Electric bicycle digital identity information and multi-video identification data matching method

Abstract

According to the electric bicycle digital identity information and multi-video identification data matching method provided by the invention, one or more key point locations are set in the monitoring target area, the radio frequency identification equipment is set at each key point location, the radio frequency identification equipment does not need to be installed at other point locations, and only a low-cost pure video camera needs to be installed to capture video images, so that the monitoring efficiency is greatly improved. The digital identity information of the electric bicycle in the monitoring target area can be effectively identified by sharing the data acquired by the approval reading equipment of the key point location, so that the implementation cost of the scheme is greatly reduced.

Description

Technical Field

[0001] The invention relates to the technical field of electronic identity recognition of electric bicycles, and in particular to a method for matching digital identity information of electric bicycles with multi-video recognition data. Background Art

[0002] The current digital license plate for electric bicycles is designed based on 920MHz~925MHz ultra-high frequency RFID technology. The electronic identity information of the electric bicycle, such as the electronic identity code and license plate, is written into the ultra-high frequency RFID chip and embedded into the digital license plate of the electric bicycle. At the same time, the chip information and license plate information are bound and associated when the electric bicycle is registered. Therefore, the layout and chip of the digital license plate of the electric bicycle contain the license plate number at the same time.

[0003] With the popularization of traffic video surveillance equipment, there are currently a large number of video cameras installed on the road. Many cities have used video recognition to determine the license plate number and behavior of electric bicycles. However, the pure video method can only identify the license plate and the behavior of electric bicycles, and cannot determine the electronic identity of electric bicycles. In addition, due to the clustering of electric bicycles, the license plate characters are generally small, resulting in a low recognition success rate. There are also technicians who combine ultra-high frequency RFID and video recognition to improve the recognition effect. For example, the patent I proposed and is in use is application number: 202111207455.7, which discloses a method for identifying the digital identity information of electric bicycles. Radio frequency reading and radio frequency video surveillance equipment are set at traffic intersections in the city to identify the license plates of electric bicycles passing through the intersections, and the identified radio frequency data and video recognition data are queued. Combined with the full matching operation and fuzzy matching operation methods, the recognition effect of the digital identity of electric bicycles is greatly improved.

[0004] However, in actual use, it is found that the current method of identifying the digital identity of electric bicycles is to collect radio frequency data and video data in real time, and then combine the two data to match and identify the digital identity through calculation in the front-end device; for this purpose, radio frequency identification and video identification modules must be set up at each intersection, or radio frequency and video integrated identification equipment must be installed. This leads to a high implementation cost of the current method, which is not suitable for all intersections in all cities. Summary of the invention

[0005] In order to solve the problem of high cost of existing electric bicycle digital identity recognition methods, the present invention provides an electric bicycle digital identity information and multi-video recognition data matching method, which can use existing video equipment to realize the recognition of electric bicycle digital identity information in a specified area at a relatively low cost.

[0006] The technical solution of the present invention is as follows: a method for matching digital identity information of an electric bicycle with multi-video recognition data, characterized in that it comprises the following steps: S1: confirming the monitoring target area and determining key points within the monitoring target area; The key points include: intersections and any point on the road section; S2 sets a radio frequency reader for reading radio frequency data of the electric bicycle at the key point; The identification range of the radio frequency recognition device covers the electric bicycle lanes, and the electronic identity information in the digital license plates of all electric bicycles within the identification range is obtained by radio frequency recognition, and the license plate numbers of the electric bicycles are obtained by parsing; the license plates of the electric bicycles collected by the radio frequency recognition device are recorded as: radio frequency recognition license plates; At each of the key points, a video recognition device for identifying an electric bicycle is set according to a plurality of points of an associated road section within a specified range; the recognition range of the video recognition device covers the electric bicycle lane, and the license plate number of the electric bicycle can be captured and recognized by the video recognition device; the license plate number of the electric bicycle recognized by the video recognition device is recorded as: video recognition license plate; S3: Establishing data connections between all the radio frequency recognition devices and the video recognition devices and the backend server; S4: The video identification license plate and the radio frequency identification license plate collected by all the video identification devices and the radio frequency identification devices in real time are aggregated to the backend server in real time; A status mark is set for each of the video identification license plate and the radio frequency identification license plate; the status mark includes: no match, fuzzy match success, complete match success and match failure; the default value of the status mark is no match; S5: Among the license plates whose status mark is not a successful match, obtain the video recognition license plates one by one, and record them as: video license plates to be recognized; Read the identification time t1 of the license plate to be identified; set a matching time period T1 according to a preset time range; The one-time matching time period T1 is the average time for electric bicycles to pass through the monitoring target area; S6: Acquire all the RFID license plates whose license plate recognition time is within the time period T1, which are recorded as: a set of RFID license plates to be matched; S7: Fully matching the video license plate to be identified with each RFID license plate in the set of RFID license plates to be matched; If there is at least one RFID license plate that fully matches the video license plate to be identified, the status flags of the video license plate to be identified and the RFID license plate that fully matches the video license plate are set to successful matching; executing step S9; Otherwise, execute step S8; S8: Performing fuzzy matching operations on the video license plate to be identified and each RFID license plate in the set of RFID license plates to be matched; If there is at least one RFID license plate that satisfies the fuzzy matching condition with the video license plate to be identified, the status flag bits of the video license plate to be identified and the RFID license plate that has been fuzzily matched with it are both set to: fuzzy matching successful, and step S9 is executed; Otherwise, the status flag of the video license plate to be identified is set to: matching failed; the video license plate to be identified is recorded as: matching video data again, and step S10 is executed; S9: constructing a radio frequency data sharing database in the background server; The collected radio frequency identification plates with status marks of complete match success and fuzzy match success in all the monitoring target areas are stored in the radio frequency data sharing database; A time stamp is set for each of the radio frequency identification tags in the radio frequency data sharing database; the time stamp records the update time of the radio frequency identification tag stored in the radio frequency data sharing database; S10: Read the RFID license plate updated in the secondary matching time period T2 in the RFID data sharing database according to the preset secondary matching time period T2; wherein T2>T1; the secondary matching time T2 is set to the longest storage time of the vehicle license plate in the RFID data sharing database to maintain timeliness; If the data read is not empty, the read RFID license plate set is recorded as: re-matching RFID data set; executing step S11; If the read data is empty, execute step S12; S11: performing a complete matching operation on the re-matched video data and each RFID number plate in the re-matched RFID data set; If the complete matching operation is successful, the status flag bit of the re-matched video data is set to: complete matching success; If the full matching operation fails, the re-matching video data is subjected to a fuzzy matching operation with each RFID number plate in the re-matching RFID data set, and if the fuzzy matching operation succeeds, the status flag of the re-matching video data is set to: fuzzy matching success; Otherwise, execute step S12; S12: All the status flags of the video license plates to be identified that have failed matching are set to be put into the set of failed matching license plates, and this round of identification operation is ended.

[0007] It is further characterized by: For each radio frequency identification number plate stored in the radio frequency data sharing database, the following steps are performed: a1: monitor each RFID tag in real time, and once the RFID tag is completely matched or fuzzily matched, update the corresponding time mark to the latest time of complete match or fuzzy match; a2: Set the data validity period threshold; The data validity period threshold value indicates the validity period of the data in the radio frequency data sharing database; if the data validity period threshold value is exceeded, it indicates that the data has expired; a3: regularly checking the radio frequency data sharing database, finding the radio frequency identification plate whose time stamp has not been updated for a period exceeding the data validity period threshold, and deleting it; The complete match condition is: the characters and order of the two license plates involved in the match are exactly the same; The fuzzy matching conditions include: Condition 1: The number of identical characters in the RFID license plate and the video recognition license plate is greater than or equal to the preset value; Condition 2: The order of the main characters in the RFID license plate and the video recognition license plate must be the same; If the above two conditions are met at the same time, it is considered to meet the fuzzy matching conditions; The setting of the key points meets the following conditions: The ratio of the total number of electric bicycles passing through all the key points to the total number of electric bicycles collected by all video recognition devices in the target area to be monitored is greater than or equal to a preset coverage threshold; The key point setting method is: c1: Confirm all installed video recognition devices in the monitoring target area, denoted as: A 1 , A 2 , A 3 .....A n ; Where n is the total number of existing video recognition devices in the monitoring target area; c2: specify a start time ts, starting from ts, within the time range T1, count all the video recognition license plates collected by n video recognition devices, and build a video recognition result set for each video recognition device; c3: De-duplicate the data in each video recognition result set so that all the electric bicycle license plates in the set are unique; after de-duplicate, count the number of the video recognition license plates in each video recognition result set, recorded as: Sum 1 、Sum 2 、Sum 3 ...... Sum n ; c4: Sort all video recognition result sets from large to small according to the number of video recognition license plates included, denoted as: B 1 , B 2 , B 3 ......B n , the number of video recognition license plates in the corresponding set is: Reorder 1 、Reorder 2 、Reorder 3 ...... Reorder n ; c5: Set B 1 ~B n The video identification plates are placed in the collection TEMP 1 ~TEMP n middle; Initialize parameter i=1; c6: Remove TEMP i , get TEMP i All video identification license plates included in; In TEMP i+1 ~TEMP n Delete TEMP i All video identification license plates included in the video, removing duplicate license plates; c7: Sort all the video recognition result sets according to the number of video recognition license plates included from large to small, and obtain a re-sorted set TEMP 1 ~TEMP n ; i=i+1; c8: looping through steps c6 and c7 until there are no duplicate license plates in all the video recognition result sets; All video recognition result sets are re-sorted from large to small according to the number of video recognition license plates included, and are denoted as: C 1 , C 2 , C 3 ......C n , the number of video identification license plates in the corresponding set is: Final 1、Final 2 、Final 3 ...... Final n ; c9: Count the total number of video recognition license plates included in all the video recognition result sets: Total; Total = Final 1 +Final 2 +...... +Final n ; Initialize parameter j=1; c10: Calculate coverage; Coverage = C j / Total; c11: Compare Coverage and coverage threshold CT; If Coverage ≥ CT, execute step c15; Otherwise, execute step c12; c12: j = j + 1; c13:Coverage = (C 1 +....+C j ) / Total; c14: Loop through steps c11 to c13; c15: C 1 ~C j Set as the key point; The coverage threshold is set to 80%.

[0008] The present application provides a method for matching digital identity information of electric bicycles with multi-video recognition data, which sets one or more key points in the monitoring target area, sets a radio frequency reading device at each key point, and does not need to install radio frequency reading devices at other points. It only needs to install low-cost pure video cameras to capture video images. By sharing the data collected by the approval reading devices at the key points, the digital identity information of electric bicycles in the monitoring target area can be effectively identified, which greatly reduces the implementation cost of the scheme. In this method, the video license plate to be identified in the background server is first matched in the first matching time period T1, and the radio frequency reading license plate set uploaded by the radio frequency reading device installed at each key point is specified. If the match fails, the radio frequency reading license plate set is collected in the radio frequency data sharing database according to the preset secondary matching time period T2 for the second round of matching. By sharing the radio frequency data collected by the radio frequency reading device in the monitoring target area, the recognition rate of the video recognition license plate is effectively improved. In this method, all successfully matched RFID license plates are stored in the RFID data sharing database. If the match is successful, it means that the electric bicycle corresponding to the license plate is an active vehicle in the detection target area, and it can be stored in the shared database, rather than storing all the data collected by the RFID reading device in the database, which effectively controls the amount of data stored in the shared database, improves the calculation efficiency of the second round of matching, and makes this method suitable for real-time monitoring application scenarios. At the same time, in this method, a time mark is set for each RFID license plate, and a data validity period threshold is set, the data in the RFID data sharing database is screened, and the invalid data is deleted from the database in a timely manner, ensuring that only active electric bicycle license plates can be saved in the database for calculation, further effectively controlling the validity of the data in the shared database. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A flowchart of a method for matching digital identity information of an electric bicycle with multi-video recognition data; DETAILED DESCRIPTION

[0010] Because the RFID license plate data collected by the RFID reader has a high accuracy rate, and the video recognition device has a wide installation range. This application is an improved solution based on the existing technical solution of the electric bicycle digital identity information recognition method of the institute, which is used to reduce the overall implementation cost. Therefore, this method still combines the license plate data collected by the RFID reader and the data collected by the video recognition device to jointly complete the recognition of the electric bicycle digital identity information.

[0011] The electronic identity of an electric bicycle in this application refers to the digital identity data used to prove the uniqueness, legality, and authenticity of the electric bicycle, which generally includes information such as the electronic identity code, license plate number, and identification serial number. The digital license plate of an electric bicycle refers to a digital license plate that has an electronic identity chip embedded in the license plate of the electric bicycle and has both the ultra-high frequency RFID radio frequency performance and the appearance of an electric bicycle license plate. Electric bicycles that are not equipped with an electric bicycle digital license plate are not within the scope of application of this patent.

[0012] like Figure 1 As shown, the present application provides a method for matching digital identity information of an electric bicycle with multi-video recognition data, which includes the following steps.

[0013] S1: Confirm the monitoring target area and determine the key points within the monitoring target area; In actual operation, the RFID reader can be set at the intersection, the monitoring point where the video reader has been installed, the identity recognition point, the RFID reader point, or any other location on the road section. Usually, in order to ensure that there are enough electric bicycles passing through the key point, the intersection is selected as the key point. The number of key points KEYn satisfies the condition: KEYn≥1.

[0014] A matching time period T1 is set according to the size of the monitoring target area and the traffic conditions. A matching time period T1 is the average time for an electric bicycle to pass through the monitoring target area.

[0015] Usually, the maximum speed limit for electric bicycles in urban areas is 25km / h. In actual operation, based on the historical monitoring data of electric bicycles in the monitoring target area in the past and within the past six months, the average speed of the electric bicycles passing through the monitoring target area is calculated, and then the average time for the electric bicycles to pass through the monitoring target area is calculated, and then the maximum value is taken from all the times that pass through the monitoring target area under various routes. In this embodiment, T1 is set to 30 minutes, that is, based on the average speed, the electric bicycle can pass through the monitoring target area from one end of the monitoring target area to the other end under the longest route in 30 minutes.

[0016] Assuming that the maximum historical speed in the monitoring target area is 50km / H, the actual route distance between the RFID device set at any key point and all devices in the monitoring target area should be within the distance range of 50km / h*T1.

[0017] In this application, the setting of key points meets the following conditions: The ratio of the total number of electric bicycles passing through all key points to the total number of electric bicycles collected by all video recognition devices in the target area to be monitored is greater than or equal to a preset coverage threshold.

[0018] In this application, by selecting key points, RFID devices are only set at key points, and the RFID devices at key points are shared within the monitoring target area, without the need to set up RFID devices at all monitoring points, ensuring that the data collected by the device identification device can cover the monitoring area and reduce the overall cost.

[0019] The specific method for setting key points is as follows.

[0020] c1: Confirm all installed video recognition devices in the monitoring target area, denoted as: A 1 , A 2 , A 3 .....A n ; Where n is the total number of existing video recognition devices in the monitoring target area.

[0021] c2: Specify a start time ts. Starting from ts, within the time range T1, count all video recognition license plates collected by n video recognition devices, and build a video recognition result set for each video recognition device.

[0022] c3: Remove duplicate data from each video recognition result set, so that all electric bicycle license plates in the set are unique; after removing duplicate data, count the number of video recognition license plates in each video recognition result set, recorded as: Sum 1 、Sum 2 、Sum 3 ...... Sum n .

[0023] c4: Sort all video recognition result sets from large to small according to the number of video recognition license plates included, denoted as: B 1 , B 2 , B 3 ......B n , the number of video recognition license plates in the corresponding set is: Reorder 1 、Reorder 2 、Reorder 3 ...... Reorder n .

[0024] c5: Set B 1 ~B n The video identification plates are placed in the collection TEMP 1 ~TEMP n middle; Initialize parameter i=1.

[0025] c6: Remove TEMP i , get TEMP i All video identification license plates included in; In TEMP i+1 ~TEMP n Delete TEMP i All video identification license plates included in the video are removed.

[0026] During the process of removing duplicate license plates, if TEMP i+1 ~TEMP n If all the license plates in the set are deleted, the data in the set will be set to empty and the corresponding number of license plates will be set to 0.

[0027] c7: Sort all video recognition result sets according to the number of video recognition license plates included from large to small, and obtain the re-sorted set TEMP 1 ~TEMP n ; i=i+1.

[0028] c8: loop through steps c6 and c7 until there are no duplicate license plates in all video recognition result sets; All video recognition result sets are re-sorted from large to small according to the number of video recognition license plates included, and are denoted as: C 1 , C 2 , C 3 ......C n , the number of video identification license plates in the corresponding set is: Final 1 、Final 2 、Final 3 ...... Final n .

[0029] c9: count the total number of video recognition license plates included in all video recognition result sets: Total; Total = Final 1 +Final 2 +...... +Final n ; Initialize parameter j=1.

[0030] c10: Calculate coverage; Coverage = C j / Total.

[0031] c11: Compare Coverage and coverage threshold CT; If Coverage ≥ CT, execute step c15; Otherwise, execute step c12.

[0032] c12: j = j + 1.

[0033] c13:Coverage = (C 1 +....+C j ) / Total.

[0034] c14: Loop through steps c11 to c13.

[0035] c15: C 1 ~C j Set as key point.

[0036] For example, there are 30 points in the monitoring target area where video recognition equipment is installed, n=30. After calculation, 8 of them are found to meet the conditions: the ratio of the total number of electric bicycles passing through all these points to the total number of electric bicycles collected by all video recognition equipment in the target area to be monitored is greater than or equal to the preset coverage threshold, then these 8 points are set as key points.

[0037] The coverage rate threshold CT is set according to the monitoring requirements of the monitoring target area and can be set between [50%, 100%]. In this embodiment, the coverage rate threshold is set to 80%.

[0038] S2 sets a radio frequency reading device at a key point for reading radio frequency data of an electric bicycle.

[0039] The identification range of the radio frequency reading device covers the electric bicycle lanes. The electronic identity information in the digital license plates of all electric bicycles within the identification range is obtained by radio frequency reading, and the license plate numbers of the electric bicycles are obtained by parsing. The license plates of the electric bicycles collected by the radio frequency reading device are recorded as: radio frequency reading license plates; At each key point, a video recognition device for identifying electric bicycles is set up according to multiple points of the associated road section within a specified range; the recognition range of the video recognition device covers the electric bicycle lane, and the license plate number of the electric bicycle can be captured and recognized by the video recognition device; the license plate number of the electric bicycle recognized by the video recognition device is recorded as: video recognition license plate.

[0040] The RFID reading equipment in this application generally adopts a road reading and writing device with ultra-high frequency RFID reading function, or has both video recognition and RFID reading functions, which complies with the "RFID and video integrated recognition equipment" of GA / T 1976-2021. The equipment reading range covers the electric bicycle lane or the motor vehicle lane that needs to be monitored, facing the rear of the electric bicycle in the direction of travel. The actual operating frequency of the RFID reading equipment is 920~925MHz, and the ultra-high frequency RFID reading must comply with the GB / T 29768 air interface protocol to read the electronic identity code, license plate number and other information of the chip in the digital license plate of the electric bicycle.

[0041] The video recognition device in this application generally directly captures the electric bicycle within the video recognition range, and identifies the data information of its license plate number through the existing image recognition algorithm. During the video recognition process, the identification area can be specified according to the specific application. Taking the red light running snapshot application in traffic management as an example, the video recognition range can be set to the area where the electric bicycle exceeds the stop line when the red light is on. The video recognition device captures the electric bicycle when it enters the range and identifies its license plate number. Therefore, in addition to the video recognition license plate, the data recognized by the video recognition device generally also includes panoramic images of electric bicycles, behavioral images and other information.

[0042] S3: Establish data connection between all radio frequency reading devices and video recognition devices and the background server.

[0043] In this application, the data collected by the radio frequency reading devices installed at key points are reused, so all data must be aggregated to the background server for shared use.

[0044] S4: All video recognition license plates and radio frequency recognition license plates collected by video recognition devices and radio frequency recognition devices are aggregated to the backend server in real time; A status mark is set for each video identification license plate and radio frequency identification license plate; the status mark includes: no match, fuzzy match success, complete match success and match failure; the default value of the status mark is no match.

[0045] In the present application, a status mark is set for each license plate to mark the identification status of the license plate and the timeliness of the license plate.

[0046] S5: among the license plates whose status mark is not a successful match, obtain the video recognition license plates one by one, and record them as: video license plates to be recognized; Read the identification time t1 of the license plate to be identified; set a matching time period T1 according to the preset time range.

[0047] S6: Obtain all RFID license plates whose license plate recognition time is within the time period T1, recorded as: the set of RFID license plates to be matched.

[0048] S7: Completely match the video license plate to be identified with each RFID license plate in the set of RFID license plates to be matched.

[0049] The exact match condition is that the characters and sequence of the two license plates involved in the match are exactly the same.

[0050] If there is at least one RFID license plate that satisfies the complete matching condition with the video license plate to be identified, the status flag bits of the video license plate to be identified and the RFID license plate that completely matches it are set to successful matching; executing step S9; Otherwise, execute step S8.

[0051] S8: Perform fuzzy matching operations on the video license plate to be identified and each radio frequency identification license plate in the set of radio frequency license plates to be matched.

[0052] Fuzzy matching conditions include: Condition 1: The number of identical characters in the RFID license plate and the video recognition license plate is greater than or equal to the preset value; Condition 2: The order of the main characters in the RFID license plate and the video recognition license plate must be the same; If the above two conditions are met at the same time, it is considered to meet the fuzzy matching conditions; If there is at least one RFID license plate that satisfies the fuzzy matching condition with the video license plate to be identified, the status flags of the video license plate to be identified and the RFID license plate that has been fuzzily matched with it are both set to: fuzzy matching successful, and step S9 is executed; Otherwise, the status flag of the video license plate to be identified is set to: matching failed; the video license plate to be identified is recorded as: matching video data again, and executing step S10.

[0053] In this embodiment, the fuzzy matching condition is set as follows: only one character on the license plate can be different, and the rest of the characters are the same and located in the same position. Scenario 1: The RFID license plate is "Su A12345", the video recognition license plate is "A12355", and the fuzzy matching threshold is 1. According to the set conditions, there is only one digit difference, and the fuzzy matching is successful.

[0054] Scenario 2: The RFID license plate is "Su A12345", the video recognition license plate is "A12356", and the fuzzy matching threshold is 1. According to the set conditions, there are 2 differences and the fuzzy matching fails.

[0055] Of course, the actual fuzzy matching conditions need to be adjusted based on the success rate and effect of video recognition data, the number of local electric bicycles, the number of digits in the license plate number, etc.

[0056] Based on the fuzzy matching operation, the success rate of recognizing the license plates of electric bicycles in the same area will be greatly improved. It is especially suitable for the situation where the license plates are blocked due to clustering of electric bicycles and the license plate characters are small, ensuring that the technical solution of the present invention is more practical.

[0057] The specific implementation process of complete matching and fuzzy matching can be implemented based on existing technologies.

[0058] S9: In the background server, a radio frequency data sharing database is constructed; In all monitoring target areas, the collected RFID license plates with status marks of complete match success and fuzzy match success are stored in the RFID data sharing database; A time stamp is set for each RFID license plate in the RFID data sharing database; the time stamp records the update time when the RFID license plate is stored in the RFID data sharing database; S10: according to the preset secondary matching time period T2, reading the RFID number plate updated within the secondary matching time period T2 in the RFID data sharing database; wherein T2>T1; If the data read is not empty, the read RFID license plate set is recorded as: re-matching RFID data set; executing step S11; If the read data is empty, execute step S12.

[0059] The secondary matching time T2 is set to the longest storage time for the vehicle license plate to maintain validity in the radio frequency data sharing database.

[0060] The present application controls the validity of the identified electric bicycle license plate through T2, and the specific value of T2 can be set to the average value of historical data or can be set according to implementation needs. In this embodiment, T2 is set to 7 days.

[0061] S11: performing a complete matching operation on the rematched video data and each RFID number plate in the rematched RFID data set; If the complete matching operation is successful, the status flag bit of the rematched video data is set to: complete matching success; If the complete matching operation fails, the rematching video data is subjected to a fuzzy matching operation with each RFID number plate in the rematching RFID data set. If the fuzzy matching operation succeeds, the status flag of the rematching video data is set to: fuzzy matching success; Otherwise, execute step S12.

[0062] S12: All the status flags of the video license plates to be identified that have failed matching are set to be put into the set of failed matching license plates, and this round of identification operation is ended.

[0063] The video recognition license plate that fails to match can be matched again with the license plates in the matching license plate database, and the electric bicycle digital license plates that are completely matched and fuzzy matched with the matching license plate database are screened out, and the license plate status is updated. The specific process can be operated based on the method in the existing technical solution.

[0064] In specific implementation, since electric bicycles are likely to have a certain degree of repetitiveness when driving on the road, that is, they often pass a certain point at a fixed time every day or every week, so establishing a database of successfully matched license plates has a high practical value for assisting license plate matching. In the event that the electric bicycle video recognition license plate fails to match within the T1 time period, there is a way to further verify its electronic identity and match its license plate information.

[0065] For example: a certain point M in the monitoring target area is a video recognition point, but it is not an important point and no radio frequency reading equipment is installed. The license plate number of an electric bicycle is A23456. It often passes through point M, and within the previous week, its license plate matching data has been added to the database of successfully matched license plates, and the database retention time period has not exceeded. The next time the electric bicycle passes through point M, the video recognition device identifies its license plate as A23459, but does not read its radio frequency reading license plate containing electronic identity information from important points (the electric bicycle did not pass through important points this time). The video recognition license plate can be fully matched or fuzzily matched through the database of successfully matched license plates to verify its electronic identity, and correct its matched license plate number to "A23456".

[0066] In order to ensure the timeliness of the license plates stored in the shared database, the following steps are performed for each RFID license plate stored in the RFID data sharing database: a1: Monitor each RFID license plate in real time. Once the RFID license plate is completely matched or fuzzily matched, update its corresponding time stamp to the latest complete match or fuzzy match time; a2: Set the data validity period threshold; The data validity period threshold indicates the validity period of data in the radio frequency data sharing database; if the data validity period threshold is exceeded, it means that the data has expired; a3: Regularly check the RFID data sharing database, find and delete the RFID license plates whose time stamp has not been updated for a period exceeding the data validity period threshold.

[0067] After using the technical solution of the present invention, key points are selected as important points, and other video installation positions are selected as ordinary points. The two are combined, and the radio frequency reading equipment set at the key points is used to match, calibrate, and enhance the image data captured by several low-cost pure video cameras that have been built or newly built at the surrounding upstream and downstream intersections, and associate the electronic identity information of the electric bicycle. Matching the electronic identity information of the electric bicycle with the multi-point video recognition data can not only verify the identity data information of the electric bicycle, but also further enhance the accuracy and success rate of license plate number recognition. At the same time, it also makes full use of the existing road video recognition equipment, reduces the cost of equipment installation, and can provide great convenience for the management of electric bicycles.

Claims

1. A method for matching digital identity information of an electric bicycle with multi-video recognition data, characterized in that: It includes the following steps: S1: confirming the monitoring target area and determining key points within the monitoring target area; The key points include: intersections and any point on the road section; S2 sets a radio frequency reader for reading radio frequency data of the electric bicycle at the key point; The identification range of the radio frequency recognition device covers the electric bicycle lanes, and the electronic identity information in the digital license plates of all electric bicycles within the identification range is obtained by radio frequency recognition, and the license plate numbers of the electric bicycles are obtained by parsing; the license plates of the electric bicycles collected by the radio frequency recognition device are recorded as: radio frequency recognition license plates; At each of the key points, a video recognition device for identifying an electric bicycle is set according to a plurality of points of an associated road section within a specified range; the recognition range of the video recognition device covers the electric bicycle lane, and the license plate number of the electric bicycle can be captured and recognized by the video recognition device; the license plate number of the electric bicycle recognized by the video recognition device is recorded as: video recognition license plate; S3: Establishing data connections between all the radio frequency recognition devices and the video recognition devices and the backend server; S4: The video identification license plate and the radio frequency identification license plate collected by all the video identification devices and the radio frequency identification devices in real time are aggregated to the backend server in real time; A status mark is set for each of the video identification license plate and the radio frequency identification license plate; the status mark includes: no match, fuzzy match success, complete match success and match failure; the default value of the status mark is no match; S5: Among the license plates whose status mark is not a successful match, obtain the video recognition license plates one by one, and record them as: video license plates to be recognized; Read the identification time t1 of the license plate to be identified; set a matching time period T1 according to a preset time range; The one-time matching time period T1 is the average time for electric bicycles to pass through the monitoring target area; S6: Acquire all the RFID license plates whose license plate recognition time is within the time period T1, which are recorded as: a set of RFID license plates to be matched; S7: Fully matching the video license plate to be identified with each RFID license plate in the set of RFID license plates to be matched; If there is at least one RFID license plate that fully matches the video license plate to be identified, the status flags of the video license plate to be identified and the RFID license plate that fully matches the video license plate are set to successful matching; executing step S9; Otherwise, execute step S8; S8: Performing fuzzy matching operations on the video license plate to be identified and each RFID license plate in the set of RFID license plates to be matched; If there is at least one RFID license plate that satisfies the fuzzy matching condition with the video license plate to be identified, the status flag bits of the video license plate to be identified and the RFID license plate that has been fuzzily matched with it are both set to: fuzzy matching successful, and step S9 is executed; Otherwise, the status flag of the video license plate to be identified is set to: matching failed; the video license plate to be identified is recorded as: matching video data again, and step S10 is executed; S9: constructing a radio frequency data sharing database in the background server; The collected radio frequency identification plates with status marks of complete match success and fuzzy match success in all the monitoring target areas are stored in the radio frequency data sharing database; A time stamp is set for each of the radio frequency identification tags in the radio frequency data sharing database; the time stamp records the update time of the radio frequency identification tag stored in the radio frequency data sharing database; S10: Read the RFID license plate updated in the secondary matching time period T2 in the RFID data sharing database according to the preset secondary matching time period T2; wherein T2>T1; the secondary matching time T2 is set to the longest storage time of the vehicle license plate in the RFID data sharing database to maintain timeliness; If the data read is not empty, the read RFID license plate set is recorded as: re-matching RFID data set; executing step S11; If the read data is empty, execute step S12; S11: performing a complete matching operation on the re-matched video data and each RFID number plate in the re-matched RFID data set; If the complete matching operation is successful, the status flag bit of the re-matched video data is set to: complete matching success; If the full matching operation fails, the re-matching video data is subjected to a fuzzy matching operation with each RFID number plate in the re-matching RFID data set, and if the fuzzy matching operation succeeds, the status flag of the re-matching video data is set to: fuzzy matching success; Otherwise, execute step S12; S12: All the video license plates to be identified with status flags set to failed matching are put into a set of failed matching license plates, and this round of identification operation is ended.

2. According to claim 1, a method for matching digital identity information of an electric bicycle with multi-video recognition data, characterized in that: For each radio frequency identification number plate stored in the radio frequency data sharing database, the following steps are performed: a1: monitor each RFID tag in real time, and once the RFID tag is completely matched or fuzzily matched, update the corresponding time mark to the latest time of complete match or fuzzy match; a2: Set the data validity period threshold; The data validity period threshold value indicates the validity period of the data in the radio frequency data sharing database; if the data validity period threshold value is exceeded, it indicates that the data has expired; a3: Regularly check the radio frequency data sharing database, find the radio frequency identification plate whose time stamp has not been updated for a period exceeding the data validity period threshold, and delete it.

3. According to claim 1, a method for matching digital identity information of an electric bicycle with multi-video recognition data is characterized in that: The complete matching condition is that the characters and sequence of the two license plates involved in the matching are exactly the same.

4. According to claim 1, a method for matching digital identity information of an electric bicycle with multi-video recognition data, characterized in that: The fuzzy matching conditions include: Condition 1: The number of identical characters in the RFID license plate and the video recognition license plate is greater than or equal to the preset value; Condition 2: The order of the main characters in the RFID license plate and the video recognition license plate must be the same; If the above two conditions are met at the same time, it is considered to meet the fuzzy matching conditions.

5. According to claim 1, a method for matching digital identity information of an electric bicycle with multi-video recognition data, characterized in that: The setting of the key points meets the following conditions: The ratio of the total number of electric bicycles passing through all the key points to the total number of electric bicycles collected by all video recognition devices in the target area to be monitored is greater than or equal to a preset coverage threshold.

6. According to claim 1, a method for matching digital identity information of an electric bicycle with multi-video recognition data, characterized in that: The key point setting method is: c1: Confirm all installed video recognition devices in the monitoring target area, which are recorded as: A1, A2, A3...A n ; Where n is the total number of existing video recognition devices in the monitoring target area; c2: specify a start time ts, starting from ts, within the time range T1, count all the video recognition license plates collected by n video recognition devices, and build a video recognition result set for each video recognition device; c3: De-duplicate the data in each of the video recognition result sets, so that all the electric bicycle license plates in the set are unique; after de-duplicate, count the number of the video recognition license plates in each of the video recognition result sets, and record them as: Sum1, Sum2, Sum3... Sum n ; c4: Sort all video recognition result sets from large to small according to the number of video recognition license plates included, and record them as: B1, B2, B3...B n , the number of video recognition license plates in the corresponding set is: Reorder1, Reorder2, Reorder3... Reorder n ; c5: Set B1~B n The video identification license plates are placed in the collection TEMP1~TEMP n middle; Initialize parameter i=1; c6: Remove TEMP i , get TEMP i All video identification license plates included in; In TEMP i+1 ~TEMP n Delete TEMP i All video identification license plates included in the video, removing duplicate license plates; c7: Sort all the video recognition result sets according to the number of video recognition license plates included from large to small, and obtain the re-sorted sets TEMP1~TEMP n ; i=i+1; c8: looping through steps c6 and c7 until there are no duplicate license plates in all the video recognition result sets; All video recognition result sets are re-arranged from large to small according to the number of video recognition license plates included, and are recorded as: C1, C2, C3...C n , the number of video identification license plates in the corresponding set is: Final1, Final2, Final3... Final n ; c9: Count the total number of video recognition license plates included in all the video recognition result sets: Total; Total= Final1+Final2+...... +Final n ; Initialize parameter j=1; c10: Calculate coverage; Coverage = C j / Total; c11: Compare Coverage and coverage threshold CT; If Coverage ≥ CT, execute step c15; Otherwise, execute step c12; c12: j = j + 1; c13:Coverage =(C1+....+C j ) / Total; c14: Loop through steps c11 to c13; c15: C1~C j Set to the key point.

7. According to claim 5, a method for matching digital identity information of an electric bicycle with multi-video recognition data, characterized in that: The coverage threshold is set to 80%.

Images