Regional intelligent clustering information pushing method and system
Through the camera identifying pedestrians and performing intelligent regional clustering, the problem that existing information recommendation systems are difficult to accurately push information is solved, and accurate information push in designated areas is realized, reducing costs and improving results.
Patent Information
- Application Number
- CN202411915426.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-24
- Publication Date
- 2025-04-04
- Estimated Expiration
- 2044-12-24
AI Technical Summary
The existing information recommendation system is difficult to accurately push information to people in need, resulting in increased information recommendation costs and poor results.
Through the camera, pedestrians are identified and intelligently clustered in the area, quasi-user groups are determined, and information is pushed in the designated area to achieve accurate information push.
It realizes accurate determination of user groups within the region, reduces information push costs, and improves push effects.
Smart Images

Figure CN119766873A_ABST
Abstract
Description
Technical Field
[0001] The present application belongs to the technical field of information push, and in particular, relates to a method and system for regional intelligent clustering information push. Background Art
[0002] With the development of the digital economy, information recommendation (advertising) is no longer limited to traditional media, but has widely covered multiple channels such as the Internet, mobile applications, and social media. This method of information recommendation has greatly improved the publicity effect of information and provided a solid user base for the business products promoted by the recommended information.
[0003] However, many of the current information recommendation methods cannot accurately push information to the people in need, and it is difficult to accurately convert it into actual consumption, which increases the cost of information recommendation. In addition, in order to accurately recommend information, many information recommendation systems use relatively complex technical means to screen out accurate user groups, which requires a lot of manpower and material resources, and even so, they cannot accurately locate the information push group.
[0004] Therefore, how to recommend information accurately and intelligently is an urgent problem to be solved. Summary of the invention
[0005] The purpose of this application is to provide a method and system for pushing information based on regional intelligent clustering, so as to accurately determine the prospective user groups in the region, push information within the determined user groups, and achieve better push effects.
[0006] In order to achieve the above purpose, the technical solution of this application is as follows: A method for pushing information of regional intelligent clustering, comprising: All cameras are registered to the management server, and each camera joins the same multicast group through its own gateway router; The camera captures and identifies pedestrians. When the number of times the same pedestrian is identified during continuous capturing exceeds a first threshold, the identified pedestrian features and a timestamp are recorded, and a first message including the identified pedestrian features, the camera's own geographic location information, and a timestamp is multicast to other cameras. After receiving the first message, other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold. If it is within the range threshold, they check whether they have recorded the same pedestrian features. If they find that the same pedestrian features are recorded, they further determine whether the difference between the timestamp recorded by themselves and the timestamp in the first message is less than the time difference threshold. If it is less, they send a second message associated with the camera in the first message to the management server; After receiving the second message, the management server divides the associated cameras into the same area, and sends information push instructions to the cameras in the designated area according to the policy of pushing information in the designated area.
[0007] Furthermore, the information push method of regional intelligent clustering also includes: After receiving the first message, other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold, and if it exceeds the range threshold, send a rejection message to their own gateway router; After receiving the rejection message, the gateway router records the multicast suppression entry on the interface receiving the rejection message, and suppresses the first message sent by the camera that sends the first message.
[0008] Furthermore, the information push method of regional intelligent clustering also includes: If the gateway router that receives the rejection message has only one outbound interface, it continues to send the rejection message to the upstream multicast router. After receiving the rejection message, the upstream multicast router records the multicast suppression table entry to suppress the first message sent by the camera that sends the first message from being sent to the gateway router.
[0009] Furthermore, the information push method of regional intelligent clustering also includes: When the camera sends the first message, it carries the IP address of its own gateway router; After receiving the first message, the other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold, and if it exceeds the range threshold, send a rejection message to the gateway router of the camera that sent the first message; The gateway router of the camera that sends the first message records the multicast suppression entry, and suppresses the first message sent by the camera that sends the rejection message.
[0010] Furthermore, the information push method of regional intelligent clustering also includes: The camera captures and identifies the designated target, and when the number of times the designated target is identified during the continuous capturing process exceeds a first threshold, a first message including the identified designated target feature, the camera's own geographic location information and a timestamp is sent; After receiving the first message, other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold, and if within the range threshold, check whether they have recorded the same designated target feature; When it is found that the specified target feature is recorded, it is further determined whether the difference between the timestamp recorded by itself and the timestamp in the first message is less than the time difference threshold. If so, a second message associated with the camera in the first message is sent to the management server; When it is found that no designated target feature is recorded, a period of time is waited, and if the designated target is identified during the waiting period, a second message associated with the camera in the first message is sent to the management server.
[0011] The present application also proposes a regional intelligent clustering information push system, including a camera, a gateway router and a management server, wherein: All cameras are registered to the management server, and each camera joins the same multicast group through its own gateway router; The camera captures and identifies pedestrians. When the number of times the same pedestrian is identified during continuous capturing exceeds a first threshold, the identified pedestrian features and a timestamp are recorded, and a first message including the identified pedestrian features, the camera's own geographic location information, and a timestamp is multicast to other cameras. After receiving the first message, other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold. If it is within the range threshold, they check whether they have recorded the same pedestrian features. If they find that the same pedestrian features are recorded, they further determine whether the difference between the timestamp recorded by themselves and the timestamp in the first message is less than the time difference threshold. If it is less, they send a second message associated with the camera in the first message to the management server; After receiving the second message, the management server divides the associated cameras into the same area, and sends information push instructions to the cameras in the designated area according to the policy of pushing information in the designated area.
[0012] Furthermore, the regional intelligent clustering information push system further includes: After receiving the first message, other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold, and if it exceeds the range threshold, send a rejection message to their own gateway router; After receiving the rejection message, the gateway router records the multicast suppression entry on the interface that receives the rejection message, and suppresses the first message sent by the camera that sends the first message.
[0013] Furthermore, the regional intelligent clustering information push system further includes: If the gateway router that receives the rejection message has only one outbound interface, it continues to send the rejection message to the upstream multicast router. After receiving the rejection message, the upstream multicast router records the multicast suppression table entry to suppress the first message sent by the camera that sends the first message from being sent to the gateway router.
[0014] Furthermore, the regional intelligent clustering information push system further includes: When the camera sends the first message, it carries the IP address of its own gateway router; After receiving the first message, the other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold, and if it exceeds the range threshold, send a rejection message to the gateway router of the camera that sent the first message; The gateway router of the camera that sends the first message records the multicast suppression entry, and suppresses the first message sent by the camera that sends the rejection message.
[0015] Furthermore, the regional intelligent clustering information push system further includes: The camera captures and identifies the designated target, and when the number of times the designated target is identified during the continuous capturing process exceeds a first threshold, a first message including the identified designated target feature, the camera's own geographic location information and a timestamp is sent; After receiving the first message, other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold, and if within the range threshold, check whether they have recorded the same designated target feature; When it is found that the specified target feature is recorded, it is further determined whether the difference between the timestamp recorded by itself and the timestamp in the first message is less than the time difference threshold. If so, a second message associated with the camera in the first message is sent to the management server; When it is found that no designated target feature is recorded, a period of time is waited, and if the designated target is identified during the waiting period, a second message associated with the camera in the first message is sent to the management server.
[0016] The present application proposes a method and system for pushing information based on regional intelligent clustering, which realizes regional intelligent clustering through the recognition of the same pedestrian between cameras. Therefore, when information needs to be pushed to a specified area, the management server directly sends instructions to the cameras in the area to push the information, thereby realizing accurate information push according to the area. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] Figure 1 This is a flow chart of a method for pushing information based on regional intelligent clustering for this application.
[0018] Figure 2 This is a schematic diagram of the network structure of a regional intelligent clustering information push system for this application. DETAILED DESCRIPTION
[0019] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0020] Embodiment 1, as Figure 1As shown, a method for pushing information of regional intelligent clustering is provided, comprising: Step S1: All cameras are registered to the management server, and each camera joins the same multicast group through its own gateway router.
[0021] This embodiment is applicable to the information push system of smart cities. The information push can be traffic information, public safety information, or product promotion information, etc. This application is not limited to the specific type of information pushed. For a certain administrative block, it may include multiple different communities, or multiple different commercial blocks. If information is pushed indiscriminately to people in the entire administrative block, the information people receive will be too mixed, and most of it will be regarded as junk information and will not receive attention. This application uses cameras to identify pedestrians and perform regional intelligent clustering. It only pushes information in areas where information push is required, so as to achieve accurate information push.
[0022] To achieve the above purpose, each camera in this embodiment has its own number and built-in geographic location information, for example: longitude: 116.4074°E latitude: 39.9042°N. When registering with the management server, the camera can submit its own geographic location information, so the management server also has the geographic location information of each camera, or obtains the geographic location information through configuration.
[0023] The management server sends a unified multicast group address, such as 226.1.1.1, to all registered cameras to establish a multicast tree for camera synchronization messages. After registering with the management server, each camera needs to initiate an IGMP join message and send it to its own gateway router. The gateway router sends a PIM (*, G) join message to the PIM upstream router, so that the camera joins the (*, G) multicast forwarding tree for the network registration multicast address.
[0024] Step S2, the camera captures and identifies pedestrians. When the number of times the same pedestrian is identified during continuous capturing exceeds a first threshold, the identified pedestrian features and timestamp are recorded, and a first message including the identified pedestrian features, the camera's own geographic location information and timestamp is multicast to other cameras.
[0025] Specifically, the camera captures images periodically (for example, every 10 seconds), performs target recognition on the images, extracts pedestrian features, such as a facial feature vector [0.1, -0.2, 0.3, ..., 0.7, -0.8, 0.9], and establishes a record table entry about the facial feature vector.
[0026] If the number of times a same pedestrian is identified by a camera during continuous capture exceeds a first threshold, for example, exceeds 2 times, it is determined that the pedestrian is staying or wandering within the range of the camera.
[0027] For example, if the person is identified at least 2 times in 5 consecutive snapshots, it means that the person is staying or wandering under the camera, rather than just passing by. It should be noted that if the interval between two successful snapshots of the same person exceeds 4 times (not including 4 times), it does not meet the condition of "at least 2 times in 5 consecutive snapshots" and is not considered a "wandering face". This is because if the time is too long, it is possible that the pedestrian has passed by somewhere and returned.
[0028] After determining that a pedestrian stays or wanders within the range of this camera, the camera saves the pedestrian's features as a "wandering face" and adds a timestamp. The timestamp records the time of the most recent capture and is continuously updated when the person is captured again subsequently.
[0029] If the camera "wanders around a face" and the timestamp is not updated later because the corresponding face is not captured, it will be aged and deleted after being retained for a certain period of time (for example, 1 hour).
[0030] The camera will also broadcast pedestrian features that meet the "wandering face" to the entire network in a unified multicast address (for example, 226.1.1.1), including pedestrian features (for example: [0.1, -0.2, 0.3, ..., 0.7, -0.8, 0.9]), the camera's geographic location information, such as (116.4074°E latitude: 39.9042°N) and timestamp (2024-11-12, 9:55:30~10:00:00), and its own camera ID (0x1111). This message is called the "first message". Since all cameras use unified multicast, the gateway router's multicast forwarding table entry uses the (*, G) entry, where G is the network registration multicast address.
[0031] After receiving the first message, the gateway router sends it to the (*, G) multicast forwarding tree. The gateway router and other multicast routers in the network send the first message to other interfaces of the (*, G) multicast forwarding tree based on the bidirectional PIM method, but do not include the inbound interface, so as to realize the whole network transmission on the (*, G) multicast forwarding tree.
[0032] Step S3, after receiving the first message, other cameras determine whether the distance between themselves and the camera that sent the first message exceeds the range threshold. If it is within the range threshold, check whether they have recorded the same pedestrian features. When it is found that the same pedestrian features are recorded, further determine whether the difference between the timestamp recorded by themselves and the timestamp in the first message is less than the time difference threshold. If it is less, send a second message associated with the camera in the first message to the management server.
[0033] In this embodiment, the camera that sends the first message is called camera A. After receiving the first message, other cameras, such as camera B, first determine the distance between the geographic information contained in the received message and their own geographic location information. If it is within a certain distance range (which can be pre-set by the administrator, the distance range can be calculated by straight-line distance, such as 100M, or by the walking route in the actual map), it will be processed. Otherwise, it will not be processed, thereby saving consumption, because it is meaningless if the distance is too far.
[0034] If the distance between camera B and camera A is no more than 100M, camera B will further check whether there is a similar feature vector in its local "wandering face" database. If the face feature vector exists, it will then determine whether the timestamp of the face is less than a certain threshold (called the time difference threshold) with the timestamp of the first message, for example, less than 30 minutes. If so, it means that this camera is in the same area as the camera that sent the message. Therefore, camera B sends its own camera ID, such as 0x2222, and the camera ID in the first message (i.e., camera A, 0x1111) to the management server. This message is called a "camera association" message, that is, sending a second message associated with the camera in the first message to the management server. If the time difference threshold is exceeded, camera B will not process it.
[0035] This embodiment controls the time difference within a certain threshold. If the interval is too long, it may be a face that appeared the day before. At this time, although the straight-line distance is short, it may not be able to travel from camera A to camera B in a short time, and it does not necessarily belong to the same area, so it will not be processed.
[0036] Step S4: After receiving the second message, the management server divides the cameras with associated relationship into the same area, and sends information push instructions to the cameras in the designated area according to the policy of pushing information in the designated area.
[0037] The management server receives a large number of "camera association" messages, extracts the association information between the cameras from the messages, and obtains the associated camera IDs, such as "camera A (ID: 0x1111) - camera B (ID: 0x2222)".
[0038] Through this association relationship, a camera set can be established. For example, there is an association message of "Camera B-Camera C" and an association information of Camera A-Camera C, then "Camera A, Camera B, Camera C" form a set and belong to the same area, such as area 10.
[0039] If the management server also receives the association message between camera A and camera D, but does not receive the association message between camera D and other cameras, camera D is associated with camera A, but not with camera B or camera C, so camera D can also be clustered into area 10. Any two associated cameras can establish a regional cluster. Any camera can join a region as long as it is associated with one of the cameras in a regional cluster.
[0040] The associated information of the management server has an aging period. If no update message is received before the aging time, the corresponding associated item needs to be deleted.
[0041] Obviously, this set represents a business district that is close in distance and connected by foot, or an area that is extended to be connected by vehicle. This kind of regional aggregation is much more scientific than the management server aggregating from geographic information, because close distances do not necessarily belong to the same business district, and can be two areas that cannot be crossed by walls, mountains, or rivers.
[0042] In this embodiment, cameras are intelligently clustered, and cameras in the same area are aggregated into one set.
[0043] There are different strategies for information push, such as sending within a specified area. Therefore, when information needs to be pushed to a specified area, the management server directly sends instructions to the cameras in the area to push information, realizing accurate information push by area. For example, a store in a commercial area pushes product or service information to the commercial area, or the government pushes public information in a certain community to a certain community.
[0044] Embodiment 2: In order to reduce the mutual influence of multicast messages of cameras at a long distance, this embodiment is further improved on the basis of embodiment 1. The regional intelligent clustering information push method also includes:
[0045] After receiving the first message, other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold, and if it exceeds the range threshold, send a rejection message to their own gateway router; After receiving the rejection message, the gateway router records the multicast suppression entry on the interface that receives the rejection message, and suppresses the first message sent by the camera that sends the first message.
[0046] Specifically, if camera B determines that it is at a "certain distance" from camera A, it sends a "registration rejection" IGMP message to its own gateway router B, which contains the IP address of camera A (20.20.20.20) and the IP address of camera B (30.30.30.30). After receiving the message, gateway router B records the multicast suppression entry on the corresponding interface of the (*, G) entry (the outbound interface that receives the above message): the IP address of camera A (20.20.20.20). In the future, when the gateway router receives the first message with the source IP address of camera A (20.20.20.20) and the destination address of G (226.1.1.1), it will no longer send the message to the outbound interface, and camera B will no longer be able to receive the first message sent by camera A.
[0047] Embodiment 3: This embodiment is further improved on the basis of embodiment 2 to further reduce the propagation of multicast messages in the network. The regional intelligent clustering information push method also includes:
[0048] If the gateway router that receives the rejection message has only one outbound interface, it continues to send the rejection message to the upstream multicast router. After receiving the rejection message, the upstream multicast router records the multicast suppression table entry to suppress the first message sent by the camera that sends the first message from being sent to the gateway router.
[0049] Specifically, if the (*, G) multicast forwarding table entry of gateway router B has only one outbound interface connected to camera B, it will continue to send a "registration rejection" PIM message to the upstream multicast router, which contains the IP address of camera A (20.20.20.20). After receiving the above message, the upstream multicast router records the multicast suppression entry in the local (*, G) multicast forwarding table entry: the IP address of camera A1 (20.20.20.20). In the future, when the upstream multicast router receives the first message with the source IP address of camera A, it will no longer send it to gateway router B.
[0050] It should be noted that if the network router B has only one outgoing interface and the camera B refuses to receive the message, the network router B does not need to receive the first message, which can further suppress redundant multicast messages in the network.
[0051] However, when a new outbound interface is added to the multicast forwarding table of gateway router B (*, G), and the interface does not suppress the first message whose source IP address is camera A, network router B needs to send a "registration rejection cancellation" PIM message to the upstream multicast router, which contains the IP address of camera A that needs to be suppressed. Upon receiving the above message, the upstream multicast router deletes the multicast suppression entry in the local (*, G) multicast forwarding table: the mark of camera A1's IP address (20.20.20.20). In the future, the upstream multicast router will receive a characteristic multicast message whose source IP address is camera A and send it to gateway router B.
[0052] Embodiment 4: This embodiment is further improved on the basis of embodiment 2 to further reduce the propagation of multicast messages in the network. The regional intelligent clustering information push method also includes:
[0053] When the camera sends the first message, it carries the IP address of its own gateway router; After receiving the first message, the other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold, and if it exceeds the range threshold, send a rejection message to the gateway router of the camera that sent the first message; The gateway router of the camera that sends the first message records the multicast suppression entry, and suppresses the first message sent by the camera that sends the rejection message.
[0054] Specifically, when camera A multicasts the facial feature vector that meets the "wandering face" to the entire network in a unified multicast address (for example, 226.1.1.1), the IP address of the gateway router is also carried in the message, that is, the IP address of gateway router A. When camera B finds that it is too far away from camera A, in addition to sending a rejection registration message to the directly connected gateway router B, it also sends a "remote rejection registration" message to the gateway router A of camera A (the IP address of gateway router A is taken from the first message sent by camera A), which contains the IP addresses of camera A and camera B. After receiving the message, gateway router A records the multicast suppression table entry on its outbound interface connected to camera A according to the IP address of camera A in the message, and registers the IP address of camera B (30.30.30.30). In the future, when gateway router A receives the first message from camera B, it directly rejects it.
[0055] In this embodiment, since camera B interacts with the gateway router of camera A, rather than camera B telling camera A which then tells gateway router A, camera A avoids centrally processing notifications from a large number of other remote cameras and centrally registering with gateway router A.
[0056] In Example 5, in order to better verify the connected area, more accurate processing is required. For cameras that should belong to the same area, even if they are only captured once, they should be clustered together. The information push method of regional intelligent clustering described in this embodiment also includes:
[0057] The camera captures and identifies the designated target, and when the number of times the designated target is identified during the continuous capturing process exceeds a first threshold, a first message including the identified designated target feature, the camera's own geographic location information and a timestamp is sent; After receiving the first message, other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold, and if within the range threshold, check whether they have recorded the same designated target feature; When it is found that the specified target feature is recorded, it is further determined whether the difference between the timestamp recorded by itself and the timestamp in the first message is less than the time difference threshold. If so, a second message associated with the camera in the first message is sent to the management server; When it is found that no designated target feature is recorded, a period of time is waited, and if the designated target is identified during the waiting period, a second message associated with the camera in the first message is sent to the management server.
[0058] Specifically, the designated target is a staff member who verifies the connectivity area. After the camera A recognizes the designated target, it processes according to the method of embodiment 1 and may also send the first message to the management server.
[0059] If the designated target is within the range of the nearby camera B again within a short period of time, but the camera B does not capture the designated target twice within five consecutive times, then according to the processing flow of Example 1, the camera A and the camera B cannot form an area.
[0060] This embodiment has improved this. After the management server receives the first message about the designated target from camera A, it extracts the camera ID, the facial feature vector and the timestamp from it. If it finds that the facial feature vector belongs to the designated target, it sends a network-wide multicast message containing the ID of camera A, the geographical location of camera A, the facial feature vector and the timestamp. This message is called the "regional connectivity query" message, which is a type of the first message, in which the facial feature vector is the designated target. The first message can also be sent by camera A through multicast.
[0061] After receiving the information, other cameras B extract the above information. Then, the distance between the geographic information contained in the received message and its own geographic location information is determined first. Only when the distance is within a certain range is processed, so as to save consumption, because it is meaningless if the distance is too far. Then, it is determined whether there is a corresponding designated target feature locally. If so, it is determined whether the difference between the timestamp recorded by itself and the timestamp in the first message is less than the time difference threshold. If so, it sends a second message associated with itself and the camera in the first message to the management server.
[0062] If not, wait for a while to see if the designated target can be captured. The waiting time is: time difference threshold - (current time - timestamp in the message). If the designated target is captured within the waiting time, reply a second message to the management server, which contains the ID of this camera and the ID of camera A. If the designated target is not found after the waiting time, camera B clears the first message extracted to the management server. Therefore, if the waiting time is exceeded, the time will time out, which is meaningless.
[0063] When the management server receives the second message, it clusters camera B and camera A as the same area.
[0064] Camera A in this application will multicast the facial feature vector that meets the "wandering face" to the entire network in the form of a unified multicast group address. After receiving it, other cameras will first determine the distance between the geographic information contained in the received message and their own geographic location information. Within a certain distance range, they will check their local "wandering face" library. If the face exists, they will then determine whether the timestamp of the face is less than a certain threshold compared to the timestamp of the newly received message. If so, it means that this camera is in the same area as the camera that sent the message, so it will send its own camera ID and the camera ID in the message (that is, camera A) to the management server to achieve regional clustering.
[0065] The camera sends a "registration rejection" message to reduce multicast messages in the network, and also achieves the suppression of two-way messages and multicast tree pruning. In addition, the management server can actively initiate a "regional connectivity query" for specific facial feature vectors, allowing the camera to perform special processing on specific facial feature vectors and achieve analysis of specific connectivity areas.
[0066] By sending feature multicast messages to each other, cameras can locally analyze the camera information associated with themselves, thereby establishing an information table of cameras associated with themselves, so that the management server can perform regional clustering based on the associated information table, which can reduce the pressure on the management server to perform regional clustering through centralized computing.
[0067] Example 6, corresponding to the above-mentioned method for pushing information of regional intelligent clustering, this embodiment proposes an information pushing system for regional intelligent clustering, including a camera, a gateway router and a management server, wherein: All cameras are registered to the management server, and each camera joins the same multicast group through its own gateway router; The camera captures and identifies pedestrians. When the number of times the same pedestrian is identified during continuous capturing exceeds a first threshold, the identified pedestrian features and a timestamp are recorded, and a first message including the identified pedestrian features, the camera's own geographic location information, and a timestamp is multicast to other cameras. After receiving the first message, other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold. If it is within the range threshold, they check whether they have recorded the same pedestrian features. If they find that the same pedestrian features are recorded, they further determine whether the difference between the timestamp recorded by themselves and the timestamp in the first message is less than the time difference threshold. If it is less, they send a second message associated with the camera in the first message to the management server; After receiving the second message, the management server divides the associated cameras into the same area, and sends information push instructions to the cameras in the designated area according to the policy of pushing information in the designated area.
[0068] This embodiment is a regional intelligent clustering information push system, which also includes the following embodiments: In a specific embodiment, the regional intelligent clustering information push system further includes: After receiving the first message, other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold, and if it exceeds the range threshold, send a rejection message to their own gateway router; After receiving the rejection message, the gateway router records the multicast suppression entry on the interface receiving the rejection message, and suppresses the first message sent by the camera that sends the first message.
[0069] In another specific embodiment, the regional intelligent clustering information push system further includes: If the gateway router that receives the rejection message has only one outbound interface, it continues to send the rejection message to the upstream multicast router. After receiving the rejection message, the upstream multicast router records the multicast suppression table entry to suppress the first message sent by the camera that sends the first message from being sent to the gateway router.
[0070] In another specific embodiment, the regional intelligent clustering information push system further includes: When the camera sends the first message, it carries the IP address of its own gateway router; After receiving the first message, the other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold, and if it exceeds the range threshold, send a rejection message to the gateway router of the camera that sent the first message; The gateway router of the camera that sends the first message records the multicast suppression entry, and suppresses the first message sent by the camera that sends the rejection message.
[0071] In another specific embodiment, the regional intelligent clustering information push system further includes: The camera captures and identifies the designated target, and when the number of times the designated target is identified during the continuous capturing process exceeds a first threshold, a first message including the identified designated target feature, the camera's own geographic location information and a timestamp is sent; After receiving the first message, other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold, and if within the range threshold, check whether they have recorded the same designated target feature; When it is found that the specified target feature is recorded, it is further determined whether the difference between the timestamp recorded by itself and the timestamp in the first message is less than the time difference threshold. If so, a second message associated with the camera in the first message is sent to the management server; When it is found that no designated target feature is recorded, a period of time is waited, and if the designated target is identified during the waiting period, a second message associated with the camera in the first message is sent to the management server.
[0072] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.
Claims
1. A method for pushing information of regional intelligent clustering, characterized in that: The information push method of regional intelligent clustering includes: All cameras are registered to the management server, and each camera joins the same multicast group through its own gateway router; The camera captures and identifies pedestrians. When the number of times the same pedestrian is identified during continuous capturing exceeds a first threshold, the identified pedestrian features and a timestamp are recorded, and a first message including the identified pedestrian features, the camera's own geographic location information, and a timestamp is multicast to other cameras. After receiving the first message, other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold. If it is within the range threshold, they check whether they have recorded the same pedestrian features. If they find that the same pedestrian features are recorded, they further determine whether the difference between the timestamp recorded by themselves and the timestamp in the first message is less than the time difference threshold. If it is less, they send a second message associated with the camera in the first message to the management server; After receiving the second message, the management server divides the associated cameras into the same area, and sends information push instructions to the cameras in the designated area according to the policy of pushing information in the designated area.
2. The information push method of regional intelligent clustering according to claim 1, characterized in that: The method for pushing information of regional intelligent clustering further includes: After receiving the first message, other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold, and if it exceeds the range threshold, send a rejection message to their own gateway router; After receiving the rejection message, the gateway router records the multicast suppression entry on the interface receiving the rejection message, and suppresses the first message sent by the camera that sends the first message.
3. The information push method of regional intelligent clustering according to claim 2 is characterized in that: The method for pushing information of regional intelligent clustering further includes: If the gateway router that receives the rejection message has only one outbound interface, it continues to send the rejection message to the upstream multicast router. After receiving the rejection message, the upstream multicast router records the multicast suppression table entry to suppress the first message sent by the camera that sends the first message from being sent to the gateway router.
4. The information push method of regional intelligent clustering according to claim 2 is characterized in that: The method for pushing information of regional intelligent clustering further includes: When the camera sends the first message, it carries the IP address of its own gateway router; After receiving the first message, the other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold, and if it exceeds the range threshold, send a rejection message to the gateway router of the camera that sent the first message; The gateway router of the camera that sends the first message records the multicast suppression entry, and suppresses the first message sent by the camera that sends the rejection message.
5. The information push method of regional intelligent clustering according to claim 1, characterized in that: The method for pushing information of regional intelligent clustering further includes: The camera captures and identifies the designated target, and when the number of times the designated target is identified during the continuous capturing process exceeds a first threshold, a first message including the identified designated target feature, the camera's own geographic location information and a timestamp is sent; After receiving the first message, other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold, and if within the range threshold, check whether they have recorded the same designated target feature; When it is found that the specified target feature is recorded, it is further determined whether the difference between the timestamp recorded by itself and the timestamp in the first message is less than the time difference threshold. If so, a second message associated with the camera in the first message is sent to the management server; When it is found that no designated target feature is recorded, a period of time is waited, and if the designated target is identified during the waiting period, a second message associated with the camera in the first message is sent to the management server.
6. A regional intelligent clustering information push system, characterized in that: The regional intelligent clustering information push system includes a camera, a gateway router, and a management server, wherein: All cameras are registered to the management server, and each camera joins the same multicast group through its own gateway router; The camera captures and identifies pedestrians. When the number of times the same pedestrian is identified during continuous capturing exceeds a first threshold, the identified pedestrian features and a timestamp are recorded, and a first message including the identified pedestrian features, the camera's own geographic location information, and a timestamp is multicast to other cameras. After receiving the first message, other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold. If it is within the range threshold, they check whether they have recorded the same pedestrian features. If they find that the same pedestrian features are recorded, they further determine whether the difference between the timestamp recorded by themselves and the timestamp in the first message is less than the time difference threshold. If it is less, they send a second message associated with the camera in the first message to the management server; After receiving the second message, the management server divides the associated cameras into the same area, and sends information push instructions to the cameras in the designated area according to the policy of pushing information in the designated area.
7. The information push system for regional intelligent clustering according to claim 6, characterized in that: The regional intelligent clustering information push system also includes: After receiving the first message, other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold, and if it exceeds the range threshold, send a rejection message to their own gateway router; After receiving the rejection message, the gateway router records the multicast suppression entry on the interface receiving the rejection message, and suppresses the first message sent by the camera that sends the first message.
8. The information push system for regional intelligent clustering according to claim 7, characterized in that: The regional intelligent clustering information push system also includes: If the gateway router that receives the rejection message has only one outbound interface, it continues to send the rejection message to the upstream multicast router. After receiving the rejection message, the upstream multicast router records the multicast suppression table entry to suppress the first message sent by the camera that sends the first message from being sent to the gateway router.
9. The information push system for regional intelligent clustering according to claim 7, characterized in that: The regional intelligent clustering information push system also includes: When the camera sends the first message, it carries the IP address of its own gateway router; After receiving the first message, the other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold, and if it exceeds the range threshold, send a rejection message to the gateway router of the camera that sent the first message; The gateway router of the camera that sends the first message records the multicast suppression entry, and suppresses the first message sent by the camera that sends the rejection message.
10. The information push system for regional intelligent clustering according to claim 6, characterized in that: The regional intelligent clustering information push system also includes: The camera captures and identifies the designated target, and when the number of times the designated target is identified during the continuous capturing process exceeds a first threshold, a first message including the identified designated target feature, the camera's own geographic location information and a timestamp is sent; After receiving the first message, other cameras determine whether the distance between them and the camera that sent the first message exceeds the range threshold, and if within the range threshold, check whether they have recorded the same designated target feature; When it is found that the specified target feature is recorded, it is further determined whether the difference between the timestamp recorded by itself and the timestamp in the first message is less than the time difference threshold. If so, a second message associated with the camera in the first message is sent to the management server; When it is found that no designated target feature is recorded, a period of time is waited, and if the designated target is identified during the waiting period, a second message associated with the camera in the first message is sent to the management server.
Citation Information
Patent Citations
Multi-camera collaboration-based illegal parking vehicle license plate recognition method and device
CN110288837A
Distributed intelligent passenger flow statistics effective deduplication method
CN111783588A
Intelligent traffic management method for large-scale community
CN115440051A
Media intelligent publishing system for multiple terminals
CN115695547A
Intelligent video surveillance system for school zone
KR101672780B1