A trajectory collision method and system based on multiple verifications

Abstract

The present invention provides a trajectory collision method and system based on multiple verifications, wherein the method comprises: obtaining first trajectory flow data, the first trajectory flow data being the result of the previous trajectory collision between the trajectory flow data to be found and the trajectory flow data to be matched; time slicing the trajectory flow data to be matched at a first time slice granularity; determining whether there is trajectory flow data to be matched within a specified time slice; updating the number of matches of the first trajectory flow data; and outputting the trajectory collision result after a preset number of verifications. Distance, area, and similarity analysis are performed on different trajectories, and multiple data verifications in multiple dimensions such as time and space are combined to obtain the most reliable collision result. The problem that the trajectory collision result based on one-dimensional verification is unreliable and misleading for further analysis is solved, and the trajectory collision method with multiple verifications is adopted to improve the accuracy of trajectory collision, and effectively prevent and solve the occurrence of potential public safety hazards.

Description

Technical field

[0001] The present invention relates to the field of trajectory collision technology, and in particular to a trajectory collision method and system based on multiple verifications. [Background Technology]

[0002] Generally speaking, trajectory collision is based on one-dimensional verification. Due to the differences between systems, the data density and distribution of trajectories vary. If verification is done in a simple one-dimensional manner, the collision results obtained are unreliable and will be misleading for further analysis.

[0003] Taking the collision of video and Wi-Fi trajectories as an example, the common approach is to directly compare the similarity of the two trajectories, or to slice the trajectories at a certain time granularity and sort them by distance within that time unit. The matching information within that time unit is then accumulated, and the top-ranked trajectories are considered the matching set. However, because video and Wi-Fi are two different systems, their detection times are not completely synchronized. Furthermore, Wi-Fi-enabled electronic devices are also different, with varying packet transmission cycles. This can lead to some devices sending packets more frequently than others. Therefore, using this as the final collision result is unreliable and can be misleading for further analysis. [Summary of the invention]

[0004] The technical problem to be solved by the present invention is to provide a trajectory collision method and system based on multiple verifications, which can solve the problem that the trajectory collision results based on one-dimensional verification are unreliable and misleading for further analysis. By adopting the trajectory collision method with multiple verifications, the accuracy of trajectory collision is improved, and the occurrence of potential public safety incidents is effectively prevented and resolved.

[0005] To address the above technical issues, one embodiment of the present invention provides a trajectory collision method based on multiple verifications, comprising the steps of: obtaining first trajectory stream data, where the first trajectory stream data is the result of a previous trajectory collision between the trajectory stream data to be searched and the trajectory stream data to be matched; time slicing the trajectory stream data to be matched at a first time slice granularity; determining whether there is any trajectory stream data to be matched within a specified time slice; updating the number of matches for the first trajectory stream data; and outputting a trajectory collision result after a preset number of verifications.

[0006] Preferably, obtaining first trajectory stream data, where the first trajectory stream data is a result of a previous trajectory collision between the trajectory stream data to be found and the trajectory stream data to be matched, includes: time slicing the trajectory stream data to be matched at a second time slice granularity; determining whether there is trajectory stream data to be matched within a specified time slice; if so, filtering out matching information that meets the conditions within the specified time slice; updating the matching count of the matching information; and outputting a trajectory collision result.

[0007] Preferably, before time slicing the trajectory flow data to be matched at the first time slice granularity, the method further includes: inputting the first trajectory flow data to be matched.

[0008] Preferably, before time slicing the trajectory flow data to be matched at the second time slice granularity, the method further includes: inputting the second trajectory flow data to be matched.

[0009] Preferably, the first trajectory stream data and the second trajectory stream data are the same.

[0010] Preferably, the second time slice granularity is the same as the first time slice granularity.

[0011] Preferably, the second time slice granularity is different from the first time slice granularity.

[0012] Preferably, updating the matching times of the first trajectory flow data includes: for matching information not included in the specified time slice, the matching times are reduced accordingly; for matching information included in the specified time slice, the matching times remain unchanged.

[0013] Preferably, updating the matching number of the matching information includes: if the trajectory flow data to be matched matches the trajectory flow data to be found within the specified time slice successfully, then the matching number of the trajectory flow data to be matched and the trajectory flow data to be found within the original specified time slice is increased accordingly.

[0014] On the other hand, an embodiment of the present invention provides a trajectory collision system based on multiple verifications, including: a positioning device, and the system executes the above method.

[0015] Preferably, the positioning device includes a WiFi probe device, a POE module, and a server. The WiFi probe device is used to detect the device MAC; the POE module is used to power the WiFi probe device and transmit data back to the server.

[0016] Preferably, the server includes a database server and a positioning server, the database server is used to store the detected device MAC information, and the positioning server is used to perform positioning calculations on the data stored in the database server and store location information corresponding to the MAC information.

[0017] Compared with existing technologies, the above technical solution has the following advantages: by analyzing the distance, area, and similarity of different trajectories, combined with multiple data verifications in multiple dimensions such as time and space, the most reliable collision results are obtained. This solves the problem that trajectory collision results based on one-dimensional verification are unreliable and can mislead further analysis. By adopting a trajectory collision method with multiple verifications, the accuracy of trajectory collisions is improved, effectively preventing and resolving potential incidents that endanger public safety. It is universally applicable to trajectory collisions between different attributes, as well as between trajectories with the same attributes. Application scenarios include various wireless scenarios such as indoor and outdoor, and the application field can be expanded to voice recognition, face recognition, big data analysis, and so on.

Brief Description of the Drawings

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 This is a diagram of the hash matching principle in the prior art.

[0020] Figure 2 This is a diagram of the matching principle of DTW in the prior art.

[0021] Figure 3 This is a flow chart of a trajectory collision method based on multiple verifications of the present invention.

[0022] Figure 4 yes Figure 3 Flowchart of a method for obtaining a trajectory collision result from a previous trajectory collision.

[0023] Figure 5 This is a schematic diagram of key-value storage used in a trajectory collision method based on multiple verifications of the present invention.

[0024] Figure 6 yes Figure 3 Flowchart of the method for obtaining the trajectory collision result in this trajectory collision.

[0025] Figure 7 This is a diagram of a preferred embodiment of trajectory collision based on multiple verifications of the present invention.

[0026] Figure 8 This is a structural diagram of a trajectory collision system based on multiple verifications of the present invention.

[0027] Figure 9 yes Figure 8 Schematic diagram of the positioning server storage. [Specific implementation method]

[0028] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0029] Before introducing the trajectory collision method, let's first briefly introduce the hashing concept, DTW algorithm, geohash encoding, and Key-Value storage system.

[0030] Figure 1 The basic idea of ​​hashing is that after two adjacent data points in the original data space are transformed by the same mapping or projection, the probability that the two data points are still adjacent in the new data space is high, while the probability that non-adjacent data points are mapped to the same bucket is low. In other words, if we perform some hash mapping on the original data, we hope that the two originally adjacent data can be hashed into the same bucket with the same bucket number. The basic idea of ​​hashing is as shown in the attached figure. Figure 1 shown.

[0031] Figure 2 This is the matching principle diagram of DTW in the existing technology. The DTW algorithm was originally used in the field of speech recognition. It is based on the idea of ​​dynamic programming (DP) and solves the problem of template matching with different pronunciation lengths. It is an early and classic algorithm in speech recognition and is used for isolated word recognition. The matching principle diagram of DTW in the existing technology is as follows Figure 2 As shown in the figure, for people traveling simultaneously, the unequal length of their position information also creates the problem of matching multiple trajectories. Incorporating the principles of DTW, it can also be used to identify trajectory similarities. DTW is a basic algorithm that is scalable, and optimization methods include fast-DTW, SparseDTW, LB_Keogh, and LB_Improved.

[0032] Geohash converts a longitude and latitude into a sortable and comparable string. A geohash represents not a point but a rectangular area. Users can publish geocodes to indicate their location near a specific address without revealing their precise coordinates, thus enhancing privacy. Geohash is significantly more efficient than using longitude and latitude directly. This example uses the concept of geohash to implement data hash mapping.

[0033] The default geohash encoding is 12 digits, and the precision range corresponding to the first 9 digits is as follows.

[0034]

[0035]

[0036] Organizations of all sizes are increasingly faced with the need to process big data, yet relational databases have reached their scalability limits. One solution is to use a key-value store, a NoSQL (non-relational database) model in which data is organized, indexed, and stored as key-value pairs. Key-value stores are ideal for business data that doesn't involve extensive data relationships or business relationships. They also significantly reduce disk read and write times, offering better read and write performance than SQL databases.

[0037] Commonly used key-value storage systems include Redis, Hbase, Leveldb, Scalaris, HyperDex, etc.

[0038] Example 1

[0039] Figure 3 This is a flow chart of a trajectory collision method based on multiple verifications of the present invention. Figure 3 As shown, a trajectory collision method based on multiple verifications includes the following steps:

[0040] S1. Obtain first trajectory stream data, where the first trajectory stream data is a result of a previous trajectory collision between the trajectory stream data to be searched and the trajectory stream data to be matched;

[0041] S2. Time-slicing the trajectory flow data to be matched at a first time-slicing granularity;

[0042] S3. Determine whether there is any trajectory flow data to be matched within the specified time slice;

[0043] S4. Update the matching times of the first trajectory flow data;

[0044] S5. Output the trajectory collision result after a preset number of verifications.

[0045] Given that collision results obtained by one-dimensional verification methods are not very reliable and can be misleading for further analysis, this patent proposes a multi-verification trajectory collision method that combines multiple data verifications in multiple dimensions such as time and space to obtain the most reliable collision results.

[0046] This embodiment performs secondary verification in two dimensions: distance and time synchronization. In specific implementation, it is not limited to two verifications, and multiple verifications using other methods may be added. Figure 4 yes Figure 3 The flowchart of the method for obtaining the trajectory collision result in the previous trajectory collision is shown in the figure. Taking the video trajectory and Wi-Fi trajectory as examples, the trajectory collision method with multiple verifications is introduced. The flowchart of the first collision verification is shown in the figure. Figure 3 As shown, the steps include:

[0047] S11. Input the trajectory flow data to be matched

[0048] This case uses a face to find the corresponding mobile phone MAC address, so the input is a user's video stream data. During project implementation, track collision can be performed online in real time, inputting only the most recent data stream for track collision processing. Offline track collision can also be performed, inputting historical track stream data to be matched.

[0049] S12. Time slice the input trajectory flow data

[0050] Generally speaking, it's difficult to completely synchronize time processing between two independent systems. Even within the same system, it's difficult to guarantee simultaneous detection of two different users. Therefore, trajectory collisions can tolerate a certain amount of time deviation in the trajectory stream data. Time slicing involves segmenting the trajectory stream data along the time dimension to ensure that the compared trajectory positions fall within a tolerable time range. If they do, the time dimension cannot be matched, and the collision is meaningless.

[0051] Assume that the time span of the input trajectory flow data to be matched is 1 minute, the length of the time slice is 10 seconds, and the tolerable time range is 10 seconds. Then, the 1-minute data can be divided into 6 time slices.

[0052] It should be noted that:

[0053] a. In the actual system, since the detected trajectory flow data is not complete and continuous, not all time slices in the 6 time slices contain the input trajectory flow data.

[0054] b. The time tolerance range under different project requirements is different. The principle to follow is that if the tolerance range in the time dimension is loose, then the tolerance range in the space dimension must be strict; conversely, if the tolerance range in the time dimension is strict, then the tolerance range in the space dimension must be loose.

[0055] S13, determine whether there is input trajectory flow data within the time slice

[0056] If yes, proceed to step S14; if not, continue processing the next time slice and proceed to step S13 at the same time.

[0057] S14. Filter out matching information that meets the conditions within the time slice

[0058] In the time slice, the trajectory information that is closest to the input trajectory stream data is found from all the trajectory stream data to be searched, and the matching information set in the time slice is obtained.

[0059] Analysis can be performed based on distance, area, and similarity. Distance determines whether two trajectories are within a distance threshold, using spherical distance and Euclidean distance as examples. Area determines whether two trajectories belong to the same area. Similarity determines whether two trajectories are similar.

[0060] The data calculation / screening here includes a variety of methods, which can be done through direct calculation or by using the hashing idea to quickly screen the data. The following focuses on the hash screening method. The premise of hash screening is that the trajectory information is stored in the form of key-value. Taking the WiFi trajectory as an example, geohash+time is used as the key and mac set is used as the value to store it for quick search. The storage result diagram is shown in the attached figure. Figure 5 As shown, Figure 5 This is a schematic diagram of the key-value storage used in the trajectory collision method based on multiple verifications of the present invention. The method is as follows:

[0061] a. Calculate the average coordinates of the input trajectory flow in the current time slice and find the corresponding geohash value.

[0062] b. Match the spatial dimension with the trajectory flow data to be found.

[0063] The current time slice range is t0 to t1. The average coordinates of the input trajectory stream within the time slice are calculated using the geohash value. Keywords are generated for each time interval based on the geohash values. The corresponding MAC set is retrieved from the key-value system based on the keyword. This is the set of MACs that are similar to the trajectory stream data to be matched in the time slice in both temporal and spatial dimensions.

[0064] It should be noted that the accuracy of the geohash value to be matched must be consistent with the accuracy of the geohash stored in the key-value store. Since the geohash represents a rectangular area, not a circular area, the accuracy cannot be fully guaranteed at a certain location. Therefore, the neighbor geohash can be introduced to expand the area range.

[0065] S15. Update the matching times of each matching information

[0066] The number of matches is updated for the information matched in time and space in S14.

[0067] Taking the matched MAC set as an example, if the matched MAC already exists, the original matching count will be increased accordingly; if it does not exist, the matching count will be 1.

[0068] S16: Determine whether all time slices have been traversed

[0069] If the traversal is not complete, the next time slice is processed and step S13 is performed; otherwise, step S17 is performed.

[0070] S17: Output the collision result and store it.

[0071] The first collision result is recorded as res1.

[0072] This yields the collision result and the corresponding number of matches after the first verification. However, because video and Wi-Fi are two different systems, their detection times are not completely synchronized. Furthermore, Wi-Fi-enabled electronic devices are also different, with varying packet transmission cycles. This can result in some devices transmitting packets more frequently than others. Therefore, using this as the final collision result is unreliable and can be misleading for further analysis.

[0073] Based on the first verification, the collision result is subjected to a second verification with time synchronization. Figure 6 yes Figure 3 The flow chart of the method for obtaining the trajectory collision result in this trajectory collision is shown in FIG. Figure 6 As shown, the steps include:

[0074] S21. Input the trajectory flow data to be matched

[0075] The trajectory flow data here is exactly the same as that in S11 and is used for secondary verification.

[0076] S22: Obtain trajectory flow data of the collision result obtained in S17.

[0077] Read the matching information track that meets the screening conditions obtained in S17. Further analysis based on the results of the first verification can improve the reliability of the analysis results and gradually narrow the target group to reduce the computational complexity of the algorithm.

[0078] S23, time slicing the input trajectory flow data

[0079] The time slice granularity here can be the same as or different from the time slice granularity in step S12. Referring to the application scenario of the actual system, the slicing method is the same as step S12. For example, if the time slice granularity in step S12 is 10s, the time slice granularity in S23 can be 20s.

[0080] S24: Determine whether there is input trajectory flow data within the time slice

[0081] If yes, proceed to step S25; if not, continue processing the next time slice and proceed to step S24 at the same time.

[0082] S25. Update the number of matches in the collision result

[0083] Determine which matching information in the collision result res1 obtained in S17 is within the time slice. For matching information not included in the time slice, its matching count is reduced accordingly; for matching information included in the time slice, the matching count is retained and not processed.

[0084] S26: Determine whether all time slices have been traversed

[0085] If the traversal is not complete, the next time slice is processed and step S24 is performed; otherwise, step S27 is performed.

[0086] S27. Output the collision result res2 after multiple verifications and store it.

[0087] It should be noted that:

[0088] a. This solution can achieve simultaneous collision of multiple trajectory stream data in a parallel manner.

[0089] b. There may be multiple collision results. Through online operations, the number of collisions of trajectories with the same attributes will gradually become more prominent over time.

[0090] Figure 7 This is a preferred embodiment of the trajectory collision based on multiple verifications of the present invention. Figure 7 As shown in the figure, in the specific implementation, assuming that a WiFi track matching the face track uid1 is searched, the first track verification is performed with a collision verification at a time slice granularity of 10s, and the matching information that meets the conditions and the corresponding number of times are obtained: {M1: 8, M2: 6, M3: 6}. From the collision results, it can be seen that M1 is the closest, and M2 and M3 are both ranked second.

[0091] The collision result is subjected to a second time synchronization collision verification, with 20s as the time slice granularity for collision verification, and the matching information that does not meet the time slice and the corresponding number of times are obtained: {M1: 1, M2: 3, M3: 0}.

[0092] Combining the results of the two verifications, the first result minus the second result, the trajectory collision results and the corresponding times after multiple verifications are: {M1: 7, M2: 3, M3: 6}. It can be seen that M1 still ranks first, while M3 ranks higher than M2. The corresponding example is shown in the attached figure. Figure 7 shown.

[0093] As the example in the figure shows, M2 had more data in the first half, resulting in a slightly higher first verification result. However, due to a lack of data in the second half, the second verification had more out-of-time times, reducing the overall number of collision results. M3, on the other hand, had less data, but its data was always present, so its overall collision results were more reliable than M2.

[0094] In actual application scenarios, this phenomenon is not uncommon. Therefore, the collision results obtained through multiple trajectory verifications are more reliable, otherwise it will only mislead further analysis.

[0095] Example 2

[0096] Figure 8 This is a structural diagram of a trajectory collision system based on multiple verifications of the present invention. Figure 8 As shown, a trajectory collision system based on multiple verifications includes: a positioning device, a server, and the system executes the above method. The Wi-Fi probe can provide basic identity recognition data, and the collected MAC address data can be associated with the data of telecommunications companies and public security agencies to establish a multi-dimensional public safety monitoring system. The MAC address is the only identification code of the smartphone and can be used to identify identity information. Combined with the construction of video perception deployment locations, the Wi-Fi probe has a wider coverage and can collect MAC addresses within the range. At the same time, the data is not restricted, and MAC addresses can be collected in large quantities. The Wi-Fi probe can realize real-time data transmission, and the monitoring data can be sent back in real time; identity matching: the MAC address is the only identification code of the mobile phone, and it can be combined with other data to achieve identity matching.

[0097] The Wi-Fi positioning system applies Wi-Fi positioning technology to the scene of real-time tracking and identification of personnel. Through real-time positioning technology, suspicious personnel on the scene can be found and tracked in time. The Wi-Fi positioning system consists of Wi-Fi probe equipment, POE module, database server, and positioning server. The system structure diagram is as follows Figure 1 shown.

[0098] The Wi-Fi probe device has the following uses:

[0099] (1) The built-in misleading module transmits a high-connection frequency SSID to mislead devices to connect and increase the probability of capturing MAC.

[0100] (2) Scan all channels and capture device MAC addresses without missing any packets.

[0101] (3) Encrypted and transmitted information such as the strength of the marked MAC signal and the connection time difference is sent to the location calculation server for accurate location calculation.

[0102] The POE module supplies power to the Wi-Fi probe device while transmitting data back to the database server.

[0103] The database server, as a database for storing MAC addresses, quickly compares the MAC addresses captured by the WiFi probe device, transmits the successful comparison data to the positioning server, and updates the connection duration, connection time, location and other information of the MAC-marked device into the database.

[0104] The positioning server runs the positioning algorithm, calculates the real-time received signal and matches it with the fingerprint database data, and calculates the coordinates to be positioned based on the fingerprint coordinates. The schematic diagram of the positioning result storage is shown in the attached figure. Figure 9 shown. Figure 9 yes Figure 8 Schematic diagram of the positioning server storage.

[0105] The storage format of the positioning server is that each line of data represents the ID of the device to be positioned, the MAC address of the device to be positioned, the name of the device to be positioned, the X coordinate of the device to be positioned, the Y coordinate of the device to be positioned, and the reporting time.

[0106] As can be seen from the above description, the trajectory collision method and system based on multiple verifications according to the present invention achieves the most reliable collision results by analyzing different trajectories based on distance, area, and similarity, combined with multiple data verifications in multiple dimensions such as time and space. This solves the problem that trajectory collisions are typically based on one-dimensional verification. Due to differences between systems, the data density and distribution of trajectories vary. Using a single one-dimensional verification approach yields unreliable collision results, which can be misleading for further analysis. The multiple verification trajectory collision method improves the accuracy of trajectory collisions and effectively prevents and resolves potential incidents that endanger public safety. It is universally applicable to trajectory collisions between different attributes, such as Wi-Fi and video, Wi-Fi and Bluetooth, Wi-Fi and vehicle, video and vehicle, video and Bluetooth, vehicle and Bluetooth, and so on. It can also be used for trajectory collisions between the same attributes. Application scenarios include various indoor and outdoor wireless scenarios, and the application areas can be expanded to voice recognition, facial recognition, big data analysis, and more.

[0107] The embodiments of the present invention are described in detail above. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core ideas. At the same time, for those skilled in the art, according to the ideas of the present invention, there may be changes in the specific implementation methods and application scopes. In summary, the contents of this specification should not be understood as limiting the present invention.

Claims

1. A trajectory collision method based on multiple verifications, characterized in that: include: Acquire first trajectory flow data, where the first trajectory flow data is a result of a previous trajectory collision between the trajectory flow data to be found and the trajectory flow data to be matched; The obtaining of the first trajectory flow data comprises: performing time slicing on the trajectory flow data to be matched at a second time slice granularity; determining whether there is trajectory flow data to be matched within a specified time slice; if so, filtering out matching information that meets the conditions within the specified time slice; updating the matching times of the matching information; and outputting a trajectory collision result; Time slicing the trajectory flow data to be matched at a first time slice granularity; Before time slicing the trajectory flow data to be matched at the first time slice granularity, the method further includes: inputting the trajectory flow data to be matched; Before time slicing the trajectory flow data to be matched at the second time slice granularity, the method further includes: inputting the trajectory flow data to be matched; Determine whether there is trajectory flow data to be matched within the specified time slice; Update the matching times of the first track flow data: for the matching information not included in the specified time slice, its matching times are reduced accordingly; for the matching information included in the specified time slice, the matching times remain unchanged; Output the trajectory collision results after the preset number of verifications.

2. The trajectory collision method based on multiple verification according to claim 1, characterized in that: The second time slice granularity is different from the first time slice granularity.

3. The trajectory collision method based on multiple verification according to claim 1, characterized in that: Updating the number of matches of the matching information includes: if the trajectory flow data to be matched and the trajectory flow data to be found within the specified time slice are matched successfully, then the number of matches between the trajectory flow data to be matched and the trajectory flow data to be found within the original specified time slice is increased accordingly.

4. A trajectory collision system based on multiple verifications, characterized in that: include: Positioning device, the system executes the method according to any one of claims 1 to 3.

5. The trajectory collision system based on multiple verification according to claim 4, characterized in that: The positioning device includes a WiFi probe device, a POE module, and a server. The WiFi probe device is used to detect the device MAC; The POE module is used to supply power to the WiFi probe device and transmit data back to the server at the same time.

6. The trajectory collision system based on multiple verification according to claim 5, characterized in that: The server includes a database server and a positioning server. The database server is used to store the detected device MAC information. The positioning server is used to perform positioning calculation on the data stored in the database server and store the location information corresponding to the MAC information.

Images