Passing method, system and equipment of intelligent gate sentry

Abstract

The invention relates to a passing method, system and device of an intelligent gate sentry, and the method comprises the steps: obtaining the portrait data of a visitor when the visitor enters a first preset detection region; performing image matching on the portrait data and a reservation database, and judging whether the visitor is a target reservation visitor; if the visitor is the target reservation visitor, a voice prompt is generated, and the voice prompt is used for prompting the visitor to perform an identity verification action; obtaining verification action data of the visitor; matching the verification action data with a verification action set during reservation; when the verification action data is successfully matched with the verification action, controlling the access control equipment to release; if the visitor is not the target reservation visitor or verification action matching fails, controlling the virtual security interaction terminal to perform access registration on the visitor to obtain access information of the visitor; and when the access information meets the preset requirement, controlling the access control equipment to release. According to the invention, the accuracy of visitor identity authentication can be improved.

Description

Technical Field

[0001] The present application relates to the field of access control, and specifically to a method, system and equipment for accessing an intelligent gate. Background Art

[0002] With the rapid development of intelligent technology, the traditional manual gate management mode is changing towards intelligence and automation. As an important part of the construction of smart communities and smart parks, the intelligent access control system plays an increasingly important role in improving traffic efficiency and strengthening security management. At present, the intelligent gate system is widely used in office buildings, residential communities, public places and other fields. By integrating a variety of sensors and recognition technologies, it realizes the effective management and monitoring of personnel entry and exit, significantly improving traffic efficiency while improving safety.

[0003] In the related technology, the common face recognition access control system currently performs identity authentication by recognizing the face. Such systems usually install recognition equipment or scanning equipment at fixed locations, and visitors actively cooperate to authenticate their identities. However, in the prior art, identity authentication is only performed by face, and visitors with similar appearances may be misidentified. In addition, the existing face recognition system is easily deceived by photos, videos and other means for identity authentication, and cannot effectively perform identity authentication, which reduces the accuracy of visitor identity authentication. Summary of the invention

[0004] The present application provides a method, system and device for passing through a smart gate, which are used to improve the accuracy of visitor identity authentication.

[0005] In a first aspect of the present application, a method for passing through an intelligent gate is provided, which is applied to a server, and the method comprises: When a visitor enters the first preset detection area, the visitor's portrait data is obtained; the portrait data is matched with the reservation database to determine whether the visitor is the target reservation visitor; if the visitor is the target reservation visitor, a voice prompt is generated, and the voice prompt is used to prompt the visitor to perform identity authentication; the visitor's verification action data is obtained; the verification action data is matched with the verification action set during the reservation; when the verification action data and the verification action are matched successfully, the access control device is controlled to release; if the visitor is not the target reservation visitor, or the verification action fails to match, the virtual security interaction terminal is controlled to register the visitor's visit and obtain the visitor's access information. The virtual security interaction terminal is provided with a voice interface and a display screen interface; when the access information meets the preset requirements, the access control device is controlled to release.

[0006] Optionally, image matching is performed between the portrait data and the reservation database to determine whether the visitor is a target reservation visitor, specifically including: Extract multiple frames of continuous portrait data of the visitor from the portrait data; select multiple frames of images whose image clarity is greater than a preset clarity threshold from the multiple frames of continuous portrait data as images to be processed; extract a first feature data set of the visitor from the images to be processed, the first feature data set including facial feature data, body feature data, gait feature data and clothing feature data; use a deep learning model to perform scene analysis on the images to be processed, identify the environmental features of the visitor within a preset range, and obtain a second feature data set, the second feature data set including light intensity data, occlusion degree data and crowd density data; determine whether the visitor is a target appointment visitor based on the first feature data set and the second feature data set.

[0007] Optionally, judging whether the visitor is a target reservation visitor according to the first feature data set and the second feature data set specifically includes: The feature weight of each feature data in the first feature data set is determined by a preset feature weight calculation model; the feature weight is adjusted according to the second feature data set; the similarity between the first feature data set after the weight adjustment and the feature data in the reservation database is calculated to obtain a first matching result; if the first matching result is greater than a first preset threshold, the visitor is determined to be a target reservation visitor; if the first matching result is less than a second preset threshold, the visitor is determined not to be a target reservation visitor; if the first matching result is between the first preset threshold and the second preset threshold, a secondary matching is performed, and the first preset threshold is greater than the second preset threshold.

[0008] Optionally, if the first matching result is between the first preset threshold and the second preset threshold, a secondary matching is performed, specifically including: Extract dynamic temporal features from multiple frames of continuous portrait data; enhance each feature data in the first feature data set based on the dynamic temporal features to obtain target feature data, and construct an enhanced feature data set based on the target feature data; determine the temporal weight of each target feature data in the enhanced feature data set based on the second feature data set; perform weighted processing on the enhanced feature data set based on the temporal weight to obtain a dynamic feature matching score; perform weighted fusion of the dynamic feature matching score with the first matching result to obtain a second matching result; when the second matching result is greater than a second preset threshold, determine that the visitor is a target appointment visitor; when the second matching result is less than the second preset threshold, determine that the visitor is not a target appointment visitor.

[0009] Optionally, if the visitor is not a target scheduled visitor, or the verification action fails to match, the virtual security interaction terminal is controlled to register the visitor and obtain the visitor's access information, including: Determine the visitor's access registration list based on the preset registration list and portrait data; control the display screen interface of the virtual security interactive terminal to display the access registration list, and obtain the visitor's access information through the voice interface and / or the information input box of the display screen interface to obtain a registration database.

[0010] Optionally, the visitor's access registration list is determined based on a preset registration list and the portrait data, specifically including: Perform target detection analysis on the portrait data to identify whether the visitor carries any items and obtain analysis results; if the visitor does not carry any items, the preset registration list is used as the visitor's access registration list; if the visitor carries any items, a supplementary registration list for the visitor is generated based on the analysis results; the preset registration list and the supplementary registration list are combined to obtain the visitor's access registration list.

[0011] Optionally, a method for passing through an intelligent gate, the method further includes: When the target visitor enters the second preset detection area, the movement trajectory data of the target visitor in the second preset detection area is obtained, and the target visitor is any visitor who has passed the access control device; based on the reservation database and the registration database, the movement trajectory data is analyzed to obtain the movement trajectory analysis result of the target visitor; the behavior status of the target visitor is determined by the movement trajectory analysis result; if the behavior status meets the preset abnormal behavior rules, the target visitor is marked as a suspicious visitor, and the alarm device is triggered to emit a warning sound.

[0012] In a second aspect of the present application, a smart gate access system is provided, comprising: A first acquisition module, used for acquiring the visitor's portrait data when the visitor enters the first preset detection area; A judgment module is used to match the portrait data with the reservation database to determine whether the visitor is a target reservation visitor; A prompt module, used to generate a voice prompt if the visitor is a target reservation visitor, and the voice prompt is used to prompt the visitor to perform identity authentication; The second acquisition module is used to obtain the visitor's verification action data; A matching module, used to match the verification action data with the verification action set during the reservation; The first control module is used to control the access control device to release the user when the verification action data matches the verification action successfully; The registration module is used to control the virtual security interaction terminal to register the visitor if the visitor is not a target appointment visitor or the verification action fails to match, and obtain the visitor's access information. The virtual security interaction terminal is provided with a voice interface and a display screen interface; The second control module is used to control the access control device to release access when the access information meets the preset requirements.

[0013] In the third aspect of the present application, an electronic device is provided, including a processor, a memory, a user interface and a network interface, the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device performs any of the methods described above.

[0014] In a fourth aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores instructions, and when the instructions are executed, any of the above methods is executed.

[0015] In summary, one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. By acquiring the portrait data when the visitor enters the preset detection area and matching it with the reservation database, the system can quickly complete the preliminary identity recognition; on this basis, the scheduled visitor is guided to complete the preset verification action through voice prompts, and the acquired action data is matched with the verification action set at the time of appointment, so as to achieve the accuracy of identity authentication; for non-scheduled visitors or verification failures, access registration is carried out through the virtual security interactive terminal, providing an efficient visitor processing process, thereby solving the problem of low accuracy of identity authentication in traditional gate systems. This visitor management mechanism that combines portrait recognition, action verification and intelligent interaction enables the system to ensure security through double verification when scheduled visitors pass through, and quickly handles non-scheduled visitors through intelligent interaction, improving the accuracy of visitor identity authentication.

[0016] 2. By selecting high-definition images from multiple frames of continuous portrait data and extracting multi-dimensional feature data such as visitors' faces, postures, gaits, and clothing, the comprehensiveness of feature extraction is improved; at the same time, the deep learning model is used to analyze environmental features such as light intensity, occlusion degree, and crowd density, which enhances the environmental adaptability of the system; on this basis, the dynamic weight allocation of each dimensional feature is performed through the preset feature weight calculation model, and the weight is adjusted in combination with environmental features to perform hierarchical matching judgment, thus realizing dynamic optimization of feature recognition. This method combines multi-dimensional features and environmentally aware identity recognition mechanisms, enabling the system to adaptively adjust the recognition strategy, which not only improves the accuracy and robustness of identity recognition, but also realizes intelligent regulation of the recognition process.

[0017] 3. By extracting dynamic time series features from continuous portrait data and enhancing the original feature data, the time series continuity of the features is improved; at the same time, the time series weight is determined according to the environmental characteristics and weighted processing is performed to optimize the reliability of the features; finally, through the weighted fusion of the dynamic feature matching score and the initial matching result, a comprehensive evaluation of the judgment basis is achieved, thus solving the problem of insufficient accuracy of single static feature matching in fuzzy situations. This method is based on the enhanced matching mechanism of time series features, which enables the system to make full use of dynamic information for identity judgment, which not only improves the recognition accuracy in fuzzy situations, but also realizes a more comprehensive dynamic evaluation of visitor identity.

[0018] 4. By performing target detection and analysis on visitor portrait data to identify personal belongings, intelligent judgment of registration needs is achieved; for visitors without personal belongings, a preset registration list is used, and for visitors with belongings, a supplementary registration list is generated and combined with the preset list, providing a differentiated information collection solution; at the same time, a multi-modal information entry method is provided through the display interface and voice interface of the virtual security interactive terminal, which optimizes the user's registration experience, thereby solving the problems of fixed forms, incomplete information collection and single entry methods in traditional registration methods. This method is based on the intelligent registration mechanism of portrait analysis, which enables the system to realize intelligent adaptation of the registration process, which not only improves the completeness and accuracy of registration information, but also improves registration efficiency through convenient information entry methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a flow chart of a method for passing through an intelligent gate in an embodiment of the present application; Figure 2 It is a structural schematic diagram of a passage system of an intelligent gate in an embodiment of the present application; Figure 3 It is a schematic diagram of the structure of an electronic device in an embodiment of the present application.

[0020] Explanation of the accompanying drawings: 201, first acquisition module; 202, judgment module; 203, prompt module; 204, second acquisition module; 205, matching module; 206, first control module; 207, registration module; 208, second control module; 209, alarm module; 301, processor; 302, communication bus; 303, user interface; 304, network interface; 305, memory. DETAILED DESCRIPTION

[0021] In order to enable technicians in this field to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments.

[0022] In the description of the embodiments of the present application, words such as "for example" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "for example" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "for example" or "for example" is intended to present related concepts in a specific way.

[0023] In the description of the embodiments of the present application, the meaning of the term "multiple" refers to two or more. For example, multiple systems refer to two or more systems, and multiple screen terminals refer to two or more screen terminals. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. The terms "include", "comprise", "have" and their variations all mean "including but not limited to", unless otherwise specifically emphasized.

[0024] Figure 1 It is a flow chart of a method for passing through an intelligent gate in an embodiment of the present application.

[0025] See also Figure 1 In the embodiment of the present application, a method for passing through a smart gate is applied to a server, and the method includes: S101, when a visitor enters a first preset detection area, obtaining the visitor's portrait data; When a visitor enters the first preset detection area, the access system of the smart gate will automatically trigger the data collection process, and detect the visitor in real time and obtain the visitor's portrait data through the camera, depth sensor or other sensing equipment deployed in the area. The first preset detection area refers to a specific area near the gate. The first preset detection area is usually pre-set by the system and monitored by devices such as cameras, infrared sensors or radars. Its main function is to sense the arrival of visitors in real time and trigger the corresponding data collection process. The size and shape of the first preset detection area can be adjusted according to the scene requirements. For example: for the lobby access control of an enterprise, the first preset detection area can be set within 2-3 meters in front of the door. When the visitor enters the area, the camera automatically starts recognition; for the community gate, the first preset detection area can cover the doorway. When the visitor approaches the access control device, the system starts identity verification.

[0026] Portrait data mainly includes facial data and body data. Facial data is the original facial feature information extracted by the face detection algorithm, including key features such as facial structure, facial contour, and skin color; body data is used to record the visitor's overall profile information, including body proportions, posture, and external contour shape, to ensure that it can provide auxiliary recognition basis in the subsequent matching process. In addition, if the system detects that the visitor is wearing a hat, mask, or other items that may affect facial recognition, the visitor may be prompted to adjust the wearing items through voice or display to ensure the integrity and clarity of the facial image data.

[0027] S102, performing image matching between the portrait data and the reservation database to determine whether the visitor is a target reservation visitor; Specifically, multiple frames of continuous portrait data of the visitor are extracted from the portrait data; multiple frames of images whose image clarity is greater than a preset clarity threshold are selected from the multiple frames of continuous portrait data as images to be processed; a first feature data set of the visitor is extracted from the images to be processed, and the first feature data set includes facial feature data, body feature data, gait feature data and clothing feature data; a deep learning model is used to perform scene analysis on the image to be processed to identify environmental features of the visitor within a preset range to obtain a second feature data set, and the second feature data set includes light intensity data, occlusion degree data and crowd density data; and whether the visitor is a target appointment visitor is determined based on the first feature data set and the second feature data set.

[0028] Extract multiple frames of continuous portrait data from the acquired portrait data. In the process of extracting multiple frames of continuous portrait data, the system continuously captures the visitor's facial image and body image at a high frame rate to ensure that complete and clear image data can be obtained even when the visitor moves, the light changes, or there is occlusion.

[0029] After acquiring multiple frames of continuous portrait data, the system will evaluate the quality of all images and select images that meet the requirements as images to be processed based on the preset clarity threshold. Specifically, the system will calculate the contrast, edge sharpness, noise level, exposure, and visibility of facial key points of each frame, and eliminate blurred, overexposed, or severely occluded images. For example, if a frame of image is blurred due to the rapid movement of the visitor, or is overexposed due to excessive light, the system will automatically discard the image and give priority to images with higher clarity and complete facial features for subsequent processing. In addition, in order to further improve the accuracy of the data, the system may also use image super-resolution reconstruction technology to optimize the clarity of some images to ensure that sufficiently clear images to be processed are obtained even in low light or motion blur.

[0030] After selecting the images to be processed that meet the clarity requirements, the system will use computer vision and deep learning technology to perform feature analysis on the images and extract the first feature data set of the visitor, including facial feature data, body feature data, gait feature data and clothing feature data. Facial feature data is the core of identity recognition. The system will use facial key point detection and feature encoding technology to extract information such as the visitor's facial features, facial contour, skin color, eye distance, nose bridge width, etc., and convert them into facial feature vectors that can be used for matching. Body feature data includes information such as the visitor's height, shoulder width, limb proportions, etc. These data can be used as auxiliary information, especially in the case of multiple people walking together or partial occlusion, to help improve the stability of identity matching. Gait feature data analyzes the visitor's walking patterns such as stride, center of gravity change, arm swing amplitude, etc. through the gait recognition algorithm, and generates a unique gait feature vector, so that when the visitor's facial information is limited (such as wearing a mask or lowering his head), it can still provide reliable identity matching basis. Clothing feature data is used to record the color, style, and accessories (such as backpacks, hats, etc.) of the visitor's clothes when he enters the detection area as additional reference information for identity confirmation. For example, if a visitor registers that he or she is wearing a blue top and black pants when making an appointment, the system can further confirm his or her identity by comparing the clothing features. Even if the visitor's facial recognition confidence is low, the clothing features can still be used to assist in determining his or her identity.

[0031] After obtaining the first feature data set of the visitor, the system uses a deep learning model to perform scene analysis on the processed image to identify key information of the visitor's environment and generate a second feature data set, which includes light intensity data, occlusion degree data and crowd density data. The light intensity data is used to describe the brightness of the area where the visitor is located. The system will analyze whether the current lighting is clear enough through the image histogram and determine whether it is necessary to adjust the camera parameters or enable infrared fill light technology to improve the image quality. For example, if the system detects that the light in the area where the visitor is located is weak, it may automatically enhance the image brightness or adjust the exposure parameters to improve the availability of portrait data. The occlusion degree data is used to evaluate the visibility of the visitor's face or body. For example, if the visitor wears a mask, hat, sunglasses, or is partially blocked by others in a group, the system will record this information and perform weighted processing during the matching process to reduce the recognition error caused by occlusion. For example, if the visitor's face is covered by a mask, the system will reduce the weight of facial features and increase the matching proportion of gait features and body features. Crowd density data is used to analyze the number and distribution of people around visitors. For example, in high-traffic areas, the system may add multi-frame analysis to ensure that the matching results of the target visitor are not affected by interference from other people in the camera image.

[0032] After obtaining the first feature data set and the second feature data set, the system will match the visitor's identity according to the first feature data set and the second feature data set to determine whether the visitor is the target appointment visitor. Specifically, the feature weight of each feature data in the first feature data set is determined by a preset feature weight calculation model; the feature weight is adjusted according to the second feature data set; the first feature data set after the weight adjustment is similarly calculated with the feature data in the appointment database to obtain a first matching result; if the first matching result is greater than the first preset threshold, the visitor is determined to be the target appointment visitor; if the first matching result is less than the second preset threshold, the visitor is determined not to be the target appointment visitor; if the first matching result is between the first preset threshold and the second preset threshold, a second matching is performed, and the first preset threshold is greater than the second preset threshold.

[0033] The feature weight is calculated for each feature data in the first feature data set by a preset feature weight calculation model to ensure that the contribution of different features is reasonably distributed during identity matching. The preset feature weight calculation model is obtained based on a large amount of historical matching data and deep learning algorithm training, and can set weights according to the recognition of different features in identity recognition to improve the accuracy and stability of matching. For example: in an ideal environment (sufficient light, no obstruction, and low crowd density), the system usually gives the highest weight (such as 70%) to facial feature data, because face recognition is the core basis for identity matching; body feature data and gait feature data are given 15% and 10% weights respectively as auxiliary judgment basis for facial recognition, especially to provide additional matching support when facial features are limited; clothing feature data may change greatly due to the visitor's clothing changes, so it is only given 5% weight as auxiliary reference information. The preset feature weight calculation model can be calculated using statistical analysis methods (such as weighted average, feature regression) or deep learning methods (such as neural network training), so that it can automatically learn the best feature weight distribution to adapt to different identity matching needs.

[0034] After obtaining the feature weights of the first feature data set, the system will dynamically adjust the feature weights of the first feature data set in combination with the second feature data set to adapt to different environmental conditions. Specifically, the system will first analyze the current environmental status of the visitor, such as detecting whether the light is sufficient, whether the visitor's face is blocked, whether there are a large number of people around him, etc., and adjust the distribution ratio of feature weights according to these environmental factors. For example, in an ideal environment with sufficient light and no obstruction, the system will maintain a high weight of facial features (such as 70%), but if the ambient light is dim, the system may reduce the weight of facial features (such as to 50%), and increase the weight of gait features and body features accordingly to reduce the impact of insufficient light on recognition. Similarly, if the visitor wears a mask or a hat, resulting in partial obstruction of facial features, the system will appropriately reduce the weight of facial features (such as to 40%), and increase the influence of gait features (such as 30%) and body features (such as 25%) to ensure the stability of the matching results. In addition, in crowded scenes, since visitors may partially overlap with other people, the distinguishability of clothing features will be reduced. Therefore, the system may reduce the weight of clothing features (for example, from 5% to 2%), while increasing the weight of gait features to ensure that the system can accurately identify the target visitor even when multiple people are traveling together.

[0035] After completing the weight adjustment of the first feature data set, the system will use the similarity calculation algorithm to match the adjusted feature data with the target scheduled visitor feature data in the reservation database, and calculate a first matching result, that is, the matching score between the current visitor and the scheduled visitor. The matching score calculation usually uses cosine similarity, Euclidean distance, Mahalanobis distance or deep neural network embedding matching algorithm to measure the similarity between the feature data of the current visitor and the target visitor registered in the database. For example, assuming that the feature data of a visitor after weight adjustment is: facial features 50%, gait features 30%, body features 15%, clothing features 5%, then the system will calculate a matching score based on the similarity of these features, such as 87%. When calculating the matching degree, the system will comprehensively consider the similarity of multiple features to ensure that even if the availability of some features is reduced due to environmental influences (such as facial features are limited due to insufficient light), other features can still be used for effective matching.

[0036] After calculating the first matching result, the system will compare the matching score with the first preset threshold. If the matching degree is higher than the threshold, it means that the characteristic data of the current visitor is highly matched with the characteristic data of the target appointment visitor. The system can directly confirm their identity and allow them to enter the next step of the access process. For example, if the first preset threshold set by the system is 85%, and the matching degree calculation result of the visitor is 90%, it means that the credibility of their identity is extremely high. The system can directly determine that they are the target appointment visitor and allow them to pass the identity verification.

[0037] If the first matching result is lower than the second preset threshold, it indicates that the characteristic data of the current visitor has a low matching degree with the target visitor in the reservation database, and the system will directly determine that the visitor is not the target reservation visitor and refuse to pass the identity authentication.

[0038] If the first matching result is between the first preset threshold and the second preset threshold, that is, the matching degree is high but has not yet reached the standard of direct confirmation, the system will trigger a secondary matching mechanism to further improve the accuracy of identity recognition. Specifically, dynamic time series features are extracted from multiple frames of continuous portrait data; based on the dynamic time series features, each feature data in the first feature data set is enhanced to obtain target feature data, and an enhanced feature data set is constructed based on the target feature data; based on the second feature data set, the time series weight of each target feature data in the enhanced feature data set is determined; based on the time series weight, the enhanced feature data set is weighted to obtain a dynamic feature matching score; the dynamic feature matching score is weightedly fused with the first matching result to obtain a second matching result; when the second matching result is greater than the second preset threshold, the visitor is determined to be a target appointment visitor; when the second matching result is less than the second preset threshold, the visitor is determined not to be a target appointment visitor.

[0039] Extract dynamic temporal features from multiple frames of continuous portrait data. Dynamic temporal features refer to the changes in facial expressions, gait rhythm, body posture and the continuity of clothing details of visitors over a period of time. These features can provide richer identity information than single-frame images. For example, facial features can be analyzed through micro-expression changes such as the curvature of the mouth corners and the frequency of eye blinking, while gait features can be extracted through stride length, arm swing amplitude, and step consistency.

[0040] After extracting the dynamic temporal features, the system uses these temporal information to enhance the features of the first feature data set to improve the matching accuracy. The feature enhancement process includes multi-frame feature fusion, temporal stability analysis, and abnormal frame removal, that is, by comprehensively analyzing the data of multiple frames, the robustness of the feature data is improved. For example, if the facial feature extraction of a certain frame image is inaccurate due to changes in light, the system can compensate through the information of the previous and next frames to obtain a more stable facial feature vector. In addition, gait features can also be enhanced through multi-frame fusion, such as calculating the average stride and frequency of visitors in multiple gait cycles, thereby reducing the matching error caused by single-frame anomalies. Finally, after feature enhancement, the system will obtain an enhanced feature data set, in which each target feature data is optimized to ensure that the identity information used for matching is more accurate and stable.

[0041] After obtaining the enhanced feature data set, the system will further combine the second feature data set to calculate the temporal weight, that is, assign different importance weights to the target feature data at different time points to adapt to the current environment. The system will calculate the corresponding environmental impact factor E(t) for each frame of enhanced feature data. The environmental impact factor E(t) is used to measure the reliability of the feature data of the target frame in the current environment. The environmental impact factor E(t) is usually composed of the light intensity impact factor L(t), the occlusion degree impact factor O(t) and the crowd density impact factor D(t). The calculation formula of the environmental impact factor E(t) is: The closer the environmental impact factor E(t) is to 1, the higher the data quality of the target frame; the lower the value, the greater the impact of the environment on the data of the target frame. After calculating the environmental impact factor E(t) of each frame, the system will calculate its feature confidence C(t) in combination with the feature stability of the target frame. The feature confidence measures the reliability of the feature data of the target frame in the entire time series data, and is calculated as follows: , where E(t) is the environmental influencing factor of the target frame, S(t) is the feature stability of the target frame, which indicates the similarity between the feature data of the target frame and the previous and next frames (e.g., the similarity of gait features at multiple time steps), and α and β are hyperparameters for adjusting weights, usually α>β, because environmental factors have a greater impact on features. Feature stability S(t) can be calculated using cosine similarity, LSTM (Long Short-Term Memory Network), or time series regression models. For example, if the gait features of a frame are highly similar to those of the previous and next three frames, then its S(t) value is high, indicating that the features of the frame are relatively stable. Conversely, if the features of a frame fluctuate greatly (e.g., facial expressions change dramatically, or gait is abnormal), then the S(t) value is low. After calculating the feature confidence C(t) of each frame, the system normalizes the feature data of all frames to ensure that the sum of the weights is 1: , where W(t) is the temporal weight of the t-th frame, C(t) is the feature confidence of the t-th frame, C(i) is the feature confidence of the i-th frame, and N is the total number of frames, that is, the number of frames collected by the system in this time window (for example, 10 frames or 20 frames).

[0042] After calculating the temporal weight of each target feature data, the system performs weighted calculations on the enhanced feature data set based on these weights to obtain a dynamic feature matching score. Weighted calculations usually use weighted average, temporal feature fusion, or deep learning temporal modeling (such as LSTM, Transformer) to comprehensively analyze feature information at multiple time points. For example, if a visitor's facial features have a high degree of match in some frames, but a low degree of match in other frames due to lighting or occlusion, the system will use a temporal weighting method to make the matching score more dependent on high-confidence frames, thereby minimizing errors caused by environmental influences. Ultimately, the system generates a dynamic feature matching score for subsequent identity matching decisions.

[0043] After calculating the dynamic feature matching score, the system will perform weighted fusion with the first matching result (i.e., the matching degree calculated based on the single-frame features) to obtain the second matching result. The weighted fusion method can adopt linear weighting, Bayesian fusion, or neural network decision fusion to comprehensively consider the importance of static feature matching (first matching result) and dynamic feature matching (dynamic feature matching score). For example, in the case of good light and no occlusion, the weight of the first matching result may be higher (such as 70%), while the weight of the dynamic feature matching score is lower (such as 30%); but in the case of poor light or more occlusion, the weight of the dynamic feature matching score may be increased (such as 60%) to ensure that the recognition result is not affected by the error of the single-frame image. Finally, the system will calculate a comprehensive matching degree, that is, the second matching result, for final identity confirmation.

[0044] The system compares the obtained second matching result with the second preset threshold. If the second matching result is greater than the second preset threshold, it indicates that the identity characteristics of the current visitor match the characteristics of the scheduled visitor, and the system can directly confirm their identity and allow them to enter. For example, if the first preset threshold is set to 85%, and the second matching result of the visitor is 88%, the system will determine that the visitor is the target scheduled visitor. If the second matching result is less than the second preset threshold, it indicates that the characteristic information of the visitor has a low matching degree with the scheduled visitor, and the system will directly determine that the visitor is not the target scheduled visitor and refuse to pass the identity verification.

[0045] S103: If the visitor is a target reservation visitor, a voice prompt is generated, and the voice prompt is used to prompt the visitor to perform identity authentication; When the system determines that the visitor is a target appointment visitor, in order to further ensure the accuracy of the identity and prevent misidentification, the system will automatically generate a voice prompt to guide the visitor to perform the preset identity verification action. The voice prompt can be broadcasted through the voice playback system of the smart gate, for example: "Hello, your appointment information has been confirmed, please perform the identity verification action." The identity verification action is the action set by the visitor when making an appointment, such as waving, nodding, raising the hand to make an OK gesture, turning 360 degrees, etc. These actions can be used as an additional identification method to ensure that the visitor is indeed the person who has made an appointment, rather than an unauthorized person who only looks similar. For example, a company requires visitors to select a specific identity verification action (such as waving the right hand twice) when making an appointment. When the visitor arrives at the gate and is initially identified as a scheduled visitor, the system will prompt him to perform the verification action through voice: "Please wave your right hand twice to complete the identity verification." In this way, even if there are other people who look similar to the visitor, the system can still perform a second confirmation through action matching to improve security. In addition, the voice prompt can also be combined with the screen display guidance to allow visitors to more intuitively understand the actions to be performed and avoid information omissions caused by hearing impairment or noisy environment.

[0046] S104, obtaining the visitor's verification action data; After the system guides the visitor to perform the identity authentication action through voice prompts, the access system of the smart gate will capture and obtain the visitor's verification action data in real time to further ensure the authenticity of his identity. The acquisition of verification action data usually relies on high-precision cameras, depth sensors or laser radars and other devices, which can accurately capture the dynamic information of the visitor's body movement trajectory, posture changes, gesture features, etc., and convert them into action feature data that can be used for matching. For example, if the verification action set by the visitor when making an appointment is to wave his right hand twice, the system will detect the visitor's right hand trajectory, number of waves, wave direction, amplitude, etc. through the camera and depth sensor, and record these data as the current verification action data. If the visitor's verification action involves more complex dynamics (such as turning 360 degrees or crossing and unfolding both hands), the system will combine time series analysis to ensure the integrity and accuracy of the action. In addition, in order to improve the robustness of recognition, the system may also use posture estimation algorithms (such as OpenPose, MediaPipe) or deep learning models (such as LSTM, Transformer) to analyze the action data, filter out noise data, and ensure that the recognized action meets the visitor's appointment settings.

[0047] S105, matching the verification action data with the verification action set during the reservation; After obtaining the visitor's verification action data, the smart gate's access system will match the data with the verification action set by the visitor when making an appointment to further confirm the authenticity of the visitor's identity. The matching process usually uses deep learning action recognition models (such as LSTM, Transformer) or computer vision algorithms (such as dynamic time warping, cosine similarity, and Euclidean distance).

[0048] Specifically, the system will first convert the visitor's current motion data into standardized time series features, and then compare it with the visitor's preset verification motion data in the reservation database. For example, if the verification motion set by the visitor when making a reservation is to wave his right hand twice, the system will analyze the currently captured motion data to check whether there is a change in the trajectory of the right hand, whether the number of waves is 2, and whether the direction and amplitude of the wave match. If the original action set by the visitor is to turn 360 degrees, the system will use the posture estimation algorithm to analyze the changes in the rotation angles of key points of the body (such as shoulders, waist, and legs) in the time series to ensure that its motion trajectory is highly consistent with the preset trajectory in the database.

[0049] In addition, in order to improve the fault tolerance of matching, the system may allow errors within a certain range. For example, if the action performed by the visitor deviates slightly from the preset action in angle, time, or amplitude (such as an error of less than 10%), the system can still determine that the match is successful to avoid misidentification caused by differences in individual action habits or slight errors. If the match is successful, the system will enter step S106 (control the access control device to release); if the match fails, it will enter step S107 (visitor registration process) to ensure that visitors who fail to pass the verification can complete identity confirmation through other means.

[0050] S106, when the verification action data matches the verification action successfully, the access control device is controlled to release the person; When the visitor's verification action data successfully matches the verification action recorded during the appointment, the system will send a door opening command to the access control unit to control the access control device to release. For example, the smart door lock is unlocked, the gate is opened, the access door slides open automatically, etc., and may be accompanied by voice broadcast or screen display information to prompt that the visitor has successfully passed the verification. For example, the gate system may broadcast: "Authentication is successful, the access control is open, please pass." The screen may also display: "Welcome, please pay attention to safety." In addition, the system can also automatically record the time of visitor entry and exit, the number of the access control device passed, and other information while the visitor passes, and upload it to the security management database for subsequent query or management. For example, in corporate office buildings, smart parks or high-security areas, managers can view visitor entry and exit records through the background system to ensure that the access behavior of all personnel is traceable.

[0051] S107, if the visitor is not the target scheduled visitor, or the verification action fails to match, the virtual security interaction terminal is controlled to register the visitor and obtain the visit information of the visitor, and the virtual security interaction terminal is provided with a voice interface and a display screen interface; Specifically, the visitor's access registration list is determined based on the preset registration list and portrait data; the display screen interface of the virtual security interactive terminal is controlled to display the access registration list, and the visitor's access information is obtained through the voice interface and / or the information input box of the display screen interface to obtain a registration database.

[0052] First, the visitor's access registration list is determined based on the preset registration list and the portrait data. Specifically, target detection analysis is performed on the portrait data to identify whether the visitor carries any items and obtain analysis results. If the visitor does not carry any items, the preset registration list is used as the visitor's access registration list. If the visitor carries any items, a supplementary registration list for the visitor is generated based on the analysis results. The preset registration list and the supplementary registration list are combined to obtain the visitor's access registration list.

[0053] After the visitor enters the first preset detection area, the access system of the smart gate will perform target detection analysis on the visitor's portrait data to identify whether the visitor is carrying any personal belongings. This analysis process can usually be done through deep learning target detection algorithms (such as Faster R-CNN, YOLO, SSD) or computer vision technology, by detecting the visitor's body area and its surrounding environment to identify whether the visitor is carrying a backpack, briefcase, handbag, suitcase, electronic device (such as laptop, tablet), special items (such as folders, sample boxes), etc.

[0054] For example, when a visitor enters the first preset detection area, the system's camera will automatically capture his full-body image and analyze whether the visitor is carrying anything based on the object detection model. If a handbag is detected in the visitor's hand or a backpack is detected on the back, the system will further analyze the size and shape of the item and store the recognition result in the analysis result database. On the contrary, if the system does not detect that the visitor is carrying any items, it will determine that the visitor is not carrying any additional items and record the result.

[0055] When the target detection analysis shows that the visitor does not carry any items, the system will directly call the standard preset registration list as the visitor's visit registration form. The preset registration list usually contains the following specific information items: basic identity information of the visitor (name, ID number, contact information, etc.), purpose of visit (business negotiation, interview, visiting relatives, etc.), information of the visitee (department, name, position, etc.), expected stay time, visit time and other basic registration items.

[0056] When the system detects that a visitor is carrying items, it will start an intelligent supplementary registration list generation program based on the analysis results of the target detection. This program will automatically generate corresponding registration items according to different types of items. The supplementary registration list contains the following details: specific type classification of items (according to the preset item classification standards), detailed quantity statistics of each type of items, description of the appearance characteristics of the items (including identifiable information such as color, material, brand, etc.), approximate size data of the items, and notes on special items (such as fragile items, valuables, etc.). For example, for a visitor carrying "a black backpack (about 40x30cm) and a silver suitcase (about 65x45cm)", the system will generate a supplementary registration list containing these detailed information.

[0057] The preset registration list and the supplementary registration list are combined to generate a complete visit registration list. The final visit registration list can be divided into several main sections: the first part is the visitor basic information area, which contains all standard registration items; the second part is the personal belongings registration area, which lists all detected items in detail; the third part is the precautions and management requirements automatically generated by the system based on the visitor type and item situation; the fourth part is the access rights and timeliness description.

[0058] After the access registration list is generated, the system controls the display screen interface of the virtual security interactive terminal to display the access registration list, and obtains the visitor's access information through the voice interface and / or the information input box of the display screen interface to obtain the registration database. When the visitor needs to register information, the virtual security interactive terminal will intelligently guide the inquiry item by item according to the preset information registration template, such as "What is your name?", "What is the purpose of your visit this time?", "What department and personnel do you want to visit?", etc. These questions will be asked synchronously through the voice of the virtual security guard and displayed in clear text on the display screen. Visitors can choose to answer these questions directly by voice according to their personal habits (the voice answer will be automatically converted into text and displayed on the screen), or fill in the information in the corresponding input box by handwriting by touching the display screen. The system will record and store each registration information of the visitor in real time, and after the visitor confirms that all the information is correct, the complete registration data will be saved in a structured form in the registration database.

[0059] S108. When the access information meets the preset requirements, the access control device is controlled to release the access.

[0060] After the system has collected all the visitor's registration information, it will conduct an intelligent audit of this information according to the preset audit rules. The first step is to check whether the visitor is on the abnormal visit list (the source of the abnormal visit list includes: visitors with violations in historical visit records, such as damaging public property, violating access control management regulations, carrying contraband, overstaying, repeatedly violating appointment time, malicious provocation, etc.; bad visitor information shared by cooperative units or security departments; and the system automatically analyzes visitor behavior patterns to determine the existence of safety hazards. Records), then continue to review the integrity of the visitor's identity information, the rationality of the purpose of the visit, the validity of the visit object (such as confirming whether the person being visited is on duty), whether the visit time is within the permitted range (such as whether it is during working hours), whether the personal belongings meet safety requirements, and other dimensions. For example, when a visitor named Zhang San's registration information shows "from ABC Company, carrying a laptop, making an appointment to visit Manager Li Si of the Technology Department for business negotiation", the system verifies that the visitor is not on the abnormal visit list, and the information is complete, the visit time is during working hours, and the visitor Li Si is indeed on duty, etc., and all the conditions meet the preset requirements, the system will send a door opening command to the access control controller, and at the same time prompt the visitor through the LED display and voice "Verification passed, please pass", and the access control device will automatically open to allow the visitor to enter. At the same time, the system will automatically send a reminder message of the visitor's arrival to the person being visited, and record the visitor's entry time in the background.

[0061] Optional, in Figure 1 Based on the embodiment shown, the following steps may also be performed: When the target visitor enters the second preset detection area, the movement trajectory data of the target visitor in the second preset detection area is obtained, and the target visitor is any visitor who has passed the access control device; based on the reservation database and the registration database, the movement trajectory data is analyzed to obtain the movement trajectory analysis result of the target visitor; the behavior status of the target visitor is determined by the movement trajectory analysis result; if the behavior status meets the preset abnormal behavior rules, the target visitor is marked as a suspicious visitor, and the alarm device is triggered to emit a warning sound.

[0062] After the target visitor successfully passes through the access control device and enters the second preset detection area (such as the interior of an office building, the visitor waiting area, the conference room corridor, etc.), the system will automatically obtain its movement trajectory data for subsequent analysis of its behavior pattern. This process usually relies on high-definition cameras, infrared sensors, or lidar to ensure that the visitor's movement path in the area can be accurately captured. For example, if the visitor enters the public corridor of an office building, the camera and thermal sensing sensor will record its walking direction, stop position, movement speed and other information in real time, and generate a time series trajectory, which is stored in the system database.

[0063] After successfully obtaining the target visitor's movement trajectory data, the system will compare and analyze the data with the visitor information in the reservation database and registration database to determine whether the visitor's movement path and behavior are consistent with his or her reservation information and registration information.

[0064] The reservation database usually contains the visitor's purpose of visit, reservation area, visitor information, etc., while the registration database may also record the visitor's belongings, access rights, pass records, etc. For example, if the visitor's scheduled visit location is the conference room on the 3rd floor, but the system detects that he has been wandering around the R&D laboratory on the 5th floor for a long time, this behavior may be marked as abnormal. The system uses trajectory mining algorithms (such as HMM hidden Markov model, DBSCAN clustering, LSTM trajectory prediction) to analyze the visitor's movement path and generate action trajectory analysis results, which may include: whether the visitor follows the scheduled path (such as whether he goes to the registered conference room or office area); whether the visitor stays in a specific area for too long (such as whether he wanders around sensitive areas for a long time); whether the visitor's movement pattern is abnormal (such as whether there are abnormal behaviors such as repeatedly entering and exiting the same floor, suddenly changing direction, etc.).

[0065] After completing the action trajectory analysis, the system will determine the visitor's current behavior status based on the visitor's movement pattern, stay time, walking path and other information, and judge whether it meets the preset abnormal behavior rules. The behavior status usually includes: normal visitor behavior, the visitor walks along the scheduled path and completes the visit within a reasonable time (such as the visitor enters the conference room and stays for 30 minutes before leaving); wandering behavior, the visitor stays in a non-scheduled area for a long time, or repeatedly enters and exits a certain area (such as walking back and forth at the door of the R&D laboratory for 10 minutes); deviation from the access route behavior, the visitor does not follow the scheduled path, but goes to an unauthorized area (such as making an appointment to visit the 3rd floor, but actually goes to the 6th floor); fast escape behavior, the visitor suddenly accelerates and shows the characteristics of trying to evade monitoring (such as running fast and bypassing the camera area); abnormal stay behavior, the visitor stays abnormally in certain sensitive areas (such as staying at the door of the server room for more than 5 minutes without access rights).

[0066] If the visitor's behavior status meets the preset abnormal behavior rules (such as staying in sensitive areas for a long time, entering unauthorized floors or rooms, repeatedly entering and exiting the same area, and escaping quickly, etc.), the system will immediately mark the visitor as a suspicious visitor, trigger the alarm device to sound a warning, and send a real-time alarm notification to the security system. For example, if the visitor's scheduled visit location is the 305 conference room on the 3rd floor, but the system detects that he has entered the finance office on the 4th floor and stayed in the area for more than 5 minutes, the system will classify his behavior as "deviation from the access route behavior". If the visitor then wanders abnormally (such as staying at the doors of multiple offices on the same floor and trying to enter different rooms), the system will further increase the risk level and mark him as a suspicious visitor. At this time, the gate guard or security center will receive a real-time alarm, including the visitor's current location, behavior trajectory, video screenshots, etc., so that security personnel can conduct remote voice warnings or on-site verification.

[0067] See also Figure 2 , is a schematic diagram of the structure of a smart gate passage system provided in an embodiment of the present application, and a smart gate passage system 200 specifically includes: The first acquisition module 201 is used to acquire the visitor's portrait data when the visitor enters the first preset detection area; The judgment module 202 is used to perform image matching between the portrait data and the reservation database to determine whether the visitor is a target reservation visitor; Prompt module 203, used to generate a voice prompt if the visitor is a target reservation visitor, the voice prompt is used to prompt the visitor to perform identity authentication; The second acquisition module 204 is used to acquire the visitor's verification action data; A matching module 205, for matching the verification action data with the verification action set during the reservation; The first control module 206 is used to control the access control device to release the user when the verification action data matches the verification action successfully; The registration module 207 is used to control the virtual security interaction terminal to register the visitor if the visitor is not a target scheduled visitor or the verification action fails to match, and obtain the visitor's access information. The virtual security interaction terminal is provided with a voice interface and a display screen interface; The second control module 208 is used to control the access control device to release the access when the access information meets the preset requirements.

[0068] Optionally, the determination module 202 is specifically configured to: Extract multiple frames of continuous portrait data of the visitor from the portrait data; select multiple frames of images whose image clarity is greater than a preset clarity threshold from the multiple frames of continuous portrait data as images to be processed; extract a first feature data set of the visitor from the images to be processed, the first feature data set including facial feature data, body feature data, gait feature data and clothing feature data; use a deep learning model to perform scene analysis on the images to be processed, identify the environmental features of the visitor within a preset range, and obtain a second feature data set, the second feature data set including light intensity data, occlusion degree data and crowd density data; determine whether the visitor is a target appointment visitor based on the first feature data set and the second feature data set.

[0069] Optionally, the determination module 202 is further specifically configured to: The feature weight of each feature data in the first feature data set is determined by a preset feature weight calculation model; the feature weight is adjusted according to the second feature data set; the similarity between the first feature data set after the weight adjustment and the feature data in the reservation database is calculated to obtain a first matching result; if the first matching result is greater than a first preset threshold, the visitor is determined to be a target reservation visitor; if the first matching result is less than a second preset threshold, the visitor is determined not to be a target reservation visitor; if the first matching result is between the first preset threshold and the second preset threshold, a secondary matching is performed, and the first preset threshold is greater than the second preset threshold.

[0070] Optionally, the judgment module 202 is also specifically used to: extract dynamic timing features from multiple frames of continuous portrait data; enhance each feature data in the first feature data set based on the dynamic timing features to obtain target feature data, and construct an enhanced feature data set based on the target feature data; determine the timing weight of each target feature data in the enhanced feature data set based on the second feature data set; perform weighted processing on the enhanced feature data set based on the timing weight to obtain a dynamic feature matching score; perform weighted fusion of the dynamic feature matching score with the first matching result to obtain a second matching result; when the second matching result is greater than a second preset threshold, determine that the visitor is a target appointment visitor; when the second matching result is less than the second preset threshold, determine that the visitor is not a target appointment visitor.

[0071] Optionally, the registration module 207 is specifically used for: Determine the visitor's access registration list based on the preset registration list and portrait data; control the display screen interface of the virtual security interactive terminal to display the access registration list, and obtain the visitor's access information through the voice interface and / or the information input box of the display screen interface to obtain a registration database.

[0072] Optionally, the registration module 207 is further specifically used for: Perform target detection analysis on the portrait data to identify whether the visitor carries any items and obtain analysis results; if the visitor does not carry any items, the preset registration list is used as the visitor's access registration list; if the visitor carries any items, a supplementary registration list for the visitor is generated based on the analysis results; the preset registration list and the supplementary registration list are combined to obtain the visitor's access registration list.

[0073] Optionally, the system further includes an alarm module 209, which is specifically used for: When the target visitor enters the second preset detection area, the movement trajectory data of the target visitor in the second preset detection area is obtained, and the target visitor is any visitor who has passed the access control device; based on the reservation database and the registration database, the movement trajectory data is analyzed to obtain the movement trajectory analysis result of the target visitor; the behavior status of the target visitor is determined by the movement trajectory analysis result; if the behavior status meets the preset abnormal behavior rules, the target visitor is marked as a suspicious visitor, and the alarm device is triggered to emit a warning sound.

[0074] It should be noted that: when the device provided in the above embodiment realizes its function, only the division of the above functional modules is used as an example. In actual application, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the device and method embodiments provided in the above embodiment belong to the same concept, and the specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0075] This embodiment also discloses an electronic device, referring to Figure 3 The electronic device may include: at least one processor 301 , at least one communication bus 302 , a user interface 303 , a network interface 304 , and at least one memory 305 .

[0076] The communication bus 302 is used to realize the connection and communication between these components.

[0077] The user interface 303 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 303 may also include a standard wired interface and a wireless interface.

[0078] The network interface 304 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0079] Among them, the processor 301 may include one or more processing cores. The processor 301 uses various interfaces and lines to connect various parts in the entire server, and executes various functions of the server and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 305, and calling data stored in the memory 305. Optionally, the processor 301 can be implemented in at least one hardware form of digital signal processing (Digital Signal Processing, DSP), field programmable gate array (Field-Programmable Gate Array, FPGA), and programmable logic array (Programmable Logic Array, PLA). The processor 301 can integrate one or a combination of a central processing unit (Central Processing Unit, CPU), a graphics processing unit (Graphics Processing Unit, GPU) and a modem. Among them, the CPU mainly processes the operating system, user interface and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 301, and it can be implemented separately through a chip.

[0080] Among them, the memory 305 may include a random access memory (Random Access Memory, RAM) and may also include a read-only memory (Read-Only Memory). Optionally, the memory 305 includes a non-transitory computer-readable storage medium. The memory 305 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 305 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 305 may also be optionally at least one storage device located away from the aforementioned processor 301. As Figure 3 As shown, the memory 305 as a computer storage medium may include an operating system, a network communication module, a user interface module, and an application program for a method of passing through an intelligent gate.

[0081] exist Figure 3In the electronic device shown, the user interface 303 is mainly used to provide an input interface for the user and obtain data input by the user; and the processor 301 can be used to call an application program storing a method for passing through a smart gate in the memory 305. When executed by one or more processors 301, the electronic device executes one or more methods in the above-mentioned embodiments.

[0082] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the order of the actions described, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required for the present application.

[0083] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0084] In the several embodiments provided in this application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0085] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0086] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0087] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory 305. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a memory 305 and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the various embodiments of the present application. The aforementioned memory 305 includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a magnetic disk or an optical disk.

[0088] The above is only an exemplary embodiment of the present disclosure and cannot be used to limit the scope of the present disclosure. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the disclosure of the specification, those skilled in the art will easily think of other embodiments of the present disclosure. This application is intended to cover any modification, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the technical field that are not recorded in the present disclosure. The description and examples are only regarded as exemplary, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A method for passing through an intelligent gate, characterized in that: Applied in a server, the method comprises: When a visitor enters the first preset detection area, obtaining the visitor's portrait data; Perform image matching between the portrait data and the reservation database to determine whether the visitor is a target reservation visitor; If the visitor is a target reservation visitor, a voice prompt is generated, and the voice prompt is used to prompt the visitor to perform identity authentication; Obtaining verification action data of the visitor; Matching the verification action data with the verification action set during the reservation; When the verification action data successfully matches the verification action, the access control device is controlled to release the person; If the visitor is not the target scheduled visitor, or the verification action fails to match, the virtual security interaction terminal is controlled to register the visitor and obtain the visit information of the visitor, wherein the virtual security interaction terminal is provided with a voice interface and a display screen interface; When the access information meets the preset requirements, the access control device is controlled to release the access; The image matching of the portrait data with the reservation database to determine whether the visitor is a target reservation visitor specifically includes: Extracting multiple frames of continuous portrait data of the visitor from the portrait data; Selecting a plurality of frames of images whose image clarity is greater than a preset clarity threshold from the plurality of frames of continuous portrait data as images to be processed; Extracting a first feature data set of the visitor from the image to be processed, wherein the first feature data set includes facial feature data, body feature data, gait feature data and clothing feature data; Using a deep learning model to perform scene analysis on the image to be processed, identifying environmental features of the visitor within a preset range, and obtaining a second feature data set, wherein the second feature data set includes light intensity data, occlusion degree data, and crowd density data; Determine whether the visitor is the target reservation visitor according to the first feature data set and the second feature data set.

2. The method according to claim 1, characterized in that The determining whether the visitor is the target reservation visitor according to the first feature data set and the second feature data set specifically includes: Determine the feature weight of each feature data in the first feature data set by using a preset feature weight calculation model; Adjusting the feature weight according to the second feature data set; Calculate the similarity between the weight-adjusted first feature data set and the feature data in the reservation database to obtain a first matching result; If the first matching result is greater than a first preset threshold, determining that the visitor is the target reservation visitor; If the first matching result is less than a second preset threshold, it is determined that the visitor is not the target appointment visitor; If the first matching result is between the first preset threshold and the second preset threshold, a secondary matching is performed, and the first preset threshold is greater than the second preset threshold.

3. The method according to claim 2, characterized in that If the first matching result is between the first preset threshold and the second preset threshold, performing a secondary matching specifically includes: Extracting dynamic time series features from the multiple frames of continuous portrait data; Based on the dynamic time series feature, each feature data in the first feature data set is enhanced to obtain target feature data, and an enhanced feature data set is constructed according to the target feature data; Determining, according to the second feature data set, a temporal weight of each of the target feature data in the enhanced feature data set; Based on the temporal weight, weighted processing is performed on the enhanced feature data set to obtain a dynamic feature matching score; Performing weighted fusion of the dynamic feature matching score and the first matching result to obtain a second matching result; When the second matching result is greater than the second preset threshold, determining that the visitor is the target reservation visitor; When the second matching result is less than the second preset threshold, it is determined that the visitor is not the target scheduled visitor.

4. The method according to claim 1, characterized in that: If the visitor is not the target scheduled visitor, or the verification action fails to match, the virtual security interaction terminal is controlled to register the visitor to obtain the visit information of the visitor, specifically including: Determine the visitor's access registration list based on a preset registration list and the portrait data; The display screen interface of the virtual security interactive terminal is controlled to display the access registration list, and the access information of the visitor is obtained through the voice interface and / or the information input box of the display screen interface to obtain a registration database.

5. The method according to claim 4, characterized in that The determining the visitor's access registration list based on the preset registration list and the portrait data specifically includes: Performing target detection analysis on the portrait data to identify whether the visitor carries any items and obtaining analysis results; If the visitor does not carry the item, the preset registration list is used as the visit registration list of the visitor; If the visitor carries the item, generating a supplementary registration list of the visitor based on the analysis result; The preset registration list and the supplementary registration list are combined to obtain the visit registration list of the visitor.

6. The method according to claim 1, characterized in that The method further comprises: When a target visitor enters a second preset detection area, obtaining movement trajectory data of the target visitor in the second preset detection area, the target visitor being any visitor who has passed through the access control device; Based on the reservation database and the registration database, the movement trajectory data is analyzed to obtain the movement trajectory analysis result of the target visitor; Determine the behavior state of the target visitor through the action trajectory analysis result; If the behavior status meets the preset abnormal behavior rules, the target visitor is marked as a suspicious visitor, and the alarm device is triggered to emit a warning sound.

7. A smart gate access system, characterized in that: include: A first acquisition module, used for acquiring the portrait data of the visitor when the visitor enters the first preset detection area; A judgment module, used for performing image matching between the portrait data and the reservation database to judge whether the visitor is a target reservation visitor; A prompt module, used for generating a voice prompt if the visitor is a target reservation visitor, wherein the voice prompt is used to prompt the visitor to perform identity authentication; A second acquisition module, used to acquire the verification action data of the visitor; A matching module, used to match the verification action data with the verification action set during the reservation; A first control module, used to control the access control device to release the user when the verification action data successfully matches the verification action; A registration module, for controlling the virtual security interaction terminal to register the visitor if the visitor is not the target appointment visitor or the verification action fails to match, and obtaining the visit information of the visitor, wherein the virtual security interaction terminal is provided with a voice interface and a display screen interface; A second control module is used to control the access control device to release the access when the access information meets the preset requirements; The judgment module is also used to extract multiple frames of continuous portrait data of the visitor from the portrait data; and select multiple frames of images whose image clarity is greater than a preset clarity threshold from the multiple frames of continuous portrait data as images to be processed; Extracting a first feature data set of visitors from the image to be processed, the first feature data set including facial feature data, body feature data, gait feature data and clothing feature data; A deep learning model is used to perform scene analysis on the processed image to identify the environmental characteristics of the visitor within a preset range to obtain a second feature data set, which includes light intensity data, occlusion degree data and crowd density data; based on the first feature data set and the second feature data set, it is determined whether the visitor is a target appointment visitor.

8. A smart gate access device, characterized in that: include: one or more processors and memory; The memory is coupled to the one or more processors, and the memory is used to store computer program codes, wherein the computer program codes include computer instructions, and the one or more processors call the computer instructions to enable the access device of the smart gate to execute the method as described in any one of claims 1-6.

9. A computer-readable storage medium comprising instructions, characterized in that: When the instruction is executed on a passage device of an intelligent gate, the passage device of the intelligent gate executes the method as described in any one of claims 1-6.

Images