KR20240021671A - Suspicious object monitoring device based on false image probability - Google Patents

Abstract

The artificial intelligence-based integrated boundary method according to the technical idea of the present disclosure includes the steps of acquiring a plurality of radar images for an area of interest, generating an overlapping image by overlapping the plurality of radar images, a plurality of overlapping images, and a plurality of overlapping images. Obtaining a real image probability for a target included in an overlapping image using an artificial intelligence model learned based on the targets, and identifying whether the target is an unidentified object based on location information about objects included in the area of interest. and selectively monitoring the unidentified object based on the real probability when the target is identified as an unidentified object. When the target is identified as an unidentified object, selectively monitoring the unidentified object based on the real probability. When the virtual image probability is within the setting range, obtaining a movement path of the unidentified object using location information about the target, acquiring a camera image of the unidentified object according to the movement path, and a plurality of It includes obtaining a probability that the unidentified object is a suspicious object using an artificial intelligence model learned based on camera images and a plurality of unidentified objects.

Description

Suspicious object monitoring device based on false image probability}

The technical idea of the present disclosure relates to an integrated perimeter system using artificial intelligence, and more specifically, to an integrated perimeter system that detects unidentified objects based on radar image analysis and camera image analysis using artificial intelligence.

In general, coastal surveillance radars are installed to monitor the coastal area to prevent intruders from occurring near the coast. Coastal surveillance radar is equipment installed on the ground to monitor the coast and is operated for the purpose of early detection of enemies infiltrating through the sea. When operating a radar, the radar operator reports vessels not identified through the radar to the decision maker.

A large amount of manpower and budget are required to operate the radar, and the working environment is vulnerable due to remote locations, so problems may arise in coastal surveillance when military forces are reduced.

The technical idea of the present disclosure provides an integrated perimeter system that automatically monitors unidentified objects by performing artificial intelligence analysis based on radar images and camera images acquired through radar and cameras.

In order to achieve the above object, the artificial intelligence-based integrated boundary method according to an aspect of the present disclosure includes the steps of acquiring a plurality of radar images for an area of interest, and generating an overlapping image by overlapping the plurality of radar images. , obtaining a real image probability for a target included in the overlapping image using an artificial intelligence model learned based on a plurality of overlapping images and a plurality of targets, based on location information about objects included in the area of interest. It includes a step of identifying whether the target is an unidentified object, and a step of selectively monitoring the unidentified object based on real probability when the target is identified as an unidentified object.

An object monitoring device according to another aspect of the present disclosure uses an artificial intelligence model that is learned based on a plurality of radar images and whether the target in each radar image is real, to overlap radar images obtained from the radar and overlap the overlapped radar images. Based on the signals received from the radar signal processor that calculates the probability of the real image of the target included in the image and the device that provides information on the identified object, the target is identified as an unidentified object, and the location information of the target identified as an unidentified object is provided. It includes a radar information processing unit that selectively outputs information to the control center based on the real image probability.

According to an embodiment of the present disclosure, the integrated surveillance system can provide a precise coastal surveillance function by calculating the actual probability of objects in the radar image through artificial intelligence analysis of the radar image acquired through radar.

In addition, according to an embodiment of the present disclosure, the integrated surveillance system calculates the probability of a suspicious object in the camera image through artificial intelligence analysis of the camera image acquired through the camera, thereby providing a rapid coastal surveillance function according to priority. can do.

Figure 1 is a main configuration diagram of an integrated border system according to an embodiment of the present invention.
Figure 2 is a block diagram illustrating the operation of the integrated border system 10 according to an exemplary embodiment of the present disclosure.
Figure 3 is a flowchart explaining the operation of a radar signal processor according to an exemplary embodiment of the present disclosure.
Figure 4 is a flowchart explaining the operation of the radar information processing unit according to an exemplary embodiment of the present disclosure.
Figure 5 is a block diagram illustrating the operation of the integrated border system 10 according to an exemplary embodiment of the present disclosure.
Figure 6 is a flowchart illustrating the operation of a camera information processing unit according to an exemplary embodiment of the present disclosure.
Figure 7 is a flowchart illustrating the operation of a TOD and a camera according to an exemplary embodiment of the present disclosure.
Figure 8 is a flowchart explaining the operation of the camera information processing unit according to an exemplary embodiment of the present disclosure.

Hereinafter, various embodiments of the present invention are described with reference to the accompanying drawings.

Figure 1 is a main configuration diagram of an integrated border system according to an embodiment of the present invention.

Referring to FIG. 1, the integrated border system 10 according to an embodiment of the present invention includes a control device 110, a display device 120, a Thermal Observation Device (TOD) 130, a camera 140, and object information. It may include a provision device 150, a radar 160, and a control center 200.

The control device 110 receives radar information and external information related to the object from the radar 160 and the object information providing device 150, respectively, and displays it on the display device 120, and when an event occurs (for example, an unidentified If an object is detected), related information may be transmitted to the control center 200. In this specification, an object may mean a target, fishing boat, ship, missile, projectile, aircraft, or moving object. The control device 110 may be referred to as an object monitoring device.

The control device 110 can receive data from the TOD 130 and the camera 140 to display more detailed information about the object on the display device 120 and precisely monitor whether an event has occurred.

The TOD 130 is a thermal image detection device and may be a device that detects the heat of an object and displays it as an image. TOD (130) can use far-infrared rays for long-distance observation. TOD (130) can have a longer observation distance than night vision goggles that amplify light. The radar 151 may operate as a long-distance detection device, the TOD 130 may operate as a mid-range surveillance device, and the camera 140 may operate as a short-range surveillance device. According to an exemplary embodiment of the present disclosure, an unidentified object may be detected by comprehensively using the camera 140, TOD 130, and radar 151.

The object information providing device 150 may transmit external information to the control device 110. Specifically, the object information providing device 150 may include a vessel pass device (V-PASS) 151, an automatic vessel identification system (AIS) 152, and a maritime navigation (e-NAVI) 153.

Radar 160 may be a coastal surveillance radar that detects long distances. Radar 160 is deployed in major coastal and island areas and can detect ships and aircraft moving at sea. Radar 160 can share detection information in the waters surrounding the Korean Peninsula by linking with the naval tactical C4I system and port surveillance system. In some embodiments, radar 160 may be an air defense radar for detecting aircraft, missiles, etc.

The radar 160 operates to detect objects within a relatively long distance of 20 km from the coast, measures the distance, azimuth, speed, coordinates, etc. of the detected object and outputs it to the control device 110 as radar signal information. For example, the radar 151 may output a B-scope signal indicating the distance to the detection object and the azimuth of the detection object to the control device 110. The radar 160 may have an automatic alarm generation function when an object intrudes into a set boundary area, and may display the tracks of all detected objects. The signal output by the radar 160 may be referred to as a radar signal.

V-PASS 151 can provide the location of the fishing boat and emergency rescue signals to the control device 110. V-PASS (151) is an automatic fishing vessel location transmitting device (Vessel-Pass) that can transmit the location of fishing vessels and emergency rescue signals to ensure safe operation of fishing vessels and to respond quickly in the event of a marine accident. The signal output by the V-PASS 151 may be referred to as a V-PASS signal.

AIS (152) can automatically transmit related information such as the type, location, and course of the vessel to a nearby navigation beacon management office or navigation beacon. AIS (152) is an Automatic Identification System that can determine navigation information through the location of a ship, and is an identification device for nearby ships through wireless signals for the purpose of preventing ships from colliding with each other in open seas. information can be shared. The contents of the information transmitted from AIS (152) can be recorded together with various information coming from the bridge and in the Vessel Voyage Recording Device (VDR), which can be called the black box of the ship. All AIS data (including from other ships) received during navigation can be recorded in the VDR. In some embodiments, the AIS signal may be provided to the vessel including the control device 110 through a navigational beacon management station or navigational beacon. The signal output by the AIS 152 may be referred to as an AIS signal.

e-NAVI (153) can continuously provide maritime route information, ship collision, automatic stranding prediction warning, electronic chart real-time information, maritime safety information, marine weather information, etc. to the control device (110). The signal output by the e-NAVI 153 may be referred to as an e-NAVI signal. In this specification, the V-PASS signal, AIS signal, and e-NAVI signal may be referred to as external information.

Although not shown, the object information providing device 150 may include a system that provides real-time information about an aircraft in operation or a launched missile. The control device 110 can receive information about the type, location, speed, direction, origin, destination, etc. of the aircraft or missile from the system, and signals output by the system can also be referred to as external information. That is, the object information providing device 150 may provide external information to the control device 110, and the external information may mean real-time information about the identified object.

The signal transmitted from the device including the control device 110 may be transmitted to the control center 200 including the maritime traffic control center (VTS) through a navigational beacon management center on land and a nearby unmanned lighthouse.

The control unit 110 may receive radar signals and external information from the object information providing device 150 and display confirmed and unconfirmed objects on the display device 120. For example, the external information processing unit 114 may assign a unique identification number to the identified object using external information (V-Pass signal, AIS signal, e-NAVI signal, etc.). The external information processing unit 114 displays confirmed and unconfirmed objects with unique identification numbers on the display device 120 to facilitate tracking, and at the same time transmits related information to the control center 200 when an event occurs. It operates so that subsequent processing takes place.

To this end, the control device 110 may include a radar signal processing unit 111, a radar information processing unit 112, a camera information processing unit 113, a storage unit 114, and a communication unit 115.

The radar signal processing unit 111 can distinguish between a real image and a virtual image of an object detected by the radar based on the radar image received from the radar 151.

Specifically, the radar signal processing unit 111 may generate an overlapping image by overlapping a plurality of radar images. At this time, the plurality of radar images may be images acquired at preset time intervals. If the object included in the radar image is a moving object, a trace may be formed in the overlapping image in the opposite direction of the moving path of the moving object.

The radar signal processing unit 111 may output a real image probability or a virtual image probability for an object included in an overlapping image based on a pre-learned artificial intelligence model. At this time, the preset artificial intelligence model uses overlapping images (or multi-channel images, which will be described later) as input data, and is a model learned to output the real or virtual probability of the object in the image based on the trace, shape of the object, size of the object, etc. It may be, but it is not limited to this. If the object in the overlapping image is a real image, the object in the overlapping image may be a moving object, and if it is a virtual image, the object in the overlapping image may not be a moving object. Real probability or imaginary probability may be referred to as real probability.

According to another embodiment, the radar signal processing unit 111, in addition to using an overlapping image through a plurality of radar images, inputs the plurality of radar images into an artificial intelligence model through separate channels (multi-channel) and calculates the probability of a real image. It can be obtained.

The radar signal processing unit 111 according to an embodiment of the present disclosure may set a ROI (Region of Interest) of the radar image received from the radar 151. The radar signal processing unit 111 can set an ROI based on the center point of objects included in the radar image and output the real image probability of the object within the ROI.

Specifically, the radar signal processing unit 111 can extract the center point of a plurality of objects included in the received first radar image through center point filtering, and obtain an area of a preset size as an ROI based on the center point. You can. The radar signal processing unit 111 may clip a plurality of ROIs in the first radar image.

Likewise, the radar signal processing unit 111 can extract the center point for each of the plurality of objects included in the received second radar image, and obtain an area of a preset size as an ROI based on the center point.

Thereafter, the radar signal processing unit 111 may identify the ROI clipped in the first radar image and the ROI clipped in the second radar image corresponding to the ROI clipped in the first radar image, and overlap the corresponding ROIs. Specifically, the radar signal processing unit 111 may acquire location information of the center point of the object in the first radar image, and determine the center point of the object in the second radar image located within a preset distance as the center point for the same object. . The radar signal processing unit 111 may overlap ROIs corresponding to the center point determined to be the same object.

Radar images such as B-scope have large data sizes and most areas do not contain objects, so overlapping and analyzing them in real time has been an unnecessary waste of time and space. According to the above embodiment, there is an effect of being able to analyze in real time by inputting only the desired ROI based on the center point into the artificial intelligence model.

*The radar information processing unit 112 can classify objects into confirmed objects and unconfirmed objects based on external information. That is, if the object in the radar image is an object identified based on external information, it may be determined as a confirmed object, and if it is an object not identified based on external information, it may be determined as an unidentified object. Unidentified objects can be classified into simple objects, objects of interest, and suspicious objects. A simple object may refer to an unidentified object that has not been initially identified, an object of interest may refer to an unidentified object that has not been identified for a certain period of time, and a suspicious object may refer to an unidentified object such as a spy ship, an unknown projectile, etc.

The storage unit 115 can automatically store all information about the objects of the radar signal processing unit 111 and the radar information processing unit 112 by accumulating them in a DB to enable tracking. The storage unit 115 may be a device that stores all data analyzed and tracked by the control device 110 in a DB, including tracking contents by unique identification number, surveillance screen of the radar 151, TOD 130, and camera ( 140) to save monitoring screens, etc. to manage history.

The camera information processing unit 113 manages information on cooperative organizations including equipment such as the camera 140 and TOD 130 and related organizations, and operates in conjunction to identify and track set objects.

For example, the camera information processing unit 113 can monitor unconfirmed or confirmed objects set in the radar information processing unit 112 in more detail using the TOD 130 and the coastal complex surveillance camera 140. The camera information processing unit 113 may calculate the probability that the unidentified object is a suspicious object (or a suspicious ship) by performing artificial intelligence analysis based on the camera image. A suspicious object (or a suspicious ship) may mean a ship similar to an enemy ship or spy ship.

The communication unit 116 enables the control device 110 to transmit and receive data with the object information providing device 150, the camera 140, the TOD 130, and the display device 120, and sends monitoring data to the control center 200. It may refer to a communication device that transmits .

The display device 120 may be a display device connected to the control device 110 by wire or wirelessly, or may be a terminal on which an application owned by an administrator is installed.

Figure 2 is a block diagram illustrating the operation of the integrated border system 10 according to an exemplary embodiment of the present disclosure.

Referring to FIG. 2, the radar signal processing unit 111 may calculate the probability of a real image of an object included in the radar image based on the radar image.

Specifically, the radar signal processing unit 111 may calculate the probability of a real image of an object included in a radar image based on an artificial intelligence learned model. The artificial intelligence model can be learned based on a plurality of learning radar images and whether the object included in each learning radar image is real. The radar signal processing unit 111 may output object location information (azimuth, distance, etc.) and real image probability.

The radar information processing unit 112 may output unidentified object information based on object location information, real image probability, and external information. Unidentified object information may be information about an unidentified object that has a real probability greater than a set probability.

Specifically, the radar information processing unit 112 can manage a list of objects assigned unique identification numbers based on external information. An object assigned a unique identification number may be a confirmed object. The radar information processing unit 112 may determine whether the object included in the radar image is the same as the confirmed object identified by external information. If the object included in the radar image is different from the confirmed object, the radar information processing unit 112 may determine the object included in the radar image to be an unconfirmed object. If the actual probability of an unidentified object is greater than or equal to a set probability, the radar information processing unit 112 may output information about the unidentified object.

The radar information processing unit 112 may provide information about unidentified objects to the control center 200. The control center 200 may perform follow-up processing based on information about the unidentified object. For example, the control center 200 can display unidentified objects on the screen and continuously monitor them.

Figure 3 is a flowchart explaining the operation of a radar signal processor according to an exemplary embodiment of the present disclosure. FIG. 3 can be explained with reference to FIG. 1 .

Referring to FIG. 3, in step S310, the radar signal processing unit 111 may receive a plurality of radar images from the radar 151.

In step S320, the radar signal processing unit 111 may process a plurality of radar images. Specifically, the radar signal processing unit 111 may process each of the plurality of radar images into a radar image representing the distance from the radar 151 to the object and the azimuth of the object.

In step S330, the radar signal processing unit 111 may overlap a plurality of processed radar images. Specifically, the radar signal processing unit 111 may generate an overlapping image by overlapping a plurality of processed radar images based on the center point within the ROI.

In step S340, the radar signal processing unit 111 may obtain location information and real image probability of the object based on the overlapping image using a pre-trained artificial intelligence model. The artificial intelligence model can be learned based on a plurality of learning radar images, the location of the object in each learning image, and whether the object is a real image.

According to an exemplary embodiment of the present disclosure, location information and real image probability of an object in the radar image can be obtained by analyzing the radar image based on an artificial intelligence model. The integrated perimeter system 10 can provide an automatic and precise perimeter function through artificial intelligence analysis by using real probability to identify unidentified objects.

Figure 4 is a flowchart explaining the operation of the radar information processing unit according to an exemplary embodiment of the present disclosure. FIG. 4 can be explained with reference to FIG. 1 .

Referring to FIG. 4, in step S410, the radar information processing unit 112 may receive location information and real image probability of an object from the radar signal processing unit 111, and may receive external information from the object information providing device 150. .

In step S420, the radar information processing unit 112 may check whether the object in the overlapping image is an unidentified object. Specifically, the radar information processing unit 112 determines the location of the object based on the location information of the object and determines whether the location of the object is substantially the same as the location of the object to which a unique identification number has been assigned by external information, so that the object in the overlapping image is unidentified. You can check whether it is an object.

If the object in the overlapping image is an unidentified object, in step S430, the radar information processing unit 112 may filter the unidentified object based on the real image probability. That is, the radar information processing unit 112 may perform different processing for unidentified objects based on the probability of a real image. For example, if the reality probability is less than a set value, the radar information processing unit 112 may determine that the object is an illusion and stop the unidentified object monitoring procedure, and if the reality probability is more than a set value, it may determine that the object is an illusion and follow up. The procedure can be performed.

In step S440, the radar information processing unit 112 may output information about the unidentified object. For example, the radar information processing unit 112 may output the type of unidentified object, location information of the unidentified object, probability of a real image, direction of travel of the unidentified object, speed of the unidentified object, etc.

According to an embodiment of the present disclosure, the radar information processing unit 112 may give priority to unidentified objects. Specifically, the radar information processing unit 112 may assign priority based on actual probability and distance to an unidentified object. For example, when the actual probability is greater than the first value and the distance between the control device 110 or the control center 200 and the unidentified object is less than or equal to the second value, the radar information processing unit 112 determines that the risk is high and takes priority. A higher ranking can be given.

As another example, the radar information processing unit 112 may assign priority based on at least one of the actual probability, the distance to the unidentified object, the direction of the unidentified object, and the speed of the unidentified object. Specifically, the radar information processing unit 112 may assign priorities by setting different weights to the actual probability, the distance to the unidentified object, the direction of the unidentified object, and the speed of the unidentified object. For example, if it is determined that the direction of the unidentified object is a direction approaching the control device 110 or the control center 200, priority can be given by resetting the weight to the speed of the unidentified object and the distance to the unidentified object to be high. there is. On the other hand, if the direction of the unidentified object is not in a direction that approaches the control device 110 or the control center 200, priority may be given by resetting the weight to the actual probability to be high. As shown above, by resetting the weight and assigning priority according to the direction of progress of the unidentified object, there is an effect of identifying high-risk unidentified objects in real time.

That is, the radar information processing unit 112 can list unidentified objects by priority and output information about unidentified objects with relatively high priorities.

The control center 200 may receive information about unidentified objects and continuously monitor suspected unidentified objects. Therefore, the integrated perimeter system 10 according to an exemplary embodiment of the present disclosure automatically detects unidentified objects by calculating the real image probability for objects in the radar image through an artificial intelligence model, and monitors the detected unidentified objects. function can be provided.

Figure 5 is a block diagram illustrating the operation of the integrated border system 10 according to an exemplary embodiment of the present disclosure. Referring to FIG. 5, the operations of the radar signal processing unit 111, the external information linking unit 150, and the radar information processing unit 112 as described above in FIG. 2 may be omitted.

The camera information processing unit 113 may receive information about unidentified objects. The camera information processing unit 113 can check information about the TOD 130 and the camera 140 that can photograph an unidentified object. For example, the camera information processing unit 113 may obtain information about the location and number of the TOD 130 and the camera 140. The camera information processing unit 113 may identify a location where the TOD 130 and the camera 140 will take pictures based on the location information of the unidentified object. That is, the camera information processing unit 113 can create a shooting path so that the TOD 130 and the camera 140 take photos according to the location of the unidentified object. The camera information processing unit 113 may provide a shooting path to the TOD 130 and the camera 140.

The TOD 130 and the camera 140 may receive a photographing path and capture an image of an unidentified object along the photographing path. The TOD 130 and the camera 140 may provide object images to the camera information processing unit 110.

The camera information processing unit 110 may calculate the probability of an unidentified object as a suspicious object based on the object image using a pre-trained artificial intelligence model. An artificial intelligence model can be learned based on a plurality of learning images and whether an object in each learning image is a suspicious object.

The control center 200 can take action on the unidentified object by combining information related to the unconfirmed object, such as suspicious object information, probability of a real image, and external information. According to an exemplary embodiment of the present disclosure, the probability of a suspicious object can be obtained by performing secondary artificial intelligence analysis through a camera image for an object with a high actual probability, thereby enabling rapid action by priority. there is. That is, the control center 200 can take active measures against unidentified objects with a high probability of being suspicious objects, and can take relatively passive measures with respect to unidentified objects with a low probability of being suspicious objects.

Figure 6 is a flowchart illustrating the operation of a camera information processing unit according to an exemplary embodiment of the present disclosure. Figure 6 can be explained with reference to Figures 1 and 5.

In step S610, the camera information processing unit 113 may receive information about the unidentified object. In some embodiments, the camera information processing unit 113 may receive information about an unidentified object corresponding to an actual probability higher than a set probability. For example, the camera information processing unit 113 may receive location information of an unidentified object.

In step S620, the camera information processing unit 113 may determine whether the TOD 130 and the camera 140 are available. Specifically, the camera information processing unit 113 selects the operable TOD 130 and the camera 140 among the plurality of TODs including the TOD 130 and the plurality of cameras 140 including the camera 140. You can control it.

In step S620, the camera information processing unit 113 may identify the path of the unidentified object. Specifically, the camera information processing unit 113 may calculate the movement path of the unidentified object based on the location information of the unidentified object.

In step S640, the camera information processing unit 113 may calculate a photographing path and output the photographing path so that photographing is performed according to the movement path of the unidentified object.

In some embodiments, the camera information processing unit 113 may operate when the real image probability is within a set range. Specifically, the camera information processing unit 113 may control the TOD 130 and the camera 140 to identify the type of unidentified object when the real image probability is within a set range. More specifically, the camera information processing unit 113 receives captured images from the TOD 130 and the camera 140, and an artificial intelligence model learned based on a plurality of captured images and the types of objects included in the captured images. Using , you can identify the type of unidentified object included in the captured image.

Figure 7 is a flowchart illustrating the operation of a TOD and a camera according to an exemplary embodiment of the present disclosure. FIG. 7 can be explained with reference to FIGS. 1 and 6 .

In step S710, the TOD 130 and the camera 140 may receive a shooting path.

In step S720, the TOD 130 and the camera 140 may capture images along the capturing path. Specifically, the TOD 130 and the camera 140 can capture a plurality of images of a moving unidentified object by adjusting the field of view along the capturing path. For example, the TOD 130 may capture an infrared image of an unidentified object, and the camera 140 may capture a visible light image of an unidentified object.

In step S730, the TOD 130 and the camera 140 may output captured images.

According to an exemplary embodiment of the present disclosure, precise monitoring of an unidentified object may be possible by performing photography with the TOD 130 and the camera 140 along an expected path for an object presumed to be an unidentified object.

Figure 8 is a flowchart explaining the operation of the camera information processing unit according to an exemplary embodiment of the present disclosure. FIG. 8 can be explained with reference to FIGS. 1, 6, and 7.

In step S810, the camera information processing unit 113 may receive an image of an unidentified object from the TOD 130 and the camera 140.

In step S820, the camera information processing unit 113 may obtain a suspicious object probability for the unidentified object based on the image of the unidentified object using a pre-trained artificial intelligence model. The artificial intelligence model can be learned based on TOD captured images or camera captured images, which are learning images, and whether the unidentified object corresponding to each captured image is a suspicious object.

In step S830, the camera information processing unit 113 may calculate the risk of the unidentified object based on the probability of a suspicious object. Specifically, the camera information processing unit 113 may output a higher risk level as the probability of a suspicious object increases.

In some embodiments, the control center 200 may closely monitor or take active action against unidentified objects with a high risk and take relatively passive actions with respect to unidentified objects with a low risk.

According to an exemplary embodiment of the present disclosure, the probability of a suspicious object and the risk of an unidentified object are calculated based on a TOD captured image or a camera captured image using an artificial intelligence model, so a system for automatically monitoring an unidentified object can be provided. .

As above, exemplary embodiments are disclosed in the drawings and specifications. In this specification, embodiments have been described using specific terms, but this is only used for the purpose of explaining the technical idea of the present disclosure and is not used to limit the meaning or scope of the present disclosure as set forth in the patent claims. . Therefore, those skilled in the art will understand that various modifications and other equivalent embodiments are possible. Therefore, the true technical protection scope of the present disclosure should be determined by the technical spirit of the attached patent claims.

Claims (1)

Obtaining a plurality of radar images for an area of interest;
generating an overlapping image by overlapping the plurality of radar images;
Obtaining a real image probability for a target included in the overlapping images using an artificial intelligence model learned based on a plurality of overlapping images and a plurality of targets;
identifying whether the target is an unidentified object based on location information about objects included in the area of interest;
When the target is identified as an unidentified object, selectively monitoring the unidentified object based on the real image probability;
If the actual image probability is within a set range, obtaining a movement path of the unidentified object using location information about the target;
Obtaining a camera image of the unidentified object along the movement path; and
An artificial intelligence-based integrated boundary method comprising obtaining a probability that the unidentified object is a suspicious object using an artificial intelligence model learned based on a plurality of camera images and a plurality of unidentified objects.