An image recognition method and system
By marking similar individuals whose facial recognition scores exceed a threshold in the airport's facial recognition system, similar passengers can be pre-screened and boarded using other identity verification methods. This solves the problem of erroneous boarding and improves the accuracy and security of airport access.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- TRAVELSKY TECHNOLOGY LIMITED
- Filing Date
- 2023-01-10
- Publication Date
- 2026-06-30
Smart Images
Figure CN116012920B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of image recognition technology, and in particular to an image recognition method and system. Background Technology
[0002] Facial recognition is a biometric technology that identifies individuals based on their facial features. It uses cameras or webcams to capture real-time images or videos containing faces, automatically detects and tracks faces, and then identifies the detected faces to achieve rapid identity authentication.
[0003] With the widespread adoption of facial recognition technology at domestic airports, passenger boarding time has been shortened, the process has become simpler, and self-service has made things more convenient for passengers while also saving airport manpower.
[0004] However, the use of facial recognition for boarding at airports can lead to mis-boarding. For example, passenger 2 may be on flight B but arrive at gate A. If passenger 2's facial image is similar to that of passenger 1 on flight A, and passenger 1 hasn't boarded yet, passenger 2 can board flight A through facial recognition at gate A, but the system will show passenger 1 as already boarded. When passenger 1 arrives to board, the system will notify them that passenger 1 has already boarded, increasing the time spent locating passenger 2 and resolving system issues, potentially causing flight delays. Furthermore, if passenger 1 ultimately doesn't board, passenger 2 might go to the wrong destination, leading to even more serious consequences.
[0005] Therefore, reducing the occurrence of passengers boarding the wrong plane or being misidentified has become an urgent technical problem to be solved. Summary of the Invention
[0006] This application provides an image recognition method and system that can reduce the occurrence of passengers boarding incorrectly or being misidentified.
[0007] This application discloses the following technical solution:
[0008] In a first aspect, this application discloses an image recognition method applied to an airport intelligent cloud system, the method comprising:
[0009] Acquire passenger travel information, flight status data, and passenger facial images;
[0010] Obtain the feature values of the passenger's facial image;
[0011] The feature values of the passenger's face image are compared with the feature values of other passenger face images in the first face database to obtain a first similarity. The first face database is a database of all registered passenger face images.
[0012] Determine whether there are multiple similar persons in the first face database whose first similarity to the passenger is greater than a first threshold. If so, mark the passenger and the similar persons.
[0013] Check whether the passenger has been marked. If so, adjust the first threshold to form a second threshold, where the second threshold is greater than the first threshold.
[0014] Determine whether the second threshold is greater than the highest threshold. If not, compare the feature value of the passenger's face image with the feature value of other passenger face images in the second face database to obtain the second similarity. The second face database is a database formed by filtering the first face database based on whether the passenger's itinerary information data and flight dynamic information data are the same.
[0015] If a second facial database is found to contain a single person with a second similarity score greater than a second threshold that is similar to the passenger, then the passenger is allowed to pass through the boarding gate.
[0016] Optionally, after obtaining passenger travel information data, flight dynamic information data, and passenger facial images, the method further includes:
[0017] Select passenger facial images that meet the requirements.
[0018] Optionally, before obtaining the feature values of the passenger's facial image, the method further includes:
[0019] Determine whether the passenger's facial image has been registered in the first facial database;
[0020] If the passenger's facial image is not registered in the first facial database, obtaining the feature value of the passenger's facial image includes: directly obtaining the feature value of the passenger's facial image, and after determining whether there are multiple similar persons in the first facial database whose first similarity to the passenger is greater than a first threshold, registering them in the first facial database.
[0021] If the passenger's facial image has been registered in the first facial database, then directly query whether the passenger has been marked.
[0022] Optionally, after determining whether the second threshold is greater than the highest threshold, the method further includes:
[0023] If not, the passenger will be denied passage through the boarding gate.
[0024] Optionally, the method further includes:
[0025] The system stores the passenger's travel information, flight dynamic information, and passenger facial images.
[0026] Secondly, this application discloses an image recognition system applied to an airport intelligent cloud system, the system comprising: a first acquisition module, a second acquisition module, a comparison module, a first judgment module, a query module, a second judgment module, and a third judgment module;
[0027] The first acquisition module is used to acquire passenger travel information data, flight dynamic information data, and passenger facial images;
[0028] The second acquisition module is used to acquire the feature values of the passenger's face image;
[0029] The comparison module is used to compare the feature values of the passenger face image with the feature values of other passenger face images in the first face database to obtain a first similarity. The first face database is a database of all registered passenger face images.
[0030] The first judgment module is used to determine whether there are multiple similar persons in the first face database whose first similarity to the passenger is greater than a first threshold. If so, the passenger and the similar persons are marked.
[0031] The query module is used to query whether the passenger has been marked. If so, the first threshold is adjusted to form a second threshold, and the second threshold is greater than the first threshold.
[0032] The second judgment module is used to determine whether the second threshold is greater than the highest threshold. If not, the feature value of the passenger's face image is compared with the feature value of other passenger face images in the second face database to obtain the second similarity. The second face database is a database formed by filtering the first face database based on whether the passenger's itinerary information data and flight dynamic information data are the same.
[0033] The third judgment module is used to determine whether there is a unique person in the second face database whose second similarity to the passenger is greater than the second threshold. If so, the passenger is allowed to pass through the boarding gate.
[0034] Optionally, the system further includes: a filtering module;
[0035] The filtering module is used to filter passenger facial images that meet the requirements.
[0036] Optionally, the system further includes: a registration module;
[0037] The registration module is used to determine whether the passenger's facial image has been registered into the first facial database;
[0038] If the passenger's facial image is not registered in the first facial database, obtaining the feature value of the passenger's facial image includes: directly obtaining the feature value of the passenger's facial image, and after determining whether there are multiple similar persons in the first facial database whose first similarity to the passenger is greater than a first threshold, registering them in the first facial database.
[0039] If the passenger's facial image has been registered in the first facial database, then directly query whether the passenger has been marked.
[0040] Optionally, the third determination module can also be used for:
[0041] If not, the passenger will be denied passage through the boarding gate.
[0042] Optionally, the system further includes: a storage module;
[0043] The storage module is used to store the passenger's itinerary information data, flight dynamic information data, and passenger facial images.
[0044] Compared with the prior art, this application has the following beneficial effects:
[0045] This application pre-screens similar passengers by marking individuals whose feature similarity exceeds a threshold, allowing facial recognition services to be provided only to legitimate passengers. Similar passengers can then use other identity verification methods to board, significantly reducing the probability of accidental boarding. Furthermore, facial recognition technology, along with fingerprint, iris, finger vein, and voiceprint recognition technologies, uses advanced technology to combine inherent physiological characteristics (such as fingerprints, facial images, irises, and voiceprints) and behavioral features (such as handwriting, voice, and gait) for personal identification. If different individuals possess similar physiological or behavioral characteristics, it can interfere with the identification results of biometric technologies. This application is also applicable to other biometric technologies, such as fingerprint recognition. Individuals with similar fingerprints or no fingerprints can be marked during fingerprint registration before subsequent operations. Attached Figure Description
[0046] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0047] Figure 1 A flowchart of an image recognition method provided in an embodiment of this application;
[0048] Figure 2 This is a schematic diagram of an image recognition system provided in an embodiment of this application. Detailed Implementation
[0049] The technical terms used in this application will be introduced below.
[0050] The Airport Intelligence Cloud System (PIC) serves as an integration and service platform for airports, creating smooth information exchange channels between internal and external systems. It enables various business systems to interact with data in a unified and standardized manner, allowing system services to be freely published and recombined to support various new business function requirements and processes.
[0051] Airport intelligent cloud systems are widely used in major airports, but there are still instances of passengers mistakenly boarding using facial recognition. For example, passenger 2 is on flight B but goes to gate A. If passenger 2's facial image is similar to that of passenger 1 on flight A, and passenger 1 hasn't boarded yet, passenger 2 can board flight A using facial recognition, but the system will show passenger 1 as already boarded. When passenger 1 arrives to board, the system will notify them that passenger 1 has already boarded, increasing the time spent locating passenger 2 and resolving system issues, potentially causing flight delays. Furthermore, if passenger 1 ultimately doesn't board, passenger 2 might go to the wrong destination, leading to even more serious consequences.
[0052] In view of this, this application provides an image recognition method and system that can pre-screen similar passengers by marking similar groups whose feature similarity exceeds a threshold, providing facial recognition services only to legitimate passengers. Similar passengers can then use other identity verification methods to board, significantly reducing the probability of passengers boarding incorrectly. Furthermore, facial recognition technology, along with fingerprint recognition, iris recognition, finger vein recognition, and voiceprint recognition, all utilize high-tech methods to combine inherent physiological characteristics (such as fingerprints, facial images, irises, and voiceprints) and behavioral characteristics (such as handwriting, voice, and gait) for personal identification. If different individuals possess similar physiological or behavioral characteristics, it may interfere with the identification results of biometric technologies. This application is also applicable to other biometric technologies, such as fingerprint recognition. For individuals with similar fingerprints or no fingerprints, marking can be performed during fingerprint registration before subsequent operations.
[0053] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.
[0054] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0055] See Figure 1 The figure is a flowchart of an image recognition method provided in an embodiment of this application.
[0056] S101: The airport's intelligent cloud system acquires passenger travel information data, flight dynamic information data, and passenger biometric data.
[0057] Passenger travel information data can include flight number, flight date, departure station, destination, seat number, boarding status, and verification status, among other departure-related information. Flight dynamic information data can include flight segment information, estimated departure time at each station, and flight verification status. Both passenger travel information and flight dynamic information data can be used to determine business scenarios and to establish a facial database when comparing passenger biometric data using feature values.
[0058] Passenger biometric data includes biometric information and authorization information. Biometric information can include the passenger's facial information, fingerprint information, etc., while authorization information can include the passenger's identification information, facial image information, image authorization information, etc. Specifically, image authorization information refers to which airports and airlines can use the passenger's facial photo, and the effective usage period, recording the valid time for which the passenger's facial image is authorized for use.
[0059] The airport's intelligent cloud system acquires passenger travel information, flight status data, and passenger biometric data, and stores this information in its database. In some specific implementations, the data can be obtained through security check systems, self-service check-in, airport mini-programs, etc., and then stored in the airport's intelligent cloud system's database.
[0060] This application makes passenger passage more accurate, safe, and efficient by collecting and processing passenger travel data, flight status data, and biometric data, thereby improving airport inspection rates and operational efficiency and providing greater assurance for security management.
[0061] S102: The airport's intelligent cloud system filters passenger facial images that meet the requirements.
[0062] Airport intelligent cloud systems acquire passenger facial images using cameras or video cameras and then filter those that meet requirements. Because facial images taken in different scenes, with varying emotions, and from different angles have inconsistent quality, blurry or obscured images can lower the facial recognition rate, requiring multiple attempts and wasting passenger time. Therefore, filtering out low-quality images and sending only those meeting the standards to the next stage significantly improves the recognition rate.
[0063] In one implementation, the criterion for selecting qualified passenger facial images is image clarity. Specifically, the criteria are: no blurring due to lens defocus or movement; no obstructions such as sunglasses or masks; avoidance of low light, strong light, or backlighting; image size ≤ 100KB (kilobytes); image format jpg; image resolution 120*120 (20000*2000 pixels); distance between the eyes not less than 80 pixels; and posture requirements: planar rotation, pitch, and side tilt all within the range of -15° to 15°. Passenger facial images are selected based on these requirements. If any passenger facial image does not meet the above requirements, the airport intelligent cloud system will report an error message.
[0064] S103: The airport's intelligent cloud system determines whether a passenger's facial image has been registered in the facial database.
[0065] After filtering passenger facial images that meet the requirements, the system checks whether the passenger's facial image information has already been registered in the facial database. Registering the passenger's facial image means that the passenger's facial image was extracted using a camera or video camera during the first use of the system.
[0066] If the passenger's facial image information has not been registered, proceed to step S104; if the passenger's facial image information has been registered, proceed directly to step S108.
[0067] S104: If the passenger's facial image information has not been registered, the airport's intelligent cloud system will obtain the feature values of the passenger's facial image.
[0068] Obtaining facial image feature values involves using a face detection algorithm to detect the facial contour in the original facial image, extracting feature points from the facial contour, and calculating their feature values as the feature values of the facial image. The face detection algorithm is a pre-set algorithm used by the system, and the feature points are pre-set pixel positions in the facial image, such as eyes, nose, and mouth; this application does not limit the specific features. The process of calculating the feature values of the facial image is also the process of feature modeling based on pre-set features of the facial image. This can typically be achieved using feature extraction algorithms, such as geometric feature-based methods, statistical methods, elastic graph matching methods, neural network methods, support vector basis methods, and hidden Markov models; this application does not limit the specific features used.
[0069] S105: The airport's intelligent cloud system determines whether there are multiple people in the first face database whose first similarity value to the passenger is greater than the first threshold.
[0070] The airport's intelligent cloud system performs face comparison services between passenger facial images registered in a face database and other passenger facial images in a first face database. Specifically, it determines whether there are multiple individuals in the first face database whose first similarity value exceeds a first threshold. The first face database is a database of all registered passenger facial images. The first threshold is a pre-set value based on the facial similarity value; for example, a default threshold represents general similarity and is used to judge the degree of similarity between generally similar faces. The similarity value is calculated by comparing facial feature values, i.e., the degree of overlap of facial image information.
[0071] If there are multiple people in the first face database who are similar to the passenger and whose first similarity is greater than the first threshold, then execute S106.
[0072] If there is a single person in the first face database whose first similarity to the passenger is greater than the first threshold, then S107 is executed directly.
[0073] S106: The airport's intelligent cloud system will tag passengers and similar individuals.
[0074] If there are multiple people in the first face database whose first similarity to the passenger is greater than the first threshold, the passenger and the similar people will be marked as similar passengers. When similar passengers use the face recognition service, the threshold will be dynamically adjusted according to the similarity value.
[0075] In one implementation, this tag can be stored as a field in a database.
[0076] S107: The airport's intelligent cloud system registers the passenger's facial image information into the first facial database.
[0077] S108: The airport's intelligent cloud system checks whether the passenger has been flagged.
[0078] The airport's intelligent cloud system queries whether the passenger has been tagged. If the passenger has been tagged, step S109 is executed; otherwise, step S111 is executed. In one implementation, the tagged status can be determined by checking whether the corresponding field is stored in the database.
[0079] S109: The airport's intelligent cloud system will dynamically adjust the first threshold to the second threshold.
[0080] The threshold can be set to a default threshold, a false recognition avoidance threshold, and a maximum threshold, primarily to handle scenarios of general similarity, very similarity, and twins. The default threshold represents generally similar faces, the false recognition avoidance threshold represents very similar faces, and the maximum threshold represents twin-level similarity, with the thresholds increasing sequentially. When a passenger is detected to have been flagged, the airport's intelligent cloud system dynamically adjusts the threshold, increasing it.
[0081] S110: Check whether the adjusted second threshold of the airport intelligent cloud system is greater than the highest threshold.
[0082] For similar passengers that have been marked, the airport's intelligent cloud system compares the highest similarity value among the similar passengers with the avoidance of misidentification value. If the highest similarity value is lower than the avoidance of misidentification value, the default threshold is adjusted to the avoidance of misidentification threshold, and S111 is executed. If the highest similarity value is higher than the avoidance of misidentification value, it further determines whether the highest similarity value is higher than the highest threshold. If the highest similarity value is lower than the highest threshold, the default threshold is adjusted to the highest similarity value, and S111 is executed. If the highest similarity value is higher than the highest threshold, S113 is executed directly.
[0083] S111: The airport's intelligent cloud system selects a second facial database based on passenger travel information data and flight dynamic information data.
[0084] Passengers who have completed the preceding steps, as well as those whose identities have been dynamically adjusted after being flagged, will use facial recognition services to verify their identities. Before comparing facial features, it is necessary to determine the passenger's current business scenario. Based on the passenger's travel information data and flight dynamic information data collected by the airport's intelligent cloud system, the comparison scope is narrowed down and a corresponding facial database is selected, i.e., the second facial database.
[0085] S112: The airport's intelligent cloud system determines whether there is a unique person in the second facial database whose second similarity value is greater than the second threshold that is similar to the passenger.
[0086] Based on passenger travel information and flight dynamic information collected by the airport's intelligent cloud system, a corresponding second face database is selected, and a comparison set n is obtained. The similarity ratio of the comparison results is checked. The feature values of the passenger's face image are compared with the feature values of other passenger face images in the second face database to obtain the second similarity. Then, it is determined whether there is a unique person in the second face database whose second similarity value is greater than the second threshold.
[0087] If the second face database contains multiple individuals whose second similarity to the passenger exceeds the second threshold, or if the second face database does not contain any individuals whose second similarity to the passenger exceeds the second threshold, then the passenger will be denied boarding without face recognition, and S113 will be executed, i.e., the airport intelligent cloud system will deny the passenger access to the boarding gate; if the second face database contains only one individual whose second similarity to the passenger exceeds the second threshold, then the passenger will be allowed to board through face recognition, and S114 will be executed.
[0088] S113: The airport's smart cloud system refused passengers passage through the boarding gate.
[0089] S114: The airport's smart cloud system allows passengers to pass through the boarding gate.
[0090] If the similarity ratio is greater than the adjusted second threshold, after the labeling and recognition step, there should only be one comparison result, namely the passenger himself. Then the passenger can pass through the boarding gate through facial recognition.
[0091] Furthermore, facial recognition technology, along with fingerprint recognition, iris recognition, finger vein recognition, and voiceprint recognition, all utilize advanced technology to combine inherent physiological characteristics (such as fingerprints, facial images, irises, and voiceprints) and behavioral features (such as handwriting, voice, and gait) to identify an individual. If different individuals possess similar physiological or behavioral characteristics, it may interfere with the identification results of biometric technologies. The above embodiments primarily aim to reduce false identification by marking similar cases before using facial recognition technology. Similarly, this application is also applicable to other biometric technologies, such as fingerprint recognition. For individuals with similar fingerprints or no fingerprints, marking can be performed during fingerprint registration, similar to the embodiments described above, followed by subsequent operations.
[0092] Using the above description as an example, in one scenario, there are two passengers at the airport, such as passenger 1 and passenger 2, who both need to board on the same day, and passenger 1 and passenger 2 look similar. On that day, there are flights A and B.
[0093] First, we proceed with the data collection phase. This involves collecting data on Passenger 1 and Passenger 2, including their travel itinerary data, data on all flights at the airport that day, and the passengers' biometric data.
[0094] Next, the data processing begins. Passengers 1 and 2 have their photos taken on-site while checking in, and the photos must meet the required standards. If Passenger 1 is not registered in the facial information database, but Passenger 2 is, and both have not yet passed through the final gate before boarding, Passenger 1's facial image is captured, facial feature values are extracted, and then registered in the database, while simultaneously comparing these feature values. If the similarity between Passenger 1 and Passenger 2 exceeds a threshold, they will be marked as similar passengers by the system. Because Passengers 1 and 2 are marked, they belong to a group that the facial recognition system refuses to recognize. Therefore, if Passenger 2 mistakenly goes to a gate for a particular flight and uses facial recognition for verification, it will detect that Passenger 2 has been marked. Passenger 2 will then need to adjust the similarity threshold before comparison, and the system will determine whether facial recognition is allowed for boarding based on the comparison result. Passengers who are not marked are considered normal passengers, and their thresholds do not need to be changed; they directly proceed to the facial recognition stage. Using the default threshold is not suitable for comparing the similarity values of passenger 1 and passenger 2. Instead, the highest similarity value between passenger 1 and passenger 2 is compared with the threshold to avoid false identification. The highest value is then used as the threshold to further distinguish whether a passenger can use facial recognition for boarding, for example in the following scenario:
[0095] Scenario 1: Passenger 1 and Passenger 2 are very similar passengers. If it is confirmed that the two are on the same day but different flights, but the specific flight is unknown, when Passenger 1 checks in, a photo is taken on-site and compared with a database containing information on passengers on that day's flights, as well as Passenger 2's facial photo. The similarity value is compared with a threshold to avoid false identification, and the highest value is selected as the threshold for comparison. This allows for precise location of Passenger 1 and exclusion of Passenger 2.
[0096] If it is determined that passenger 1 belongs to flight A and passenger 2 belongs to flight B, when passenger 2 walks to the boarding gate of flight A (where passenger 1 has not yet boarded) to have their photo taken, the default threshold will be dynamically adjusted to avoid misidentification. Then, it will be compared with the highest similarity value in the facial database of flight A, and the highest value will be selected as the threshold. If passenger 2 is not passenger 1 or a twin, it is impossible for them to have such a high similarity value, so passenger 2 and passenger 1 can be distinguished more accurately, preventing passenger 2 from boarding flight A. The same applies to passenger 1 walking to flight B; the adjusted threshold will also be used for facial recognition.
[0097] Passengers in this scenario can enter the facial recognition process by adjusting the dynamic threshold.
[0098] Scenario 2: Passenger 1 and Passenger 2 are twins or look-alikes, with negligible differences in facial features, resulting in a high degree of similarity. For example, when Passenger 1 takes a photo while processing transactions, it is compared with photos in a database. If this database includes a photo of Passenger 2, the similarity score compared to the threshold for avoiding misidentification is very likely to exceed the maximum threshold. Moreover, this comparison result may include not only Passenger 1 but also Passenger 2, making misidentification unavoidable.
[0099] If a passenger has confirmed a specific flight, but passenger 1, who is supposed to be on flight A, goes to flight B, which passenger 2 has not yet boarded, even if a false identification threshold is used for comparison, the facial database for flight B, which does not contain passenger 1's image, will still cause a false identification because the similarity between passenger 1's image and passenger 2's image exceeds the highest threshold.
[0100] Passengers in this scenario belong to a group that is denied boarding by facial recognition and cannot enter the facial recognition process by adjusting the dynamic threshold.
[0101] Finally, the facial recognition process begins. Based on the passenger's flight information, a comparison set *n* is determined in the facial database. If the flight information is uncertain, a comparison is made with the facial database of the same day's flights, for example, during check-in. If the flight information is known, a comparison is made with the facial database of the current flight, for example, during boarding. If only one data point in the comparison result shows a similarity ratio greater than a threshold (meaning only the passenger's face in the database resembles their own), the passenger is allowed to board. If the similarity ratio is less than the threshold, the system does not allow the passenger to board.
[0102] This application pre-screens similar passengers by marking individuals whose feature similarity exceeds a threshold, allowing facial recognition services to be provided only to legitimate passengers. Similar passengers can then use other identity verification methods to board, significantly reducing the probability of accidental boarding. Furthermore, facial recognition technology, along with fingerprint, iris, finger vein, and voiceprint recognition technologies, uses advanced technology to combine inherent physiological characteristics (such as fingerprints, facial images, irises, and voiceprints) and behavioral features (such as handwriting, voice, and gait) for personal identification. If different individuals possess similar physiological or behavioral characteristics, it can interfere with the identification results of biometric technologies. This application is also applicable to other biometric technologies, such as fingerprint recognition. Individuals with similar fingerprints or no fingerprints can be marked during fingerprint registration before subsequent operations.
[0103] See Figure 2The figure is a schematic diagram of an image recognition system provided in an embodiment of this application. The system 200 includes at least: a first acquisition module 201, a second acquisition module 202, a comparison module 203, a first judgment module 204, a query module 205, a second judgment module 206, and a third judgment module 207.
[0104] The first acquisition module 201 is used to acquire passenger itinerary information data, flight dynamic information data, and passenger facial images;
[0105] The second acquisition module 202 is used to acquire the feature values of the passenger's face image;
[0106] The comparison module 203 is used to compare the feature values of the passenger face image with the feature values of other passenger face images in the first face database to obtain the first similarity. The first face database is a database of all registered passenger face images.
[0107] The first judgment module 204 is used to determine whether there are multiple similar persons in the first face database whose first similarity to the passenger is greater than a first threshold. If so, the passenger and the similar persons are marked.
[0108] The query module 205 is used to query whether a passenger has been marked. If so, the first threshold is adjusted to form a second threshold, and the second threshold is greater than the first threshold.
[0109] The second judgment module 206 is used to determine whether the second threshold is greater than the highest threshold. If not, the feature value of the passenger's face image is compared with the feature value of other passenger face images in the second face database to obtain the second similarity. The second face database is a database formed by filtering the first face database based on whether the passenger's itinerary information data and flight dynamic information data are the same.
[0110] The third judgment module 207 is used to determine whether there is a unique person in the second face database whose second similarity to the passenger is greater than the second threshold. If so, the passenger is allowed to pass through the boarding gate.
[0111] This application pre-screens similar passengers by marking individuals whose feature similarity exceeds a threshold, allowing facial recognition services to be provided only to legitimate passengers. Similar passengers can then use other identity verification methods to board, significantly reducing the probability of accidental boarding. Furthermore, facial recognition technology, along with fingerprint, iris, finger vein, and voiceprint recognition technologies, uses advanced technology to combine inherent physiological characteristics (such as fingerprints, facial images, irises, and voiceprints) and behavioral features (such as handwriting, voice, and gait) for personal identification. If different individuals possess similar physiological or behavioral characteristics, it can interfere with the identification results of biometric technologies. This application is also applicable to other biometric technologies, such as fingerprint recognition. Individuals with similar fingerprints or no fingerprints can be marked during fingerprint registration before subsequent operations.
[0112] It should be noted that the various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. The embodiments described above are merely illustrative. Units described as separate components may or may not be physically separate, and components indicated as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.
[0113] The above description is merely one specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
Claims
1. An image recognition method, characterized in that, The method, applied to an airport intelligent cloud system, includes: Obtain the passenger's facial image to be verified; Obtain the feature values of the passenger's facial image; The feature value is compared with the feature values of other passenger face images in the first face database to obtain the first similarity between the passenger face image and other passenger face images in the first face database. The first face database is a database of all registered passenger face images. If there are multiple similar individuals in the first face database whose first similarity is greater than the first threshold, then the passenger to be verified and the similar individuals are marked as similar passengers. If the passenger to be verified has been marked as a similar passenger, the feature value is compared with the feature values of other passenger face images in the second face database to obtain a second similarity between the passenger face image and other passenger face images in the second face database; the second face database is a database formed by filtering the first face database based on the travel information data and flight dynamic information data of the passenger to be verified. If there is a single person in the second face database whose second similarity score is greater than the second threshold, then the passenger to be verified is allowed to pass through the boarding gate; the second threshold is adjusted based on the first threshold, the second threshold is greater than the first threshold, and the second threshold is less than the highest threshold; the highest threshold is the critical value at which the airport intelligent cloud system determines that the face recognition is unreliable; If there are zero or more similar individuals in the second facial database whose similarity score is greater than the second threshold, then the passenger to be verified will be denied passage through the boarding gate.
2. The method according to claim 1, characterized in that, After obtaining the passenger's facial image to be verified, the method further includes: Select passenger facial images that meet the requirements.
3. The method according to claim 1, characterized in that, The method further includes: The system stores the travel information data, flight dynamic information data, and passenger facial images of the passengers to be verified.
4. An image recognition system, characterized in that, The system is applied to an airport intelligent cloud system, which includes: a first acquisition module, a second acquisition module, a comparison module, a first judgment module, a second judgment module, and a third judgment module. The first acquisition module is used to acquire the passenger's facial image to be verified; The second acquisition module is used to acquire the feature values of the passenger's face image; The comparison module is used to compare the feature value with the feature values of other passenger face images in the first face database to obtain the first similarity between the passenger face image and other passenger face images in the first face database. The first face database is a database of all registered passenger face images. The first judgment module is used to mark the passenger to be verified and the similar persons as similar passengers if there are multiple similar persons in the first face database whose first similarity is greater than the first threshold. The second judgment module is used to compare the feature value with the feature value of other passenger face images in the second face database if the passenger to be verified has been marked as a similar passenger, to obtain a second similarity between the passenger face image and other passenger face images in the second face database; the second face database is a database formed by filtering the first face database based on the travel information data and flight dynamic information data of the passenger to be verified; The third judgment module is used to allow the passenger to be verified to pass through the boarding gate if there is only one similar person in the second face database whose second similarity is greater than the second threshold; the second threshold is adjusted based on the first threshold, the second threshold is greater than the first threshold, and the second threshold is less than the highest threshold; the highest threshold is the critical value at which the airport intelligent cloud system determines that the face recognition is unreliable; if there are zero or more similar persons in the second face database whose second similarity is greater than the second threshold, the passenger to be verified is refused to pass through the boarding gate.
5. The system according to claim 4, characterized in that, The system also includes: a filtering module; The filtering module is used to filter passenger facial images that meet the requirements.
6. The system according to claim 4, characterized in that, The system also includes: a storage module; The storage module is used to store the travel information data, flight dynamic information data, and passenger facial images of the passenger to be verified.