Intelligent pre-check self-service baggage check-in terminal system and method based on visual image technology

Abstract

The application relates to an intelligent pre-check self-service luggage consigning terminal system based on visual image technology, and relates to the field of image processing.The terminal system comprises a size detection module which is used for detecting whether the shape and size of target luggage meet consigning standards by using a depth visual camera; and an intelligent identification module which is used for collecting a current scanning picture obtained by performing scanning on target luggage by an X-ray scanning machine, and performing intelligent identification on whether each object in the current scanning picture belongs to a prohibited article type.The application also relates to an intelligent pre-check self-service luggage consigning method.By the application, intelligent identification of whether each piece of luggage contains a prohibited article can be performed by using a customized intelligent identification model, and the intelligent identification of the prohibited article is organically combined with detection of the appearance, size, type and weight of the luggage, so that all detection parameters of the luggage can be detected by a passenger at one place.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of image processing, and in particular to an intelligent pre-check self-service baggage check-in terminal system and method based on visual image technology. BACKGROUND

[0002] In the process of self-service baggage check-in of passengers at the airport check-in, the luggage carried by the passengers needs to be detected to obtain the relevant information of the luggage and to determine whether the luggage meets the check-in requirements. For example, various detection technologies such as appearance, size, type and weight of the luggage are used to detect and determine whether the appearance, size, type and weight of each piece of luggage carried by each passenger meet the requirements respectively. Only when all the detection indicators meet the requirements, each piece of luggage carried by each passenger is allowed to be released and transported to the conveying belt machine to be sent to the conveying system leading to the airport cargo vehicle.

[0003] For example, the Chinese utility model patent publication CN205486359U proposes a multifunctional self-service baggage check-in intelligent terminal, which includes a box body, a main box embedded in the front end side of the box body, and a display device, a baggage label scanning device, a baggage conveying device and a baggage specification detection device connected in sequence with the main box. The box body of the multifunctional self-service baggage check-in intelligent terminal is a traditional manual counter. The display device includes a support arm installed on one side of the box body and a display installed on the support arm. The display can rotate around the support arm as the center. The terminal realizes the organic combination of manual counter and self-service check-in. When the self-service baggage check-in terminal is operating, the manual counter can provide services such as accepting consultation and handling boarding pass for passengers. The counter staff only needs to rotate the display device to the appropriate position to assist passengers in handling self-service check-in. In this way, it is convenient for passengers to handle check-in, saves time, improves passenger experience, and maximizes benefits.

[0004] For example, the Chinese invention patent publication CN112164186A proposes a check-in and baggage check-in method of an online terminal of an ATIS intelligent terminal of a terminal building. The method uses an online terminal of an ATIS intelligent terminal of a terminal building, a passenger client, a city terminal building check-in service client, a handheld terminal client, a monitor, a backup intelligent terminal and a baggage storage and check-in self-service site. The passenger scans a check-in two-dimensional code to reserve check-in service. The online terminal of the ATIS intelligent terminal of the terminal building judges the source of the baggage check-in two-dimensional code scanned by the passenger and sends the passenger information to the city terminal building check-in service client. The city terminal building service client performs online seat selection check-in, reduces passenger waiting time, reduces pressure and risk of channel congestion caused by centralization, and effectively improves the travel efficiency of passengers. The baggage storage and check-in self-service site provides baggage check-in service for passengers, improving the efficiency and experience of passenger travel.

[0005] Currently, the self-service baggage check-in device is responsible for interacting with the airport departure system and the baggage conveying system, and prompts the dangerous goods of the passenger's baggage at the device end while completing the baggage check-in service of the passenger. If the passenger is not clear about the requirements of carrying dangerous goods or forgets whether he carries dangerous goods, he needs to move to the baggage unpacking room after completing the baggage check-in service process to unpack and check the baggage that does not pass the security check and remove the dangerous goods. As can be seen, the self-service baggage check-in device in the prior art can only detect the appearance, size, type and weight of the baggage, and has no detection and prompting function for prohibited articles. This separate detection mode and the detection difficulty of prohibited articles increase the business handling link of the passenger and reduce the speed and efficiency of the passenger boarding service process. SUMMARY

[0006] In order to solve the technical problems in the prior art, the present application provides an intelligent pre-check self-service baggage check-in terminal system and method based on visual image technology, which can use a customized intelligent recognition model to perform intelligent recognition processing on whether there are prohibited articles in each piece of baggage carried by each passenger based on the targeted screening of each item of basic data. On this basis, the intelligent recognition processing of the prohibited articles of the baggage is organically combined with the detection of the appearance, size, type and weight of the baggage to ensure that the passenger can complete the detection of all detection parameters of the baggage at one place, thereby improving the speed and efficiency of the passenger boarding service process.

[0007] According to one aspect of the present application, an intelligent pre-check self-service baggage check-in terminal system based on visual image technology is provided, which comprises:

[0008] A certificate scanning module is configured to scan the identity certificate of the current passenger to obtain the identity information of the current passenger when the current passenger checks in baggage, and print a boarding pass and a baggage tag corresponding to the obtained identity information;

[0009] A barcode recognition module is connected with the certificate scanning module and configured to use a high-definition industrial camera to recognize the barcode of the baggage tag of the target baggage of the current passenger located in the baggage detection area to obtain the baggage tag code corresponding to the target baggage;

[0010] A size detection module is connected with the barcode recognition module and configured to use a depth vision camera to detect the shape and size of the target baggage, and send a first detection signal when the shape and size of the target baggage meet the check-in standard, or send a second detection signal otherwise;

[0011] The intelligent identification module is connected with the size detection module, and is configured to collect a current scanning picture obtained by performing scanning on the target luggage by the X-ray scanner, analyze each image area occupied by each object in the current scanning picture, and perform the following intelligent identification operation on each object as a to-be-identified object: adopting a convolutional neural network learned for multiple times, based on each pixel point in the image area corresponding to the to-be-identified object corresponding to each part of cyan component value, each part of magenta component value, each part of yellow component value and each part of black component value, each pixel point in the image area corresponding to the to-be-identified object corresponding to each part of vertical coordinate value and each part of horizontal coordinate value, multiple edge pixel points in the image area corresponding to the to-be-identified object corresponding to multiple parts of pixel value gradient, the number of pixel rows of the current scanning picture, the number of pixel columns of the current scanning picture and the signal-to-noise ratio of the current scanning picture, intelligently identifying the object type corresponding to the to-be-identified object; the intelligent identification module is further configured to send a first identification signal when there is more than one type of prohibited article in each object type corresponding to each object in the current scanning picture, or send a second identification signal.

[0012] According to another aspect of the present application, an intelligent pre-check self-service luggage check-in method based on visual image technology is provided, the method comprising:

[0013] When the current passenger checks in the luggage, the identity document of the current passenger is scanned to obtain the identity information of the current passenger, and the boarding pass and the luggage tag corresponding to the obtained identity information are printed;

[0014] The high-definition industrial camera is used to identify the bar code of the luggage tag of the target luggage of the current passenger in the luggage detection area, so as to obtain the luggage tag code corresponding to the target luggage;

[0015] The depth vision camera is used to detect the shape and size of the target luggage, and a first detection signal is sent when the shape and size of the target luggage meet the check-in standard, or a second detection signal is sent when the shape and size of the target luggage do not meet the check-in standard;

[0016] The current scanning picture obtained by the acquisition X-ray scanning machine performing scanning on the target luggage is parsed to analyze the image area occupied by each object in the current scanning picture, and each object is taken as a to-be-identified object to perform the following intelligent identification operation: the to-be-identified object is identified based on the following information: the respective cyan component value, the respective magenta component value, the respective yellow component value and the respective black component value of each pixel point in the image area corresponding to the to-be-identified object, the respective vertical coordinate value and the respective horizontal coordinate value of each pixel point in the image area corresponding to the to-be-identified object, the respective pixel value gradient of a plurality of edge pixel points in the image area corresponding to the to-be-identified object, the number of pixel rows and the number of pixel columns of the current scanning picture, and the signal-to-noise ratio of the current scanning picture.

[0017] When there is more than one type of prohibited article in the respective types of objects corresponding to each object in the current scanning picture, a first identification signal is sent, otherwise, a second identification signal is sent.

[0018] When the second detection signal or the first identification signal is received, the target luggage is returned to the luggage detection area, and when the first detection signal or the second identification signal is received, the target luggage is transmitted to the conveyor belt machine to be sent to the row conveying system leading to the airport cargo vehicle.

[0019] When there is more than one type of prohibited article in the respective types of objects corresponding to each object in the current scanning picture, a first identification signal is sent, otherwise, a second identification signal is sent.

[0020] Therefore, the present application has at least the following four key points:

[0021] Point one: the intelligent pre-check self-service luggage check-in terminal system for the airport, the intelligent identification module is supplemented to complete the intelligent identification of whether the current passenger's each piece of luggage as the target luggage carries the prohibited article, and when the target luggage carries the prohibited article or the shape or size of the target luggage does not meet the check-in standard, the target luggage is returned to the luggage detection area, otherwise, the target luggage is transmitted to the conveyor belt machine to be sent to the luggage sorting system, so that the appearance and size of the luggage are detected when the self-service luggage check-in business is handled, and the luggage is checked and prompted for the prohibited article, so as to help the airport management party to obtain more comprehensive luggage information data, improve the airport operation safety, and reduce the subsequent business handling link of the passenger, improve the efficiency of the business process of the passenger before boarding.

[0022] The second point is that the intelligent identification model with a customized structure is used to complete the intelligent identification of the type of the prohibited article, and the structure customization of the intelligent identification model mainly reflects that the intelligent identification model is a convolutional neural network after multiple learning, and the number of learning times of the convolutional neural network is positively correlated with the scanning resolution of the X-ray scanner, so that different structures of the intelligent identification model are built for different X-ray scanners, and the stability and reliability of the intelligent identification of the prohibited article are ensured.

[0023] The third point is that sufficient and comprehensive multiple basic data are introduced to perform the intelligent identification of the type of the prohibited article, the multiple basic data include the respective cyan component values, magenta component values, yellow component values and black component values of each pixel point of the image area corresponding to the to-be-identified article, the respective vertical coordinate values and horizontal coordinate values of each pixel point of the image area corresponding to the to-be-identified article, the respective pixel value gradients of multiple edge pixel points in the image area corresponding to the to-be-identified article, the number of pixel rows and the number of pixel columns of the current scanning picture, and the signal-to-noise ratio of the current scanning picture, so as to further ensure the stability and reliability of the intelligent identification of the prohibited article.

[0024] The fourth point is that in each learning action performed on the convolutional neural network, the type of the identified article in the historical scanning picture obtained by the historical scanning of the X-ray scanner is taken as the single output content of the convolutional neural network, and the respective cyan component values, magenta component values, yellow component values and black component values of each pixel point of the image area occupied by the identified article in the historical scanning picture, the respective vertical coordinate values and horizontal coordinate values of each pixel point of the image area occupied by the identified article in the historical scanning picture, the respective pixel value gradients of multiple edge pixel points in the image area occupied by the identified article in the historical scanning picture, the number of pixel rows and the number of pixel columns of the historical scanning picture, and the signal-to-noise ratio of the historical scanning picture are taken as the item-by-item output content of the convolutional neural network, the current learning action performed on the convolutional neural network is completed, and the learning effect of each learning action of the convolutional neural network is ensured. BRIEF DESCRIPTION OF DRAWINGS

[0025] The embodiments of the present application will be described below with reference to the accompanying drawings, in which:

[0026] Figure 1 It is an outline view of the hardware structure based on which the intelligent pre-check self-service baggage check-in terminal system and method based on visual image technology according to the present application are based.

[0027] Figure 2A technical flow chart of an intelligent pre-check self-service baggage check-in terminal system and method based on visual image technology according to the present application.

[0028] Figure 3 An internal structure diagram of an intelligent pre-check self-service baggage check-in terminal system based on visual image technology according to the first embodiment of the present application.

[0029] Figure 4 An internal structure diagram of an intelligent pre-check self-service baggage check-in terminal system based on visual image technology according to the second embodiment of the present application.

[0030] Figure 5 An internal structure diagram of an intelligent pre-check self-service baggage check-in terminal system based on visual image technology according to the third embodiment of the present application.

[0031] Figure 6 An internal structure diagram of an intelligent pre-check self-service baggage check-in terminal system based on visual image technology according to the fourth embodiment of the present application.

[0032] Figure 7 An internal structure diagram of an intelligent pre-check self-service baggage check-in terminal system based on visual image technology according to the fifth embodiment of the present application.

[0033] Figure 8 A step flow chart of an intelligent pre-check self-service baggage check-in method based on visual image technology according to the sixth embodiment of the present application. DETAILED DESCRIPTION

[0034] Figure 1 An external appearance diagram of a hardware structure based on which an intelligent pre-check self-service baggage check-in terminal system and method based on visual image technology according to the present application is based.

[0035] As Figure 2 shown, a technical flow chart of an intelligent pre-check self-service baggage check-in terminal system and method based on visual image technology according to the present application is given.

[0036] In Figure 2 particular, the specific technical flow of the present application is as follows:

[0037] Technical flow A: design a customized structure intelligent recognition model to complete the intelligent recognition of whether the target baggage carries prohibited articles when each piece of baggage of the current passenger is taken as the target baggage;

[0038] Specifically, the structure customization of the intelligent recognition model mainly manifests in the following aspects:

[0039] Firstly, the intelligent recognition model is a convolutional neural network after multiple learning;

[0040] Second: The number of times the convolutional neural network has undergone learning is positively correlated with the scanning resolution of the X-ray scanner;

[0041] Third: In each learning action performed on the convolutional neural network, the object type of the identified object in the historical scan image obtained from the X-ray scanner is used as the single output content of the convolutional neural network. The values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel point in the image area occupied by the identified object in the historical scan image, the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel point in the image area occupied by the identified object in the historical scan image, the multiple pixel value gradients corresponding to multiple edge pixels in the image area occupied by the identified object in the historical scan image, the number of pixel rows, the number of pixel columns in the historical scan image, and the signal-to-noise ratio of the historical scan image are used as the item-by-item output content of the convolutional neural network to complete the current learning action performed on the convolutional neural network.

[0042] The design of the above-mentioned customized structures allows for the construction of intelligent recognition models with different structures for different X-ray scanners, thereby ensuring the stability and reliability of intelligent recognition of prohibited items.

[0043] Technical Process B: Introduce comprehensive and sufficient basic data to intelligently identify whether each piece of luggage of the current passenger contains prohibited items when it is the target luggage;

[0044] Specifically, the multiple basic data include the values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel in the image area corresponding to the object to be identified in the current scan image of the luggage; the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel in the image area corresponding to the object to be identified; the multiple pixel value gradients corresponding to multiple edge pixels in the image area corresponding to the object to be identified; the number of pixel rows and the number of pixel columns in the current scan image; and the signal-to-noise ratio of the current scan image.

[0045] The comprehensive introduction of the aforementioned basic data further ensures the stability and reliability of intelligent identification of prohibited items;

[0046] Technical Process C: The intelligent recognition model, which adopts the customized structure design of Technical Process A, is based on multiple basic data selected by Technical Process B to intelligently identify whether each piece of luggage of the current passenger contains prohibited items when it is considered as target luggage.

[0047] Technical Process D: The intelligent identification module that executes Technical Process C will be added to the existing intelligent pre-inspection self-service baggage check-in terminal system of the airport to help airport management obtain baggage information data more comprehensively;

[0048] Specifically, in the intelligent pre-inspection self-service baggage check-in terminal system after the intelligent identification module is added, when the target baggage is found to contain prohibited items or the shape or size of the target baggage does not meet the check-in standards, the target baggage is returned to the baggage inspection area; otherwise, the target baggage is transferred to the conveyor belt to be sent to the baggage sorting system. Thus, when processing self-service baggage check-in, the appearance and size of the baggage are checked, and prohibited items are inspected and prompted at the same time.

[0049] In this way, all parameters of the target baggage are checked in one place, which improves the safety of airport operations, reduces the number of subsequent business processing steps for passengers, and improves the efficiency of the pre-boarding business process for passengers.

[0050] The key points of this invention are: the organic integration of the intelligent identification module with the existing intelligent pre-inspection self-service baggage check-in terminal system of airports, the design of multiple customized structures of the intelligent identification model, and the full and comprehensive introduction of multiple basic data for the intelligent identification of prohibited items.

[0051] The present invention will now be described in detail by way of embodiments of an intelligent pre-inspection self-service baggage check-in terminal system and method based on visual image technology.

[0052] First Embodiment

[0053] Figure 3 This is an internal structural diagram of an intelligent pre-inspection self-service baggage check-in terminal system based on visual image technology, as shown in the first embodiment of the present invention.

[0054] like Figure 3 As shown, the intelligent pre-inspection self-service baggage check-in terminal system based on visual image technology includes the following components:

[0055] The document scanning module is used to scan the current passenger's ID document to obtain the current passenger's identity information when the passenger is checking in baggage, and to print the passenger's boarding pass and baggage tag corresponding to the obtained identity information;

[0056] For example, when a passenger is checking in their baggage, scanning the passenger's ID to obtain the passenger's identity information and printing the passenger boarding pass and baggage tag corresponding to the obtained identity information includes: scanning the passenger's ID to obtain the passenger's identity information, accessing a network database to collect the passenger boarding pass and baggage tag corresponding to the obtained identity information, and printing the passenger boarding pass and baggage tag corresponding to the obtained identity information.

[0057] A barcode recognition module, connected to the document scanning module, is used to use a high-definition industrial camera to recognize the barcode of the baggage tag of the current passenger's target baggage located in the baggage inspection area, so as to obtain the baggage barcode corresponding to the target baggage.

[0058] Specifically, a high-definition industrial camera is used to identify the baggage tag of the current passenger's target baggage located in the baggage detection area to obtain the baggage barcode corresponding to the target baggage. The high-definition industrial camera has a built-in high-definition camera mechanism, an image processing mechanism, a barcode output mechanism, and a parallel data bus.

[0059] The size detection module, connected to the barcode recognition module, is used to detect the shape and size of the target luggage using a depth vision camera, and to issue a first detection signal when the shape and size of the target luggage meet the baggage check-in standards; otherwise, it issues a second detection signal.

[0060] For example, a depth vision camera is used to detect the shape and size of the target luggage, and a first detection signal is issued when the shape and size of the target luggage meet the baggage check-in standards; otherwise, a second detection signal is issued. This includes: the depth vision camera has a built-in depth sensor and a vision sensor, and the depth sensor and the vision sensor are integrated together to form the sensing component of the depth vision camera.

[0061] For example, a depth vision camera is used to detect the shape and size of the target luggage, and a first detection signal is issued when the shape and size of the target luggage meet the baggage check-in standards; otherwise, a second detection signal is issued. The depth vision camera also has a built-in image detection component for intelligently detecting the shape and size of the target luggage based on the collected visual data. Since the intelligent detection of the shape and size of objects in the imaging area is a routine technology in the field of artificial intelligence, it will not be elaborated here.

[0062] The intelligent recognition module, connected to the size detection module, is used to acquire the current scan image obtained by the X-ray scanner scanning the target luggage, analyze the image regions occupied by each object in the current scan image, and perform the following intelligent recognition operation on each object as the object to be recognized: using a convolutional neural network that has been learned multiple times, based on the values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel point in the image region corresponding to the object to be recognized, the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel point in the image region corresponding to the object to be recognized, the gradient values ​​of multiple edge pixels in the image region corresponding to the object to be recognized, the number of pixel rows, the number of pixel columns in the current scan image, and the signal-to-noise ratio of the current scan image, the intelligent recognition of the object type corresponding to the object to be recognized;

[0063] For example, using a programmable logic device, a convolutional neural network that has undergone multiple learning iterations intelligently identifies the object type corresponding to the object to be identified based on the values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel point in the image region corresponding to the object to be identified; the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel point in the image region corresponding to the object to be identified; the pixel value gradients corresponding to multiple edge pixels in the image region corresponding to the object to be identified; the number of pixel rows and columns in the current scanned image; and the signal-to-noise ratio of the current scanned image.

[0064] The intelligent recognition module is also used to issue a first recognition signal when there is more than one prohibited item type in each of the item types corresponding to each object in the current scanned image; otherwise, it issues a second recognition signal.

[0065] The method employs a convolutional neural network that has undergone multiple learning iterations to intelligently identify the object type based on the values ​​of cyan, magenta, yellow, and black components corresponding to each pixel in the image region corresponding to the object to be identified, the vertical and horizontal coordinates corresponding to each pixel in the image region corresponding to the object to be identified, the pixel value gradients corresponding to multiple edge pixels in the image region corresponding to the object to be identified, the number of pixel rows and columns in the current scanned image, and the signal-to-noise ratio of the current scanned image. The method further includes a positive correlation between the number of learning iterations of the convolutional neural network and the scanning resolution of the X-ray scanner.

[0066] For example, the positive correlation between the number of times the convolutional neural network undergoes learning and the scanning resolution of the X-ray scanner includes: the higher the scanning resolution of the X-ray scanner, the more times the convolutional neural network undergoes learning;

[0067] The method of intelligently identifying the object type based on the convolutional neural network after multiple learning iterations, using the values ​​of cyan, magenta, yellow, and black components corresponding to each pixel in the image region corresponding to the object to be identified, the vertical and horizontal coordinates corresponding to each pixel in the image region corresponding to the object to be identified, the pixel gradients corresponding to multiple edge pixels in the image region corresponding to the object to be identified, the number of pixel rows and columns in the current scan image, and the signal-to-noise ratio of the current scan image, further includes: in each learning action performed on the convolutional neural network, the previously identified objects in the historical scan images obtained from the X-ray scanner are included. The object type is used as a single output of the convolutional neural network. The values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel point in the image area occupied by the identified object in the historical scan image, the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel point in the image area occupied by the identified object in the historical scan image, the values ​​of multiple pixel gradients corresponding to multiple edge pixels in the image area occupied by the identified object in the historical scan image, the number of pixel rows, the number of pixel columns in the historical scan image, and the signal-to-noise ratio of the historical scan image are used as the item-by-item output of the convolutional neural network to complete the current learning action performed on the convolutional neural network.

[0068] Specifically, for each pixel, the value of any color component, including the single cyan component, single magenta component, single yellow component, and single black component, is between 0 and 255.

[0069] Specifically, each pixel in the CMYK color space has a corresponding single cyan component value (i.e., a single C component value), a single magenta component value (i.e., a single M component value), a single yellow component value (i.e., a single Y component value), and a single black component value (i.e., a single K component value).

[0070] The method of using a depth vision camera to detect the shape and size of the target luggage, and issuing a first detection signal when the shape and size of the target luggage meet the baggage check-in standards; otherwise, issuing a second detection signal includes: using a depth vision camera to detect the shape of the target luggage based on a deep learning mode.

[0071] The method further includes using a depth vision camera to detect the shape and size of the target luggage, and issuing a first detection signal when the shape and size of the target luggage meet the baggage handling standards; otherwise, issuing a second detection signal also includes using a depth vision camera to detect the size of the target luggage based on the depth information of the target luggage.

[0072] Second Embodiment

[0073] Figure 4 This is an internal structural diagram of an intelligent pre-inspection self-service baggage check-in terminal system based on visual image technology, as shown in the second embodiment of the present invention.

[0074] like Figure 4 As shown, compared to Figure 3 The intelligent pre-inspection self-service baggage check-in terminal system based on visual image technology also includes:

[0075] The baggage conveying module is connected to the size detection module and the intelligent recognition module respectively, and is used to return the target baggage to the baggage detection area when the second detection signal or the first recognition signal is received;

[0076] For example, the baggage conveying module is connected to the size detection module and the intelligent recognition module respectively, and is used to return the target baggage to the baggage detection area when the second detection signal or the first recognition signal is received. The baggage conveying module includes a built-in signal receiving component, a signal analysis component, a conveying execution component and a baggage return component.

[0077] The baggage conveying module is also used to transmit the target baggage to the conveyor belt for delivery to the baggage sorting system when it receives the first detection signal or the second identification signal.

[0078] The intelligent recognition module is also used to issue a first recognition signal when there is more than one prohibited item type in each of the item types corresponding to each object in the current scanned image; otherwise, it issues a second recognition signal, including: the prohibited item types include lighters, oversized cosmetics and power banks.

[0079] Third Embodiment

[0080] Figure 5 This is an internal structural diagram of an intelligent pre-inspection self-service baggage check-in terminal system based on visual image technology, as shown in the third embodiment of the present invention.

[0081] like Figure 5 As shown, compared to Figure 3 The intelligent pre-inspection self-service baggage check-in terminal system based on visual image technology also includes:

[0082] The on-site warning module is connected to the size detection module and the intelligent recognition module respectively, and is used to broadcast the warning information corresponding to the identification of prohibited items on-site when the second detection signal or the first recognition signal is received;

[0083] Specifically, the on-site warning module is connected to the size detection module and the intelligent recognition module respectively, and is used to broadcast the on-site warning information corresponding to the identification of prohibited items when the second detection signal or the first recognition signal is received. The on-site warning module is an acoustic alarm module or an optical alarm module.

[0084] Fourth embodiment

[0085] Figure 6 This is an internal structural diagram of an intelligent pre-inspection self-service baggage check-in terminal system based on visual image technology, as shown in the fourth embodiment of the present invention.

[0086] like Figure 6 As shown, compared to Figure 3 The intelligent pre-inspection self-service baggage check-in terminal system based on visual image technology also includes:

[0087] The interactive operation module is used to display the passenger interactive operation interface, providing an interactive operation interface for the current passenger to exchange information;

[0088] For example, the interactive operation module is used to display the passenger interactive operation interface, providing an interactive operation interface for the current passenger to interact with information, including: implementing the interactive operation module using a touch screen.

[0089] Fifth embodiment

[0090] Figure 7 This is an internal structural diagram of an intelligent pre-inspection self-service baggage check-in terminal system based on visual image technology, as shown in the fifth embodiment of the present invention.

[0091] like Figure 7 As shown, compared to Figure 3 The intelligent pre-inspection self-service baggage check-in terminal system based on visual image technology also includes:

[0092] The model storage module, connected to the intelligent recognition module, is used to receive various model parameters of the convolutional neural network, i.e., the intelligent recognition model, after multiple learning iterations, and to complete the model storage of the intelligent recognition model by storing the various model parameters of the intelligent recognition model.

[0093] Specifically, the model storage module can be implemented using FLASH flash memory or MMC storage chip, and connected to the intelligent recognition module. It is used to receive various model parameters of the convolutional neural network, i.e., the intelligent recognition model, after multiple learning iterations, and to complete the model storage of the intelligent recognition model by storing the various model parameters of the intelligent recognition model.

[0094] Next, various embodiments of the present invention will be further described.

[0095] Optionally, within the above embodiments, in the intelligent pre-inspection self-service baggage check-in terminal system based on visual image technology:

[0096] The method of intelligently identifying the object type of the object to be identified using a convolutional neural network that has undergone multiple learning iterations further includes: inputting the values ​​of each cyan component, magenta component, yellow component, and black component corresponding to each pixel point in the image region corresponding to the object to be identified, the vertical coordinate values ​​and horizontal coordinate values ​​corresponding to each pixel point in the image region corresponding to the object to be identified, the pixel value gradients corresponding to multiple edge pixels in the image region corresponding to the object to be identified, the number of pixel rows and columns in the current scanned image, and the signal-to-noise ratio of the current scanned image into the convolutional neural network that has undergone multiple learning iterations in parallel;

[0097] For example, the following steps are taken to input in parallel into the convolutional neural network after multiple learning iterations: inputting the values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel point in the image region corresponding to the object to be identified; inputting the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel point in the image region corresponding to the object to be identified; inputting the values ​​of multiple pixel gradients corresponding to multiple edge pixels in the image region corresponding to the object to be identified; inputting the number of pixel rows and columns in the current scanned image; inputting the signal-to-noise ratio of the current scanned image; inputting the values ​​of each pixel value corresponding to each pixel point in the image region corresponding to the object to be identified; inputting the values ​​of each pixel gradient ... The toolbox implements the testing and simulation of the data processing process of inputting the values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel point of the image region corresponding to the object to be identified, the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel point of the image region corresponding to the object to be identified, the values ​​of multiple pixel gradients corresponding to multiple edge pixels in the image region corresponding to the object to be identified, the number of pixel rows and columns of the current scanned image, and the signal-to-noise ratio of the current scanned image into the convolutional neural network after multiple learnings in parallel.

[0098] The method of intelligently identifying the object type of the object to be identified by using a convolutional neural network that has undergone multiple learning iterations, based on the values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel point in the image region corresponding to the object to be identified, the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel point in the image region corresponding to the object to be identified, the gradient values ​​of multiple edge pixels in the image region corresponding to the object to be identified, the number of pixel rows and columns in the current scanned image, and the signal-to-noise ratio of the current scanned image, further includes: running the convolutional neural network that has undergone multiple learning iterations to obtain the object type corresponding to the object to be identified output by the convolutional neural network that has undergone multiple learning iterations.

[0099] And, optionally within the above embodiments, in the intelligent pre-inspection self-service baggage check-in terminal system based on visual image technology:

[0100] The positive correlation between the number of times the convolutional neural network has undergone learning and the scanning resolution of the X-ray scanner includes: using an information conversion function to represent the information conversion relationship between the number of times the convolutional neural network has undergone learning and the scanning resolution of the X-ray scanner;

[0101] The information conversion function used to represent the positive correlation between the number of times the convolutional neural network has undergone learning and the scanning resolution of the X-ray scanner includes: in the information conversion function, the scanning resolution of the X-ray scanner is the input information of the information conversion function;

[0102] The information conversion function, which represents the positive correlation between the number of times the convolutional neural network has undergone learning and the scanning resolution of the X-ray scanner, further includes the following: in the information conversion function, the number of times the convolutional neural network has undergone learning corresponding to the scanning resolution of the X-ray scanner is the output information of the information conversion function.

[0103] Sixth Embodiment

[0104] Figure 8 This is a flowchart illustrating the steps of an intelligent pre-inspection self-service baggage check-in method based on visual image technology according to the sixth embodiment of the present invention.

[0105] like Figure 8 As shown, the intelligent pre-inspection self-service baggage check-in method based on visual image technology includes the following steps:

[0106] When a passenger checks in their baggage, their ID is scanned to obtain their identity information, and a boarding pass and baggage tag corresponding to the obtained identity information are printed.

[0107] For example, when a passenger is checking in their baggage, scanning the passenger's ID to obtain the passenger's identity information and printing the passenger boarding pass and baggage tag corresponding to the obtained identity information includes: scanning the passenger's ID to obtain the passenger's identity information, accessing a network database to collect the passenger boarding pass and baggage tag corresponding to the obtained identity information, and printing the passenger boarding pass and baggage tag corresponding to the obtained identity information.

[0108] A high-definition industrial camera is used to identify the baggage tag of the current passenger's target baggage located in the baggage inspection area in order to obtain the baggage barcode corresponding to the target baggage.

[0109] Specifically, a high-definition industrial camera is used to identify the baggage tag of the current passenger's target baggage located in the baggage detection area to obtain the baggage barcode corresponding to the target baggage. The high-definition industrial camera has a built-in high-definition camera mechanism, an image processing mechanism, a barcode output mechanism, and a parallel data bus.

[0110] A depth vision camera is used to detect the shape and size of the target luggage. If the shape and size of the target luggage meet the baggage handling standards, a first detection signal is issued; otherwise, a second detection signal is issued.

[0111] For example, a depth vision camera is used to detect the shape and size of the target luggage, and a first detection signal is issued when the shape and size of the target luggage meet the baggage check-in standards; otherwise, a second detection signal is issued. This includes: the depth vision camera has a built-in depth sensor and a vision sensor, and the depth sensor and the vision sensor are integrated together to form the sensing component of the depth vision camera.

[0112] For example, a depth vision camera is used to detect the shape and size of the target luggage, and a first detection signal is issued when the shape and size of the target luggage meet the baggage check-in standards; otherwise, a second detection signal is issued. The depth vision camera also has a built-in image detection component for intelligently detecting the shape and size of the target luggage based on the collected visual data. Since the intelligent detection of the shape and size of objects in the imaging area is a routine technology in the field of artificial intelligence, it will not be elaborated here.

[0113] The system acquires the current scan image obtained by scanning the target luggage with an X-ray scanner, analyzes the image regions occupied by each object in the current scan image, and performs the following intelligent recognition operation on each object as the object to be identified: using a convolutional neural network that has been trained multiple times, the system intelligently identifies the object type corresponding to the object to be identified based on the values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel in the image region corresponding to the object to be identified; the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel in the image region corresponding to the object to be identified; the pixel value gradient corresponding to multiple edge pixels in the image region corresponding to the object to be identified; the number of pixel rows and columns in the current scan image; and the signal-to-noise ratio of the current scan image.

[0114] For example, using a programmable logic device, a convolutional neural network that has undergone multiple learning iterations intelligently identifies the object type corresponding to the object to be identified based on the values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel point in the image region corresponding to the object to be identified; the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel point in the image region corresponding to the object to be identified; the pixel value gradients corresponding to multiple edge pixels in the image region corresponding to the object to be identified; the number of pixel rows and columns in the current scanned image; and the signal-to-noise ratio of the current scanned image.

[0115] If there is more than one prohibited item type among the various item types corresponding to each item in the current scanned image, a first identification signal is issued; otherwise, a second identification signal is issued.

[0116] Upon receiving the second detection signal or the first identification signal, the target baggage is returned to the baggage detection area, and upon receiving the first detection signal or the second identification signal, the target baggage is transferred to the conveyor belt to be sent to the transportation system leading to the airport cargo vehicle;

[0117] Specifically, when there is more than one prohibited item type in each item type corresponding to each object in the current scanned image, a first identification signal is issued; otherwise, a second identification signal is issued, including: the prohibited item type includes lighters, firearms, explosives, power banks, or knives.

[0118] The method employs a convolutional neural network that has undergone multiple learning iterations to intelligently identify the object type based on the values ​​of cyan, magenta, yellow, and black components corresponding to each pixel in the image region corresponding to the object to be identified, the vertical and horizontal coordinates corresponding to each pixel in the image region corresponding to the object to be identified, the pixel value gradients corresponding to multiple edge pixels in the image region corresponding to the object to be identified, the number of pixel rows and columns in the current scanned image, and the signal-to-noise ratio of the current scanned image. The method further includes a positive correlation between the number of learning iterations of the convolutional neural network and the scanning resolution of the X-ray scanner.

[0119] For example, the positive correlation between the number of times the convolutional neural network undergoes learning and the scanning resolution of the X-ray scanner includes: the higher the scanning resolution of the X-ray scanner, the more times the convolutional neural network undergoes learning;

[0120] The method of intelligently identifying the object type based on the convolutional neural network after multiple learning iterations, using the values ​​of cyan, magenta, yellow, and black components corresponding to each pixel in the image region corresponding to the object to be identified, the vertical and horizontal coordinates corresponding to each pixel in the image region corresponding to the object to be identified, the pixel gradients corresponding to multiple edge pixels in the image region corresponding to the object to be identified, the number of pixel rows and columns in the current scan image, and the signal-to-noise ratio of the current scan image, further includes: in each learning action performed on the convolutional neural network, the previously identified objects in the historical scan images obtained from the X-ray scanner are included. The object type is used as a single output of the convolutional neural network. The values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel point in the image area occupied by the identified object in the historical scan image, the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel point in the image area occupied by the identified object in the historical scan image, the values ​​of multiple pixel gradients corresponding to multiple edge pixels in the image area occupied by the identified object in the historical scan image, the number of pixel rows, the number of pixel columns in the historical scan image, and the signal-to-noise ratio of the historical scan image are used as the item-by-item output of the convolutional neural network to complete the current learning action performed on the convolutional neural network.

[0121] Specifically, for each pixel, the value of any color component, including the single cyan component, single magenta component, single yellow component, and single black component, is between 0 and 255.

[0122] Specifically, each pixel in the CMYK color space has a corresponding single cyan component value (i.e., a single C component value), a single magenta component value (i.e., a single M component value), a single yellow component value (i.e., a single Y component value), and a single black component value (i.e., a single K component value).

[0123] The method of using a depth vision camera to detect the shape and size of the target luggage, and issuing a first detection signal when the shape and size of the target luggage meet the baggage check-in standards; otherwise, issuing a second detection signal includes: using a depth vision camera to detect the shape of the target luggage based on a deep learning mode.

[0124] The method further includes using a depth vision camera to detect the shape and size of the target luggage, and issuing a first detection signal when the shape and size of the target luggage meet the baggage handling standards; otherwise, issuing a second detection signal also includes using a depth vision camera to detect the size of the target luggage based on the depth information of the target luggage.

[0125] Furthermore, in the intelligent pre-inspection self-service baggage check-in terminal system and method based on visual image technology according to the present invention:

[0126] The process of inputting the values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel in the image region corresponding to the object to be identified, the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel in the image region corresponding to the object to be identified, the gradient values ​​of each pixel value corresponding to multiple edge pixels in the image region corresponding to the object to be identified, the number of pixel rows and columns in the current scanned image, and the signal-to-noise ratio of the current scanned image into the convolutional neural network after multiple learning iterations includes: converting the values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel in the image region corresponding to the object to be identified, the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel in the image region corresponding to the object to be identified, the gradient values ​​of each pixel value corresponding to multiple edge pixels in the image region corresponding to the object to be identified, the number of pixel rows and columns in the current scanned image, and the signal-to-noise ratio of the current scanned image into octal values ​​before inputting them into the convolutional neural network after multiple learning iterations.

[0127] For example, the values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel in the image region corresponding to the object to be identified; the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel in the image region corresponding to the object to be identified; the pixel value gradients corresponding to multiple edge pixels in the image region corresponding to the object to be identified; the number of pixel rows and columns in the current scanned image; and the signal-to-noise ratio of the current scanned image are all converted into octal values ​​and then input in parallel into the image region after multiple learning iterations. The convolutional neural network includes: firstly, using a numerical conversion module to perform octal numerical conversion on the following: the values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel point in the image region corresponding to the object to be identified; the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel point in the image region corresponding to the object to be identified; the pixel value gradients corresponding to multiple edge pixels in the image region corresponding to the object to be identified; the number of pixel rows and columns in the current scanned image; and the signal-to-noise ratio of the current scanned image.

[0128] For example, the following steps further include converting the values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel point in the image region corresponding to the object to be identified; the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel point in the image region corresponding to the object to be identified; the pixel value gradients corresponding to multiple edge pixels in the image region corresponding to the object to be identified; the number of pixel rows and columns in the current scanned image; and the signal-to-noise ratio of the current scanned image into octal values ​​before inputting them in parallel into the convolutional neural network after multiple learning iterations: Then, a parallel control module is used to input the values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel in the image region corresponding to the object to be identified after octal value conversion, the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel in the image region corresponding to the object to be identified, the gradient values ​​of multiple edge pixels in the image region corresponding to the object to be identified, the number of pixel rows and columns in the current scanned image, and the signal-to-noise ratio of the current scanned image into the convolutional neural network after multiple learning iterations.

[0129] And wherein, running the convolutional neural network after multiple learning iterations to obtain the object type corresponding to the object to be identified output by the convolutional neural network after multiple learning iterations includes: the object type corresponding to the object to be identified output by the convolutional neural network after multiple learning iterations is in octal numerical representation.

[0130] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0131] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments for apparatus / electronic devices / computer-readable storage media / computer program products are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0132] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of protection of the present invention.

Claims

1. A smart pre-inspection self-service baggage check-in terminal system based on visual image technology, characterized in that, The terminal system includes: The document scanning module is used to scan the current passenger's ID document to obtain the current passenger's identity information when the passenger is checking in baggage, and to print the passenger's boarding pass and baggage tag corresponding to the obtained identity information; A barcode recognition module, connected to the document scanning module, is used to use a high-definition industrial camera to recognize the barcode of the baggage tag of the current passenger's target baggage located in the baggage inspection area, so as to obtain the baggage barcode corresponding to the target baggage. The size detection module, connected to the barcode recognition module, is used to detect the shape and size of the target luggage using a depth vision camera, and to issue a first detection signal when the shape and size of the target luggage meet the baggage check-in standards; otherwise, it issues a second detection signal. The intelligent recognition module, connected to the size detection module, is used to acquire the current scan image obtained by the X-ray scanner scanning the target luggage, analyze the image regions occupied by each object in the current scan image, and perform the following intelligent recognition operation on each object as the object to be recognized: using a convolutional neural network that has been trained multiple times, based on the values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel point in the image region corresponding to the object to be recognized, the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel point in the image region corresponding to the object to be recognized, the pixel value gradient corresponding to multiple edge pixels in the image region corresponding to the object to be recognized, the number of pixel rows, the number of pixel columns in the current scan image, and the signal-to-noise ratio of the current scan image; the intelligent recognition module is also used to issue a first recognition signal when there is more than one prohibited item type among the object types corresponding to each object in the current scan image, otherwise, issue a second recognition signal; The number of times the convolutional neural network is trained is positively correlated with the scanning resolution of the X-ray scanner. In each learning action performed on the convolutional neural network, the object type of the identified object in the historical scan image obtained from the X-ray scanner is used as a single output of the convolutional neural network. The following outputs are used sequentially: the values ​​of each pixel in the image area occupied by the identified object in the historical scan image (cyan, magenta, yellow, and black components), the vertical and horizontal coordinates of each pixel in the image area occupied by the identified object in the historical scan image, the pixel gradients of multiple edge pixels in the image area occupied by the identified object in the historical scan image, the number of pixel rows and columns in the historical scan image, and the signal-to-noise ratio of the historical scan image. This completes the current learning action performed on the convolutional neural network. Specifically, the values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel point in the image region corresponding to the object to be identified, the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel point in the image region corresponding to the object to be identified, the values ​​of multiple pixel gradients corresponding to multiple edge pixels in the image region corresponding to the object to be identified, the number of pixel rows and columns in the current scanned image, and the signal-to-noise ratio of the current scanned image are converted into octal values ​​and then input in parallel into the convolutional neural network that has been learned multiple times.

2. The intelligent pre-inspection self-service baggage check-in terminal system based on visual image technology as described in claim 1, characterized in that: in, The shape and size of the target luggage are detected using a depth vision camera, and a first detection signal is issued when the shape and size of the target luggage meet the baggage handling standards; otherwise, a second detection signal is issued, including: detecting the shape of the target luggage using a depth vision camera based on a deep learning mode. The method involves using a depth vision camera to detect the shape and size of the target luggage, and issuing a first detection signal when the shape and size of the target luggage meet the baggage handling standards; otherwise, issuing a second detection signal further includes using a depth vision camera to detect the size of the target luggage based on the collected depth information of the target luggage.

3. The intelligent pre-inspection self-service baggage check-in terminal system based on visual image technology as described in claim 2, characterized in that, The terminal system also includes: The baggage conveying module is connected to the size detection module and the intelligent recognition module respectively, and is used to return the target baggage to the baggage detection area when the second detection signal or the first recognition signal is received; The baggage conveying module is also used to transfer the target baggage to a conveyor belt to be sent to the transportation system leading to the airport cargo vehicle when the first detection signal or the second identification signal is received. The intelligent recognition module is also used to issue a first recognition signal when there is more than one prohibited item type in each of the item types corresponding to each object in the current scanned image; otherwise, it issues a second recognition signal, including: the prohibited item type includes lighters, firearms, explosives, power banks, or knives.

4. The intelligent pre-inspection self-service baggage check-in terminal system based on visual image technology as described in claim 2, characterized in that, The terminal system also includes: The on-site warning module is connected to the size detection module and the intelligent recognition module respectively, and is used to broadcast the warning information corresponding to the identification of prohibited items on-site when the second detection signal or the first recognition signal is received.

5. The intelligent pre-inspection self-service baggage check-in terminal system based on visual image technology as described in claim 2, characterized in that, The terminal system also includes: The interactive operation module is used to display the passenger interactive operation interface, providing an interactive operation interface for the current passenger to exchange information.

6. The intelligent pre-inspection self-service baggage check-in terminal system based on visual image technology as described in claim 2, characterized in that, The terminal system also includes: The model storage module, connected to the intelligent recognition module, is used to receive various model parameters of the convolutional neural network (i.e., the intelligent recognition model) after multiple learning iterations, and to complete the model storage of the intelligent recognition model by storing these model parameters.

7. The intelligent pre-inspection self-service baggage check-in terminal system based on visual image technology as described in any one of claims 2-6, characterized in that: The method of intelligently identifying the object type of the object to be identified using a convolutional neural network that has undergone multiple learning iterations further includes: inputting the values ​​of each cyan component, magenta component, yellow component, and black component corresponding to each pixel point in the image region corresponding to the object to be identified, the vertical coordinate values ​​and horizontal coordinate values ​​corresponding to each pixel point in the image region corresponding to the object to be identified, the pixel value gradients corresponding to multiple edge pixels in the image region corresponding to the object to be identified, the number of pixel rows and columns in the current scanned image, and the signal-to-noise ratio of the current scanned image into the convolutional neural network that has undergone multiple learning iterations in parallel; The method of intelligently identifying the object type of the object to be identified by using a convolutional neural network that has undergone multiple learning iterations, based on the values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel point in the image region corresponding to the object to be identified, the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel point in the image region corresponding to the object to be identified, the gradient values ​​of multiple edge pixels in the image region corresponding to the object to be identified, the number of pixel rows and columns in the current scanned image, and the signal-to-noise ratio of the current scanned image, further includes: running the convolutional neural network that has undergone multiple learning iterations to obtain the object type corresponding to the object to be identified output by the convolutional neural network that has undergone multiple learning iterations.

8. The intelligent pre-inspection self-service baggage check-in terminal system based on visual image technology as described in any one of claims 2-6, characterized in that: The positive correlation between the number of times the convolutional neural network has undergone learning and the scanning resolution of the X-ray scanner includes: using an information conversion function to represent the information conversion relationship between the number of times the convolutional neural network has undergone learning and the scanning resolution of the X-ray scanner; The information conversion function used to represent the positive correlation between the number of times the convolutional neural network has undergone learning and the scanning resolution of the X-ray scanner includes: in the information conversion function, the scanning resolution of the X-ray scanner is the input information of the information conversion function; The information conversion function, which represents the positive correlation between the number of times the convolutional neural network has undergone learning and the scanning resolution of the X-ray scanner, further includes the following: in the information conversion function, the number of times the convolutional neural network has undergone learning corresponding to the scanning resolution of the X-ray scanner is the output information of the information conversion function.

9. A smart pre-inspection self-service baggage check-in method based on visual image technology, characterized in that, The method includes: When a passenger checks in their baggage, their ID is scanned to obtain their identity information, and a boarding pass and baggage tag corresponding to the obtained identity information are printed. A high-definition industrial camera is used to identify the baggage tag of the current passenger's target baggage located in the baggage inspection area in order to obtain the baggage barcode corresponding to the target baggage. A depth vision camera is used to detect the shape and size of the target luggage. If the shape and size of the target luggage meet the baggage handling standards, a first detection signal is issued; otherwise, a second detection signal is issued. The system acquires the current scan image obtained by scanning the target luggage with an X-ray scanner, analyzes the image regions occupied by each object in the current scan image, and performs the following intelligent recognition operation on each object as the object to be identified: using a convolutional neural network that has been trained multiple times, the system intelligently identifies the object type corresponding to the object to be identified based on the values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel in the image region corresponding to the object to be identified; the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel in the image region corresponding to the object to be identified; the pixel value gradient corresponding to multiple edge pixels in the image region corresponding to the object to be identified; the number of pixel rows and columns in the current scan image; and the signal-to-noise ratio of the current scan image. If there is more than one prohibited item type among the various item types corresponding to each item in the current scanned image, a first identification signal is issued; otherwise, a second identification signal is issued. Upon receiving the second detection signal or the first identification signal, the target baggage is returned to the baggage detection area, and upon receiving the first detection signal or the second identification signal, the target baggage is transferred to the conveyor belt to be sent to the transportation system leading to the airport cargo vehicle; Specifically, if more than one type of prohibited item exists in the various categories corresponding to each object in the current scanned image, a first identification signal is issued; otherwise, a second identification signal is issued, including: the prohibited item type includes lighters, firearms, explosives, power banks, or knives; The number of times the convolutional neural network is trained is positively correlated with the scanning resolution of the X-ray scanner. In each learning action performed on the convolutional neural network, the object type of the identified object in the historical scan image obtained from the X-ray scanner is used as a single output of the convolutional neural network. The following outputs are used sequentially: the values ​​of each pixel in the image area occupied by the identified object in the historical scan image (cyan, magenta, yellow, and black components), the vertical and horizontal coordinates of each pixel in the image area occupied by the identified object in the historical scan image, the pixel gradients of multiple edge pixels in the image area occupied by the identified object in the historical scan image, the number of pixel rows and columns in the historical scan image, and the signal-to-noise ratio of the historical scan image. This completes the current learning action performed on the convolutional neural network. Specifically, the values ​​of each cyan component, each magenta component, each yellow component, and each black component corresponding to each pixel point in the image region corresponding to the object to be identified, the values ​​of each vertical coordinate and each horizontal coordinate corresponding to each pixel point in the image region corresponding to the object to be identified, the values ​​of multiple pixel gradients corresponding to multiple edge pixels in the image region corresponding to the object to be identified, the number of pixel rows and columns in the current scanned image, and the signal-to-noise ratio of the current scanned image are converted into octal values ​​and then input in parallel into the convolutional neural network that has been learned multiple times.

Images