KR20220144179A - Method for operating optimal open parking lot employing technology of image recognition by CCTV - Google Patents

Abstract

An embodiment relates to an optimal parking lot operation method using CCTV-based image recognition technology. Specifically, the system provides technical solutions related to a parking lot, which is not parking control equipment, to provide a CCTV- and application-based video data-oriented lightweight model. Therefore, in relation to a B2B parking lot, parking lot utilization plans for various types of buildings are provided to a parking lot owner and building owner. In addition, since the parking lot management model is lightweight based on CCTV, maintenance costs required for infrastructure are reduced and inefficiencies of existing parking industry are eliminated. For example, the parking lot management model provides a stable matching between space demanders and suppliers, reduces wasted idle parking spaces, and responds to a variety of needs such as autonomous driving and electric vehicles.

Description

{Method for operating optimal open parking lot employing technology of image recognition by CCTV}

The content disclosed in the present specification relates to the field of parking lot operation technology, and more specifically, when operating the parking lot, since the image of the parking lot is analyzed based on CCTV to obtain empty parking space information, etc., the technology for optimally operating the parking lot is about

Unless otherwise indicated herein, the material described in this section is not prior art to the claims of this application, and inclusion in this section is not an admission that it is prior art.

In general, the scale of a related field is quite large in a situation where many people own a car, and there is room for further satisfaction of the demand for a parking space with the advent of electric vehicles and car-sharing services in the future.

And, most parking lot managers currently do not have internal SW and data dedicated personnel, and there is also a demand for a parking lot management system based on vehicle access records. In addition, parking control equipment is too expensive and bulky, so it is not easy to apply it to all parking lots. Therefore, the existing parking control equipment is limited to small parking lot for this reason.

In particular, in this regard, parking lot owners and building owners in the B2B field face various difficulties, for example, it is not easy to know the optimal use of their parking lot, and the optimal management and promotion of their parking lot is also similar.

Also, there are difficulties experienced by parking applicants in the B2C field, it is not easy to know the available parking space for each time period, and there are spaces where existing B2C parking services are omitted.

In addition, in addition to these points, even though there is currently an idle space, such a parking lot is not open to the general public, and there is a need to provide the owners of the idle space with an optimal management and promotion plan for the parking lot.

So, in order to solve these problems, there is room to provide an infrastructure that can use the parking lot optimally without introducing expensive parking control equipment, and introduce a lightweight management model based on CCTV.

As a similar technology related to this background, there is an image-based parking space recognition technology that has been released recently, and related research is being conducted in several places.

However, since the CCTV image data pool, which is the most important basis in relation to the contents disclosed in this specification, is limited, the technology is still mature enough for people to actually use it.

For example, as such a technique, there is a method in which a person directly performs labeling at a specific location and inputs it into an algorithm, which cannot immediately respond to a large number of unspecified image data and requires manpower.

So, in order to solve these points, in the present specification, deep learning is applied in the labeling step, so that the algorithm recognizes the parking space by itself and checks the presence or absence of a vehicle.

Therefore, for this reason, by automating the process of labeling and inputting a parking space by a person, it is possible to input and learn a lot of data in a short time.

In addition, the current technology also allows learning based on pre-labeled parking lot and vehicle image data.

Through this, it is easy to recognize whether there is a vehicle in the space entered in advance, but the efficiency is low compared to the input resources because manpower is required in the labeling process. Also, there is a problem in image recognition, and it is not easy to respond to the following situations.

1. There are cases where the parking line is not recognized well by the camera because it is blurred.

2. Due to the installation location and angle of the CCTV, the parking lot may look like a non-rectangular shape (trapezoid, etc.).

3. In some cases, it is difficult to recognize the parking line due to conditions such as night, rain, and snow.

And, the patent literature of such a background is also as follows.

(Patent Document 0001) KR102019031 Y1

(Patent Document 0002) KR1020190134018 A

The technique of Patent Document 1 is to provide parking information in real time by checking whether or not parking spaces are parked in a specific parking lot when a parking space is identified. For reference, this technology is to the extent of recognizing a parking space using two-dimensional information of a parking space in relation to the disclosure of the present specification.

And, the technique of Patent Document 2 is disclosed that calculates a parking space based on deep learning.

The disclosed content provides a technology solution, not a parking control equipment, in relation to parking lots such as B2B, that is, a lightweight model centered on video data based on CCTV and apps, so parking lot owners and clients can utilize parking lots for various types of buildings It is intended to provide an optimal parking lot operation method using CCTV-based image recognition technology to provide

The optimal operation method of parking lot using CCTV-based image recognition technology according to the embodiment is,

In relation to parking lots, we provide a technology solution rather than a parking control device, that is, a lightweight model centered on video data based on CCTV and apps.

Specifically, when a parking space is first identified in order to manage a parking lot, etc., an empty parking space that can be easily parked in the parking lot is checked in real time and provided to the user.

And in this case, since an empty parking space for easy parking is identified by matching with various types of parking lots, types of parking spaces, and types of parking environments, parking space information that is practically helpful to the user is provided.

For example, the parking lot type is an on-street parking lot, a building (eg, apartment) parking lot, etc., the parking space type is a general parking space, a disabled parking space, a women's parking space, etc., and the parking environment type is a parking day, For example, weekdays or weekends, time of day such as morning, afternoon, and night, weather, and parking lot.

In addition, at this time, since the empty parking space for easy parking is provided by reflecting the on-site situation by MOT (multi object tracking) technology and a machine learning-based parking space recognition model, accurate information with a sense of presence is also provided.

And, in addition to this, this MOT technology analyzes CCTV images for a long time (at least one week), tracks the vehicle and records the coordinates, analyzes the distribution of the vehicle location based on the recorded coordinates, and parks a place with a high distribution. It is characterized by recognition as a space.

According to the embodiments, by providing a technology solution, not a parking control equipment, in relation to a B2B parking lot, that is, a lightweight model centered on image data based on CCTV and apps, parking lots for various types of building owners and owners It provides suggestions on how to use it.

For example, we provide technical solutions to developers who need to build new parking lots, managers who own existing buildings, and individuals who are struggling with their own parking lot. And furthermore, it provides customized parking spaces for car sharing services, electric vehicles, and shuttle companies.

And, in relation to B2C, more diverse parking spaces and parking products such as weekly/monthly parking are provided.

In addition, since the parking lot management model is lightweight based on CCTV, the maintenance cost required for infrastructure is reduced, and the inefficiency of the existing parking industry is eliminated. For example, it provides stable matching between space consumers and suppliers, thereby reducing wasted idle parking space, and responding to more diverse needs such as autonomous driving and electric vehicles.

1 and 2 are diagrams for conceptually explaining an optimal parking lot operation method using CCTV-based image recognition technology according to an embodiment; 3 is a view showing the overall system to which an optimal parking lot operation method using CCTV-based image recognition technology is applied according to an embodiment; 4 is a block diagram showing the configuration of a management information processing device applied to an optimal method for operating a parking lot using a CCTV-based image recognition technology according to an embodiment; 5 is a flow chart showing the operation of an optimal parking lot operation method using CCTV-based image recognition technology according to an embodiment;

1 and 2 are diagrams for conceptually explaining a method for optimally operating a parking lot using a CCTV-based image recognition technology according to an embodiment.

Specifically, FIG. 1 is a schematic diagram showing an optimal parking lot operation method according to an embodiment, and FIG. 2 is a conceptual diagram of such an optimal parking lot operation method.

As shown in Figures 1 and 2, the optimal operating method of the parking lot of one embodiment provides a technology solution, not a parking control equipment, in relation to a parking lot, that is, it provides a lightweight model centered on video data based on CCTV and apps, It provides parking lot owners and building owners with ways to utilize parking lots for various types of buildings.

So, the optimal parking lot operation method according to an embodiment is to expand the image image set for the parking lot, form a management/analysis algorithm application case based on the image, and provide a solution for performing quantitative analysis of the parking lot based on CCTV.

For example, such a solution analyzes the remaining parking space information by time zone only with CCTV images. Then, the number of vehicles entering and leaving and the residence time are analyzed, and an average remaining space database of residential/shopping malls/business districts is constructed.

To this end, the optimal parking lot operation method according to an embodiment provides an open parking lot-related technology in the form of software and a web service based on CCTV and cloud servers. And, it also provides a CCTV-based parking lot image recognition solution provided in the form of an API. So, in reality, it is provided to demanding parties such as each parking management company in the form of an API.

And, as a specific solution in relation to this solution, in one embodiment, it is to analyze the CCTV image of the parking lot by performing image analysis based on MOT technology and machine learning.

In addition, it manages vehicle flow and entry/exit status with only CCTV video data for crime prevention.

So, one embodiment provides a parking management solution suitable for buildings with 50 or less parking spaces, which is difficult to introduce parking control equipment.

In addition, it helps to manage B2B parking spaces for small building owners, and provides, for example, control solutions to small shopping malls, buildings, and apartment complexes where it is not easy to introduce a parking control solution.

In more detail, the optimal operating method according to this embodiment (refer to the details of Fig. 1) is a case in which a parking space is first identified in order to manage a parking lot, etc. Collect surveillance video. So, these surveillance images are analyzed, and empty parking spaces that are easy to park in the parking lot are checked in real time and provided to the user.

In this case, this optimal operation method provides parking space information that is practically helpful to the user, since an empty parking space for easy parking is identified by matching each parking space type, parking space type, and parking environment type.

For example, the parking lot type is an on-street parking lot, a building (eg, apartment) parking lot, etc., the parking space type is a general parking space, a disabled parking space, a women's parking space, etc., and the parking environment type is a parking day, For example, weekdays or weekends, time of day such as morning, afternoon, and night, weather, and parking lot.

In addition, at this time, the optimal operation method provides such an empty parking space that is easy to park by reflecting the on-site situation by MOT technology and a machine learning-based parking space recognition model (the MOT technology is applied), so accurate information with a sense of presence provision is also made.

For example, as shown in FIG. 2 , the parking space recognition model to which this MOT technology is applied is only an empty parking space where parking of the parking lot is easy in the current monitoring image when a surveillance image is taken from a video surveillance device, for example, CCTV. It calculates for different parking areas and provides integrated recommendations.

For reference, the parking space recognition model to which the MOT technology is applied according to this embodiment will be described in more detail below.

3 is a view showing the overall system to which an optimal parking lot operation method using CCTV-based image recognition technology is applied according to an embodiment.

As shown in FIG. 3, the system of one embodiment includes a plurality of image monitoring devices 100-1 to 100-n installed for a plurality of different parking areas in a specific parking lot, and an empty parking space recognition model with the parking space recognition model. It includes a central management information processing device 200 that provides parking information.

Additionally, the system according to an embodiment provides a variety of overall services by connecting the management information processing device 200 to an external linkage, and provides a police station management device 300-1, a hospital management device 300-2, and the like. include

In this case, such a system connects a plurality of video monitoring devices 100-1 to 100-n and the management information processing device 200 through their own networks to interwork individually for each group. In this case, the self-network may use, for example, a wireless communication network such as Wi-Fi or Ethernet.

Also, on the other hand, this system provides an empty parking space for easy parking in the management information processing device 200 to the mobile terminal of the parking lot customer when the parking lot is optimally operated in this way, so that the customer visiting the building can easily access the parking lot. make it easy to find.

And, on the other hand, the system installs a display (not shown) for each parking area when there is an empty parking space that is easy to park in the parking lot, that is, when parking recommendation information for each empty parking space is calculated, parking according to the currently empty parking space Recommendation information is displayed on a corresponding display to provide information about an empty parking space for easy parking to a user entering the parking lot. In this case, such a display may be made by, for example, different colors in the order in which there are many empty parking spaces for easy parking, or different colors in the order of lightness.

The video monitoring devices 100-1 to 100-n are installed for a number of different parking areas in a specific parking lot, and by capturing a monitoring image in each parking area and transmitting it to the management information processing device 200, management information processing In the device 200, an empty parking surface for easy parking is found for each of several parking areas.

The management information processing device 200 preferentially registers a parking space recognition model to which the MOT technology is applied in advance, and when this model identifies a parking space in a parking lot, it selects an empty parking space that is easy to park in the parking lot in a number of different ways. It is calculated and provided for each type of parking lot, type of parking space, and type of parking environment. So, when the above-mentioned monitoring image is input, the management information processing apparatus 200 calculates information about an empty parking space in which parking of a parking lot is easy in the current monitoring image from the parking space recognition model for each different parking area, and in real time It provides the actual parking space information in an integrated way.

Additionally, in this case, the parking space recognition model is made as follows.

a) First, when the parking space recognition model recognizes the parking space of the parking lot, the parking space in the parking lot (including shape and environment (environment includes long-time coordinate distribution of objects by MOT)) and parking recommendation degree ( Ease of parking) is defined as a learning model by classifying the number of different parking lot types, parking space types, and parking environment types.

In this regard, a specific configuration is, for example, as follows.

In the case of learning the above-mentioned parking space, in one embodiment, all vehicles caught on the screen are recognized and coordinates are tracked through a MOT (multi-object tracking) algorithm.

And, in this case, the moving vehicle may know the path. Conversely, the coordinates of a stationary (parked) vehicle are fixed.

So, if you look at the distribution of vehicle coordinates after learning a lot of data, you can find that the distribution is bitten at specific coordinates, and it can be inferred that this is a parking lot.

In summary, it is as follows:

1) Analyze CCTV images for a long time (at least one week) and record the coordinates by tracking the vehicle.

2) And, based on the recorded coordinates, the distribution of the vehicle location is analyzed.

3) Recognize a place with a high distribution as a parking space.

Additionally, the level of detection is precise enough to allow continuous tracking even if only 10% of the vehicle is visible.

And, even if the overall pixel value changes through Optical Flow technology, it can be recognized and corrected.

Therefore, it can respond even when the shooting screen changes like a moving CCTV.

b) Then, a data set indicating the shape and environment of the parking lot and the ease of parking for each type of parking lot, parking space type, and parking environment type is extracted.

c) Next, the data set thus extracted is attributeized by reflecting information about a number of different parking lot locations, parking time zones, and parking environment.

d) And, based on these results, properties such as the shape and environment of the parking lot and the degree of ease of parking are determined for each of the different learning models defined above.

e) Next, the result thus determined is normalized.

f) So, based on these normalized results, the shape and environment of the parking lot and the degree of parking ease are set for each learning model. And, through this, independent (empty parking space and parking recommendation) and dependent (parking) for generating information for calculating a number of different parking lot types, parking space types, and parking environment types, and empty parking spaces for each type of parking environment Including cell shape and environment (environment includes long-time coordinate distribution of objects by MOT, degree of ease of parking) Set as variables.

g) Then, the results set in this way are generated as learning and training data.

h) So, from these results, a deep learning-based parking space recognition model according to an embodiment is generated.

In addition, the parking space recognition model will be described a little more as follows.

First, since such a parking space recognition model is different from the usual parking state (or parking pattern) in the case of a weekend, for example, a different learning model is generated for each weekday, weekend, public holiday, and the like. In addition, a model can be created for each parking area, or a model can be created in several bundles by setting standards. This allows us to determine the appropriate method according to the characteristics of the data.

And, when data for generating a recognition model is transmitted in this way, there may be cases where a large number of data is not collected due to errors in communication and equipment, etc. . So, after creating a basic dataset through this, attributes are added for additionally necessary related information.

Also, if some data is not collected due to data interruption, the data is removed.

So, next, we determine the valid attributes for each model, generate normal values, and then determine the independent and dependent variables. For example, in the model, the shape and environment of a parking space, and the degree of parking ease may be effective dependent variables, and an empty parking space and parking recommendation information may be set as independent variables.

Then, in order to generate a learning model, learning and training data are generated from the entire data. In general, 70% of the entire dataset is used as training data, and 30% is used as training data to test the model after model creation. Next, we create a learning model. At this stage, it is decided which learning model to use. For example, it refers to a configuration in which the necessary layers are configured based on deep learning and the input and output layers are configured to set the maximum number of outputs. Evaluate the generated model, and if the model is satisfied with the error rate, the model is simulated with new data. If the model update is not required, the training model is saved and used as the model.

4 is a block diagram illustrating a configuration of a management information processing apparatus applied to an optimal method for operating a parking lot using a CCTV-based image recognition technology according to an embodiment.

As shown in FIG. 4 , the management information processing apparatus 200 according to an embodiment includes an interface unit 201 , a DSP unit 202 , a CPU 203 , a storage unit 204 , and a key signal input unit 205 . ), and a display unit 206 .

The interface unit 201 receives a surveillance image from an image monitoring device, etc., and receives the surveillance image for each different parking area in an integrated manner. In addition, when the management information processing device 200 calculates information about an empty parking space for easy parking, the interface unit 201 also transmits and provides this information to a mobile terminal of a pre-registered parking lot customer. Additionally, in this case, the interface unit 201 transmits various control information related to the optimal operation of the parking lot of the management information processing device 200 to the corresponding parking lot safety bar driving device through the D/O unit and the A/O unit. It is also transmitted to several control devices, including In addition, the interface unit 201 may include an A/I unit and a D/I unit in order to receive various analog and digital information related to the optimal parking lot operation.

When a monitoring image is input through the interface unit 201, the DSP unit 202 performs digital signal processing on the current monitoring image, so that parking is easy through the parking space recognition model in the current monitoring image. To obtain information related to the output.

The CPU 203 controls each unit, and when receiving the digital monitoring image output by the DSP unit 202, an empty parking space in which parking of the parking lot is easy in the current digital monitoring image from the parking space recognition model That is, the parking recommendation information for each empty parking space is calculated for each different parking area, and the actual parking space information is integrally provided in real time.

The storage unit 204 stores various information related to the optimal operation of the parking lot, including the above-described parking space recognition model. In this case, the parking space recognition model calculates an empty parking space and a parking recommendation degree in the parking lot by applying the MOT technology to a number of different parking lot types, parking space types, and parking environment types when identifying a parking space in a parking lot. Therefore, parking recommendation information is provided for each empty parking space.

The key signal input unit 205 receives various setting commands related to the optimal operation of the parking lot under the control of the CPU 203 .

The display unit 207 displays various information related to the optimal operation of the parking lot to the manager and the operator under the control of the CPU 204 .

5 is a flowchart illustrating an operation of an optimal parking lot operation method using CCTV-based image recognition technology according to an embodiment (see FIGS. 2 and 3).

As shown in Fig. 5, the system of one embodiment first registers the customer information using the parking lot and the above-mentioned parking space recognition model when identifying the parking space of the parking lot, and the parking space recognition model is a lot of parking lot monitoring images. By accumulating and learning, we create.

In this state, the system according to an embodiment receives each monitoring image from the image monitoring apparatus 100 installed for each of a plurality of different parking areas in the parking lot from the management information processing apparatus 200 .

Then, when the monitoring image is input in this way, the management information processing device 200 calculates parking recommendation information for each empty parking space of the parking lot in the current monitoring image from the parking space recognition model for each different parking area, and in real time Provides integrated parking space information.

Next, the management information processing device 200, when the registered customer using the parking lot visits the corresponding building, the pre-registered customer mobile terminal (M) parking recommendation information for each currently vacant parking space, that is, a vacant parking easy By providing a parking space, customers can park easily and conveniently.

As described above, in one embodiment, when a parking space is first identified in order to manage a parking lot, an empty parking space for easy parking in the parking lot is checked in real time and provided to the user.

In this case, one embodiment provides parking space information that is practically helpful to the user, since an empty parking space for easy parking is identified by matching each parking space type, parking space type, and parking environment type.

For example, the parking lot type is an on-street parking lot, a building (eg, apartment) parking lot, etc., the parking space type is a general parking space, a disabled parking space, a women's parking space, etc., and the parking environment type is a parking day, For example, weekdays or weekends, time of day such as morning, afternoon, and night, weather, and parking lot.

In addition, at this time, since an embodiment provides such an empty parking space for easy parking by reflecting the on-site situation by MOT technology and a machine learning-based parking space recognition model (the above MOT technology is applied), accurate information with a sense of presence provision is also made.

Therefore, through this, one embodiment provides a technology solution, not a parking control equipment, in relation to 2B parking lot, that is, a lightweight model centered on image data based on CCTV and apps, so that parking lot owners and clients can use various types of buildings. It provides a plan for using the parking lot.

In addition, this optimal parking lot operation method provides a better service by easily providing the relevant information in the form of a map in the mobile app when an empty parking space for easy parking is provided to the customer.

To this end, this optimal operating method is performed as follows.

a) First, when parking recommendation information for each vacant parking lot of a parking lot is calculated, parking recommendation information for each vacant parking lot is provided to a pre-registered mobile terminal.

b) At this time, the mobile terminal (M) maps an empty parking lot for easy parking from the mobile app below, and the corresponding empty parking lot and the movement path of the building entrance for a plurality of different open parking lots.

c) So, to guide the user entering the parking lot and the corresponding building.

The mobile app is

a) When moving to a building related to the parking lot, a map guiding the parking lot and the building entrance is registered in advance for a plurality of different parking lots.

b) Then, the map of the parking lot to which the user moves is activated when parking recommendation information for each empty parking space of the corresponding parking lot is provided.

c) Next, an empty parking space for easy parking is provided on the activated map according to the parking recommendation information for each empty parking space.

d) In addition, the corresponding empty parking lot on the map and the movement route of the building entrance are provided.

In addition, when parking recommendation information for each empty parking space of the parking lot is provided by the mobile terminal (M) in addition to this, the parking recommendation information for each empty parking space as a pre-registered user type is provided differently for each parking space of the corresponding user type. , recommends empty parking spaces according to users.

In this case, the user types are ordinary people, women, people with disabilities, etc., and general parking spaces are recommended for ordinary people, and women and disabled parking spaces are preferentially recommended for women and people with disabilities.

On the other hand, this optimal parking lot operation method notifies that when a customer parks in an actual parking lot through this empty parking space information, a beacon terminal is installed for each parking area (including other types) and has arrived at the corresponding location, Knowing that the customer is there.

For reference, in relation to this configuration, reference may be made to FIG. 1 (a beacon terminal disclosed).

To this end, this optimal operating method installs a beacon terminal for each parking area when parking recommendation information for each empty parking space of the parking lot is provided to the mobile terminal (M), and arrives at the location of the corresponding empty parking space with the mobile app. Each time the mobile terminal is located within a preset radius from the corresponding beacon terminal, it notifies the user of the arrival of an empty parking lot.

In addition, in this case, in one embodiment, when the arrival of the location of an empty parking space is notified from the beacon terminal, the parking fee is calculated from the mobile app as a beacon notification time, and parking fee information is provided to the user in real time.

At this time, the beacon notification time for calculating the parking fee is provided, for example, in the form of a list, so that when the user actually parks or leaves the parking lot after work, only the corresponding beacon notification time is selected to calculate the exact parking fee. .

Therefore, when the parking fee is calculated through this, the payment of the parking fee from the mobile app is processed through communication when the mobile terminal M is located within a preset radius in the parking fee calculator, so that the user is informed in advance before the vehicle entry crackdown. Settle the parking fee.

100: video monitoring device 200: management information processing device 201: interface unit 202: DSP unit 203: CPU 204: storage 205: key signal input unit 206: display unit

Claims (7)

When determining the parking space of the parking lot, the empty parking space in the parking lot and the parking recommendation degree are calculated by a number of different parking lot types, parking space types, and parking environment types based on MOT (multi object tracking), and empty parking spaces A first step of pre-registering a parking space recognition model that provides parking recommendation information for each; a second step of receiving each monitoring image from an image monitoring device installed for a plurality of different parking areas in the parking lot; and when the monitoring image is input, calculating parking recommendation information for each empty parking space of the parking lot in the current monitoring image from the parking space recognition model for each different parking area, a third step of providing real parking space information in real time ; Including, The parking space recognition model is, a) When recognizing the parking space of the parking lot, a parking space including a shape in the parking lot, and an environment including a long-time coordinate distribution of an object by MOT, and easy parking Define a model that learns by classifying parking recommendations, including degree, by a number of different parking lot types, parking space types, and parking environment types; extracting a data set indicating the shape, environment, and parking ease characteristics, c) attribute the extracted data set by reflecting information on a number of different parking lots, parking time, and parking environment, d) the attribute Based on the results, determine the properties of the parking lot shape, environment, and parking ease for each of the different learning models defined above, e) normalize the determined results, f) each learning based on the normalized results By setting the shape, environment, and parking ease of each model, a number of different parking lot types, an empty parking space for each parking space type and parking environment type, and an empty parking space for generating information for calculating parking recommendation , set as dependent variables including independent variables including parking recommendation, parking lot shape, environment including long-time coordinate distribution of objects by MOT, and ease of parking, g) setting the results as learning and training data generating, h) generating a deep learning-based parking space recognition model from the generated result; An optimal parking lot operation method using CCTV-based image recognition technology, characterized by The parking lot according to claim 1, wherein, after the third step, when the parking recommendation information for each empty parking space of the parking lot is calculated, a display is installed for each parking area to display the parking recommendation information for each currently empty parking space on the corresponding display, a fourth step of providing information about an empty parking space for easy parking to a user entering the . The optimal operating method of a parking lot using CCTV-based image recognition technology, characterized in that it further comprises a. The method according to claim 2, After the third step, a) when the parking recommendation information for each empty parking space of the parking lot is calculated, to provide parking recommendation information for each empty parking space to a pre-registered mobile terminal, b) the mobile terminal From the mobile app below, the user entering the parking lot and the corresponding building by mapping the empty parking lot for easy parking, the corresponding empty parking lot, and the movement path of the building entrance by a number of different open parking lots, c) 5th step to guide you; Further comprising, The mobile app, a) When moving to a building related to the parking lot, by registering in advance a map guiding the parking lot and the building entrance by connecting the parking lot and the building entrance for a number of different parking lots, b) the parking lot the user moves Activate the map of when parking recommendation information for each empty parking space of the corresponding parking lot is provided, c) provide an empty parking space for easy parking according to the parking recommendation information for each empty parking space on the activated map, d) providing a movement route of the corresponding empty parking lot and the building entrance on the map; An optimal parking lot operation method using CCTV-based image recognition technology, characterized by The method according to claim 3, The fourth step, When the parking recommendation information for each empty parking space of the parking lot is provided to the mobile terminal, the parking recommendation information for each empty parking space as a pre-registered user type for each parking space of the corresponding user type Recommending an empty parking space according to the user by providing different customization; An optimal operation method for parking lots using CCTV-based image recognition technology, characterized by The method according to claim 4, After the fourth step, When parking recommendation information for each empty parking space of the parking lot is provided to the mobile terminal, a beacon terminal is installed for each parking area, and the location of the empty parking space corresponding to the mobile app A fifth step of notifying the user of the arrival of an empty parking space because the mobile terminal notifies the arrival whenever the mobile terminal is located within a preset radius from the corresponding beacon terminal; The optimal operating method of a parking lot using CCTV-based image recognition technology, characterized in that it further comprises a. The method according to claim 5, After the fifth step, when the arrival of the location of the empty parking lot is notified from the beacon terminal, the parking fee is calculated from the mobile app as a beacon notification time, and parking fee information is provided to the user in real time a sixth step; The optimal operating method of a parking lot using CCTV-based image recognition technology, characterized in that it further comprises a. The method according to claim 6, After the sixth step, when the parking fee is calculated, the payment of the parking fee from the mobile app is processed through communication when the mobile terminal is located within a preset radius in the parking fee calculator, so that the user can receive the vehicle to settle the parking fee in advance before the entry crackdown; An optimal operation method for parking lots using CCTV-based image recognition technology, characterized by

Images