Parking lot management system

Abstract

To analyze an aerial image of a parking lot captured from above by a drone, to implement simple, quick and reliable patrol operation, thereby intensively and efficiently patrolling the parking lot to check vehicles newly parked.SOLUTION: In a parking lot management system, a drone 1 equipped with a camera 1a that images a parking lot from above, a controller 2 which manipulates the drone 1, a tablet terminal 3 which receives vehicle image data in the parking lot captured by the camera 1a, and a data analysis server 6 which receives the vehicle image data from the tablet terminal 3 communicate with each other. The data analysis server 6 analyzes two vehicle images received periodically from the tablet terminal 3, generates new parking image data showing a vehicle newly parked, and transmits the generated new parking image data to the tablet terminal 3.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to a parking lot management system that manages parking conditions by analyzing image information obtained by photographing parked vehicles using a drone from above the parking lot. [Background technology]

[0002] Traditionally, in suburban malls and other stores, parking spaces for a certain number of cars have been reserved adjacent to the store to accommodate the convenience of customers who come by car. In suburban stores, parking spaces for an average of 400 cars may be reserved per store.

[0003] When a customer parks their vehicle in a designated spot in the parking lot to enter a store, leaves the vehicle, and then returns to the vehicle, there is a risk that the parked vehicle may be involved in a collision due to the movement of other vehicles, or that someone taking a break in the parked vehicle may suddenly become ill, or that the vehicle may be broken into by a suspicious person.

[0004] As an example of this type of accident, it is well known that there have been many reports of young children left in cars, especially in the summer, dying of heat stroke due to lack of moisture caused by the rising temperature inside the car. For this reason, store managers have their employees patrol the parking lot at regular intervals to check whether the vehicles of their valued customers have been involved in any accidents or incidents.

[0005] Meanwhile, Patent Document 1 below discloses a technology in which a surveillance robot (walking vehicle type) patrols a parking lot. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] JP 2008-310568 A Summary of the Invention [Problem to be solved by the invention]

[0007] Patent Document 1 describes a parking lot management method that includes a recognition device that identifies cars parked in a parking area and transmits identifying information, and a mobile device equipped with an imaging device that moves around the parking lot, capturing images of the inside and outside of parked cars and transmitting the image information, and by managing information from the recognition device and information from the mobile device with a management device in an amusement park, it is possible to monitor the inside and outside of cars in the parking lot in real time, prevent young children in cars from dying from heatstroke, and detect suspicious people.

[0008] However, in the above-mentioned management method, the walking patrol process by the surveillance robot requires securing an aisle for the robot to move around in advance, and depending on the parking position relationship between adjacently parked vehicles, the width of the robot for the surveillance robot to move around may be larger than the width of the aisle, resulting in parking positions that the surveillance robot cannot patrol, which not only prevents normal patrol but also creates a risk of contact between the surveillance robot and the parked vehicles. In addition, because the robot is an expensive and delicate precision device, it is expensive to purchase and maintain, and there is a risk of losing money.

[0009] The present invention has been made to solve the above-mentioned problems, and an object of the present invention is to provide an inexpensive parking lot management system that can easily, quickly, and reliably perform parking lot patrols, efficiently patrol and confirm the safety of parked vehicles, and prevent incidents and accidents in the parking lot by analyzing images obtained by a drone taking aerial photographs of the parking lot from above. [Means for solving the problem]

[0010] The parking lot management system of the present invention that achieves the above object has the following configuration.

[0011] The parking lot management system of the present invention is a parking lot management system in which a drone equipped with a camera that photographs a parking lot from the sky, a controller that operates the drone, a tablet terminal that receives parking lot image data photographed by the camera, and a server device that receives the parking lot image data from the tablet terminal communicate with each other, the tablet terminal includes a transfer means that transfers the parking lot image data received from the camera to the server device, a receiving means that receives new parking lot image data analyzed by the server device, and a display means that displays an image of the new parking lot image data received by the receiving means, and the server device includes an analysis means that analyzes two parking lot image data periodically received from the tablet terminal to create the new parking lot image data that displays a newly parked vehicle, and a transfer means that transfers the new parking lot image data created by the analysis means to the tablet terminal. The drone can also fly a set flight route to photograph the parking lot from the air. Furthermore, the server device can also centrally manage the parking lot image data transmitted from a plurality of the tablet terminals. The analysis means has a new parking determination means for, when a vehicle exists in a parking space of the later of the two parking lot image data, determining that the vehicle existing in the parking space of the later parking lot image data corresponds to a newly entered vehicle if there is no vehicle in the parking space of the earlier of the two parking lot image data, and determining that the vehicle existing in the parking space of the later parking lot image data corresponds to either a replacement vehicle or a continuing vehicle if a vehicle exists in the parking space of the earlier of the two parking lot image data, and a same vehicle determination means for determining whether the vehicle existing in the parking space of the later parking lot image data corresponds to a replacement vehicle or a continuing vehicle depending on the degree of similarity between the vehicle existing in the parking space of the earlier of the two parking lot image data and the vehicle existing in the parking space of the later of the parking lot image data that has been determined to correspond to either a replacement vehicle or a continuing vehicle. Effect of the Invention

[0012] According to the present invention, a relatively inexpensive drone can take aerial photographs of a parking lot from above, analyze the captured images, and perform patrol work of the parking lot simply, quickly, and reliably, thereby enabling focused and efficient patrols to check for newly parked vehicles. In addition, the system can patrol large parking lots with a small number of personnel without missing newly parked vehicles.

[0013] Furthermore, even if the patrol frequency increases, vehicles that have not been moved can be excluded from the patrol targets, so the time required to identify newly parked vehicles can be significantly reduced. [Brief description of the drawings]

[0014] The drawings show a particular embodiment of a parking management system according to the present invention, and include essential features of the invention as well as alternative and preferred embodiments. [Figure 1] 1 is a schematic diagram showing an example of a parking lot management system according to the present invention. [Diagram 2] FIG. 1 is a block diagram showing an example of a parking lot management system according to the present invention. [Diagram 3] FIG. 3 is a diagram showing an example of an image analysis process performed by the data analysis server shown in FIG. 2. [Figure 4] (a) Parking detection image diagram (before time), (b) Parking detection image diagram (after time), (c) Parking judgment image diagram. [Diagram 5] 1A is a diagram showing an example of a same vehicle determination model by the image analysis module during replacement, and FIG. 1B is a diagram showing an example of a same vehicle determination model by the image analysis module during continuation. [Figure 6] FIG. 13 is a diagram showing parking determination results in the cases of continuation and new entry. [Figure 7] FIG. 13 is a diagram showing a parking determination result in the case of switching; [Figure 8] 4 is a flowchart showing an example of a patrol processing procedure in the parking lot management system according to the embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0015] Each of the following embodiments includes not only essential configurations of the present invention, but also configurations that can be selectively adopted, configurations that can be adopted in combination, and preferred configurations.

[0016] [First embodiment] <System configuration> Fig. 1 is a schematic diagram of a parking lot management system DS according to the present invention, and Fig. 2 is a block diagram showing an example of the parking lot management system DS according to the present invention. According to the parking lot management system DS according to the present invention, it is possible to easily grasp the information of vehicles entering and leaving a parking lot PA in an amusement store such as a pachinko parlor, and to prevent accidents caused by vehicles being parked for long periods of time. Specifically, it is possible to grasp the parking time of the same vehicle by comparing and analyzing over time images taken by a drone 1 equipped with a camera 1a that takes aerial photographs of the parking lot from above.

[0017] The parking lot management system DS comprises a drone 1 equipped with a camera 1a that takes aerial photographs of the parking lot from above, a controller (also called a radio control) 2 that controls the drone 1, a tablet terminal 3 that receives vehicle image data of the parking lot photographed by the camera 1a, and a data analysis server 6 that receives the vehicle images from the tablet terminal 3, all of which communicate with each other.

[0018] The basic design of the parking lot management system DS according to the present invention will be explained. In the flight of the drone 1, an automatic flight path can be set according to the building shape and parking lot area and shape of each gaming parlor. In other words, by registering the flight path, stopping position and shooting position of the drone 1 from the longitude, latitude, altitude and aircraft speed, and setting the drone 1 to move automatically according to the registration order, it is possible to design it so that all flights are automatic and manual operation is used only in emergencies.

[0019] In order to ensure that all vehicles parked in the parking lot PA are photographed without any omissions, the number of images to be taken and the camera angle of view are set for each store according to the building shape and parking lot area and shape of each amusement store, and images are taken automatically at a specified position (the shooting position is fixed by registering longitude, latitude and altitude), making it possible to design the system so that basically no manual photography is required.

[0020] In this way, by combining the automatic flight path and automatic photography, it is possible to eliminate errors caused by manual operation. Even in this case, only the takeoff and landing settings of drone 1 must be manually set before automatic flight by setting the coordinate recognition registration for takeoff and landing. This coordinate recognition registration is a mechanism in which the aircraft is held and rotated 360 degrees to have the aircraft recognize the takeoff and landing points, and the aircraft automatically returns to the recognized takeoff and landing points after automatic photography is completed. In this case, the takeoff and landing error of the registered points is within 30 cm.

[0021] The drone 1 can photograph the parking lot PA from the sky, and communicates with the controller 2 via WiFi FPV (2.4 GHz band) and is operated automatically or manually. The image of the parking lot PA photographed from the sky by the camera 1a mounted on the drone body is received by the tablet terminal 3 and transmitted to the data analysis server 6 via the WiFi transceiver 4. The reason for using the 2.4 GHz frequency band here is that the wireless transmission distance is long and radio waves are unlikely to be interrupted even if there are obstacles such as shields. For the drone 1 and camera 1a according to the present invention, devices normally used in this type of field can be used. The WiFi transceiver 4 may be portable or may be mounted on the drone 1 or the tablet terminal 3.

[0022] In this embodiment, a drone that uses the 2.4 GHz band in transmission and reception mode is used as an example, but if the pilot is a Class 1 land-based special radio operator, a drone using the 5.8 GHz band can be used. This allows for more advanced parking lot image capture.

[0023] The controller 2 is equipped with a stick bar for controlling the flight path of the drone 1. An operator (pilot) can operate the stick bar to instruct the drone 1 to ascend, descend, move left and right, hover, etc. The camera 1a captures images of the entire parking lot area while the drone 1 is hovering above the parking lot PA at a predetermined altitude. The captured parking lot images are captured into a memory device, such as an SD memory, mounted on the drone 1.

[0024] By connecting the SD memory detached from the drone 1 to the SD memory connector of the tablet terminal 3, the installed program is executed and the parking lot image is transmitted to the data analysis server 6 via the WiFi transceiver unit 4 and the Internet 5.

[0025] A drone photography flight app, a patrol app, and a management app are pre-installed on the tablet terminal 3.

[0026] The drone photography flight app has, as drone photography flight functions, a pre-flight check function for drone 1, a flight start function, an in-flight emergency control function, and a post-flight check and image upload function. The patrol app has, as patrol functions, a patrol target display function, a patrol check function, and a patrol record list function. The management app has a system management function. Each app may have other functions.

[0027] The patrol function consists of patrol target display → patrol check → patrol performance list. The patrol target display confirms the location of the vehicle to be checked based on the analysis results of the parking lot photo data. It is also possible to add a control to display the area of ​​the vehicle to be checked on the screen.

[0028] The patrol check function is a function for checking the results of checking the vehicles to be patrolled (applying a checked mark), and a function for handing over the patrol to another patrol officer may be added. The patrol performance list is a function for displaying a list of past patrol results on the tablet terminal 3.

[0029] The data analysis server 6 includes an image analysis module 7, a CPU 8, an information storage unit 9, and a transmission / reception unit 10.

[0030] The image analysis module 7 includes a recognition module that recognizes vehicles in a parking lot as objects, and an extraction module that extracts the difference between two parking lot images recognized by the recognition module.

[0031] The CPU 8 controls the data analysis server 6 overall, and connects the input unit (keyboard) 11 and the display unit (display) 12 to control the vehicle determination process in the AI ​​image analysis. The CPU 8 performs functional processing to import parking lot photograph data captured by the drone 1 into the image analysis module 7, and causes the image analysis module 7 to analyze the imported parking lot photograph data, while storing in the information storage unit 9 the data to be sent to the manager operating the tablet terminal 3 as information to be used in patrol checks.

[0032] The information storage unit 9 includes a plurality of frame buffers for storing image information (parking lot photographed data, analysis result data, etc.) exchanged with the tablet terminal 3 via the Internet 5, and a data buffer. The plurality of frame buffers can store two images of parking lot photographed data in chronological order. The data buffer can also store analysis result data to be transmitted to the tablet terminal 3 by image analysis, linked to a time stamp.

[0033] The transmitting / receiving unit 10 receives image information from the tablet terminal 3 via the Internet 5, and transmits patrol target information that has been analyzed and processed to the tablet terminal 3.

[0034] FIG. 3 shows an example of an image analysis process by the data analysis server 6 shown in FIG. 2. FIG. 4 shows (a) a parking detection image (previous time), (b) a parking detection image (later time), (c) a parking judgment result image, and FIGS. 5(a) and (b) are diagrams showing an example of a same vehicle judgment model by an image analysis module. FIG. 6 shows the parking judgment result in the case of continued parking and new entry. FIG. 7 shows the parking judgment result in the case of replacement. Each process is realized by the CPU 8 of the data analysis server 6 executing the image analysis module 7.

[0035] The image analysis process includes an image analysis pre-processing 21, a vehicle detection model 22, a new parking determination program 23, and a same vehicle determination model 25.

[0036] In Fig. 5, the image analysis module 7 compares vehicle images of the same parking position from before and after in time to determine whether the vehicle is a replacement vehicle (newly parked) or a continuing vehicle. That is, images of vehicles in matching parking areas are cut out from the previous image and the current image as a pair of images, and a replacement vehicle is detected from the similarity of the pair of images. The pair of vehicle images cut out from the previous image and the current image are input, the similarity of the output result is calculated, and whether the vehicle is a replacement vehicle or a continuing vehicle is determined from the similarity.

[0037] FIG. 5(a) is a pair of images of a replaced vehicle, and when comparing vehicle images of the same parking position, the position of the bonnet is reversed from left to right. In this case, since the similarity is low as Output1 of the same vehicle determination model 25, it is determined that the previously parked vehicle has moved and a new vehicle has been parked in the same parking position, that is, the currently parked vehicle is a replaced vehicle (newly parked). FIG. 5(b) is a pair of images of a continuing vehicle, and since the similarity is high as Output2 of the same vehicle determination model 25, it is determined that the previously parked vehicle continues to be parked, that is, the currently parked vehicle is a continuing vehicle.

[0038] In the image analysis process of FIG. 3, image analysis is performed using two photographs taken by the drone 1 over time (for example, every hour or 30 minutes). First, as pre-processing 21 for image analysis, the parking lot photograph data transmitted from the tablet terminal 3 and stored in the information storage unit 9 is subjected to tilt correction of the photographed image and cut out comparison areas (detection areas) of the two images taken before and after. Specifically, in the image pre-processing 21, as shown in FIG. 4, an arbitrary parking space area S in the parking lot PA is cut out, and an image 41 taken before the parking space area S and an image 51 taken after the parking space area S are cut out from the overall image taken of the entire parking lot PA from above at the previous and current times (last time and this time), and compared.

[0039] Tilt correction, for example, detects the longest straight line of a building or parking lot line, calculates the angle between the horizontal and the longest straight line, and rotates the entire image so that the straight line is horizontal. The detection area is cut out, for example, as follows: (1) A template file that is commonly used for each store in the overhead image is prepared. (2) The range of the area to be cut out for each shooting angle is set in advance. (3) Stores that match the template file are detected from the captured image, and the area that will be subject to new parking judgment is cut out according to the cut-out range.

[0040] Next, the vehicle detection model 22 detects the location where the vehicle is parked in the captured image (vehicle bounding box coordinates). Then, the position and characteristics (category, shape, size, etc.) of the vehicle parked in each parking space captured in the cut-out images 41, 51 are identified. Specifically, the parking lot photographed data captured in the information storage unit 9 in chronological order is read onto a memory (not shown), and the trained vehicle detection model detects the vehicle bounding box coordinates from the cut-out captured images 41, 51. This ensures that no parked vehicle is overlooked.

[0041] The new parking judgment program 23 judges whether it is a "new entry" or a "replacement or continuation". Specifically, in the new parking judgment program 23, the parking lot photograph data captured in the information storage unit 9 in chronological order is read out onto a memory (not shown), and the coordinates of the cutout images 41, 51 are compared. If it is recognized that there is "no vehicle" in the previous time and that there is "vehicle" in the later time (the bounding box coordinates of the detected vehicle are present only in the current image), it judges it as a new entry 24, creates analysis result data 23A that corresponds to the photographed image and displays the location of the new parking, and displays the new parking location in the photographed image. This makes it possible to patrol only new vehicles. On the other hand, if it is not recognized as above, that is, if it is recognized that there is "vehicle" in both the previous time and the later time (the bounding box coordinates of the detected vehicle are present in both the previous image and the current image), it judges it as "replacement 26 or continuation 27", and creates analysis result data 23B that corresponds to the photographed image and displays the location of the "replacement or continuation" parking.

[0042] The same vehicle determination model (similarity determination model) 25 detects whether a vehicle is the same when a vehicle is detected at the same position in an image taken one hour ago, for example. That is, it determines whether the vehicles before and after the new parking determination program 23 determines that the vehicle is "replacement or continuation" ("vehicle present" in the previous image → "vehicle present" in the current image) are the same vehicle, and judges whether the vehicle corresponds to "replacement" or "continuation". Specifically, in the same vehicle determination model (similarity determination model) 25, for analysis result data 23B determined by the new parking determination program 23 as "replacement 26" meaning that a vehicle has been replaced or "continuation 27" meaning that a vehicle has been continuously parked, it determines whether the vehicles before and after the analysis result data 23B are the same vehicle in time series. On the other hand, in the case of analysis result data 23A determined by the new parking determination program 23 as "new entry 24" ("no vehicle" in the previous image → "vehicle present" in the current image), meaning that a vehicle has been newly entered, the processing of the same vehicle determination model 25 is not performed. In addition, if the previous image shows "vehicle present" and the current image shows "vehicle absent" (the bounding box of the detected vehicle exists only in the previous image), this means "leaving the depot" and is therefore not included.

[0043] 4 and 6, the new parking judgment program 23 and the same vehicle judgment model 25 will be specifically described. The vehicle 40A parked in the parking space P1 in the previous cutout image 41 is judged as "continue 27" in the same vehicle judgment model 25 because the judgment elements such as the shape, the vehicle direction, and the inclination with respect to the parking line match the vehicle 40A parked in the parking space P1 in the later cutout image 51.

[0044] On the other hand, while no vehicle was parked in parking space P2 in the cropped photographed image 41 from the previous time, vehicle 50A is parked in parking space P2 in the cropped image 51 from the later time, so the new parking judgment program 23 judges it to be a "new entry 24" and processing of the same vehicle judgment model 25 is not performed.

[0045] With reference to Fig. 4 and Fig. 7, the image analysis process in another parking pattern will be described. The vehicle 40A parked in the parking space P1 in the cut-out photographed image 41 at the previous time has the same form as the vehicle 40B parked in the parking space P1 in the cut-out photographed image 51 at the later time, but the vehicle's orientation and inclination with respect to the parking line are different. In this case, the same vehicle determination model 25 judges it as "swap 26". Also, the vehicle 50A parked in the parking space P2 in the cut-out photographed image 41 at the previous time and the vehicle 50B parked in the parking space P2 in the cut-out photographed image 51 at the later time have different forms, so it is judged as "swap 26".

[0046] In the same vehicle determination model 25, identity is determined based on the vehicle's design, color, size, and other aspects, as well as on other determination factors (determination criteria), such as the direction in which the vehicle is parked, its inclination relative to the parking line, and its position. In the image analysis process, it is preferable to perform image analysis using a machine learning model using AI (artificial intelligence). Here, the machine learning model has the function of extracting useful rules, knowledge expressions, and judgment criteria from a set of data input to the device through analysis, outputting the judgment results, and learning knowledge. There are various methods, but they can be broadly divided into "supervised learning," "unsupervised learning," and "reinforcement learning." Furthermore, in order to realize these methods, there is also a method called "deep learning" that learns to extract the feature amount itself.

[0047] 8 is a flow chart showing an example of a patrol procedure by a patrol person (operator, store clerk) using the parking lot management system according to this embodiment, in which S101 to S120 indicate each step.

[0048] First, the patrol officer turns on the tablet terminal 3 and logs in as an administrator (S101).

[0049] Next, the drone 1 is set at a takeoff and landing location that will be a preset home position within the parking lot PA (S102). After checking the safety of the surroundings, the patrol officer turns on the drone 1 (S103) and checks on the controller screen whether there is a problem with the remaining battery power based on the battery voltage of the drone 1.

[0050] If it is determined that the remaining battery power is insufficient (S104), a power receiving unit (not shown) is connected to the drone power receiving connector to start charging the lithium battery (S105). After a predetermined time has passed, the charging voltage is checked to determine whether to end charging. If it is determined that the charging is insufficient, the charging process is continued, and if it is determined that the charging is sufficient, this process is terminated (S106). After that, the process proceeds to S107, as in the case where the remaining battery power is sufficient.

[0051] Next, the patrol officer checks the environment in which the drone 1 will be flown, such as weather conditions (wind speed, amount of rain, presence or absence of snowfall) and the safety of the surrounding area (S107). Next, the patrol officer operates the stick of the controller 2 to take off the drone 1 (S108), and navigates the drone 1 by automatic flight set in the follow-up mode (S109). Except in an emergency, the drone 1 navigates the automatically set flight route to capture images of the entire parking lot PA from the sky, and manual flight is not required except in an emergency, so there is no risk of camera shake or human error due to manual flight and the patrol officer does not need to be an experienced drone operator.

[0052] Depending on the shape of the parking lot, if it is possible to capture an image of the entire parking lot PA, the drone 1 is made to hover at a constant altitude, with the center of the parking lot as the base, and the entire parking lot PA is captured as a still image by the camera 1a (S110).

[0053] The patrol officer confirms that drone 1 has automatically taken photos using camera 1a, and then operates the stick of controller 2 to store the captured images in a recording medium such as an SD memory equipped in drone 1 (S111), and then returns drone 1 to home position H (S112).

[0054] Next, the photographed image data stored in a recording medium such as an SD memory is automatically or manually uploaded to the data analysis server 6 via the Internet, and it is determined whether data transmission is complete (S113). Note that a configuration is also possible in which the photographed image is automatically transferred to the data analysis server 6 without going through an SD memory or the like.

[0055] The tablet terminal 3 receives the analysis data from the data analysis server 6 (S114), and after receiving the analysis data from the data analysis server 6, displays an image identifying the newly parked vehicle (new entry or replacement) on the liquid crystal screen of the tablet terminal 3 (S115). This specific image is a difference from the previously transmitted photographed data, and therefore excludes vehicles that have already moved out of the parking lot PA and continuing vehicles transmitted before the photographing. Therefore, even if there is one new vehicle and the number of parked vehicles before that was 20, for example, the patrol staff only needs to confirm the one newly parked vehicle to complete the patrol process, and the patrol work burden is significantly reduced.

[0056] Next, the patrol officer starts the patrol confirmation work for the new vehicle while visually checking the liquid crystal screen of the tablet terminal 3 (S116). When the patrol confirmation work for the new vehicle is completed, a "check completed" is entered into the management table managed on the tablet terminal 3 (S117). Thereafter, vehicles that are determined to be "the same" as the vehicle in question are evaluated as "continued" or "already patrolled" and are excluded from the patrol targets. If there is a server that centrally manages the management table, by sending the latest management table from each store to the server, it is also possible for the general manager at headquarters to centrally grasp the latest parking situations of all stores in the case of multiple stores.

[0057] Next, it is determined whether the current time is the closing time (S118), and if it is determined that it is the closing time, the drone 1 is retrieved and stored in a predetermined position (S119), and this process ends. On the other hand, if it is determined that it is not the closing time, it is determined after a predetermined time has passed whether a predetermined time for the next surveillance shooting has passed (S120), and if the predetermined time has passed, preparation for the next patrol is promptly performed, and if the predetermined time has not passed, the drone waits until the next patrol.

[0058] After the drone 1 has completed its patrol, the patrol officer can visually inspect the vehicles by narrowing down the number of vehicles to be inspected using the screen displayed on the liquid crystal display of the tablet terminal 3. If necessary, the patrol officer can check the interior of the vehicle and input the inspection process (whether there is a problem or not) of the newly parked vehicle into the management table. In addition, if the vehicle is judged as "continue 27" in the same vehicle judgment model 25 as a result of the analysis reception data and a predetermined time has passed, the officer can check the interior of the vehicle again and input the result into the management table.

[0059] This reduces the time and effort required for patrolling the parking lot thoroughly to check whether new vehicles are parked there, and enables patrolling to focus on only necessary new vehicles, making it possible to quickly check whether children or other persons have been left behind in cars.

[0060] If a patrol officer discovers an emergency situation, such as an infant left behind in a vehicle, through a visual inspection, he or she will contact in-store staff via a communication means such as a tablet terminal 3 or a mobile phone, and make an announcement over the store's PA system requesting the vehicle owner to take immediate action, and if necessary, contact the local police station.

[0061] By periodically repeating patrols using drone 1 until the store's business hours end, patrol staff can complete patrol work in a shorter time than with conventional patrol work, even if there are only a small number of patrol staff and a large number of vehicles parked in a large parking lot, as the patrol staff can identify and patrol newly parked vehicles.

[0062] In another embodiment, the navigation of the drone 1, the aerial photography of the parking lot PA, and the takeoff and landing of the drone 1 can all be performed fully automatically. In such a case, the flight of the drone 1 can be confirmed by a notification sent to the tablet terminal 3 that a new vehicle check is required.

[0063] As another embodiment, an infrared camera may be installed and the temperature of the rooftop of the parked vehicle may be detected to identify vehicles that should be patrolled. In such a case, by adding an additional patrol process suitable for the summer season when the temperature inside the vehicle is high, it becomes possible to focus on patrolling dangerous vehicles.

[0064] Each step of the present invention can also be realized by executing software (programs) acquired via a network or various storage media on a processing device (CPU, processor) such as a personal computer (computer).

[0065] The present invention is not limited to the above-described embodiments, and various modifications (including organic combinations of the respective embodiments) are possible based on the spirit of the present invention, and these are not excluded from the scope of the present invention.

[0066] The disclosure regarding the present invention described above can be summarized at least as follows. A parking lot management system in which a drone equipped with a camera that takes aerial photographs of a parking lot from above, a controller that controls the drone, a tablet terminal that receives vehicle image data of the parking lot photographed by the camera, and a server device that receives the vehicle image data from the tablet terminal communicate with each other, wherein the tablet terminal is equipped with a transfer means that transfers the vehicle image data taken from the camera to the server device, a receiving means that receives new parking image data analyzed by the server device, and a display means that displays the new parking image data received by the receiving means, and the server device is equipped with an analysis means that analyzes two vehicle images, one before and one after the other, received from the tablet terminal periodically or at an arbitrary time interval, to create new parking image data showing a newly parked vehicle, and a transfer means that transfers the new parking image data created by the analysis means to the tablet terminal.

[0067] The present invention can include at least the following embodiments. (1) The drone flies a set flight route and takes aerial photographs of the parking lot. (2) The server device centrally manages the parking lot image data transmitted from the multiple tablet terminals. (3) The analysis means is configured to, in a case where a vehicle exists in a parking space of the later of the two parking lot image data, determine that the vehicle existing in the parking space of the later parking lot image data corresponds to a newly entered vehicle if there is no vehicle in the parking space of the earlier of the two parking lot image data, and determine that the vehicle existing in the parking space of the later parking lot image data corresponds to either a replacement vehicle or a continuing vehicle if a vehicle exists in the parking space of the earlier of the two parking lot image data, and to determine whether the vehicle existing in the parking space of the later parking lot image data corresponds to a replacement vehicle or a continuing vehicle based on the degree of similarity between the vehicle existing in the parking space of the earlier of the two parking lot image data and the vehicle existing in the parking space of the later parking lot image data that has been determined to correspond to either a replacement vehicle or a continuing vehicle. The drone having the above-mentioned is configured to fly a set flight route and take an aerial photograph of the parking lot. [Explanation of symbols]

[0068] 1. Drone 1a Camera 3. Tablet devices 6 Data analysis server 7 Image Analysis Module 8 CPU

Claims

1. A parking lot management system in which a drone equipped with a camera that photographs a parking lot from above, a controller that controls the drone, a tablet terminal that receives parking lot image data photographed by the camera, and a server device that receives the parking lot image data from the tablet terminal communicate with each other, The tablet terminal includes: a transfer means for transferring the parking lot image data received from the camera to the server device; A receiving means for receiving new parking image data analyzed by the server device; a display means for displaying an image of the new parking image data received by the receiving means, The server device includes: an analysis means for analyzing the two parking lot image data periodically received from the tablet terminal and creating the new parking image data showing a newly parked vehicle; A transfer means for transferring the new parking image data created by the analysis means to the tablet terminal; A parking lot management system comprising:

2. The parking lot management system according to claim 1, characterized in that the drone flies a set flight route to take aerial photographs of the parking lot.

3. The parking lot management system according to claim 1 , wherein the server device centrally manages the parking lot image data transmitted from the plurality of tablet terminals.

4. The analysis means When a vehicle is present in a parking space section of the latter parking space image data of the two parking space image data, a new parking determination means for determining that the vehicle existing in the parking space of the subsequent parking space image data corresponds to a newly entered vehicle if there is no vehicle in the parking space of the previous parking space image data of the two parking space image data, and determining that the vehicle existing in the parking space of the subsequent parking space image data corresponds to either a replacement vehicle or a continuing vehicle if there is a vehicle in the parking space of the previous parking space image data; an identical vehicle determination means for determining whether the vehicle present in the parking lot section of the subsequent parking lot image data corresponds to a replacement vehicle or a continuing vehicle based on the degree of similarity between the vehicle present in the parking lot section of the previous parking lot image data and the vehicle present in the parking lot section of the subsequent parking lot image data that has been determined to correspond to either a replacement vehicle or a continuing vehicle; 2. The parking lot management system according to claim 1, further comprising:

Images