An intelligent management method and system for the access control of animal houses based on image recognition

Abstract

The present invention relates to the technical field of access control management, and discloses an intelligent management method and system for animal house access control based on image recognition. The system includes: a data acquisition module, which is used to obtain historical individual images of animals and personnel through monitoring devices and collect the access control status; a data storage module, which stores the individual image sets of animals and personnel; a data processing module, which calculates the similarity by comparing real-time images with historical images and assigns individual numbers to the real-time images; an access control management module, which controls the opening and closing of the outer door, cage door, activity field hanging door and connecting door of the animal house. The access control mechanism includes: when there is no animal in the corridor image and the personnel permission at the outer door meets the conditions, the outer door is opened for a limited time; the cage door determines the opening conditions according to the status of the connecting door and the hanging door and the presence of animals; when the cage door is opened, the connecting door and the hanging door are locked and closed, and the door status is adjusted according to the number of personnel in the cage. The present invention can dynamically adjust the access control permissions and optimize the security and management efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of access control management, and more particularly, to an intelligent management method for animal house access control based on image recognition. Background Art

[0002] The animal house access control management system not only needs to ensure the safety and health of animals, but also needs to effectively monitor the entry and exit of personnel to prevent unauthorized personnel from entering, thereby protecting animals from potential disease transmission, theft or other hazards. Through intelligent access control management, precise control of access to different areas can be achieved, ensuring that only authorized personnel can enter each area. At the same time, real-time data monitoring is provided for managers, greatly improving the management efficiency of animal houses, reducing human operation errors, and ensuring the safety of personnel and animals.

[0003] However, in the existing animal house access control management technology, traditional access control systems rely on manual verification or fixed password recognition, which not only brings security risks but also increases the complexity of management. Existing access control systems usually lack the ability to intelligently identify and analyze real-time images and cannot dynamically judge whether there are potential security threats. Summary of the Invention

[0004] In view of this, the present invention proposes an intelligent management method and system for animal house access control based on image recognition, aiming to solve the problem of the lack of an intelligent access control mechanism in animal house management in the current technology.

[0005] On the one hand, an intelligent management system for animal house access control based on image recognition proposed by the present invention includes:

[0006] A data acquisition module configured to obtain historical individual images of each animal through monitoring devices in the animal house, obtain individual images of each person through pre-entered user appearance feature information, and collect the opening and closing states of each door in the intelligent access control management system.

[0007] A data storage module configured to establish and store an individual image set according to the historical individual images of each animal and person, where the individual image set includes an individual number, a permission level, an image entry time, and the corresponding historical individual image.

[0008] A data processing module configured to compare a real-time individual image with each historical individual image and calculate the similarity, select the highest similarity between the real-time individual image and the historical individual images, and assign an individual number to the real-time individual image.

[0009] The access control management module is configured to control the opening and closing of the outer door of the animal house, the cage door, the hanging door of the activity field, and the connecting door according to the access control mechanism. Among them, the outer door of the animal house communicates with the corridor for personnel activities, the cage door is used to control the connection between the cage and the corridor, the hanging door of the activity field is used to control the connection between the cage and the activity field, and the connecting door is used to control the connection between each cage.

[0010] The access control mechanism includes: obtaining the real-time image of the corridor part and the real-time image of the outer door part of the animal house through the monitoring device. When there is no animal in the real-time image of the corridor part, there is a person in the real-time image of the outer door part of the animal house, and the permission level of the current person is higher than or equal to the preset level of the outer door of the animal house, the outer door of the animal house is opened for a limited time. When there is a person at one end of the corridor part close to the outer door of the animal house in the real-time image, the outer door of the animal house is opened for a limited time.

[0011] When the cage door receives an opening instruction, obtain the opening and closing states of the connecting door and the hanging door of the activity field and whether there is an animal in the cage. If it is judged that both the connecting door and the hanging door of the activity field are in the closed state, there is no animal in the cage, and the permission level of the current person is greater than or equal to the preset level of the cage door, the cage door is opened.

[0012] When the cage door is in the open state, the connecting door and the hanging door of the activity field are locked in the closed state and the number of people in the cage is obtained and the individual number of each person is identified. When the number of people in the cage is zero, the cage door is adjusted to the state of being opened for a limited time.

[0013] When the connecting door is in the open state, the hanging door of the activity field is locked in the closed state.

[0014] Furthermore, the data storage module is also configured to.

[0015] Set a storage quantity threshold for all historical individual images in the data storage module. When each time interval period is satisfied, judge the relationship between the current storage quantity and the storage quantity threshold.

[0016] When the current storage quantity is less than or equal to the storage quantity threshold, obtain new individual images through the data acquisition module and supplement them to the data storage module.

[0017] When the current storage quantity is greater than the storage quantity threshold, trigger the deletion mechanism. After completing the deletion mechanism, supplement new individual images to the data storage module.

[0018] Furthermore, the deletion mechanism includes: sorting the historical individual images from large to small according to the image entry time, and sequentially selecting the historical individual images from large to small, which are recorded as the images to be confirmed.

[0019] Set a preset input time threshold, obtain all the highest similarities of each historical individual image in the individual image set corresponding to the image to be confirmed, and calculate a similarity threshold. Determine whether to perform a deletion operation based on the relationship between the latest highest similarity of the image to be confirmed and the similarity threshold, and the relationship between the image input time and the input time threshold. When the latest highest similarity is less than the similarity threshold, delete the image to be confirmed. When the latest highest similarity is greater than the similarity threshold and the image input time is less than or equal to the input time threshold, retain the image to be confirmed. When the latest highest similarity is greater than the similarity threshold and the image input time is greater than the input time threshold, obtain the deletion of the image to be confirmed.

[0020] Further, when the time interval period is satisfied and individual images are supplemented into the data storage module, the supplementary quantity is the difference between the storage quantity threshold and the current storage quantity.

[0021] Further, when the deletion mechanism is triggered, start calculating the similarity threshold, and the similarity threshold is calculated and obtained through the following relationship:

[0022]

[0023] Where S is the similarity threshold; N is the number of all the highest similarities of the current historical individual images; μ and σ are respectively the average value and the standard deviation of all the highest similarities of the current historical individual images; k is a threshold strictness coefficient, which affects the value of the similarity threshold, and the value range of k is (0.5, 2]; tanh(x) is the hyperbolic tangent function.

[0024] Further, when the data processing module calculates the similarity, it includes: using a trained convolutional neural network to determine whether the moving target in the real-time individual image is a person or an animal. When the judgment result is a person, select all the individual image sets corresponding to the person. When the judgment result is an animal, select all the individual image sets corresponding to the animal. Compare the real-time individual image with each historical individual image, and calculate the similarity based on the pose estimation algorithm. The similarity satisfies the following relationship:

[0025]

[0026] Where s(I real , I history ) is the similarity, i is, is the angle of the i-th joint in the real-time individual image, is the angle of the i-th joint in the historical individual image, L real is the average step length of the individual in the real-time individual image, L history is the average step length of the individual in the historical individual image, Treal Let \(T\) be the average gait cycle of an individual in the real-time individual image. history Let \(T'\) be the average gait cycle of an individual in the historical individual image.

[0027] Furthermore, the access control mechanism for the time-limited opening of the outer door of the animal house specifically includes: when it is determined that there is a moving target in the area of the real-time image of the outer door of the animal house, taking the real-time image of the outer door of the animal house as the real-time individual image and identifying the individual number according to the similarity, comparing the permission level of the current person with the preset level of the outer door of the animal house to determine whether to open the outer door of the animal house; when it is satisfied that there is a person in the real-time image of the corridor, determining whether there is a person within two meters of the outer door of the animal house. If the determination result is yes, the outer door of the animal house is opened for a limited time.

[0028] When there is no moving target in the area of the real-time image of the outer door of the animal house, the outer door of the animal house is closed.

[0029] After the outer door of the animal house is opened, the number of people entering is counted and the individual number of each person is recorded according to the real-time image of the corridor.

[0030] Furthermore, the access control management module is also configured to adjust the permission levels of each door according to the current time. If the current time is during non-working hours, the preset level of each door is increased by one level. When the working hours start, the preset level of each door is adjusted back by one level. If the current time is during working hours, the preset level remains unchanged.

[0031] Furthermore, it also includes: a warning module, configured to obtain the real-time image of the corridor and issue an alarm when an animal is identified in the real-time image of the corridor.

[0032] An access control display module is set outside each cage door. The access control display module is configured to obtain the opening and closing states of the outer door of the animal house, the cage door, and the hanging door of the activity field and perform visual display.

[0033] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0034] Through image recognition and comparison with real-time images, the system can accurately judge the identities of personnel and animals, and combine the permission levels to control the opening and closing of different doors, ensuring the safety management inside the animal house. This intelligent control greatly improves the management efficiency and safety, and avoids the omissions that may be caused by traditional manual control.

[0035] The system compares and stores historical individual images, and uses a similarity calculation mechanism to update and manage the individual image set in a timely manner, ensuring the accuracy and real-time performance of the system. The deletion mechanism avoids the accumulation of expired and useless data, ensuring the rationality and effectiveness of data storage.

[0036] According to the real-time time and work requirements, the system can automatically adjust the permission levels of each door to ensure that the management strategies in different time periods meet the actual needs. For example, the permission level is automatically increased during non-working hours to enhance security, and the normal management strategy is restored during working hours to flexibly adapt to different scenarios.

[0037] By monitoring the images in the corridor area in real time, the system can detect abnormalities in a timely manner, especially the presence of animals, and issue an alarm through the warning module to help the staff take timely countermeasures to prevent accidents.

[0038] By using a convolutional neural network and a pose estimation algorithm to compare the real-time individual images with the historical individual images, the system can accurately identify the identity of personnel or animals, calculate the similarity based on features such as angle differences and gait, and provide more accurate identity recognition and behavior analysis.

[0039] When storing images, the system maintains the freshness and relevance of the data through a periodic replenishment and deletion mechanism, avoiding the impact of overfilled or outdated data on the system performance and ensuring the long-term validity of the image data.

[0040] On the other hand, an intelligent management method for the animal house access control based on image recognition proposed by the present invention includes: obtaining the historical individual images of each animal through the monitoring devices in the animal house, obtaining the individual images of each person through the pre-entered user appearance feature information, and collecting the switch states of each door in the access control intelligent management system.

[0041] An individual image set is established and stored according to the historical individual images of each animal and person, and the individual image set includes individual numbers, permission levels, image entry times, and corresponding historical individual images.

[0042] Compare the real-time individual image with each historical individual image and calculate the similarity, select the highest similarity between the real-time individual image and the historical individual images, and assign an individual number to the real-time individual image.

[0043] Control the opening and closing of the outer door of the animal house, the cage door, the hanging door of the activity field, and the connecting door according to the access control mechanism; wherein, the outer door of the animal house is connected to the corridor for personnel activities, the cage door is used to control the connection between the cage and the corridor, the hanging door of the activity field is used to control the connection between the cage and the activity field, and the connecting door is used to control the connection between each cage.

[0044] Among them, the access control mechanism includes: obtaining the real-time image of the corridor part and the real-time image of the outer door part of the animal house through a monitoring device. When there is no animal in the real-time image of the corridor part, there is a person in the real-time image of the outer door part of the animal house, and the permission level of the current person is higher than or equal to the preset level of the outer door of the animal house, the outer door of the animal house is opened within a limited time; when there is a person at one end of the corridor part close to the outer door of the animal house in the real-time image, the outer door of the animal house is opened within a limited time.

[0045] When the cage door receives an opening instruction, obtain the switch states of the access door and the hanging door of the activity field, and whether there is an animal in the cage. If it is determined that both the access door and the hanging door of the activity field are in the closed state, there is no animal in the cage, and the permission level of the current person is greater than or equal to the preset level of the cage door, the cage door is opened.

[0046] When the cage door is in the open state, the access door and the hanging door of the activity field are locked in the closed state, and the number of people in the cage is obtained and the individual number of each person is identified; when the number of people in the cage is zero, the cage door is adjusted to the state of being opened within a limited time.

[0047] When the access door is in the open state, the hanging door of the activity field is locked in the closed state.

[0048] It can be understood that the above-mentioned method and system for intelligent management of animal house access control based on image recognition have the same beneficial effects and will not be elaborated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] By reading the detailed description of the preferred embodiments below, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present invention. Moreover, throughout the drawings, the same reference numerals are used to represent the same components.

[0050] Figure 1 It is a schematic plan view of an animal house provided by an embodiment of the present invention.

[0051] Figure 2 It is a functional framework diagram of an intelligent management system for animal house access control based on image recognition provided by an embodiment of the present invention.

[0052] Figure 3 It is a flowchart of a method for intelligent management of animal house access control based on image recognition provided by an embodiment of the present invention.

[0053] Among them, 1. Outer door of the animal house; 2. Cage door; 3. Hanging door of the activity field; 4. Access door; 5. Corridor; 6. Cage; 7. Activity field. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0054] Exemplary embodiments disclosed in the present application will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments disclosed in the present application are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be fully conveyed to those skilled in the art. It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the drawings and in combination with the embodiments.

[0055] Referring to Figure 1 As shown, an embodiment of the present invention provides a schematic plan view of an animal house. Specifically, the animal house includes a corridor 5 and a plurality of cages 6. The outer door 1 of the animal house is connected to the corridor 5. The plurality of cages 6 are adjacent to the corridor 5 and the activity field 7 respectively, and the cages 6 are located between the corridor 5 and the activity field 7. Each cage 6 is connected to the corridor 5 through a cage door 2 and is connected to the activity field 7 through a movable hanging door 3. Each adjacent cage 6 is provided with a connecting door 4.

[0056] Referring to Figure 2 As shown, an embodiment of the present invention provides an intelligent management system for the access control of an animal house based on image recognition, including: a data acquisition module configured to obtain the historical individual images of each animal through the monitoring devices in the animal house, obtain the individual images of each person through the pre-entered user appearance feature information, and collect the opening and closing states of each door in the access control intelligent management system.

[0057] A data storage module configured to establish and store an individual image set according to the historical individual images of each animal and person. The individual image set includes an individual number, a permission level, an image entry time, and the corresponding historical individual image.

[0058] A data processing module configured to compare the real-time individual image with each historical individual image and calculate the similarity, select the highest similarity between the real-time individual image and the historical individual image, and assign an individual number to the real-time individual image.

[0059] An access control management module configured to control the opening and closing of the outer door of the animal house, the cage door, the hanging door of the activity field, and the connecting door according to the access control mechanism; wherein, the outer door of the animal house is connected to the corridor for personnel activities, the cage door is used to control the connection between the cage and the corridor, the hanging door of the activity field is used to control the connection between the cage and the activity field, and the connecting door is used to control the connection between each cage.

[0060] The access control mechanism includes: obtaining the real-time image of the corridor area and the real-time image of the outer door of the animal house through monitoring devices. When there is no animal in the real-time image of the corridor area, there is a person in the real-time image of the outer door of the animal house, and the permission level of the current person is higher than or equal to the preset level of the outer door of the animal house, the outer door of the animal house is opened for a limited time; when there is a person at one end of the corridor area close to the outer door of the animal house in the real-time image, the outer door of the animal house is opened for a limited time.

[0061] When the cage door receives an opening instruction, obtain the switch states of the connecting door and the suspended door of the activity field, and whether there is an animal in the cage. If it is determined that both the connecting door and the suspended door of the activity field are in the closed state, there is no animal in the cage, and the permission level of the current person is greater than or equal to the preset level of the cage door, the cage door is opened.

[0062] When the cage door is in the open state, the connecting door and the suspended door of the activity field are locked in the closed state, and the number of people in the cage is obtained and the individual number of each person is identified; when the number of people in the cage is zero, the cage door is adjusted to the state of being opened for a limited time.

[0063] When the connecting door is in the open state, the suspended door of the activity field is locked in the closed state.

[0064] Specifically, each door in the access control intelligent management system includes: the outer door of the animal house, the cage door, the suspended door of the activity field, and the connecting door. In this embodiment, the person refers to the animal house staff; the user appearance feature information includes the appearance images or videos of the staff during employment and subsequent periods; locking means that the switch state is fixed to the open or closed state and cannot be adjusted; the image entry time is the duration when the historical individual image is moved into the individual image set. The method for obtaining the highest similarity is to compare the real-time individual image with each historical individual image and calculate the similarity, compare the magnitudes of the calculated similarities, and select the largest similarity as the highest similarity. When there are two or more highest similarities with the same value, select the historical individual image with the latest image entry time, and then assign the individual number corresponding to the historical individual image to the real-time individual image; when there is only one highest similarity, directly assign the individual number corresponding to the historical individual image to the real-time individual image.

[0065] It can be understood that the system performs real-time identity verification on personnel and animals through image recognition technology, combines permission management to control the access control in the animal house, and ensures that only verified personnel with appropriate permissions can enter the designated area. Compared with the traditional access control system, this image recognition-based management method improves the intelligence and security of management, and avoids the omissions and potential risks of manual management.

[0066] By establishing an individual image set and comparing it with real-time images, the system can accurately identify the identities of people and animals. The assignment of individual numbers is based on the highest similarity of historical images, ensuring the accurate matching of real-time images with historical data, and improving the accuracy and processing speed of identification.

[0067] The system can monitor the access control status in real time and dynamically control the opening and closing of doors according to different situations. For example, the outer door of the animal house is opened for a limited time according to the real-time images of the corridor part and the outer door part, and the cage door is opened or adjusted to be opened for a limited time under specific conditions, improving the flexibility and security of management.

[0068] Through the precise monitoring and control of the access control status, the system can ensure that only personnel meeting specific permission requirements can enter specific areas at different times and under different conditions. For example, when the cage door is opened, the system will ensure that the connecting door and the hanging door of the activity field are locked and closed, thus avoiding unnecessary area crossings and reducing potential safety risks.

[0069] The system uses the comparison of real-time images and historical images to calculate the similarity, and adopts the method of selecting the highest similarity to match the identities of people, ensuring the accuracy of identification. When there are multiple same similarities, the system further improves the effectiveness of identification by selecting the individual image with the latest image entry time, ensuring the timeliness and accuracy of data.

[0070] The system automatically processes multiple links such as data collection, storage, identification, and access control, reducing manual intervention and operation errors. Especially the dynamic management and comparison mechanism of images ensure that the system can continuously optimize identification and management over time and adapt to changing usage scenarios.

[0071] In some embodiments of the present application, the data storage module is further configured to: set a storage quantity threshold for all historical individual images in the data storage module, and when each time interval period is satisfied, judge the relationship between the current storage quantity and the storage quantity threshold.

[0072] If the current storage quantity is less than or equal to the storage quantity threshold, new individual images are obtained through the data collection module and supplemented into the data storage module.

[0073] If the current storage quantity is greater than the storage quantity threshold, a deletion mechanism is triggered. After the deletion mechanism is completed, new individual images are supplemented into the data storage module.

[0074] In some embodiments of the present application, the deletion mechanism includes: sorting the historical individual images from large to small according to the image entry time, and sequentially selecting the historical individual images from large to small, which are recorded as the images to be confirmed.

[0075] Set a preset input time threshold, obtain all the highest similarities of each historical individual image in the individual image set corresponding to the image to be confirmed, and calculate the similarity threshold. Determine whether to perform a deletion operation based on the relationship between the latest highest similarity of the image to be confirmed and the similarity threshold, as well as the image input time and the input time threshold; when the latest highest similarity is less than the similarity threshold, delete the image to be confirmed; when the latest highest similarity is greater than the similarity threshold and the image input time is less than or equal to the input time threshold, retain the image to be confirmed; when the latest highest similarity is greater than the similarity threshold and the image input time is greater than the input time threshold, obtain the deletion of the image to be confirmed.

[0076] It can be understood that by setting a storage quantity threshold, the system can dynamically manage data storage according to the current storage situation, ensuring that the system will not affect performance or cause storage overflow due to over-full storage. This automated storage management method improves the stability and efficiency of the system.

[0077] When the storage quantity exceeds the threshold, the system will trigger a deletion mechanism, and determine which images should be deleted by analyzing the similarity and time information of the images. This not only avoids unnecessary storage occupancy but also ensures that the stored data is always the latest and most useful.

[0078] By sorting according to the image input time and judging the similarity, the system can automatically clean up historical images with high repetition and high ineffectiveness, and retain the most representative and valuable data. This intelligent deletion mechanism can ensure that data storage is both efficient and accurate.

[0079] By combining the similarity threshold and the input time threshold to determine whether to delete an image, it avoids the deficiencies of simply relying on time or simply relying on similarity, and improves the accuracy of the deletion decision. If the similarity of the latest image is high and the input time is relatively late, it is preferentially retained, further enhancing the intelligence and flexibility of the system in image management.

[0080] By regularly cleaning up redundant data in historical images (such as images with too high similarity), the system can keep the data set concise and avoid redundant images occupying storage space. This not only improves the data access efficiency but also prevents the storage space from being filled with unnecessary data.

[0081] The optimization of storage management can reduce the dependence on outdated or duplicate images, enabling faster search and access to valid data during data processing, thereby improving the overall performance of the system.

[0082] For those images with high similarity and relatively new input time, the system will choose to retain them, ensuring that only the most important and relevant data is retained in the storage. This mechanism effectively improves the utilization rate of data storage while reducing data redundancy and the interference of outdated data.

[0083] In some embodiments of the present application, when the time interval period is satisfied and individual images are supplemented into the data storage module, the supplementary quantity is the difference between the storage quantity threshold and the current storage quantity.

[0084] In some embodiments of the present application, when the deletion mechanism is triggered, the similarity threshold is calculated. The similarity threshold is obtained through the following relationship:

[0085]

[0086] Where S is the similarity threshold; N is the quantity of all the highest similarities of the current historical individual images; μ and σ are respectively the average value and the standard deviation of all the highest similarities of the current historical individual images; k is the threshold strictness coefficient, which affects the value of the similarity threshold, and the value range of k is (0.5, 2]; tanh(x) is the hyperbolic tangent function.

[0087] It should be noted that the calculation method of the similarity threshold takes into account the distribution of the highest similarities of the current historical individual images, including its quantity, average value, and standard deviation. By comprehensively considering these factors, the threshold can be dynamically adjusted according to the diversity of data in the actual image library, rather than using a fixed standard. This method can flexibly cope with the changes in image data and avoid misdeleting valid data due to too high or too low similarity.

[0088] The introduction of the threshold strictness coefficient can adjust the strictness of the similarity according to actual needs. For example, when the coefficient is relatively high, a more strict standard is set, and only when the similarity is high enough will the image be retained, which helps the system to accurately screen out the most representative data; while when the coefficient is low, the threshold is relaxed, allowing more images to be stored. In this way, the system has flexible robustness and can adapt to different application scenarios and requirements.

[0089] By using the hyperbolic tangent function, the system can smoothly calculate the similarity threshold, avoid directly causing a sharp fluctuation of the threshold when the change in similarity is large, and ensure that the deletion mechanism is more accurate. For example, when the similarity of an image approaches the threshold, using the hyperbolic tangent function can make the change in the similarity threshold more gentle, thus avoiding misdeletion.

[0090] By comprehensively considering statistical features such as the average value and standard deviation, as well as the threshold strictness coefficient, the system can more accurately evaluate the importance of each image and ensure that important image data will not be misdeleted. This mechanism not only improves the accuracy of image deletion but also makes the data management of the system more intelligent and efficient.

[0091] In some embodiments of the present application, when calculating the similarity, the data processing module includes: using a trained convolutional neural network to determine whether the moving target in the real-time individual image is a person or an animal. When the determination result is a person, all individual image sets corresponding to the person are selected. When the determination result is an animal, all individual image sets corresponding to the animal are selected; comparing the real-time individual image with each historical individual image, and calculating the similarity based on the pose estimation algorithm. The similarity satisfies the following relationship:

[0092]

[0093] where s(I real , I history ) is the similarity, i is, is the angle of the i-th joint in the real-time individual image, is the angle of the i-th joint in the historical individual image, L real is the average step length of the individual in the real-time individual image, L history is the average step length of the individual in the historical individual image, T real is the average gait cycle of the individual in the real-time individual image, T history is the average gait cycle of the individual in the historical individual image.

[0094] It can be understood that for moving target recognition: a convolutional neural network (CNN) is used to determine whether the target in the real-time individual image is a person or an animal. The convolutional neural network is a deep learning method, which is good at image classification and object recognition. Through training on images, the CNN can identify the target in the real-time image and select the corresponding individual image set according to the classification result.

[0095] If the determination result is a person, the system will select the corresponding individual image set of the person; if the determination result is an animal, the system will select the corresponding individual image set of the animal.

[0096] For pose estimation: the pose estimation algorithm is used to extract key point information from the real-time individual image and the historical individual image, such as features like the angles of each joint, step length, and gait cycle. The core idea of pose estimation is to estimate the angles of each joint (such as the bending angles of the elbow and knee) and other motion features by analyzing the skeletal structure of the human body (or animal) in the image.

[0097] By calculating the differences in multi-dimensional features such as joint angles, step length, and gait cycle, the similarity between the real-time image and the historical image can be more accurately reflected. This formula is not only applicable to human pose recognition but also can process animal images, with strong generality. Through high-precision similarity calculation, it ensures that the identity recognition of a person or an animal can be accurately linked with the access control management system, guaranteeing the security and intelligence of the animal house access control system.

[0098] In some embodiments of the present application, the access control mechanism for the limited-time opening of the outer door of the animal house specifically includes: when it is determined that there is a moving target in the area of the real-time image of the outer door of the animal house, the real-time image of the outer door of the animal house is used as the real-time individual image, and the individual number is identified according to the similarity. The access level of the current person is compared with the preset outer door level of the animal house to determine whether to open the outer door of the animal house; when it is satisfied that there is a person in the real-time image of the corridor part, it is judged whether there is a person within two meters of the outer door of the animal house. If the judgment result is yes, the outer door of the animal house is opened for a limited time.

[0099] When there is no moving target in the area of the real-time image of the outer door of the animal house, the outer door of the animal house is closed.

[0100] After the outer door of the animal house is opened, the number of people entering is counted, and the individual number of each person is recorded according to the real-time image of the corridor part.

[0101] It can be understood that through the analysis of the real-time image and similarity recognition, the system can intelligently judge whether the moving target is a person or an animal, ensuring that only legitimate personnel can open the outer door of the animal house under the condition of meeting the permission conditions, thus effectively preventing illegal personnel from entering.

[0102] Through real-time image acquisition and similarity comparison, the system can quickly identify the identity of personnel, ensuring a rapid response to access control. In practical applications, it can greatly improve work efficiency and reduce delays caused by manual inspections.

[0103] The system compares the access level of personnel with the outer door level of the animal house to ensure that only personnel with sufficient permissions can open the outer door of the animal house. This permission-based access control effectively prevents unauthorized personnel from entering the animal house and enhances security.

[0104] By judging the position of personnel in the real-time image of the corridor part, the system can not only control the opening and closing of the door, but also effectively track the number and identity of personnel entering the animal house, ensuring accurate and reliable records of personnel entry and exit. This provides strong data support for personnel management and security supervision.

[0105] When there is no moving target in the area of the real-time image of the outer door of the animal house, the system will automatically close the outer door of the animal house to avoid unnecessary energy waste and equipment load. This automatic control not only improves the efficiency of the system but also saves energy.

[0106] By setting the "limited-time opening" mechanism, the system can provide temporary openings within a specific time period to ensure the normal entry and exit of personnel or equipment, while restricting the opening of the access control during non-essential time periods, enhancing the flexibility of access control management.

[0107] In some embodiments of the present application, the access control management module is further configured to adjust the permission levels of each door according to the current time. When the current time is during non-working hours, the preset level of each door is increased by one level. When the working hours start, the preset level of each door is adjusted back by one level. When the current time is during working hours, the preset level remains unchanged.

[0108] It should be noted that the system will determine which time period the current time is in according to the pre-set working hours arrangement. In this embodiment, the working hours are from 9:00 am to 6:00 pm, and the rest of the time is non-working hours.

[0109] The working hours are defined as the normal working hours of daily life, usually the time period when employees work in the animal house. The non-working hours include periods such as night time and holidays that are not within the normal working hours. The system will perform corresponding permission management according to this time period.

[0110] When the system recognizes that the current time is during non-working hours, to improve the security of the animal house, the system will automatically increase the preset permission level of each door by one level. In this way, only those with higher permissions can pass through the door, preventing unauthorized personnel from entering. It prevents non-staff or unauthorized people from entering sensitive areas during non-working hours.

[0111] When the working hours start, the system will automatically adjust the permission level of each door back by one level according to the time, restoring it to the standard permission level during daily work. In this way, staff and authorized personnel can smoothly enter the relevant areas for daily work. It restores the normal working state, enabling staff to quickly enter and carry out relevant work, while avoiding unnecessary permission restrictions.

[0112] During working hours, the permission level of the door will remain unchanged, ensuring that personnel enter each area according to the normal permission management. The system automatically detects the current time through scheduled tasks or real-time monitoring and adjusts the permission level of the door according to the predetermined rules. The permission adjustment is automatically completed by the system without manual intervention. The system will judge according to the time and timely adjust the permission settings of the door. The change of time immediately triggers the increase or callback of the permission level, ensuring the timeliness and accuracy of access control management. The specific time period of working hours can be customized by the system administrator and adjusted according to different work requirements or scenario changes. If certain special circumstances cause changes in working hours, such as holidays or temporary overtime, the administrator can temporarily adjust the time settings of the system, thereby automatically adjusting the permission level. The preset permission level of each door can be different according to its function and location. By increasing the permission during non-working hours, the risk of unauthorized personnel entering is reduced, thus effectively protecting the animals and staff in the animal house. During working hours, the system automatically restores the normal permission level, and staff do not need to wait or manually adjust the permission, improving work efficiency.

[0113] In some embodiments of the present application, it further includes: a warning module configured to obtain real-time images of the corridor area and issue an alarm when an animal is recognized in the real-time images of the corridor area.

[0114] An access control display module is provided outside each cage door. The access control display module is configured to obtain the opening and closing states of the outer door of the animal house, the cage door, and the hanging door of the activity field and visually display them.

[0115] It should be noted that the warning module is designed to monitor the animal behavior in the corridor area in real time. Once an abnormal situation is detected, an alarm will be immediately issued to ensure that the staff can take corresponding measures in time to prevent animals from getting lost or entering restricted areas. To enhance the reliability and flexibility of the system, the warning condition judgment includes:

[0116] Condition 1: An animal is recognized in the real-time image, and the animal enters a preset sensitive area or restricted area, such as the position near the access control in the corridor area, such as the outer door of the animal house or the cage door.

[0117] Condition 2: The behavior of the animal does not match the data during normal historical activities, and uncommon abnormal actions occur.

[0118] Once the above conditions are met, the warning module will trigger an alarm to notify the staff to intervene.

[0119] Refer to Figure 3 As shown, an embodiment of the present invention provides an intelligent management method for the access control of an animal house based on image recognition, which includes: S1, obtaining historical individual images of each animal through monitoring devices in the animal house, obtaining individual images of each person through pre-entered user appearance feature information, and collecting the opening and closing states of each door in the access control intelligent management system.

[0120] S2, establishing and storing an individual image set according to the historical individual images of each animal and person. The individual image set includes individual numbers, permission levels, image entry times, and corresponding historical individual images.

[0121] S3, comparing the real-time individual image with each historical individual image and calculating the similarity, selecting the highest similarity between the real-time individual image and the historical individual images, and assigning an individual number to the real-time individual image.

[0122] S4, controlling the opening and closing of the outer door of the animal house, the cage door, the hanging door of the activity field, and the connecting door according to the access control mechanism; wherein, the outer door of the animal house is connected to the corridor for personnel activities, the cage door is used to control the connection between the cage and the corridor, the hanging door of the activity field is used to control the connection between the cage and the activity field, and the connecting door is used to control the connection between each cage.

[0123] Among them, the access control mechanism includes: obtaining the real-time images of the corridor part and the real-time images of the outer door part of the animal house through monitoring devices. When there is no animal in the real-time image of the corridor part, there is a person in the real-time image of the outer door part of the animal house, and the permission level of the current person is higher than or equal to the preset level of the outer door of the animal house, the outer door of the animal house is opened for a limited time; when there is a person at one end of the corridor part close to the outer door of the animal house in the real-time image, the outer door of the animal house is opened for a limited time.

[0124] When the cage door receives an opening instruction, obtain the switch states of the access door and the suspended door of the activity field and whether there is an animal in the cage. If it is determined that both the access door and the suspended door of the activity field are in the closed state, there is no animal in the cage, and the permission level of the current person is greater than or equal to the preset level of the cage door, the cage door is opened.

[0125] When the cage door is in the open state, the access door and the suspended door of the activity field are locked in the closed state, and the number of people in the cage is obtained and the individual number of each person is identified; when the number of people in the cage is zero, the cage door is adjusted to the state of being opened for a limited time.

[0126] When the access door is in the open state, the suspended door of the activity field is locked in the closed state.

[0127] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications or equivalent replacements can still be made to the specific implementation manners of the present invention, and any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the protection scope of the claims of the present invention.

Claims

1. An intelligent management system for the access control of animal houses based on image recognition, characterized in that, Including: A data acquisition module, configured to obtain historical individual images of each animal through monitoring devices in the animal house, obtain individual images of each person through pre-entered user appearance feature information, and collect the on / off states of each door in the access control intelligent management system; A data storage module, configured to establish and store an individual image set according to the historical individual images of each animal and person, where the individual image set includes an individual number, a permission level, an image entry time, and the corresponding historical individual image; A data processing module, configured to compare a real-time individual image with each historical individual image and calculate the similarity, select the highest similarity between the real-time individual image and the historical individual image, and assign an individual number to the real-time individual image; An access control management module, configured to control the opening and closing of the outer door of the animal house, the cage door, the hanging door of the activity field, and the connecting door according to the access control mechanism; where the outer door of the animal house is connected to the corridor for personnel activities, the cage door is used to control the connection between the cage and the corridor, the hanging door of the activity field is used to control the connection between the cage and the activity field, and the connecting door is used to control the connection between each cage; The access control mechanism includes: Obtaining the real-time image of the corridor part and the real-time image of the outer door part of the animal house through the monitoring device. When there is no animal in the real-time image of the corridor part, there is a person in the real-time image of the outer door part of the animal house, and the permission level of the current person is higher than or equal to the preset level of the outer door of the animal house, the outer door of the animal house is opened for a limited time; when there is a person at one end of the corridor part close to the outer door of the animal house in the real-time image of the corridor part, the outer door of the animal house is opened for a limited time; When the cage door receives an opening instruction, obtain the on / off states of the connecting door and the hanging door of the activity field and whether there is an animal in the cage. If it is judged that both the connecting door and the hanging door of the activity field are in the closed state, there is no animal in the cage, and the permission level of the current person is greater than or equal to the preset level of the cage door, the cage door is opened; When the cage door is in the open state, the connecting door and the hanging door of the activity field are locked in the closed state and the number of people in the cage is obtained and the individual number of each person is identified; when the number of people in the cage is zero, the cage door is adjusted to the state of being opened for a limited time; When the connecting door is in the open state, the hanging door of the activity field is locked in the closed state.

2. The intelligent management system for the animal house access control based on image recognition according to claim 1, wherein The data storage module is further configured to: Set a storage quantity threshold for all historical individual images in the data storage module, and when each time interval period is satisfied, judge the relationship between the current storage quantity and the storage quantity threshold; When the current storage quantity is less than or equal to the storage quantity threshold, obtain new individual images through the data acquisition module and supplement them to the data storage module; When the current storage quantity is greater than the storage quantity threshold, trigger a deletion mechanism. After the deletion mechanism is completed, supplement new individual images to the data storage module.

3. The intelligent management system for the animal house access control based on image recognition according to claim 2, characterized in that, The deletion mechanism includes: sorting the historical individual images from large to small according to the image entry time, and sequentially selecting the historical individual images from large to small, which are recorded as images to be confirmed; Set an input time threshold in advance, obtain all the highest similarities of each historical individual image in the individual image set corresponding to the image to be confirmed, and calculate a similarity threshold. Determine whether to perform a deletion operation based on the relationship between the latest highest similarity of the image to be confirmed and the similarity threshold, and the relationship between the image input time and the input time threshold. When the latest highest similarity is less than the similarity threshold, delete the image to be confirmed. When the latest highest similarity is greater than the similarity threshold and the image input time is less than or equal to the input time threshold, retain the image to be confirmed. When the latest highest similarity is greater than the similarity threshold and the image input time is greater than the input time threshold, obtain and delete the image to be confirmed.

4. The intelligent management system for the animal house access control based on image recognition according to claim 3, wherein, When the time interval period is satisfied and individual images are supplemented into the data storage module, the supplementary quantity is the difference between the storage quantity threshold and the current storage quantity.

5. The intelligent management system for the animal house access control based on image recognition according to claim 3, characterized in that, When the deletion mechanism is triggered, start calculating the similarity threshold, and the similarity threshold is calculated and obtained through the following relationship: Where S is the similarity threshold; N is the quantity of all the highest similarities of the current historical individual images; μ and σ are respectively the average value and standard deviation of all the highest similarities of the current historical individual images; k is the threshold strictness coefficient, which affects the value of the similarity threshold, and the value range of k is (0.5, 2]; tanh(x) is the hyperbolic tangent function.

6. The intelligent management system for the animal house access control based on image recognition according to claim 1, wherein When calculating the similarity, the data processing module includes: Use a trained convolutional neural network to determine whether the moving target in the real-time individual image is a person or an animal. When the determination result is a person, select all the individual image sets corresponding to the person. When the determination result is an animal, select all the individual image sets corresponding to the animal. Compare the real-time individual image with each historical individual image, and calculate the similarity based on the pose estimation algorithm. The similarity satisfies the following relationship: where s(I real , I history ) is the similarity, is the angle of the i-th joint in the real-time individual image, is the angle of the i-th joint in the historical individual image, L real is the average step length of the individual in the real-time individual image, L history is the average step length of the individual in the historical individual image, T real is the average gait cycle of the individual in the real-time individual image, T history is the average gait cycle of the individual in the historical individual image.

7. The intelligent management system for the animal house access control based on image recognition according to claim 6, wherein The access control mechanism for the limited-time opening of the outer door of the animal house specifically includes: When it is determined that there is a moving target in the area of the real-time image of the outer door of the animal house, use the real-time image of the outer door of the animal house as the real-time individual image and identify the individual number according to the similarity. Compare the current personnel's permission level with the preset outer door level of the animal house to determine whether to open the outer door of the animal house. When it is satisfied that there is a person in the real-time image of the corridor part, determine whether there is a person within two meters of the outer door of the animal house. If the determination result is yes, the outer door of the animal house is opened for a limited time; When there is no moving target in the area of the real-time image of the outer door of the animal house, close the outer door of the animal house; After the outer door of the animal house is opened, count the number of people entering and record the individual number of each person according to the real-time image of the corridor part.

8. The intelligent management system for the animal house access control based on image recognition according to claim 7, wherein, The access control management module is further configured to adjust the permission levels of each door according to the current time. If the current time is during non-working hours, raise the preset level of each door by one level. When the working time starts, call back the preset level of each door by one level. If the current time is during working hours, the preset level remains unchanged.

9. The intelligent management system for the animal house access control based on image recognition according to claim 8, wherein It also includes: An early warning module, configured to obtain real-time images of the corridor area and issue an alarm when an animal is recognized in the real-time images of the corridor area; An access control display module, arranged outside each cage door, configured to obtain the opening and closing states of the outer door of the animal house, the cage door, and the hanging door of the activity field and visually display them.

10. A method for intelligent management of the access control of animal houses based on image recognition, which is applied to the intelligent management system for the access control of animal houses based on image recognition according to any one of claims 1-9, characterized in that, Including: Obtaining historical individual images of each animal through monitoring devices inside the animal house, obtaining individual images of each person through pre-entered user appearance feature information, and collecting the opening and closing states of each door in the access control intelligent management system; Establishing and storing an individual image set according to the historical individual images of each animal and person, where the individual image set includes individual numbers, permission levels, image entry times, and corresponding historical individual images; Comparing the real-time individual image with each historical individual image and calculating the similarity, selecting the highest similarity between the real-time individual image and the historical individual images, and assigning an individual number to the real-time individual image; Controlling the opening and closing of the outer door of the animal house, the cage door, the hanging door of the activity field, and the connecting doors according to the access control mechanism; where the outer door of the animal house is connected to the corridor for personnel activities, the cage door is used to control the connection between the cage and the corridor, the hanging door of the activity field is used to control the connection between the cage and the activity field, and the connecting doors are used to control the connection between each cage; Wherein, the access control mechanism includes: Obtaining real-time images of the corridor area and the area outside the outer door of the animal house through monitoring devices. When the real-time image of the corridor area does not have an animal, the real-time image of the area outside the outer door of the animal house has a person, and the permission level of the current person is higher than or equal to the preset level of the outer door of the animal house, the outer door of the animal house is opened for a limited time; when there is a person at one end of the corridor area close to the outer door of the animal house in the real-time image, the outer door of the animal house is opened for a limited time; When the cage door receives an opening instruction, obtaining the opening and closing states of the connecting doors and the hanging door of the activity field and whether there is an animal in the cage. If it is determined that both the connecting doors and the hanging door of the activity field are in the closed state, there is no animal in the cage, and the permission level of the current person is greater than or equal to the preset level of the cage door, the cage door is opened; When the cage door is in the open state, the connecting doors and the hanging door of the activity field are locked in the closed state and the number of people in the cage is obtained and the individual number of each person is identified; when the number of people in the cage is zero, the cage door is adjusted to the state of being opened for a limited time; When the connecting door is in the open state, the hanging door of the activity field is locked in the closed state.

Images