Image recognition method for realizing accurate parking

Abstract

The image recognition method for realizing accurate parking is characterized by comprising a shooting module, a comparison and analysis module, a comparison data uploading module, a data storage unit and a communication module, wherein the shooting module and the comparison and analysis module are connected, the comparison and analysis module is connected with the comparison data uploading module and the communication module respectively, and the data storage unit is connected with the comparison and analysis module. The application mainly collects the ground mark and identifies the single. The scheme ensures the uniqueness of the vehicle through the synchronous identification of the vehicle mark and the ground mark, eliminates cheating, and finally ensures the fixed-point and directional parking of the vehicle. The application adopts a ground mark information identification unit, a vehicle coding information identification unit, a ground mark angle acquisition unit, a vehicle mark angle acquisition unit and an offset calculation unit to process the vehicle mark and the ground mark, and analyze the exact parking position, thereby ensuring the accurate parking.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of image recognition, in particular to an image recognition method for realizing precise parking. BACKGROUND

[0002] The existing image recognition method mainly recognizes ground markings. The Chinese patent with application number CN202110007623.1, a method for fixed-point parking of shared electric vehicles, points out that "when the user performs the parking operation on the terminal device including a mobile phone, the server acquires the current GPS positioning of the shared electric vehicle, searches for the nearest parking point P point from the shared electric vehicle, judges whether the shared electric vehicle is within the P point range, if not, the server returns the information that parking is not allowed to the user; if the shared electric vehicle is within the P point range, the server sends a shooting instruction to the camera device, the camera device captures the current image, and transmits it to the server; the server imports the captured image into the pre-trained model, and divides the parking area and its markings in the image through the full convolution network FCN model; if the divided image does not detect the parking area markings, the server returns the information that parking is not allowed to the user." The above technology captures the current image by the camera device, the camera belongs to external equipment, and this technology cannot prevent cheating. The user can easily cheat the camera by using a photo, and the server cannot confirm whether the received photo is used by the user, which is easy to lead to misjudgment, so it cannot guarantee the accuracy of parking. SUMMARY

[0003] According to the above technical problem, the present application provides an image recognition method for realizing precise parking to ensure the fixed-point and directional parking of vehicles.

[0004] The image recognition method for realizing precise parking comprises a shooting module, a comparative analysis module, a comparative data uploading module, a data storage unit, and a communication module. The shooting module and the comparative analysis module are connected. The comparative analysis module is connected with the comparative data uploading module and the communication module respectively. The data storage unit is connected with the comparative analysis module.

[0005] The comparative analysis module comprises a ground marking information recognition unit, a vehicle coding information recognition unit, a ground marking angle acquisition unit, a vehicle marking angle acquisition unit, an offset calculation unit, and an error reporting unit. The error reporting unit is connected with the ground marking information recognition unit, the vehicle coding information recognition unit, the vehicle marking angle acquisition unit, the ground marking angle acquisition unit, and the offset calculation unit respectively. The vehicle marking angle acquisition unit and the ground marking angle acquisition unit are connected with the offset calculation unit respectively. The data storage unit is connected with the ground marking information recognition unit, the vehicle coding information recognition unit, the ground marking angle acquisition unit, and the vehicle marking angle acquisition unit respectively. The error reporting unit is connected with the communication module.

[0006] The ground mark information recognition unit performs mapping processing on the pixel value of the image to be recognized by using a mapping function to obtain an image pixel matrix; converts the original image color space into HSV, sets the upper and lower limits of HSV according to the base color, performs threshold segmentation, obtains the minimum circumscribed rectangle, and compares it with the internally stored rectangle size to obtain the specific ground mark information.

[0007] The vehicle coding information recognition unit performs convolution operation on the image pixel matrix and the Gaussian Laplace edge monitoring operator, compares the feature map with the internally stored features of the vehicle identification information recognition unit, and obtains the vehicle information.

[0008] The vehicle identification angle acquisition unit compares the vehicle identification in the obtained picture with the vehicle identification recorded in the system, and acquires the inclination angle of the vehicle identification in the picture.

[0009] The ground mark angle acquisition unit compares the ground mark in the obtained picture with the ground mark recorded in the system, and acquires the inclination angle of the ground mark in the picture.

[0010] The offset calculation unit processes the inclination angles acquired by the vehicle identification angle acquisition unit and the ground mark angle acquisition unit, extracts the edge data of the ground mark and the vehicle identification, calculates the offset angle and offset displacement of the vehicle identification and the ground mark according to the edge data.

[0011] The image recognition method for precise parking is realized, and the specific operation method is as follows:

[0012] The image recognition system for precise parking guides the user to take a picture through the shooting module and sends the picture data to the shooting module.

[0013] The shooting module sends the picture to the comparative analysis module, and the comparative analysis module analyzes the picture. When the ground mark information recognition unit obtains the ground mark, the ground mark information recognition unit identifies the ground mark, matches the color bar, divides the color bar into color 1 and color bar 2, calculates the width of each color bar and records it as w1 and w2, and judges whether there is a ground mark feature in the image area when the color of the color bar and the initial setting data color are consistent and w1 and w2 are within 20% of the initial setting width. When the color of the color bar and the initial setting data color are inconsistent or w1 and w2 exceed the setting width of 20%, the ground mark information recognition unit returns error data to the communication module of the image recognition system for precise parking, and the communication module prompts the user to take a picture again.

[0014] The ground mark information recognition unit obtains the ground mark recognition correct, and identifies the vehicle mark. Each set of mark on the vehicle is represented by four colors. The vehicle coding information recognition unit obtains the four-color coding, and the color number is red 0x00, green 0x01, blue 0x02, and pink 0x03. The data storage unit of the image recognition system for precise parking transmits the four-color coding corresponding to the vehicle to the comparative analysis module.

[0015] The ground mark information recognition unit of the comparative analysis module identifies the four-color line arrangement of the vehicle mark, matches the color, area, and width.

[0016] The vehicle mark recognition is correct, the four-color coding is obtained according to the color line arrangement sequence.

[0017] The coding is compared with the comparative analysis module initially transmitted by the image recognition system for precise parking.

[0018] If the coding cannot be matched, the error data is returned to the image recognition system for precise parking, and the image recognition system for precise parking prompts the user to take a new photo.

[0019] The color arrangement data of the ground mark is confirmed, and the parking direction of the vehicle is confirmed.

[0020] After normal matching, the edge data of the ground mark and the vehicle mark is extracted.

[0021] According to the ground mark edge data, the ground mark angle obtaining unit obtains the ground mark angle A1 according to the included angle between the edge data and the horizontal line. The ground mark information recognition unit maps the pixel value of the to-be-recognized image by using a mapping function to obtain an image pixel matrix. The original image color space is converted into HSV, threshold segmentation is performed according to the upper and lower limits of the base color, a minimum circumscribed rectangle is obtained, and the middle coordinate point B1 (x1, y1) of the rectangle is obtained. The vehicle mark angle obtaining unit obtains the vehicle mark angle A2 according to the included angle between the vehicle mark edge data and the horizontal line. The vehicle coding information recognition unit performs convolution operation on the image pixel matrix and the Gaussian Laplace edge monitoring operator, compares the feature atlas with the features stored in the vehicle mark information recognition unit, obtains the comparison information, obtains the minimum circumscribed rectangle, and obtains the middle coordinate point B2 (x2, y2) of the rectangle. The offset angle of the vehicle mark and the ground mark is obtained by A2-A1, and the offset displacement is obtained by the distance between the points B2 and B1.

[0022] The ground mark angle obtaining unit and the vehicle mark angle obtaining unit return the data to the comparative analysis module.

[0023] The comparative analysis module confirms whether the precise parking requirement is met according to the offset angle and the offset displacement threshold value actually set by the user, and transmits the information to the communication module to notify the user.

[0024] The application has the advantages that: the application mainly collects the ground mark, and the recognition is single, the scheme ensures the uniqueness of the vehicle through the synchronous recognition of the vehicle mark and the ground mark, prevents cheating, and finally ensures the fixed-point and directional parking of the vehicle. The ground mark information recognition unit, the vehicle coding information recognition unit, the ground mark angle acquisition unit, the vehicle mark angle acquisition unit and the offset calculation unit are used to process the vehicle mark and the ground mark, and the exact parking position is obtained through analysis, so that the accurate parking is ensured. BRIEF DESCRIPTION OF DRAWINGS

[0025] Figure 1 It is a ground mark and parking space schematic diagram of the application.

[0026] Figure 2 It is a vehicle mark position schematic diagram on the vehicle of the application.

[0027] Figure 3 It is a vehicle mark schematic diagram of the application.

[0028] Figure 4A It is a schematic diagram of the front part of the working process of the comparative analysis module of the application.

[0029] Figure 4B It is a schematic diagram of the rear part of the working process of the comparative analysis module of the application.

[0030] Figure 5 It is a structural schematic diagram of the application.

[0031] Figure 6 It is a schematic diagram of the internal structure of the comparative analysis module of the application. EMBODIMENT

[0032] The application is described according to the drawings:

[0033] The example provides a system image recognition method, and solves the fixed-point and directional parking of a vehicle, and is directed to electric vehicles, bicycles, three-wheeled electric vehicles, four-wheeled electric vehicles and the like.

[0034] Embodiment 1

[0035] A ground mark line is arranged on the parking space, and a vehicle mark is arranged on the vehicle, when the user parks the vehicle on the parking space, the image recognition system for realizing accurate parking on the mobile phone is used for parking recognition, and the specific operation method is as follows:

[0036] The shared bicycle is parked in the parking space, and then the image recognition system for precise parking on the mobile phone guides the user to take a photo and sends the picture data to the shooting module. The shooting module sends the picture to the comparative analysis module. The comparative analysis module analyzes the above picture. The ground mark information recognition unit does not obtain the ground mark. The comparative analysis module feeds back information to the communication module. The communication module sends a message that the parking is unqualified. Please park correctly.

[0037] Embodiment 2

[0038] The ground mark line is set on the parking space, and the vehicle mark is set on the vehicle. When the user parks the vehicle on the parking space, the image recognition system for precise parking on the mobile phone is used for parking recognition. The specific operation method is as follows:

[0039] The shared bicycle is parked in the parking space, and then the image recognition system for precise parking on the mobile phone guides the user to take a photo and sends the picture data to the shooting module. The shooting module sends the picture to the comparative analysis module. The comparative analysis module analyzes the above picture. The ground mark information recognition unit does not obtain the ground mark. Figure 2 ,

[0040] The shooting module sends the picture to the comparative analysis module. The comparative analysis module analyzes the above picture. The ground mark information recognition unit obtains the ground mark.

[0041] When identifying the vehicle mark, each group of mark lines on the vehicle is represented by four colors. The vehicle mark information recognition unit obtains the four-color code, whose color number is red 0x00, green 0x01, blue 0x02, and pink 0x03. The data storage unit of the image recognition system for precise parking transmits the corresponding four-color code of the vehicle to the comparative analysis module. The ground mark information recognition unit of the comparative analysis module identifies the four-color line arrangement of the vehicle mark, matches the color, area, and width. The vehicle mark is correctly identified. The four-color code sequence is obtained from the color line arrangement sequence. The code sequence is compared with the code transmitted by the image recognition system for precise parking to the comparative analysis module.

[0042] The encoding is matched. After the matching is completed, the vehicle identity is confirmed. The comparative analysis module confirms the parking direction of the vehicle by confirming the color arrangement data of the ground mark.

[0043] After normal matching, edge data of ground mark and vehicle mark are extracted; according to the edge data of ground mark, a ground mark angle obtaining unit obtains a ground mark angle A1 according to an included angle between the edge data and a horizontal line, a ground mark information recognition unit performs mapping processing on pixel values of a to-be-recognized image by using a mapping function to obtain an image pixel matrix; an original image color space is converted into HSV, threshold segmentation is performed according to upper and lower limits of HSV set by a basic color to obtain a minimum circumscribed rectangle, and a middle coordinate point B1 (x1, y1) of the rectangle is obtained, a vehicle mark angle obtaining unit obtains a vehicle mark angle A2 according to an included angle between edge data of the vehicle mark and the horizontal line, a vehicle coding information recognition unit performs convolution operation on the image pixel matrix and a Gaussian Laplace edge monitoring operator, a feature map and internally stored features of the vehicle mark information recognition unit are compared to obtain comparison information, a minimum circumscribed rectangle is obtained, and a middle coordinate point B2 (x2, y2) of the rectangle is obtained, a deviation angle of the vehicle mark and the ground mark is obtained from A2-A1, and a deviation displacement is obtained from a distance between the points B2 and B1.

[0044] The ground mark angle obtaining unit and the vehicle mark angle obtaining unit return data to a comparison and analysis module;

[0045] The comparison and analysis module confirms that a precise parking requirement is met according to a deviation angle and a deviation displacement threshold value set by a user, and transmits information to a communication module to notify the user.

[0046] The above shows and describes the basic principles and main features of the present application and the advantages of the present application. The various components mentioned in the present application are common technologies in the prior art, and those skilled in the art should understand that the present application is not limited by the above embodiments, and the above embodiments and descriptions in the specification are only to illustrate the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the present application. The scope of protection of the present application is defined by the appended claims and their equivalents.

Claims

1. An image recognition method for realizing precision parking, characterized in that The operation is carried out by an image recognition system for realizing accurate parking, which comprises a shooting module, a comparative analysis module, a comparative data uploading module, a data storage unit and a communication module. The shooting module and the comparative analysis module are connected, the comparative analysis module is connected with the comparative data uploading module and the communication module respectively, and the data storage unit is connected with the comparative analysis module. The comparative analysis module comprises a ground identification information recognition unit, a vehicle coding information recognition unit, a ground identification angle acquisition unit, a vehicle identification angle acquisition unit, an offset calculation unit and an error reporting unit. The error reporting unit is connected with the ground identification information recognition unit, the vehicle coding information recognition unit, the vehicle identification angle acquisition unit, the ground identification angle acquisition unit and the offset calculation unit respectively. The vehicle identification angle acquisition unit and the ground identification angle acquisition unit are connected with the offset calculation unit respectively. The data storage unit is connected with the ground identification information recognition unit, the vehicle coding information recognition unit, the ground identification angle acquisition unit and the vehicle identification angle acquisition unit respectively. The error reporting unit is connected with the communication module. The specific operation method is as follows: before the image recognition system for realizing accurate parking is used, the staff implants the recognition data into the system. The image recognition system for realizing accurate parking guides the user to take a picture through the shooting module and sends the picture data to the shooting module. The shooting module sends the picture to the comparative analysis module. The comparative analysis module analyzes the picture. When the ground identification information recognition unit obtains the ground identification, the ground identification information recognition unit identifies the ground identification, matches the color bar of the ground identification, divides the color bar into color bar 1 and color bar 2, calculates the width of each color bar and marks it as w1 and w2 respectively, judges whether there is a ground identification feature in the image area when the color of the color bar is consistent with the initial data color implanted in the image recognition system for realizing accurate parking and w1 and w2 are within 20% of the initial setting width, and returns error data to the communication module of the image recognition system for realizing accurate parking when the color of the color bar is inconsistent with the initial setting data color or w1 and w2 exceed the setting width of 20%. The communication module prompts the user to take a picture again. When the ground identification information recognition unit correctly obtains the ground identification, the vehicle identification is identified. Each group of identification on the vehicle is represented by four colors. The vehicle coding information recognition unit obtains the four-color coding, the color number of which is red 0x00, green 0x01, blue 0x02 and pink 0x03. The data storage unit of the image recognition system for realizing accurate parking transmits the corresponding four-color coding of the vehicle to the comparative analysis module. The ground identification information recognition unit of the comparative analysis module identifies the four-color line arrangement of the vehicle identification, matches the color, area and width. If the corresponding identification is not identified, error data is returned to the communication module of the image recognition system for realizing accurate parking, and the communication module prompts the user to take a picture again. The vehicle identification is correctly identified by the color line arrangement sequence, and the four-color coding is obtained. The coding is compared with the coding transmitted by the image recognition system for realizing accurate parking to the comparative analysis module. If it cannot be matched, return error data to the image recognition system for precise parking, and the image recognition system for precise parking prompts the user to take a new photo; Confirm the color arrangement data of the ground mark, and confirm the parking direction of the vehicle; After normal matching, the edge data of the ground mark and the vehicle mark are extracted; According to the edge data of the ground mark, the ground mark angle acquisition unit obtains the ground mark angle A1 according to the angle between the edge data and the horizontal line, the ground mark information recognition unit performs mapping processing on the pixel value of the to-be-recognized image by using a mapping function to obtain an image pixel matrix, converts the color space of the original image into HSV, sets the upper and lower limits of HSV according to the basic color to perform threshold segmentation, obtains the minimum circumscribed rectangle, and obtains the middle coordinate point B1 (x1, y1) of the rectangle, the vehicle mark angle acquisition unit obtains the vehicle mark angle A2 according to the angle between the edge data of the vehicle mark and the horizontal line, the vehicle coding information recognition unit performs convolution operation on the image pixel matrix and the Gaussian Laplace edge monitoring operator, compares the feature atlas with the features stored in the vehicle mark information recognition unit, obtains the minimum circumscribed rectangle, and obtains the middle coordinate point B2 (x2, y2) of the rectangle, and the offset angle of the vehicle mark and the ground mark is obtained from A2-A1, and the offset displacement is obtained from the distance between the points B2 and B1. The ground mark angle acquisition unit and the vehicle mark angle acquisition unit return the data to the comparison and analysis module; The comparison and analysis module confirms whether the precise parking requirement is met according to the offset angle and the offset displacement threshold value set by the user, and transmits the information to the communication module to notify the user.

2. The image recognition method for achieving precision parking according to claim 1, wherein The ground mark information recognition unit performs mapping processing on the pixel value of the to-be-recognized image by using a mapping function to obtain an image pixel matrix, converts the color space of the original image into HSV, sets the upper and lower limits of HSV according to the basic color to perform threshold segmentation, obtains the minimum circumscribed rectangle, and compares the internal storage rectangle size to obtain the specific information of the ground mark.

3. The image recognition method for achieving precision parking according to claim 2, wherein The vehicle coding information recognition unit performs convolution operation on the image pixel matrix and the Gaussian Laplace edge monitoring operator, compares the feature atlas with the features stored in the vehicle mark information recognition unit, and obtains the vehicle information.

4. The image recognition method for achieving precision parking according to claim 2, wherein The vehicle mark angle acquisition unit compares the vehicle mark in the already obtained picture with the vehicle mark recorded in the system, and obtains the inclination angle of the vehicle mark in the picture. The ground mark angle acquisition unit compares the ground mark in the already obtained picture with the ground mark recorded in the system, and obtains the inclination angle of the ground mark in the picture. The offset calculation unit processes the inclination angles obtained by the vehicle mark angle acquisition unit and the ground mark angle acquisition unit, extracts the edge data of the ground mark and the vehicle mark, and calculates the offset angle and the offset displacement of the vehicle mark and the ground mark according to the edge data.

Images