A method for identifying vehicles parked in internal passages of hub parking lots

Abstract

The present invention discloses a method for identifying vehicles parked in the internal passage of a hub parking lot, including step 1, obtaining dynamic and static data of a target parking lot; step 2, generating a shortest path time table; step 3, establishing a candidate set of vehicles parked in parking spaces; step 4, identifying the license plate numbers of parked vehicles; step 5, calculating and evaluating the recognition success rate; step 6, calculating and evaluating the average time error; step 7, identifying vehicles parked in the internal passage: comparing the occupancy data of the successfully identified parking spaces with the in-and-out data of the parking lot gate, and filtering the complement of the two to obtain the data of vehicles parked in the internal passage of the parking lot. The present invention can make up for the problem that the existing hub parking lot is not equipped with detection equipment specifically for vehicles parked in the internal passage, can support the data information identification of vehicles parked in the internal passage of the hub parking lot, and can provide data support for the characteristic analysis of parking demand in the internal passage of the hub parking lot and further refined management and control.

Description

Technical Field

[0001] The invention relates to the field of urban parking traffic management and control, and in particular to a method for identifying vehicles parked in an inner passage of a hub parking lot. Background Art

[0002] With the deepening construction of a multi-level, integrated comprehensive transportation hub system, higher requirements are placed on the optimization of hub facilities. A hub is a type of transportation facility built to balance passenger flow when various modes of transportation coexist. As an important part of the hub, the normal operation of hub parking facilities plays a very important role in the hub. Parking facilities that lack convenience and efficiency will lead to negative impacts such as reduced driver willingness to use, increased difficulty in finding parking spaces, increased impedance of internal parking lots, and increased queue lengths at parking lot entrances and exits, which will in turn worsen road traffic conditions around the hub.

[0003] At present, the Internet and the sharing economy are developing rapidly. In order to improve the convenience of passengers taking taxis and getting on and off, hub parking facilities choose to set up temporary parking areas or temporary parking passages inside the parking lot to facilitate efficient pick-up and drop-off of passengers. This increases the demand for passage parking while there is a demand for parking spaces inside the hub parking lot. During peak demand periods, the space required for passage parking demand exceeds the temporary parking area set up in the parking lot, thereby occupying the driving space of vehicles normally looking for parking in the parking lot. At the same time, vehicles parked in the passage will affect the effective use of vacant parking spaces, resulting in a decrease in the operating efficiency of the entire parking lot. Therefore, it is necessary to study the number of passage parking demands.

[0004] For parking lots that have cameras installed throughout the internal passages of the parking lot and record the license plate numbers and other related information of vehicles parked in the passages, the identification of vehicles parked in the passages can be achieved directly through video detection related technologies; for parking lots that are equipped with video shooting functions above the parking spaces and can clearly capture vehicle license plate number detection equipment, vehicles parked in the passages can be directly screened out by comparing the data of the parking lot gates in and out. However, most hub parking lots are currently equipped with intelligent detection equipment including gate equipment at the entrances and exits of the parking lot and parking space detectors above the parking spaces in the parking lot, and are not equipped with detection equipment specifically for vehicles parked in the internal passages of the parking lot. Among them, the data recorded by the parking lot gates at the entrances and exits includes information such as the vehicle license plate number and the time of entry and exit, and the parking space occupancy data recorded by the parking space detector includes information such as the parking space number, the time of starting occupancy and the time of stopping occupancy (excluding the vehicle license plate number information). It is impossible to directly compare the two to screen out the data of vehicles parked in the internal passages of the parking lot.

[0005] Therefore, the present invention proposes a method for identifying vehicles parked in the internal channels of hub parking lots based on the parking lot gate entry and exit data containing license plate information and the parking space occupancy data not containing license plate information. The proposed method can identify the data information of vehicles parked in the internal channels of hub parking lots, provide data support for the characteristic analysis of parking demand in the internal channels of hub parking lots and further refined management and control, and is a key technology that urgently needs to be studied in the process of optimizing the operating efficiency of hub parking lots. Summary of the invention

[0006] The technical problem to be solved by the present invention is to address the deficiencies of the above-mentioned prior art and to provide a method for identifying vehicles parked in internal passages of a hub parking lot. The method for identifying vehicles parked in internal passages of a hub parking lot can solve the problem that the information of vehicles parked in internal passages of a hub parking lot is difficult to determine when the hub parking lot records gate entry and exit data containing license plate information and parking space occupancy data not containing license plate information.

[0007] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0008] A method for identifying vehicles parked in an internal passage of a hub parking lot includes the following steps.

[0009] Step 1, obtain the dynamic and static data of the target parking lot: first, select the hub parking lot with channel parking as the target parking lot; then, obtain the dynamic and static data of the target parking lot within the set survey period; among them, the dynamic and static data include static layout data and dynamic parking data; the dynamic parking data includes parking lot gate entry and exit data containing license plate number information and parking space occupancy data containing parking space number information but not license plate number information.

[0010] Step 2: Generate a shortest path time table: Based on the static layout data obtained in step 1, generate an ideal shortest path time table from each parking space to the parking lot entrance and exit.

[0011] Step 3: Establish a candidate set of vehicles parked in parking spaces: Read parking space occupancy data in chronological order, identify qualified vehicle records from the parking lot gate entry and exit data, and add them to the candidate set of vehicles parked in the corresponding parking spaces.

[0012] Step 4: Parked vehicle license plate number recognition: The license plate number of the parked vehicle is recognized for each parking space in the target parking lot. The specific recognition method is as follows:

[0013] A. When the number of vehicle records identified in the candidate set of parked vehicles in parking space i is 0, it is judged that the license plate number recognition is unsuccessful.

[0014] B. When there is one vehicle record identified in the candidate set of parked vehicles for parking space i, it is directly determined that the license plate number recognition is successful.

[0015] C. When there are two or more vehicle records identified in the candidate set of parked vehicles for parking space i, it is necessary to identify again, and the vehicle record with the highest probability is taken as the identification result, and it is judged that the license plate number recognition is successful.

[0016] Step 5. Calculate and evaluate the recognition success rate: Calculate the license plate number recognition success rate SR of all vehicles parked in parking spaces, and compare SR with the set license plate number recognition success rate threshold SR0; when SR ≥ SR0, the license plate number recognition result is considered acceptable, and proceed to step 6; otherwise, the license plate number recognition result is considered unacceptable, and repeat steps 1 to 5.

[0017] Step 6, calculate and evaluate the average time error: for all vehicles successfully identified in step 4, calculate the average value of the difference between the actual time and the ideal time of the shortest path in step 2, and then obtain the average time error; then, compare the average time error with the set time error threshold; when the average time error does not exceed the time error threshold, it is judged that the average time error is acceptable; otherwise, repeat steps 1 to 6.

[0018] Step 7, identifying vehicles parked in the internal channel: When the license plate number recognition result in step 5 is acceptable and the average time error in step 6 is acceptable, the successfully identified parking space occupancy data and the parking lot gate entry and exit data are compared, and the complement of the two is used to filter out the data of vehicles parked in the internal channel of the parking lot.

[0019] In step 2, the ideal time table of the shortest path includes the ideal value of the search time Arrive_ideal i and the ideal departure time Leave_ideal i ; Among them, Arrive_ideal i Leave_ideal is the ideal time for the vehicle to find the shortest path from the parking lot entrance to parking space i; i is the ideal time value of the shortest path from parking space i to the parking lot exit; in step 6, the average time error includes the average parking time error DIF_arrive average and average departure time error DIF_leave average .

[0020] In step 6, DIF_arrive average The calculation formula is:

[0021]

[0022] in:

[0023] DIF_arrive ij =(park_ini -time_in j )*24*60*60-Arrive_ideal i

[0024] Where DIF_arrive ij It is the time error of searching for parking space i and vehicle j corresponding to the successfully identified parking space occupancy data, unit: s.

[0025] S is the set of all successfully identified parking space occupancy data.

[0026] n is the number of parking space occupancy data that are successfully identified.

[0027] park_in i is the time when parking space i starts to be occupied.

[0028] time_in j is the time when vehicle j enters the target parking lot.

[0029] In step 6, DIF_leave average The calculation formula is:

[0030]

[0031] in:

[0032] DIF_leave ij =(time_out j -park_out i )*24*60*60-Leave_ideal i

[0033] In the formula, DIF_leave ij The time error between parking space i and vehicle j corresponding to the successfully identified parking space occupancy data, unit: s.

[0034] park_out i is the time when parking space i stops occupying.

[0035] time_out j is the time when vehicle j leaves the target parking lot.

[0036] In step 4C, the vehicle record with the highest probability of being identified is DIF ij The vehicle record corresponding to the minimum; among them, DIF ij The calculation formula is:

[0037] DIF ij =DIF_arrive ij+DIF_leave ij

[0038] Where DIF ij It is the sum of the time errors of searching and leaving parking space i for vehicle j, unit: s.

[0039] In step 5, the license plate number recognition success rate threshold SR0 is set to 80%.

[0040] In step 5, the calculation formula of SR is:

[0041]

[0042] Where N is the total number of parking space occupancy data.

[0043] In step 1, the static layout data includes the location of the parking lot entrance and exit, the parking space numbering layout, the ground markings inside the parking lot, the length of a single parking space, the width of a single parking space, the width of a single lane, and the thickness of the walls inside the parking lot.

[0044] The present invention has the following beneficial effects:

[0045] The present invention can make up for the problem that the existing hub parking lot is not equipped with detection equipment specifically for vehicles parked in the internal channels. It fully considers the current situation that the parking demand composition of hub social vehicle parking lots is becoming increasingly complex and the demand for channel parking is constantly increasing. Based on the gate entry and exit data containing license plate information and the parking space occupancy data not containing license plate information in the parking lot, an identification method for vehicles parked in the internal channels of the hub parking lot is formed.

[0046] Therefore, the present invention can support the data information identification of vehicles parked in the internal passages of the hub parking lot, and can provide data support for the characteristic analysis of parking demand in the internal passages of the hub parking lot and further refined management and control. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 The present invention is a flow chart of a method for identifying vehicles parked in an internal passage of a hub parking lot.

[0048] Figure 2 Schematic diagram of the internal layout of a parking lot according to an embodiment of the present invention.

[0049] Figure 3 This is a clustering result diagram of the data of vehicles parked in the internal passages of a parking lot according to an embodiment of the present invention (taking the data on June 17 as an example).

[0050] Figure 4 This is a parking demand prediction result diagram for the internal passage of a parking lot according to an embodiment of the present invention (daytime period).

[0051] Figure 5This is a parking demand prediction result diagram for the internal passage of a parking lot according to an embodiment of the present invention (night time period). DETAILED DESCRIPTION

[0052] The present invention is further explained below in conjunction with the accompanying drawings and specific embodiments. It should be understood that these examples are only used to illustrate the present invention and are not used to limit the scope of the present invention. After reading the present invention, various equivalent forms of modifications to the present invention by those skilled in the art all fall within the scope defined by the claims attached to this application.

[0053] like Figure 1 As shown, a method for identifying vehicles parked in an internal passage of a hub parking lot includes the following steps.

[0054] Step 1, obtain the dynamic and static data of the target parking lot: first, select the hub parking lot with channel parking as the target parking lot; then, through field investigation, obtain the dynamic and static data of the target parking lot within the set investigation period; among them, the dynamic and static data include static layout data and dynamic parking data; dynamic parking data includes parking lot gate entry and exit data containing license plate number information and parking space occupancy data containing parking space number information but not license plate number information.

[0055] Static layout data includes the location of parking lot entrances and exits, parking space numbering layout, ground markings inside the parking lot, length of a single parking space, width of a single parking space, width of a single lane, and thickness of the walls inside the parking lot.

[0056] In this embodiment, the target parking lot selected is a social vehicle parking lot in a hub in Nanjing. The parking lot has a total of 263 parking spaces, 1 entrance, 1 exit, and 3 online car-hailing waiting areas. The direction of the arrow on the ground is the driving direction of the channel, and the parking space number is the parking space number. The length of a single parking space is 5.5 meters, the width is 2.5 meters, the width of a single lane is 3.65 meters, and the thickness of other walls is obtained according to on-site measurements. The specific layout diagram is as follows: Figure 2 As shown in the figure, the gate entry and exit data and parking space occupancy data of the case parking lot are obtained, and the survey period is preferably set to 42 days.

[0057] The parking lot gate entry and exit data records the vehicle license plate number, the time of entry and exit, and the import and export information, and the parking space occupancy data records the parking space number, the detection time of the detection equipment, and the occupancy status. The gate entry and exit data are matched based on the license plate number to obtain the entry and exit time of each vehicle, and the parking space occupancy data is matched based on the parking space number to obtain the start and stop occupancy time of each parking space. Some matching results are shown in Tables 1 and 2.

[0058] Table 1 Parking lot gate in and out data (partial)

[0059]

[0060] Table 2 Parking space occupancy data (partial)

[0061]

[0062] Step 2: Generate a shortest path time table: Based on the static layout data obtained in step 1, generate an ideal shortest path time table from each parking space to the parking lot entrance and exit.

[0063] The above shortest path ideal time table includes the ideal value of the search time Arrive_ideal i and the ideal departure time Leave_ideal i ; Among them, Arrive_ideal i Leave_ideal is the ideal time for the vehicle to find the shortest path from the parking lot entrance to parking space i; i is the ideal time for the shortest path for a vehicle to leave parking space i and reach the parking lot exit.

[0064] Referring to the internal layout diagram of the target parking lot in the embodiment, according to the distribution positions of each parking space in the parking lot, the distribution positions of the parking lot entrance and exit, the parking spaces, passages, wall dimensions and ground signs and markings, the shortest paths from the parking lot entrance to each parking space and from the parking space to the parking lot exit are obtained under the premise that the vehicle follows the ground signs and markings. The average speed of the vehicle parking and leaving is obtained through on-site survey statistics, and the ideal time taken for the vehicle to park from the parking lot entrance to the parking space i is calculated. i , and the ideal time taken for a vehicle to leave parking space i and reach the parking lot exit Leave_ideal i , forming the ideal time consumption table of the shortest path, as shown in Table 3.

[0065] Table 3 Shortest path ideal time table (partial)

[0066]

[0067] Step 3: Establish a candidate set of vehicles parked in parking spaces: Read parking space occupancy data in chronological order, identify qualified vehicle records from the parking lot gate entry and exit data, and add them to the candidate set of vehicles parked in the corresponding parking spaces.

[0068] In the embodiment, Python code is written to traverse the matched parking lot gate entry and exit data and parking space occupancy data, and vehicle records whose entry time is earlier than the parking space start occupation time and exit time is later than the parking space stop occupation time are identified and added to the candidate set. The number of vehicle records that meet the conditions corresponding to a parking space occupancy data in the candidate set may be 0, 1, or multiple.

[0069] Step 4: Parked vehicle license plate number recognition: The license plate number of the parked vehicle is recognized for each parking space in the target parking lot. The specific recognition method is as follows:

[0070] A. When the number of vehicle records identified in the candidate set of parked vehicles in parking space i is 0, it is judged that the license plate number recognition is unsuccessful.

[0071] B. When there is one vehicle record identified in the candidate set of parked vehicles for parking space i, it is directly determined that the license plate number recognition is successful.

[0072] C. When there are two or more vehicle records identified in the candidate set of parked vehicles for parking space i, it is necessary to identify again, and the vehicle record with the highest probability is taken as the identification result, and it is judged that the license plate number recognition is successful.

[0073] The vehicle record with the highest probability is DIF ij The vehicle record corresponding to the minimum; among them, DIF ij The calculation formula is:

[0074] DIF ij =DIF_arrive ij +DIF_leave ij

[0075] in:

[0076] DIF_arrive ij =(park_in i -time_in j )*24*60*60-Arrive_ideal i

[0077] DIF_leave ij =(time_out j -park_out i )*24*60*60-Leave_ideal i

[0078] Where DIF ij It is the sum of the time errors of searching and leaving parking space i for vehicle j, unit: s.

[0079] DIF_arrive ij It is the error in the time taken to find a parking space i and a vehicle j in the successfully identified parking space occupancy data, in seconds.

[0080] DIF_leave ij The time error between parking space i and vehicle j in the successful parking space occupancy data is identified, in seconds.

[0081] park_in i is the time when parking space i starts to be occupied.

[0082] time_in j is the time when vehicle j enters the target parking lot.

[0083] park_out i is the time when parking space i stops occupying.

[0084] time_out j is the time when vehicle j leaves the target parking lot.

[0085] In this embodiment, take a piece of parking space occupancy data of parking space 9 as an example:

[0086] Arrive_ideal for parking space 9 i 12.1s, Leave_ideal i The time when parking space 9 starts to be occupied is 2022 / 6 / 15 7:24:00, and the time when it stops to be occupied is 2022 / 6 / 15 17:30:00. The vehicle information in the candidate set is shown in Table 4.

[0087] Table 4 Candidate set of vehicles parked in parking space 9

[0088]

[0089] According to the above formula, the time error of arriving and leaving the parking space for all vehicles in the candidate set of parking vehicles corresponding to the parking occupancy data is calculated.

[0090] Candidate Set 1 Vehicle:

[0091] DIF_arrive 91 =(park_in9-time_in1)*24*60*60-Arrive_ideal9=45s

[0092] DIF_leave 91 =(time_out1-park_out9)*24*60*60-Leave_ideal9=515s

[0093] DIF 91 =DIF_arrive 91 +DIF_leave 91 =45+515=560s

[0094] Candidate Set 2 Vehicle:

[0095] DIF_arrive 92 =40s

[0096] DIF_leave 92 =10s

[0097] DIF 92 =DIF_arrive 92 +DIF_leave 92 =40+10=50s

[0098] Candidate Set 3 Vehicle:

[0099] DIF_arrive 93 =36s

[0100] DIF_leave 93 =14s

[0101] DIF 93 =DIF_arrive 93 +DIF_leave 93 =36+14=50s

[0102] According to the calculation results, the time error of candidate set vehicles 2 and 3 in arriving and leaving the parking space is 50s, which is less than that of vehicle 1, and the time of candidate set vehicle 2 entering the parking lot is earlier than that of vehicle 3, so candidate set vehicle 2 is the final recognition result of this occupancy data of parking space 9.

[0103] That is, when the minimum DIF in the candidate set ij When there are two or more vehicles, the vehicle record that enters the target parking lot earliest is selected as the re-recognition result.

[0104] Step 5, calculate and evaluate the recognition success rate: calculate the license plate number recognition success rate SR of all vehicles parked in the parking spaces, and compare SR with the set license plate number recognition success rate threshold SR0 (preferably 80%); when SR≥SR0, the license plate number recognition result is considered acceptable, and proceed to step 6; otherwise, the license plate number recognition result is considered unacceptable, and repeat steps 1 to 5.

[0105] The calculation formula of the above SR is:

[0106]

[0107] Where n is the number of parking space occupancy data that are successfully identified.

[0108] N is the total number of parking space occupancy data.

[0109] In this embodiment, the recognition success rate of the target parking lot data is calculated by the following formula:

[0110]

[0111] According to the above calculation results, the recognition success rate of the target parking lot data in the embodiment is 82.756%, which exceeds the license plate number recognition success rate threshold SR0=80%. It is considered that the recognition success rate meets the requirements and the recognition result is acceptable.

[0112] Step 6, calculate and evaluate the average time error: for all vehicles successfully identified in step 4, calculate the average value of the difference between the actual time and the ideal time of the shortest path in step 2, and then obtain the average time error; then, compare the average time error with the set time error threshold; when the average time error does not exceed the time error threshold, it is judged that the average time error is acceptable; otherwise, repeat steps 1 to 6.

[0113] The average time error includes the average time error DIF_arrive average and average departure time error DIF_leave average .

[0114] DIF_arrive average The calculation formula is:

[0115]

[0116] Where S is the set of all successfully identified parking space occupancy data.

[0117] DIF_leave average The calculation formula is:

[0118]

[0119] In this embodiment, based on the parking space data successfully identified in the target parking lot, the average parking time error is calculated as:

[0120]

[0121] In this embodiment, based on the parking space data successfully identified in the target parking lot, the average departure time error is calculated as:

[0122]

[0123] According to the above calculation results, the average parking time error and the average departure time error of the target parking lot in the embodiment are 34.3s and 45.8s respectively. Taking into account the large volatility of parking demand in the parking lot in the embodiment and the large differences in the internal state of the parking lot at different time periods, there is a certain error in the calculation of the shortest path time table using the statistically obtained average speed of vehicles parking and leaving, but this average parking time error is acceptable. In addition, the parking lot in the embodiment adopts QR code payment, and the time required for different drivers to pay by mobile phone operation varies, so the average departure time error in the embodiment is slightly larger than the average parking time error.

[0124] Step 7, identifying vehicles parked in the internal channel: When the license plate number recognition result in step 5 is acceptable and the average time error in step 6 is acceptable, the successfully identified parking space occupancy data and the parking lot gate entry and exit data are compared, and the complement of the two is used to filter out the data of vehicles parked in the internal channel of the parking lot.

[0125] Based on the successfully identified parking space occupancy data, that is, the data of vehicles parked in the parking spaces, and the parking lot gate entry and exit data, the complement set is obtained by comparing the two, that is, the data of vehicles parked in the internal passages of the case parking lot, as shown in Table 5.

[0126] Table 5 Data of vehicles parked in the internal passage of the case parking lot (partial)

[0127]

[0128]

[0129] Based on the above steps, the parking lot internal channel parking vehicle data is successfully identified and divided into two types of parking demand through the K-Means clustering algorithm, such as Figure 3 As shown. In terms of the time distribution of the number of parking demands, the parking demands of the purple cluster are mainly distributed between 7:45 and 22:00. During this period, there are more hub-to-hub trips, which mainly belong to the daytime period; while the parking demands of the yellow cluster are mainly distributed between 0:00 and 8:00 and 21:30 to 13:45. During this period, there are fewer hub-to-hub trips, which mainly belong to the nighttime period. From the perspective of the distribution characteristics of parking duration, both categories show very significant short-term parking characteristics. From the analysis of the distribution characteristics of the number of entries, the proportion of single entry in the nighttime parking demand is higher than that in the daytime, showing a more significant single entry characteristic.

[0130] Combining the clustering results and classification features, the LSTM algorithm is used to predict the parking demand of the internal passages of these two types of parking lots. The results are as follows: Figure 4 and Figure 5As shown, the changing trend of parking demand in the internal aisles of the parking lot can be grasped, and the refined management and control of parking demand in the internal aisles of the parking lot can be achieved from a quantitative perspective.

[0131] In summary, the present invention selects a hub parking lot with channel parking, obtains the internal layout and size data of the parking lot, gate entry and exit data, and parking space occupancy data; performs preliminary matching on the parking lot gate entry and exit data and the parking space occupancy data, and generates a shortest path time table from each parking space to the parking lot entrance and exit; reads the parking space occupancy data in chronological order, identifies qualified vehicle records from the gate entry and exit data, adds them to a candidate set, and establishes a candidate set of vehicles parked in the parking space; for parking space occupancy data that identifies multiple qualified vehicle records, further identifies the one with the highest probability as the result; calculates the recognition success rate of all parking space occupancy data, the average value of the difference between the time for successfully identified vehicles to find and leave the parking space and the ideal value in the shortest path time table; if the evaluation index does not meet the requirements, re-collects relevant data for identification; if the evaluation index meets the requirements, compares the successfully identified parking space occupancy data with the gate entry and exit data, and uses the complement of the two to filter out the internal channel parked vehicle data of the parking lot, and through clustering and prediction algorithms, grasps the characteristics and changing trends of the parking demand for accessibility inside the parking lot.

[0132] The present invention takes into account the aisle parking phenomenon existing inside the hub parking lot, and identifies vehicles parked in the aisles inside the parking lot based on the gate entry and exit data containing license plate information and the parking space occupancy data not containing license plate information in the parking lot, thereby providing data support for the characteristic analysis of the parking demand inside the hub parking lot and further refined management and control, and is a new means of vehicle identification in the context of the increasingly complex parking demand composition of the hub parking lot.

[0133] The preferred embodiments of the present invention are described in detail above; however, the present invention is not limited to the specific details in the above embodiments. Within the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all belong to the protection scope of the present invention.

Claims

1. A method for identifying vehicles parked in an internal passage of a hub parking lot, characterized by: The steps include: Step 1, obtain the dynamic and static data of the target parking lot: first select the hub parking lot with channel parking as the target parking lot; then obtain the dynamic and static data of the target parking lot within the set survey period; the dynamic and static data include static layout data and dynamic parking data; the dynamic parking data includes parking lot gate entry and exit data with license plate number information and parking space occupancy data with parking space number information but without license plate number information; Step 2: Generate the shortest path time table: Generate the ideal shortest path time table from each parking space to the parking lot entrance and exit according to the static layout data obtained in step 1; Step 3: Establish a candidate set of vehicles parked in parking spaces: Read parking space occupancy data in chronological order, identify vehicle records that meet the conditions from the parking lot gate in and out data, and add them to the candidate set of vehicles parked in the corresponding parking spaces; Step 4: Parked vehicle license plate number recognition: Parked vehicle license plate numbers are recognized for each parking space in the target parking lot. The specific recognition method is as follows: A. When there are 0 vehicle records identified in the candidate set of parked vehicles in parking space i, it is judged that the license plate number recognition is unsuccessful; B. When there is one vehicle record identified in the candidate set of parked vehicles in parking space i, it is directly determined that the license plate number recognition is successful; C. When there are two or more vehicle records identified in the candidate set of parked vehicles in parking space i, it is necessary to identify again, and the vehicle record with the highest probability is taken as the identification result, and it is judged that the license plate number recognition is successful; Step 5, calculate and evaluate the recognition success rate: calculate the license plate recognition success rate SR of all vehicles parked in the parking spaces, and compare SR with the set license plate recognition success rate threshold SR0; when SR ≥ SR0, the license plate recognition result is considered acceptable, and proceed to step 6; otherwise, the license plate recognition result is considered unacceptable, and repeat steps 1 to 5; Step 6, calculate and evaluate the average time error: for all vehicles successfully identified in step 4, calculate the average value of the difference between the actual time and the ideal time of the shortest path in step 2, and then obtain the average time error; then, compare the average time error with the set time error threshold; when the average time error does not exceed the time error threshold, it is judged that the average time error is acceptable; otherwise, repeat steps 1 to 6; Step 7, identifying vehicles parked in the internal channel: When the license plate number recognition result in step 5 is acceptable and the average time error in step 6 is acceptable, the successfully identified parking space occupancy data and the parking lot gate entry and exit data are compared, and the complement of the two is used to filter out the data of vehicles parked in the internal channel of the parking lot.

2. The method for identifying vehicles parked in the inner passage of a hub parking lot according to claim 1, characterized in that: In step 2, the ideal time table of the shortest path includes the ideal value of the search time Arrive_ideal i and the ideal departure time Leave_ideal i ; Among them, Arrive_ideal i Leave_ideal is the ideal time for the vehicle to find the shortest path from the parking lot entrance to parking space i; i is the ideal time value of the shortest path from parking space i to the parking lot exit; in step 6, the average time error includes the average parking time error DIF_arrive average and average departure time error DIF_leave average .

3. The method for identifying vehicles parked facing the inner passage of a hub parking lot according to claim 2, characterized in that: In step 6, DIF_arrive average The calculation formula is: in: DIF_arrive ij =|(park_in i -time_in j )*24*60*60-Arrive_ideal i | Where DIF_arrive ij The error in the parking time for parking space i and vehicle j corresponding to the successfully identified parking space occupancy data, unit: s; S is the set of all parking space occupancy data that are successfully identified; n is the number of parking space occupancy data that are successfully identified; park_in i is the time when parking space i starts to be occupied; time_in j is the time when vehicle j enters the target parking lot.

4. The method for identifying vehicles parked facing the inner passage of a hub parking lot according to claim 3 is characterized by: In step 6, DIF_leave average The calculation formula is: in: DIF_leave ij =|(time_out j -park_out i )*24*60*60-Leave_ideal i | In the formula, DIF_leave ij The time error between parking space i and vehicle j corresponding to the successfully identified parking space occupancy data, unit: s; park_out i The time when parking space i stops occupying; time_out j is the time when vehicle j leaves the target parking lot.

5. The method for identifying vehicles parked facing the inner passage of a hub parking lot according to claim 4 is characterized in that: In step 4C, the vehicle record with the highest probability of being identified is DIF ij The vehicle record corresponding to the minimum; among them, DIF ij The calculation formula is: DIF ij =DIF_arrive ij +DIF_leave ij Where DIF ij It is the sum of the time errors of searching and leaving parking space i for vehicle j, unit: s.

6. The method for identifying vehicles parked facing the inner passage of a hub parking lot according to claim 1, characterized in that: In step 5, the license plate number recognition success rate threshold SR0 is set to 80%.

7. The method for identifying vehicles parked facing the inner passage of a hub parking lot according to claim 3, characterized in that: In step 5, the calculation formula of SR is: Where N is the total number of parking space occupancy data.

8. The method for identifying vehicles parked in the inner passage of a hub parking lot according to claim 1, characterized in that: In step 1, the static layout data includes the location of the parking lot entrance and exit, the parking space numbering layout, the ground markings inside the parking lot, the length of a single parking space, the width of a single parking space, the width of a single lane, and the thickness of the walls inside the parking lot.

Images