatial axis according to the motion trajectory image, screen out clusters with a recognition accuracy greater than 50% and perform verification, and the screening and verification unit 51 receives the motion trajectory image sent by the vehicle distribution calculation module 40;

[0063] The parking space selection unit 52 is used to receive the data sent by the screening and verification unit 51 and select a parking space closest to the vehicle. The parking space selection unit 52 sends a parking space signal to the route planning module 60 .

[0064] It should be noted that, in the embodiment, the parking space selection unit 52 can select the parking space closest to the vehicle, and there is no vehicle in motion near the parking space. The screening and verification unit 51 is convenient for binarizing and clustering the obtained motion trajectory image on the spatial axis, filtering out clusters with an identification accuracy rate of less than 50%, extracting features for each remaining cluster, and performing accuracy verification on the time axis to determine whether the vehicle is parked in the parking space, and then judging whether there is a vehicle in the parking space scene captured by the camera based on the result of the accuracy verification. If a vehicle enters the current position, it is filtered out in the parking area data to obtain the remaining parking spaces.

[0065] For examples, please refer to the attached Figure 1 , the receiving client module 80 includes:

[0066] The vehicle-mounted signal receiving unit 81 is used to receive the signal sent by the signal sending module 70;

[0067] The route navigation unit 82 is used to receive the signal sent by the vehicle-mounted signal receiving unit 81 to perform navigation operations. The route navigation unit 82 sends a positioning signal to the mobile phone terminal 90.

[0068] It should be noted that, in the embodiment, the route navigation unit 82 can navigate according to the route planned by the route planning module 60, so as to facilitate the driver to find the location of the parking space.

[0069] For examples, please refer to the attached Figure 1 The background screening module 30 uses a Selective Search algorithm to perform logical screening on the candidate parking space areas.

[0070] For examples, please refer to the attached Figure 2-3 The rotating mechanism 3 includes a rotating motor 31 connected to the pillar 2 and one side of the C-shaped plate 4, a worm 32 connected to the driving end of the rotating motor 31 and a worm wheel 33 meshingly connected to one side of the worm 32, one bottom of the worm wheel 33 is rotatably connected to the pillar 2, and one side of the other worm wheel 33 is connected to the camera 5.

[0071] It should be noted that, in the embodiment, the worm wheel 33 and the worm 32 cooperate to achieve a self-locking function, and the rotating motor 31 drives the worm 32 to rotate, thereby driving the worm wheel 33 to rotate, and then driving the C-type plate 4 and the camera 5 to rotate, thereby adjusting the angle of the camera 5.

[0072] The specific operation mode of the present invention is as follows: the worm 32 is driven to rotate by the rotating motor 31, thereby driving the worm wheel 33 to rotate, and then driving the C-shaped plate 4 and the camera 5 to rotate, so as to adjust the angle of the camera 5, so that the camera 5 can collect images in the parking lot 1, and transmit the images to the image sorting unit 21 for sorting and detection, and then the parking lot 1 background modeling is performed by the background modeling unit 22 to obtain the candidate parking space area;

[0073] The background screening module 30 uses the Selective Search algorithm to perform logical screening on the candidate parking space area, find the vehicles around the parking space, and filter out the artificial landscape images around the parking space. Then, the vehicle distribution calculation module 40 uses the continuous monitoring results of several frames on the time axis to calculate the distribution of vehicles and form a dynamic vehicle motion trajectory image.

[0074] The obtained motion trajectory image is binarized and clustered on the spatial axis by the screening verification unit 51, and clusters with recognition accuracy less than 50% are filtered out. Features are extracted for each remaining cluster, and accuracy verification is performed on the time axis to determine whether the vehicle is parked in the parking space. Then, based on the result of the accuracy verification, it is determined whether there is a vehicle in the parking space scene captured by the camera. If a vehicle enters the current position, the remaining parking spaces are filtered out in the parking area data, and then the parking space selection unit 52 can select the parking space closest to the vehicle, and the route navigation unit 82 can navigate the parking space to facilitate the driver to quickly find a parking space.

[0075] The above is an exemplary description of the present invention in combination with the accompanying drawings. It is obvious that the specific implementation of the present invention is not limited to the above-mentioned method. As long as such non-substantial improvements are made by adopting the method concept and technical solution of the present invention, or the concept and technical solution of the present invention are directly applied to other occasions without improvement, they are all within the protection scope of the present invention.

Claims

1. A parking space retrieval system based on machine vision, comprising a parking lot (1), characterized in that: One side of the parking lot (1) is connected to a support (2), the top of the support (2) is connected to a rotating mechanism (3), the top of the rotating mechanism (3) is connected to a C-shaped plate (4), the top of the C-shaped plate (4) is rotatably connected to a camera (5), one side of the camera (5) is connected to the rotating mechanism (3), and one side of the top of the camera (5) is connected to a parking space retrieval system (6); The parking space retrieval system (6) comprises: A data acquisition module (10) for acquiring images of parking spaces and vehicles in the parking lot (1); A model building module (20) is used for image classification, image detection and background modeling to obtain a candidate parking space area; A background screening module (30) is used to find vehicles around the parking space and filter the artificial landscape images around the parking space; A vehicle distribution calculation module (40) is used to calculate the distribution of vehicles on a time axis and form a dynamic vehicle motion trajectory image; A parking space determination module (50) is used to screen and verify the correctness of the calculation and select a parking space; A route planning module (60) for providing a route for a vehicle to enter a parking space; A signal sending module (70), used for sending the signal provided by the route planning module (60); The client module (80) is used to receive the signal sent by the signal sending module (70) and perform navigation.

2. The parking space retrieval system based on machine vision according to claim 1, characterized in that: The parking space retrieval system (6) also includes a mobile phone terminal (90) for receiving a vehicle positioning signal sent by the client module (80).

3. The parking space retrieval system based on machine vision according to claim 1, characterized in that: The data acquisition module (10) comprises: A car entry and exit collection unit (11) is used to obtain the number of cars located in the parking lot (1), and the car entry and exit collection unit (11) sends a signal to a model building module (20); A visual monitoring unit (12) is used to obtain images of the parking lot (1) in real time. The visual monitoring unit (12) sends images to a model building module (20). The visual monitoring unit (12) is electrically connected to the camera (5).

4. The parking space retrieval system based on machine vision according to claim 3, characterized in that: The model building module (20) comprises: An image collating unit (21) is used to sort the images collected by the visual monitoring unit (12) and detect the images; A background modeling unit (22) is used to perform background modeling of the parking lot (1) based on the image sorted by the image sorting unit (21); the background modeling unit (22) sends data to a background screening module (30); and the background screening module (30) sends data to a vehicle distribution calculation module (40).

5. The parking space retrieval system based on machine vision according to claim 1, characterized in that: The parking space determination module (50) comprises: A screening and verification unit (51) is used to perform binarization and clustering operations on the spatial axis according to the motion trajectory image, screen out clusters with a recognition accuracy greater than 50% and perform verification, and the screening and verification unit (51) receives the motion trajectory image sent by the vehicle distribution calculation module (40); The parking space selection unit (52) is used to receive the data sent by the screening and verification unit (51) and select the parking space closest to the vehicle. The parking space selection unit (52) sends a parking space signal to the route planning module (60).

6. The parking space retrieval system based on machine vision according to claim 2, characterized in that: The receiving client module (80) comprises: An on-vehicle signal receiving unit (81), used for receiving a signal sent by the signal sending module (70); The route navigation unit (82) is used to receive the signal sent by the vehicle-mounted signal receiving unit (81) to perform navigation operations, and the route navigation unit (82) sends a positioning signal to the mobile phone terminal (90).

7. The parking space retrieval system based on machine vision according to claim 1, characterized in that: The background screening module (30) uses a Selective Search algorithm to perform logical screening on the candidate parking space areas.

8. The parking space retrieval system based on machine vision according to claim 1, characterized in that: The rotating mechanism (3) comprises a rotating motor (31) connected to the support (2) and one side of the C-shaped plate (4), a worm (32) connected to the driving end of the rotating motor (31) and a worm wheel (33) meshingly connected to one side of the worm (32), the bottom of one of the worm wheels (33) being rotatably connected to the support (2), and one side of the other worm wheel (33) being connected to a camera (5).

Images