JP7361012B2 - Entry/exit judgment device - Google Patents

Description

The present invention relates to an entrance/exit determination device , and particularly to a device that recognizes objects from image frame data sequentially acquired over time.

Research and development is being conducted on technology for recognizing specific objects shown in images. Among such image recognition techniques, there is a technique that recognizes a target object in each of images sequentially acquired over time and tracks the target object.

For example, Patent Document 1 below describes an image recognition device that detects a person from a camera image and tracks the detected person. This image recognition device detects a person as a detection target from images sequentially acquired over time, and tracks the person by determining the position of the person in each image. The process of detecting a person calculates the likelihood that a specific area in the image to be processed indicates a person, and if the likelihood is greater than a predetermined likelihood threshold, an object corresponding to that specific area is detected. This is done by confirming the person as a person.

Patent Document 2 describes an object recognition device that recognizes a target on an image. This device tracks objects as targets by recognizing them as targets in images sequentially acquired over time. When recognizing an object as a target, the likelihood of the object relative to the target is used. This document describes an embodiment in which the target is a car.

Note that an object recognized on an image actually refers to an image of an object that occupies a specific area on the image, but in this specification, to simplify the expression, it refers to an image of an object that occupies a specific area on the image. An image of an object is expressed simply as an object on an image.

Japanese Patent Application Publication No. 2016-162232 JP2019-207523A

In devices that track objects such as automobiles, objects different from the object to be tracked may be recognized as overlapping on an image. In this case, the object to be tracked may be mistakenly recognized as having gone outside the photographic field of view, and accurate tracking may not be performed. In addition, if another object temporarily passes and the object to be tracked temporarily overlaps with the image, the position where the object is recognized will be blurred before and after the other object passes. may occur, making it impossible to perform highly accurate tracking.

An object of the present invention is to reliably track an object on images that are sequentially acquired over time.

The present invention sequentially acquires image frames over time, tracks an object in the image frames sequentially acquired over time, and detects when the object appears in a first area indicated by each image frame, and when the object from the appearance position where the object appears to the latest position of the object when the latest position is in a second area outside the first area and indicated by each of the image frames. Based on the displacement evaluation amount of the object, it is determined that the object has entered a proximal area adjacent to the second area , the object has appeared in the second area, and the latest position of the object is 1 area, it is determined that the object has left the adjacent area based on the displacement evaluation amount of the object from the appearance position where the object appeared to the latest position of the object. Features.

Preferably, when the displacement evaluation amount of the object from the appearance position in the first area to the latest position in the second area exceeds a predetermined threshold, the object is moved to the adjacent area. has entered the object, and the estimated displacement of the object from the appearance position in the second area to the latest position in the previous first area exceeds a predetermined threshold. , it is determined that the object has left the proximity area.

Preferably, the amount of displacement of the object in the x-axis direction from the appearance position in the first area to the latest position in the second area exceeds a predetermined x-axis displacement threshold, and When the amount of displacement in the y-axis direction perpendicular to the axial direction exceeds a predetermined y-axis direction displacement threshold, it is determined that the object has entered the proximity area , and the object is moved from the appearance position in the second area. , the amount of displacement of the object in the x-axis direction up to the latest position in the first area exceeds a predetermined x-axis displacement threshold, and the amount of displacement in the y-axis direction perpendicular to the x-axis direction is a predetermined amount. If the displacement threshold in the y-axis direction is exceeded, it is determined that the object has left the proximity area. Preferably, when the amount of displacement of the object in a specific direction from the appearance position in the first area to the latest position in the second area exceeds a predetermined displacement threshold, the proximity area It is determined that the object has entered the object, and the amount of displacement of the object in a specific direction from the appearance position in the second area to the latest position in the first area exceeds a predetermined displacement threshold. If the distance exceeds the distance, it is determined that the object has left the proximity area, and the specific direction is a direction that intersects the boundary line between the first area and the second area.

Further, the present invention sequentially acquires image frames over time, tracks an object in the image frames sequentially acquired over time, and causes the object to appear in a first area indicated by each image frame, When the latest position of the object is in a second area outside the first area and indicated by each of the image frames, from the appearance position where the object appeared to the latest position of the object . Based on the number of image frames up to that point, it is determined that the object has entered a proximal area adjacent to the second area , the object has appeared in the second area, and the latest position of the object is set to the first area. If the object is located in the nearby area, it is determined that the object has left the nearby area based on the number of image frames from the appearance position where the object appeared to the latest position of the object. Preferably, when the number of image frames from the appearance position in the first area to the latest position in the previous second area exceeds a predetermined number, it is determined that the object has entered the proximal area. If the number of image frames from the appearance position in the second area to the latest position in the previous first area exceeds a predetermined number, the object is removed from the nearby area. It is determined that the player has left the game.

Desirably, in the entrance/exit determination device according to any one of claims 1 to 4, the object is a car, and the proximity area includes an area where the car is parked.

Further, the technology related to the present invention sequentially acquires image frames over time, tracks an object with respect to the image frames sequentially acquired over time, and determines whether the object being tracked is moving or stopping. and when it is determined that the object is stationary, it is determined whether another object overlaps in front of the object within the field of view of the image frame, and if the other object overlaps the object. When it is determined that the object is present, the position of the object is stored, and the object is continued to be tracked after the overlapping state with the other object is resolved.

Preferably, the object is a car.

According to the present invention, it is possible to reliably track an object to be tracked on images that are sequentially acquired over time.

1 is a diagram showing the configuration of a vehicle management system. 1 is a diagram showing the configuration of a vehicle management device. FIG. 3 is a diagram conceptually showing a car recognized in tracking processing. FIG. 2 is a diagram conceptually showing a photographic field of view for explaining entrance/exit determination. FIG. 3 is a diagram schematically showing each bounding box obtained for two cars. FIG. 3 is a diagram conceptually illustrating an example of a photographic field of view in which a stopped vehicle to be managed and other vehicles are shown.

(1) Configuration of Vehicle Management System FIG. 1 shows the configuration of a vehicle management system 1 according to an embodiment of the present invention. The vehicle management system 1 includes a vehicle management device 10 and a base station 12. The vehicle management device 10 is installed within the parking lot 14 or beside the parking lot 14. The vehicle management device 10 sequentially photographs images of the automobile 16 (vehicle) moving within the parking lot 14 over time. Then, a management image showing the license plate of the automobile 16 is selected from images sequentially taken over time.

Further, the vehicle management device 10 recognizes the number written on the license plate of the automobile 16 from the management image. The vehicle management device 10 further measures the time at which the automobile 16 enters the parking lot in the parking lot 14 or the time at which it exits from the parking lot in the parking lot 14 (hereinafter referred to as entry/exit time) through a process described below. do. Here, the parking lot refers to a section in the parking lot 14 where the automobile 16 is parked. The vehicle management device 10 wirelessly transmits management information including entry/exit times, the number of the vehicle 16, and management image data (image data representing a management image) to the base station 12. The base station 12 stores entry/exit time, number, and management image data in association with each other.

Although an embodiment in which wireless communication is performed between the vehicle management device 10 and the base station 12 is shown here, the vehicle management device 10 and the base station 12 are connected by a communication line, and the vehicle management device 10 and Wired communication may be performed with the base station 12.

(2) Configuration of vehicle management device and processing executed by the vehicle management device (2-1) Configuration of vehicle management device FIG. 2 shows the configuration of the vehicle management device 10. The vehicle management device 10 includes an imager 24, a computing unit 18, a storage device 22, and a wireless device 20. The imager 24 includes a camera. The computing unit 18 may be configured by a processor that executes a program stored in the storage device 22. The storage device 22 may be a device such as a RAM (Random Access Memory) or a ROM (Read Only Memory). Furthermore, the storage device 22 may be a recording medium such as a hard disk. Further, the storage device 22 may be a storage on a network configured by a plurality of computers configuring a network such as the Internet.

The imager 24 sequentially captures images of the parking lot as time passes, and outputs image frame data (hereinafter referred to as image frames) obtained by the capture to the computing unit 18 sequentially as time passes. The computing unit 18 sequentially acquires image frames from the image pickup device 24 over time and performs tracking processing on each image frame. The tracking process is a process of tracking the position of a target object in the image indicated by the newly acquired latest image frame every time a new image frame is acquired. The tracking process is described, for example, in the above-mentioned Patent Document 2 as a "process for recognizing a target object."

(2-2) Tracking Process The tracking process in which the target object is a car is executed as follows, for example. The computing unit 18 recognizes objects whose likelihood of being a car exceeds a predetermined threshold from the images shown by each image frame. This process is executed based on a database stored in the storage device 22 in advance. In the following description, recognition of an object whose likelihood of being a car exceeds a predetermined threshold value is simply referred to as recognition of a car. According to the tracking process, position information of a bounding box indicating an area where a car exists is obtained.

In this embodiment, the bounding box has a rectangular shape, and each side of the bounding box is parallel to each side of the periphery of the image. The position information of the bounding box is information including the minimum coordinate value (xmin, ymin) where the x-coordinate value and y-coordinate value are the minimum, and the maximum coordinate value (xmax, ymax) where the x-coordinate value and y-coordinate value are the maximum. It may be. However, the x-coordinate value and the y-coordinate value are coordinate values on the orthogonal x-axis and y-axis defined on the image shown by the image frame. The origin of the xy coordinates is the upper left corner of the image. The positive direction of the x-axis is the right direction, and the positive direction of the y-axis is the downward direction.

FIG. 3 conceptually shows vehicles T1 to T4 recognized in the tracking process. In this figure, automobiles T1 to T4 as targets are shown within the photographic field of view 30. The car T1 is the car recognized on the image shown by the first image frame that was obtained first, and the car T2 is the car recognized on the image shown by the second image frame that was obtained next. It is. The car T3 is the car recognized on the image shown by the third image frame acquired after the second image frame, and the car T4 is the car recognized on the image shown by the fourth image frame acquired next. This is the car recognized above. Bounding boxes B1 to B4 are drawn for the cars T1 to T4, respectively, with different positions and ratios of length and width depending on the position and traveling direction of the car. The bounding box's occupied range in the x-axis direction is the range in which the car recognized on the image is projected onto the x-axis, and the bounding box's occupied range in the y-axis direction is the range in which the car recognized in the image is projected on the y-axis. This is the range projected to. Note that in FIG. 3, for convenience of explanation, the distance that a car moves in one frame period, that is, the distance between adjacent cars, is exaggerated. In an actual frame period, the distance traveled by the car in one frame period is shorter than the distance shown in FIG.

(2-3) Tracking process with multiple cars as the target In the above, tracking process with one car as the target was shown, but in the tracking process, multiple cars are the target. good. When a plurality of cars are set as targets, a process is performed for each of the plurality of cars, with one car set as the target.

(3) Selection of management image data and recognition of number plate Based on the database stored in the storage device 22, the computing unit 18 determines the number plate of the vehicle from the image frames sequentially outputted from the imager 24 over time. Image data that is likely to be shown is selected as management image data. This processing is performed, for example, by selecting an image frame representing a region whose likelihood of being a license plate exceeds a predetermined value. The computing unit 18 recognizes the number written on the license plate based on the management image data.

(4) Judgment of entry/exit (4-1) Transmission of management information The judgment of entry/exit of a car based on tracking processing will be explained. Entry/exit determination is a process for determining whether a car has entered or exited a parking lot in a parking lot. In the tracking process described above, the computing unit 18 determines the position of the bounding box as the coordinate value of the appearance position when the car is first recognized. Further, the computing unit 18 determines the position of the bounding box for the car recognized from the image frame output last from the image pickup device 24 as the coordinate values of the latest position (current position). Here, the position of the bounding box may be the position of the center of gravity of the bounding box or the position of a predetermined corner.

FIG. 4 conceptually shows a photographic field of view 30 for explaining entrance/exit determination. An entrance/exit detection range 42 is defined for the photographic field of view 30. The upper side of the photographic field of view 30 faces a public road 40. In the photographic field of view 30, an upper horizontally elongated rectangular area facing the public road 40 is an entry/exit detection range 42. A boundary line 38 between the entrance/exit detection range 42 and a region within the photographic field of view 30 that is not the entrance/exit detection range 42 extends in the x-axis direction. The left, right, and lower areas of the photographic field of view 30 that are not the entry/exit detection range 42 are areas where cars are parked, that is, parking sections. The area in the photographic field of view 30 that is not the entrance/exit detection range 42 is an area through which a car passes when entering/exiting the parking lot. Trajectories 44-1 and 44-2 indicate vehicle trajectories recognized by tracking processing. The trajectory 44-1 extends from the appearance position S1 outside the entrance/exit detection range 42 to the latest position F1 within the entrance/exit detection range 42. The trajectory 44-2 extends from the appearance position S2, which is within the entrance/exit detection range 42, to the latest position F2, which is outside the entrance/exit detection range 42.

The arithmetic unit 18 determines that the appearance position is within the entrance/exit detection range 42, the latest position is outside the entrance/exit detection range 42, the amount of displacement in the x-axis direction exceeds a predetermined x-axis displacement threshold, and When the amount of displacement in the direction exceeds a predetermined y-axis direction displacement threshold, it is determined that the automobile has entered the parking lot. In the following description, the amount of displacement in the x-axis direction and the amount of displacement in the y-axis direction will be referred to as the amount of displacement in the x-axis direction and the amount of displacement in the y-axis direction, respectively.

Further, the arithmetic unit 18 determines that the appearance position is outside the entrance/exit detection range 42, the latest position is within the entrance/exit detection range 42, the amount of displacement in the x-axis direction exceeds a predetermined x-axis displacement threshold, and When the amount of axial displacement exceeds a predetermined y-axis displacement threshold, it is determined that the automobile has left the parking lot.

The following conditions may be added to the conditions for determining that a car has entered the parking lot or that the car has left the parking lot. That is, after the image frame whose appearance position has been recognized is outputted from the imager 24 until the latest image frame is outputted from the imager 24, the number of image frames outputted from the imager 24 is A condition such as when the number exceeds a predetermined number may be added.

In addition, in the above judgment, the x/y displacement condition when the amount of displacement in the x-axis direction exceeds a predetermined x-axis direction displacement threshold and the amount of y-axis direction displacement exceeds a predetermined y-axis direction displacement threshold is as follows. In other words, the x/y displacement condition may be replaced by The condition may be replaced by a condition that the amount of displacement in the specific direction exceeds a predetermined displacement threshold. The direction intersecting the boundary line 38 in the specific direction is, for example, a direction perpendicular to the boundary line 38.

Furthermore, the x/y displacement condition may be replaced with a condition that the total amount of displacement of the vehicle from the appearance position to the latest position exceeds a predetermined displacement threshold. The total amount of displacement is defined as the square root of the sum of the square of the amount of displacement in the x-axis direction and the square of the amount of displacement in the y-axis direction.

Furthermore, the x/y displacement condition is determined by the amount of data output from the imager 24 after the image frame whose appearance position has been recognized is outputted from the imager 24 until the latest image frame is outputted from the imager 24. The condition may be replaced by the condition that the number of image frames obtained exceeds a predetermined number.

The calculator 18 further determines that when both the appearance position and the latest position are within the entrance/exit detection range 42, or when both the appearance position and the latest position are outside the entrance/exit detection range 42, the car is located in a parking lot. It is determined that the vehicle has not entered the parking lot and has not left the parking lot.

In the example shown in FIG. 4, the appearance position S1 is outside the entrance/exit detection range 42, and the latest position F1 is within the entrance/exit detection range 42. The amount of displacement in the x-axis direction of the vehicle from the appearance position S1 to the latest position F1 exceeds the x-axis displacement threshold Δx, and the amount of displacement in the y-axis direction exceeds the y-axis displacement threshold Δy. Therefore, the computing unit 18 determines that the car that drew the trajectory 44-1 has left the parking lot. Further, the appearance position S2 is within the entrance/exit detection range 42, the latest position F2 is outside the entrance/exit detection range 42, and furthermore, the amount of displacement in the x-axis direction of the car from the appearance position S2 to the latest position F2 is x The axial displacement threshold Δx is exceeded, and the y-axis displacement amount exceeds the y-axis displacement threshold Δy. Therefore, the computing unit 18 determines that the car that has drawn the trajectory 44-2 has entered the parking lot.

In this way, the computing unit 18 according to the embodiment of the present invention executes the following processes (a) to (c). (a) A process of sequentially acquiring image frames over time. (b) A process of tracking a car using image frames that are sequentially acquired over time. (c) A car appears in the entrance/exit detection range 42 (first area) or an area outside the entrance/exit detection range 42 (second area) shown by each image frame, and the latest position of the car is in the second area or the first area. If the vehicle is located in the second region, it is determined that the vehicle has entered or left the parking lot, which is a nearby region adjacent to the second region, based on the displacement evaluation amount of the vehicle from the appearance position where the vehicle appeared to the latest position. It is determined that the car has left the park.

Here, the displacement evaluation amount refers to an amount representing the degree of displacement of an object, such as an x-axis direction displacement amount, a y-axis direction displacement amount, a specific direction displacement amount, or a total displacement amount.

The computing unit 18 generates management information and outputs it to the wireless device 20 when it is determined that the car has entered the parking lot or that the car has left the parking lot. The management information is information including entry/exit time, car number, and management image data. Radio 20 transmits management information to base station 12 (FIG. 1). The base station 12 that has received the management information stores the entry/exit time, number, and management image data included in the management information in association with each other.

According to such processing, it is determined whether the car has entered the parking lot or left the parking lot for a car whose displacement evaluation amount from the appearance position to the latest position exceeds a predetermined threshold value. Therefore, an object resembling a car may be incorrectly recognized as a car over a small number of frames, and may be mistakenly recognized as a car while traveling a short distance, resulting in an erroneous determination that a car has entered or exited the park. This will prevent you from being exposed.

Furthermore, when a car is stopped on or near the boundary between the entry/exit detection range 42 and an area outside the entry/exit detection range 42, and an obstacle such as another car passes in front of the car, The recognized position of the car may become blurred before and after passing an obstacle. As a result, the vehicle may be recognized within the entry/exit detection range 42 before passing the obstacle, and may be recognized outside the entry/exit detection range 42 after passing the obstacle. Alternatively, the vehicle may be recognized outside the entry/exit detection range 42 before passing the obstacle, and may be recognized within the entry/exit detection range 42 after passing the obstacle.

In this case, if the entrance/exit judgment was made only based on whether the vehicle moved from the entrance/exit detection range 42 to the outside of the entrance/exit detection range 42 or from outside the entrance/exit detection range 42 to the entrance/exit detection range 42, Even though the user has not entered or left the venue, it is incorrectly determined that the user has entered or left the venue. In this embodiment, since the condition for determining entry/exit is that the displacement evaluation amount from the appearance position to the latest position exceeds a predetermined threshold value, such erroneous determinations are avoided.

Further, in the prior art, there is a technique in which when a car disappears from the photographic field of view, a device recognizes that the car has left the photographic field of view, and determines whether the car enters or leaves the field. In the entrance/exit determination according to this embodiment, there is no need to recognize that the vehicle has disappeared from the photographic field of view, so the entrance/exit determination can be made more quickly than in the prior art.

(4-2) Stop determination The computing unit 18 according to the present embodiment determines whether or not the vehicle is stopped based on image frames sequentially output from the imager 24 while performing tracking processing. . The calculator 18 determines the speed of the vehicle by determining the total amount of displacement of the vehicle every predetermined number of determination frames. That is, the calculator 18 calculates the latest position of the car recognized in the latest image frame and the past position of the car recognized in the image frame that is the number of determination frames before the latest image frame, and calculates the car's past position from the past position. Find the total displacement up to the latest position as the speed. The number of frames to be determined may be, for example, one frame. The arithmetic unit 18 calculates the speed determined for each number of determination frames to a predetermined stop threshold for a predetermined number of N times (including the last determined speed, where N is an integer of 1 or more). When the following conditions occur, it is determined that the vehicle is stopped. The calculator 18 may determine that the vehicle is stopped when the moving average value, which is the average value of the speeds calculated N times in the past, becomes less than or equal to the stop threshold.

(4-3) Determination of overlap with other vehicles Based on the image frames sequentially output from the imager 24, a vehicle to be managed that is determined to be stopped is recognized over a predetermined number of absent frames from a certain image frame. I will explain what happens when it disappears. In this case, there may be a case where the vehicle to be managed moves outside the photographic field of view 30 or a case where another vehicle appears in front of the vehicle to be managed. If another vehicle appears in front of the managed vehicle, even if the managed vehicle is temporarily stopped while entering or leaving the parking lot, the managed vehicle may Tracking may end.

Therefore, when a vehicle that has been determined to be stopped is no longer recognized for a predetermined number of absent frames from a certain image frame, the computing unit 18 performs an overlap determination with another vehicle. In the other vehicle overlap determination, it is determined whether or not there is another vehicle (hereinafter referred to as an overlap vehicle) that obstructs the view of the management target vehicle when viewed from the image pickup device 24 . When an overlapping vehicle exists, the computing unit 18 determines that another vehicle is overlapping, and when there is no overlapping vehicle, it determines that there is no overlap.

When the vehicle determined to be stopped is no longer recognized for a predetermined number of absent frames, and it is determined that there is no overlap, the computing unit 18 determines that the vehicle has gone out of the photographic field of view (photographing field of view). ). On the other hand, when the vehicle determined to be stopped is not recognized for a predetermined number of absent frames and an overlap is determined, the computing unit 18 determines that the vehicle is within the photographic field of view. The computing unit 18 memorizes the position of the vehicle, continues to determine whether or not other vehicles overlap with the image frames sequentially output from the imager 24, and when it is determined that there is no overlap, it starts tracking the vehicle to be managed. resume. Alternatively, the computing unit 18 restarts tracking of the managed vehicle when the managed vehicle that has been stopped starts moving.

The determination of overlap with other vehicles will be explained in detail. In the other vehicle overlap determination, when the following conditions (α) and (β) are satisfied, the computing unit 18 determines that an overlapping vehicle exists. On the other hand, when one or both of the conditions (α) and (β) are not satisfied, the arithmetic unit 18 determines that there is no overlap, meaning that no overlapping vehicle exists.

(α) The bounding box determined for the managed vehicle (managed vehicle box) overlaps the bounding box determined for another vehicle (other vehicle box). (β) The managed vehicle box is located at the same position or higher than other vehicle boxes when viewed in the vertical direction within the photographic field of view.

Specifically, the computing unit 18 determines that there is an overlap between the vehicle and one or more other vehicles when the following overlap conditions are satisfied. However, the coordinate values (xminA, yminA) are the minimum coordinate values of the bounding box of the other vehicle, and the coordinate values (xmaxA, ymaxA) are the maximum coordinate values of the other vehicle.

(xminA≦xmin≦xmaxA, or xminA≦xmax≦xmaxA), and yminA≦ymax≦ymaxA

The arithmetic unit 18 determines that there is no overlap when the overlap condition is not satisfied with any other vehicle.

FIG. 5 schematically shows bounding boxes 46-1 and 46-2 determined for two automobiles 16A and 16B, respectively. The bounding box 46-1 and the bounding box 46-2 partially overlap, and the bounding box 46-1 is located higher within the photographing field of view 30 than the bounding box 46-2. Therefore, if the vehicle 16A is a managed vehicle, an overlap determination is made. On the other hand, if the vehicle 16B is the managed vehicle, it is determined that there is no overlap.

FIG. 6 conceptually shows an example of the photographic field of view 30 in which a stopped vehicle 50 to be managed and another vehicle 52 are shown. In this example, the managed vehicle 50 is a sedan, and the other vehicle 52 is a truck. In the photographic field of view 30 in FIG. 6(a), the managed vehicle 50 traces a trajectory 44-3 from the appearance position S3 to the current position, and although it is determined that it is stopped, the tracking process continues. ing. In the photographic field of view 30 in FIG. 6(b), the managed vehicle 50 that has been determined to be stopped is no longer recognized. This is because the other vehicle 52 enters from the left side, drawing a trajectory 44-4 from the appearance position S4, and overlaps with the front side of the parked managed vehicle 50. In the process according to the present embodiment, the managed vehicle 50 is not determined to have disappeared, and tracking of the managed vehicle 50 is restarted after the other vehicle 52 has passed.

In this way, the computing unit 18 according to the embodiment of the present invention executes the following processes (i) to (vi). (i) A process of sequentially acquiring image frames over time. (ii) A process of tracking a car using image frames that are sequentially acquired over time. (iii) Processing to determine whether the car being tracked is moving or stopped. (iv) When it is determined that the vehicle being pursued is stopped, a process of determining whether another vehicle is superimposed on the near side of the vehicle being pursued within the photographic field of view 30 of the image frame. (v) Processing for storing the position of the car being chased when it is determined that another car is overlapping the car being chased. (vi) A process of continuing to track the vehicle being tracked after the overlap with other vehicles is resolved.

According to such processing, when another vehicle appears in front of the managed vehicle, the managed vehicle is temporarily stopped while entering or exiting the parking lot. However, the tracking of the managed vehicle is prevented from being terminated. Further, after another vehicle has passed, or when the vehicle to be managed starts moving again, the same vehicle to be managed is continuously tracked.

Furthermore, as described above, the conventional technology includes a method that recognizes that a car has gone out of the photographic field of view when the car disappears from the photographic field of view after it appears within the photographic field of view, and determines whether the car enters or leaves the field. There was something. In this conventional technology, if another vehicle appears in front of the managed vehicle while the managed vehicle is temporarily stopped while entering or leaving the parking lot, the managed vehicle It is recognized that the car has gone out of the field of view. As a result, it may be erroneously determined that the managed vehicle has entered or left the parking lot. According to the present embodiment, when another vehicle passes in front of the temporarily stopped managed vehicle, an erroneous determination that the managed vehicle has moved out of the photographic field of view is avoided.

Note that the process for tracking a car has been described above. The present invention may also be used for other objects such as motorcycles, bicycles, people, etc., i.e. in an environment where objects may enter or leave a specific area, the object may enter or exit a specific area; Alternatively, the present invention may be used in processing for determining whether to leave a specific area.

1 Vehicle management system, 10 Vehicle management device, 12 Base station, 14 Parking lot, 16, 16A, 16B Car, 18 Arithmetic unit, 20 Wireless device, 22 Storage device, 24 Imager, 30 Field of view, 40 Public road, 42 Entrance Exit detection range, 44-1, 44-2, 44-3, 44-4 trajectory, 46-1, 46-2 bounding box, 50 managed vehicle, 52 other vehicles.

Claims (7)

Acquire image frames sequentially over time,
Tracking the object with respect to the image frames sequentially acquired over time;
The object appears in a first region indicated by each image frame, and the latest position of the object is in a second region outside the first region and in the second region indicated by each image frame. In this case, it is determined that the object has entered a proximal area adjacent to the second area based on the displacement evaluation amount of the object from the appearance position where the object appeared to the latest position of the object . ,
When the object appears in the second area and the latest position of the object is in the first area, the estimated displacement of the object from the appearance position where the object appears to the latest position of the object is An entrance/exit determination device that determines that the object has exited from the proximity area based on the proximity area. The entrance/exit determination device according to claim 1,
When the displacement evaluation amount of the object from the appearance position in the first area to the latest position in the second area exceeds a predetermined threshold, the object enters the proximal area. It was determined that
When the displacement evaluation amount of the object from the appearance position in the second area to the latest position in the previous first area exceeds a predetermined threshold, the object leaves the proximal area. An entrance/exit determination device is characterized in that it determines that a person has entered or exited a venue. The entrance/exit determination device according to claim 1,
The amount of displacement of the object in the x-axis direction from the appearance position in the first area to the latest position in the second area exceeds a predetermined x-axis displacement threshold, and determining that the object has entered the proximity area when the amount of displacement in the vertical y-axis direction exceeds a predetermined y-axis displacement threshold;
The amount of displacement of the object in the x-axis direction from the appearance position in the second area to the latest position in the first area exceeds a predetermined x-axis displacement threshold, and An entrance/exit determination device that determines that the object has left the proximity area when the amount of displacement in the vertical y-axis direction exceeds a predetermined y-axis displacement threshold . The entrance/exit determination device according to claim 1,
If the amount of displacement of the object in a specific direction from the appearance position in the first area to the latest position in the second area exceeds a predetermined displacement threshold, the object appears in the proximal area. It is determined that the person has entered the
When the amount of displacement of the object in a specific direction from the appearance position in the second area to the latest position in the first area exceeds a predetermined displacement threshold, the object is removed from the proximal area. It was determined that he had left the venue,
The entrance/exit determination device is characterized in that the specific direction is a direction intersecting a boundary line between the first area and the second area. Acquire image frames sequentially over time,
Tracking the object with respect to the image frames sequentially acquired over time;
The object appears in a first region indicated by each image frame, and the latest position of the object is in a second region outside the first region and in the second region indicated by each image frame. In this case, it is determined that the object has entered a proximal area adjacent to the second area based on the number of image frames from the appearance position where the object appeared to the latest position of the object ,
When the object appears in the second area and the latest position of the object is in the first area, based on the number of image frames from the appearance position where the object appeared to the latest position of the object, An entry/exit determination device that determines that the object has exited from the proximity area. In the entrance/exit determination device according to claim 5,
If the number of image frames from the appearance position in the first area to the latest position in the previous second area exceeds a predetermined number, it is determined that the object has entered the proximity area. death,
If the number of image frames from the appearance position in the second area to the latest position in the previous first area exceeds a predetermined number, it is determined that the object has left the proximal area. An entry/exit determination device characterized by: In the entrance/exit determination device according to any one of claims 1 to 6 ,
the object is a car;
The entrance/exit determination device is characterized in that the proximity area includes an area where the object is parked.