Image processing device, image processing method, and image processing program

Abstract

To provide a technology that can accurately detect a presence or absence of a vehicle located in a parking space on a frame image.SOLUTION: An image processing device that is an example of the present disclosure comprises: a vehicle area setting unit that detects a vehicle imaged in a frame image and sets an area on the frame image where the detected vehicle is imaged as a vehicle area; a determination area setting unit that sets a plurality of parking determination areas for the parking space on the frame image based on the imaging position of the parking space on the frame image; a calculation unit that calculates an occupancy rate of a parking determination area in the vehicle area for each of the plurality of parking determination areas; and a determination unit that determines whether the vehicle is parked in the parking space based on the occupancy rate calculated by the calculation unit for each parking determination area.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] This invention relates to a technology for detecting whether a vehicle is parked in a parking space in a parking lot or the like. [Background technology]

[0002] Conventionally, the number of vehicles parked in a parking lot, the locations of empty parking spaces, and other information are managed by processing frame images of the parking lot and detecting the vehicles that are being photographed.

[0003] For example, there is a system that processes captured images of a parking lot to generate edge images and detects the presence or absence of vehicles in each parking area (see Patent Document 1). In Patent Document 1, in order to prevent false detection of shadows as vehicles, vehicles and shadows are distinguished by the magnitude of the edge size. [Prior art documents] [Patent Documents]

[0004] [Patent Document 1] Patent No. 3505362 [Overview of the Initiative] [Problems that the invention aims to solve]

[0005] However, in typical parking lots, parking spaces are adjacent to each other, so the detection area set on the frame image to determine the presence or absence of a vehicle for each parking space is also adjacent. Therefore, even if the target parking space for vehicle detection is empty, depending on the angle of the imaging device (camera), a part of a vehicle parked in an adjacent parking space may be captured in the frame image within the detection area set for the target parking space. This capture of a part of a vehicle parked in an adjacent parking space within the detection area could lead to a false detection that there is a vehicle in the target parking space.

[0006] The object of the present invention is to provide a technology that can accurately detect the presence or absence of a vehicle located in a parking space on a frame image. [Means for solving the problem]

[0007] An example of an image processing device according to the present disclosure includes: an image input unit that receives a frame image of a target area where a parking space is defined as the ground contact surface of a vehicle; a vehicle area setting unit that detects a vehicle captured in the frame image and sets the area on the frame image in which the detected vehicle is captured as a vehicle area; a determination area setting unit that sets a plurality of parking determination areas on the frame image for the parking space based on the image capture position of the parking space on the frame image; a calculation unit that calculates the occupancy rate of the parking determination area in the vehicle area for each of the plurality of parking determination areas set by the determination area setting unit; and a determination unit that determines whether a vehicle is parked in the parking space for each of the parking determination areas based on the occupancy rate calculated by the calculation unit, wherein the determination area setting unit sets a plurality of parking determination areas by moving the image capture position of the parking space in the height direction of the vehicle on the frame image.

[0008] In this configuration, even if a portion of a vehicle parked in an adjacent parking space is visible in the frame image of the target parking space, the presence or absence of a vehicle in the parking space on the frame image can be detected with high accuracy.

[0009] The determination unit may determine whether a vehicle is parked in a parking space for each parking determination area based on the sum of the occupancy rates calculated by the calculation unit. In this configuration, it is possible to determine whether a vehicle is parked in a parking space with simple calculations.

[0010] The determination unit may determine that no vehicle is parked in a parking space if, for each parking determination area, any of the occupancy rates calculated by the calculation unit does not reach the first threshold. This configuration reduces the possibility of mistakenly identifying a vehicle that is not parked in a target parking space as a vehicle that is parked in that space.

[0011] The frame image includes a distance calculation unit that calculates a first distance between the vehicle position defined on the boundary line between the ground surface and the vehicle area and the center of gravity of the parking space. If there are multiple vehicles that the calculation unit has determined to be parked in the parking space based on the occupancy rate, the determination unit may determine that the vehicle for which the first distance calculated by the distance calculation unit is the smallest is parked in the parking space. In this configuration, even if there are multiple vehicles that have been determined to be parked in the parking space based on the occupancy rate, the vehicle parked in the parking space can be determined.

[0012] The distance calculation unit calculates the maximum distance from the centroid of the parking space to the outer perimeter of the parking space as the second distance on the frame image, and the determination unit may determine that a vehicle is not parked in the parking space if the first distance is longer than the second distance. This configuration reduces the possibility of mistakenly identifying a vehicle that is not parked in the target parking space as a vehicle that is parked in that space.

[0013] The vehicle position may also be a position that bisects the boundary line between the ground contact surface and the vehicle area in the vehicle width direction on the frame image. [Effects of the Invention]

[0014] According to the present invention, the presence or absence of a vehicle located in a parking space on a frame image can be detected with high accuracy. [Brief explanation of the drawing]

[0015] [Figure 1] Figure 1 is a block diagram illustrating the configuration of the main parts of an image processing apparatus 1 according to an embodiment of the present invention. [Figure 2] FIG. 2 is a block diagram illustrating the configuration of the main part of the image processing unit 3. [Figure 3] FIG. 3 is a schematic diagram illustrating a frame image in which parking spaces in a parking lot are imaged. [Figure 4] FIG. 4 is a flowchart illustrating the parking determination process by the image processing unit 3. [Figure 5] FIG. 5 is a schematic diagram illustrating a vehicle area. [Figure 6] FIG. 6 is a flowchart illustrating the process of step s2 by the occupancy detection unit 33. [Figure 7] FIG. 7 is a schematic diagram illustrating a parking determination area. [Figure 8] FIG. 8 is a schematic diagram illustrating an occupied area. [Figure 9] FIG. 9 is a schematic diagram illustrating the distance L1 and the distance L2. [Figure 10] FIG. 10 is a schematic diagram illustrating a maintenance screen.

MODE FOR CARRYING OUT THE INVENTION

[0016] §1 Configuration Example FIG. 1 is a block diagram illustrating the configuration of the main part of the image processing apparatus 1 according to an embodiment of the present invention. The image processing apparatus 1 may include, for example, an image input unit 2, an image processing unit 3, and an output unit 4. This image processing apparatus 1 detects whether a vehicle is parked for each parking space provided in a parking lot.

[0017] A frame image obtained by imaging a target area defined on the ground surface of the vehicle with respect to a parking space is input to the image input unit 2. The ground surface of the vehicle is, for example, the ground, a road surface, or a floor surface. The target area is the whole or a part of the parking lot that is the target of the parking determination process described later.

[0018] An imaging device 5 is connected to the image input unit 2. The imaging device 5 may be a digital still camera that captures still images or a video camera that captures moving images. When a digital still camera is used as the imaging device 5, the image processing device 1 may input a release signal to the imaging device 5 to instruct it to capture images, and perform parking detection processing on the captured frame images. When a video camera is used as the imaging device 5, a frame rate of a few frames / sec (a high frame rate is not required) is sufficient. The image processing device 1 does not need to perform parking detection processing on all the frame images of the moving images captured by the video camera; it is sufficient to select any of the input frame images as the frame images to be processed and perform parking detection processing on them.

[0019] The image processing unit 3 processes the frame image input to the image input unit 2 and performs a parking determination process to detect the presence or absence of a vehicle for each parking space captured in the frame image.

[0020] The output unit 4 outputs to a higher-level device the presence or absence of vehicles detected by the image processing unit 3 for each parking space. This higher-level device manages the number of parked vehicles at that time, the number of available parking spaces, and the usage status of the parking lot.

[0021] Figure 2 is a block diagram illustrating the configuration of the main part of the image processing unit 3. The image processing unit 3 may include, for example, a parking space setting unit 31, a vehicle area setting unit 32, an occupancy detection unit 33, a distance calculation unit 34, and a determination unit 35.

[0022] The parking space setting unit 31 sets the area of ​​the parking space on the frame image and identification information for identifying each parking space for each parking space on the frame image. Specifically, the parking space setting unit 31 sets the coordinates of the vertices of the shape drawn by the outer perimeter of the parking space on the frame image. The coordinates on the frame image are pixel coordinates, and are a pair of horizontal and vertical coordinates of the frame image.

[0023] The parking space setting unit 31 may set the coordinates of a point appropriately selected from the outer perimeter of the parking space, instead of the coordinates of the vertices of the parking space.

[0024] The area and identification information of the parking spaces are set manually or automatically. For example, the parking space setting unit 31 may set the area of ​​the parking spaces on the frame image based on input from the operator. Alternatively, the parking space setting unit 31 may set the area of ​​the parking spaces on the frame image using edge detection technology.

[0025] To define the area and identification information of parking spaces, frame images taken at the same angle as the frame image used for parking detection processing are used. For defining the area and identification information of parking spaces, frame images showing a parking situation where no vehicles are parked in any parking spaces may be used, or multiple frame images showing a parking situation where vehicles are parked in some parking spaces may be used.

[0026] The area and identification information of the parking space are set before the parking detection process and are used each time the parking detection process is performed.

[0027] Figure 3 is a schematic diagram illustrating frame images of parking spaces in a parking lot. Figure 3 shows multiple parking spaces 61. Each parking space 61 in the frame image is a rectangle with vertices 62. For example, for each parking space 61, the coordinates of the four vertices 62 of the rectangle of the parking space 61 are set.

[0028] The vehicle area setting unit 32 detects vehicles captured in the frame image and sets the area on the frame image in which the detected vehicle is captured as the vehicle area.

[0029] The occupancy detection unit 33 sets multiple parking determination regions on the frame image for each parking space, based on the image position of the parking space on the frame image. More specifically, the occupancy detection unit 33 sets multiple parking determination regions on the frame image by moving the image position of the parking space in the vehicle height direction. The occupancy detection unit 33 also calculates the occupancy rate of each set parking determination region within the vehicle area. The occupancy detection unit 33 is an example of the "determination region setting unit" and "calculation unit" of the present invention.

[0030] The distance calculation unit 34 calculates the distance L1 between the vehicle position, which is defined on the boundary line between the vehicle's contact surface and the vehicle area, and the center of gravity of the parking space on the frame image. Distance L1 is an example of the "first distance" of the present invention.

[0031] Furthermore, the distance calculation unit 34 calculates distance L2 as the maximum distance from the centroid of each parking space to the outer perimeter of that parking space on the frame image. In particular, if the parking space is a rectangle, distance L2 is the distance between the centroid of the parking space and the vertex of the rectangle of the parking space that is furthest from the centroid. Distance L2 is an example of the "second distance" of the present invention. The centroid of the parking space may be the centroid of the figure drawn by the outer perimeter of the parking space on the frame image (e.g., the geometric centroid), or it may be a point on the frame image corresponding to the actual centroid of the parking space (e.g., the geometric centroid). Distance L2 is calculated before the parking determination process and is used each time the parking determination process is performed.

[0032] The distance L2 can be a value set as appropriate, rather than the value calculated as described above.

[0033] The determination unit 35 determines whether a vehicle is parked in a parking space for each parking determination area, based on the occupancy rate calculated by the occupancy detection unit 33.

[0034] The image processing unit 3 includes a CPU, RAM, ROM, etc., and controls each component according to information processing. The image processing unit 3 loads the program stored in the memory unit (not shown) into the RAM, and the CPU interprets and executes the program. As a result, the image processing unit 3 operates as a parking space setting unit 31, a vehicle area setting unit 32, an occupancy detection unit 33, a distance calculation unit 34, and a determination unit 35.

[0035] §2 Example of Operation Figure 4 is a flowchart illustrating the parking detection process performed by the image processing unit 3.

[0036] First, the image processing unit 3 detects vehicles captured in the frame image using the vehicle region setting unit 32 (s1). The vehicle region setting unit 32 also detects the region on the frame image in which the vehicle is captured on a pixel-by-pixel basis and sets that region as a vehicle region (s1). A vehicle region is set for each individual vehicle, and each vehicle region is distinguishable from the others. The vehicle region setting unit 32 may be implemented using a known object detection model, such as Mask RCNN, or it may be implemented using a known edge detection technique.

[0037] Figure 5 is a schematic diagram illustrating a vehicle area. Figure 5 shows a vehicle area 63 set on a frame image.

[0038] Next, the image processing unit 3 uses the occupancy detection unit 33 to move the parking spaces on the frame image to set up multiple parking determination areas, calculates the occupancy rate of each parking determination area within the vehicle area, and calculates the average value μ of these occupancy rates (s2).

[0039] Figure 6 is a flowchart illustrating the processing of step s2 by the occupancy detection unit 33. First, the occupancy detection unit 33 sets a parking determination area by moving the parking space on the frame image by a distance Δh in the direction of the vehicle's height (more precisely, from the bottom to the top of the vehicle) (s21). The parking determination area has the same shape and size as the parking space on the frame image. Next, it determines the occupied area of ​​the parking determination area within the vehicle area, in other words, the area where the vehicle area and the parking determination area overlap (s22). Next, it calculates the occupancy rate of the parking determination area within the vehicle area, in other words, the ratio of the area of ​​the occupied area to the area of ​​the vehicle area (s23). Finally, it moves the parking space by a total distance h max Repeat steps s21-s23 until the object is moved (s24). In other words, move the parking space in units of distance Δh. max Move and calculate the occupancy rate each time you move between parking spaces. Distance Δh is the number of pixels in the frame image that move once between parking spaces. max Δh is the maximum number of pixels to move between parking spaces on the frame image. Distance Δh is, for example, 1 pixel, and distance h max For example, this is 300 pixels. Next, the sum of the calculated occupancy rates is divided by the number of set parking determination areas to calculate the average value μ of the occupancy rate (s25). The occupancy detection unit 33 calculates the average value μ of the occupancy rate for each vehicle in each parking space.

[0040] distance h max This is set manually or automatically. Distance h max If this is set automatically, the image processing unit 3 uses, for example, the number of pixels on the side of the parking space on the frame image corresponding to the shorter side of the actual rectangular parking space to determine the distance h max Alternatively, the image processing unit 3 may set the distance h to the parking space by accumulating the heights of vehicles detected near the target parking space for a while and calculating the average value of those vehicle heights. max You may set it to that.

[0041] The occupancy detection unit 33 may set the area of ​​the original parking space, that is, the area of ​​the parking space before the vehicle moved, as one of the parking determination areas.

[0042] Alternatively, instead of gradually raising the parking space in the vehicle height direction, the occupancy detection unit 33 may randomly move the parking space within a predetermined range in the vehicle height direction.

[0043] Furthermore, the occupancy detection unit 33 may not only move the parking space, but also change the shape or size of the parking space to set the parking determination area.

[0044] Alternatively, instead of calculating the occupancy rate each time a parking determination area is set, the occupancy detection unit 33 may first set all parking determination areas and then calculate the occupancy rate for each parking determination area.

[0045] The occupancy detection unit 33 may calculate the root mean square of the occupancy rates by squaring each occupancy rate, averaging them, and taking the square root, instead of the average value μ of the occupancy rates. Alternatively, the occupancy detection unit 33 may calculate only the sum of the occupancy rates without calculating the average value μ of the occupancy rates.

[0046] Figure 7 is a schematic diagram illustrating an example of a parking determination area. Figure 8 is a schematic diagram illustrating an example of an occupied area. Figures 7 and 8 show enlarged portions of the frame image shown in Figure 5. Figure 7 shows the parking determination area 64 set for a parking space 61C. The parking determination area 64 is arranged in the vehicle height direction along the vertical direction of the frame image. Figure 8 shows the occupied area of ​​the parking determination area 64 in the vehicle area 63B.

[0047] Here, in order to distinguish between each parking space 61, they are referred to as parking space 61A, 61B, 61C, and 61D. Similarly, in order to distinguish between each vehicle area 63, they are referred to as vehicle area 63A, 63B, and 63C. This is also the case in Figure 9 and its explanation.

[0048] Note that in FIG. 7, only a part of the parking determination area 64 set for the parking space 61C is shown. Also, in FIG. 7, the parking determination areas 64 set for the other parking spaces 61 are not shown.

[0049] Next, the image processing unit 3 determines, by the determination unit 35, in each parking space, a vehicle whose average value μ of the occupancy rate is equal to or greater than a predetermined threshold μ TH as a candidate for a vehicle parked in that vehicle space (S3). The determination unit 35 does not include, in the candidates for vehicles parked in the vehicle space, a vehicle whose average value μ of the occupancy rate is less than the threshold μ TH in each parking space. That is, the determination unit 35 determines whether a vehicle is parked in the parking space based on the total occupancy rate.

[0050] The determination unit 35 may exclude, from the candidates for vehicles parked in the vehicle space, a vehicle whose occupancy rate does not reach the threshold γ TH regardless of whether the average value μ of the occupancy rate is equal to or greater than the threshold μ in each parking space. That is, in the target parking space, if none of the occupancy rates for the target vehicle reaches the threshold γ TH the determination unit 35 may determine that the vehicle is not parked in that parking space. The threshold γ TH is an example of the "first threshold" of the present invention. TH

[0051] Instead of the condition of whether the average value μ of the occupancy rate is equal to or greater than the threshold μ, the determination unit 35 may determine the vehicle candidates according to the following conditions. That is, in each parking space, the determination unit 35 may determine, as a candidate for a vehicle parked in that vehicle space, a vehicle whose root mean square calculated by the occupancy detection unit 33 is equal to or greater than a predetermined threshold. Alternatively, in each parking space, the determination unit 35 may determine, as a candidate for a vehicle parked in that vehicle space, a vehicle whose total occupancy rate calculated by the occupancy detection unit 33 is equal to or greater than a predetermined threshold. TH

[0052] ​​Next, the image processing unit 3 uses the distance calculation unit 34 to calculate the distance L1 between the vehicle position and the centroid of the parking space on the frame image (s4). The distance calculation unit 34 calculates the distance L1 for each parking space for the vehicle candidate determined in step s3. The vehicle position is determined on the boundary line between the vehicle's contact surface and the vehicle area. In particular, the vehicle position may be a position on the frame image that bisects the boundary line between the vehicle's contact surface and the vehicle area in the vehicle width direction.

[0053] The vehicle position may be determined within the vehicle area, rather than on the boundary line between the vehicle's contact surface and the vehicle area.

[0054] Figure 9 is a schematic diagram illustrating distances L1 and L2. Figure 9 shows an enlarged portion of the frame image shown in Figure 5. Figure 9 shows the distance L1 between the vehicle position 66 of the vehicle having the vehicle area 63B and the centroid 67 of the parking space 61C, and the distance L2 between the centroid 67 of the parking space 61C and the vertex 62 of the parking space 61C furthest from the centroid 67. The vehicle position 66 is determined to be at the center of the line segment on the outer perimeter on the near side (camera side) of the captured image in the vehicle area 63B.

[0055] Furthermore, the distance calculation unit 34 does not perform any processing on a vehicle square if there are no candidate vehicles parked in that square.

[0056] Next, the image processing unit 3 uses the determination unit 35 to determine whether a vehicle is parked in each parking space (s5). Specifically, the determination unit 35 determines the candidate vehicle for each parking space, i.e., the threshold μ TH If there is a vehicle with the above average occupancy rate μ and a distance L1 less than L2, it is determined that the vehicle is parked in that parking space. On the other hand, the determination unit 35 determines the threshold μ for each parking space. THIf the average value μ of the above occupancy rate is present and there are no vehicles with a distance L1 less than L2, it is determined that no vehicle is parked in that parking space. That is, the determination unit 35 determines that for each parking space, there are no vehicles parked in that parking space if the distance L1 is longer than the distance L2.

[0057] However, the determination unit 35 has a threshold μ TH If there are multiple vehicles that have the above average occupancy rate μ and a distance L1 less than L2, the vehicle with the shortest distance L1 among those vehicles is determined to be the vehicle parked in the parking space. However, the determination unit 35 considers the threshold μ TH If there are multiple vehicles with the same average occupancy rate μ, a distance L1 less than L2, and an equal distance L1, the vehicle assigned the smallest identification number among those vehicles is determined to be the vehicle parked in the parking space.

[0058] Note that comparing distance L1 and distance L2 is not a mandatory process. That is, the determination unit 35 may determine that the vehicle candidate determined in step s3 is the vehicle parked in the parking space. However, if there are multiple vehicle candidates, the determination unit 35 may determine that the vehicle with the shortest distance L1 among them is the vehicle parked in the parking space. However, if there are multiple vehicle candidates with the same distance L1, the determination unit 35 may determine that the vehicle assigned the smallest identification number among them is the vehicle parked in the parking space.

[0059] Furthermore, when there are multiple candidate vehicles, comparing the distance L1 between them or comparing their identification numbers is not a mandatory process. For example, when there are multiple candidate vehicles, the determination unit 35 may determine which vehicle is parked in the parking space by comparing the average value μ of the occupancy rate, the occupancy rate for each parking determination area, etc., among the candidates.

[0060] Thus, the determination unit 35 does not need to use the distance L1 for determination. In this case, the image processing unit 3 does not need to include the distance calculation unit 34.

[0061] The image processing device 1 may output maintenance image data in response to instructions from the operator. The host device of the image processing device 1 may display a maintenance screen according to the maintenance image data.

[0062] Figure 10 is a schematic diagram illustrating a maintenance screen. On this maintenance screen, the center of gravity 67 of the parking space 61, as well as the vehicle area 63, the vehicle position 66, and the matching result 68 between the parking space and the vehicle are displayed superimposed on the captured frame image. The matching result between the parking space and the vehicle is a combination of a parking space and a vehicle determined to be parked in that parking space, and is represented by a line segment connecting the center of gravity of the parking space and the vehicle position of the vehicle determined to be parked in that parking space.

[0063] According to this embodiment, for each parking determination area set by moving the parking space in the vehicle height direction on the frame image, the occupancy rate of the parking determination area within the vehicle area is calculated, and based on these occupancy rates, it is determined whether or not a vehicle is parked in the parking space. Therefore, even if a part of a vehicle parked in an adjacent parking space is captured in the frame image of the target parking space, the presence or absence of a vehicle in the parking space can be detected with high accuracy.

[0064] Furthermore, the system calculates the distance L1 between the vehicle's position and the center of gravity of the parking space, and identifies vehicles that are too far from the center of gravity of the parking space as vehicles that are not parked in that space. This reduces the error of mistakenly identifying a vehicle that is not parked in a given parking space as one that is. In particular, it reduces the error of mistakenly identifying an empty parking space as one that is parked.

[0065] Furthermore, by allowing parking status to be determined for each individual parking space, more detailed information can be provided to parking lot users, thereby enabling more efficient use of the parking lot.

[0066] In particular, at service areas and parking areas on expressways, the shortage of parking spaces for long-haul trucks during rest stops has become a serious problem in recent years. This embodiment enables efficient use of parking spaces, thus contributing to solving the shortage of parking spaces for long-haul trucks.

[0067] Furthermore, images captured by cameras are used to determine parking conditions. Cameras are easier to install than embedded sensors such as geomagnetic sensors or ultrasonic sensors installed above, and a single camera can monitor a wider area of ​​parking spaces. Therefore, the installation costs of the system for determining parking conditions can be reduced.

[0068] The correspondence between the configuration according to the present invention and the configuration according to the above embodiment can be described as follows.

[0069] <Note> The image input unit (2) receives a frame image of the target area defined as the vehicle's contact surface in the parking space (61), A vehicle area setting unit (32) detects a vehicle captured in the frame image and sets the area on the frame image in which the detected vehicle is captured as a vehicle area (63), A determination area setting unit (33) sets a plurality of parking determination areas (64) on the frame image for the parking space (61) based on the imaging position of the parking space (61) on the frame image, For each of the multiple parking determination areas (64) set by the determination area setting unit (33), a calculation unit (33) calculates the occupancy rate of the parking determination area (64) in the vehicle area (63), Each of the parking determination areas (64) includes a determination unit (35) that determines whether the vehicle is parked in the parking space (61) based on the occupancy rate calculated by the calculation unit (33), The determination area setting unit (33) is an image processing device (1) that sets a plurality of parking determination areas (64) on the frame image by moving the imaging position of the parking space (61) in the height direction of the vehicle. [Explanation of Symbols]

[0070] 1…Image processing device 2…Image input section 3…Image Processing Unit 4…Output section 5…Imaging device 31... Parking space setting section 32... Vehicle area setting unit 33... Occupancy detection unit 34... Distance calculation unit 35…Judgment section

Claims

1. An image input unit receives a frame image of a target area where the parking space is defined as the contact surface of the vehicle, A vehicle area setting unit detects vehicles captured in the frame image and sets the area on the frame image in which the detected vehicles are captured as a vehicle area. A determination area setting unit sets a plurality of parking determination areas on the frame image for the parking space based on the imaging position of the parking space on the frame image, A calculation unit calculates the occupancy rate of the parking determination area in the vehicle area for each of the multiple parking determination areas set by the determination area setting unit, and calculates the degree of occupancy of the parking determination area in the vehicle area from the occupancy rate, The system includes a determination unit that determines if the occupancy level calculated by the calculation unit is higher than a threshold, and that the vehicle is parked in the parking space. The determination area setting unit is an image processing device that sets a plurality of parking determination areas on the frame image by moving the imaging position of the parking space in the height direction of the vehicle.

2. The image processing apparatus according to claim 1, wherein the determination unit determines that a vehicle is parked in the parking space if the sum of the occupancy rates calculated by the calculation unit as the degree of occupancy is higher than a threshold.

3. The image processing apparatus according to claim 1 or 2, wherein the determination unit determines that, for each parking determination area, if any of the occupancy rates calculated by the calculation unit does not reach a first threshold, the vehicle is not parked in the parking space.

4. The frame image includes a distance calculation unit that calculates a first distance between the vehicle position defined on the boundary line between the ground surface and the vehicle area and the center of gravity of the parking space, The image processing apparatus according to any one of claims 1 to 3, wherein, if there are multiple vehicles that the calculation unit has determined to be parked in the parking space based on the degree of occupancy calculated by the calculation unit, the determination unit determines that the vehicle for which the first distance calculated by the distance calculation unit is the smallest is parked in the parking space.

5. The distance calculation unit calculates the maximum distance from the centroid of the parking space to the outer perimeter of the parking space on the frame image as the second distance. The image processing apparatus according to claim 4, wherein the determination unit determines that a vehicle whose first distance is longer than the second distance is not parked in the parking space.

6. The image processing apparatus according to claim 4 or 5, wherein the vehicle position is a position on the frame image that bisects the boundary line between the ground surface and the vehicle area in the vehicle width direction of the vehicle.

7. The process involves inputting a frame image of a target area where the parking space is defined as the contact surface of the vehicle, and The steps include detecting a vehicle captured in the frame image and setting the region on the frame image in which the detected vehicle is captured as the vehicle region, The steps include setting a plurality of parking determination regions on the frame image for the parking space based on the imaging position of the parking space on the frame image, The steps include: calculating the occupancy rate of the parking determination area in the vehicle area for each of the multiple parking determination areas set in the step of setting the parking determination area; and calculating the degree of occupancy of the parking determination area in the vehicle area from the occupancy rate; The computer performs the following steps: If the calculated occupancy level is higher than a threshold, it determines that the vehicle is parked in the parking space. An image processing method in which, in the step of setting the parking determination area, a computer sets up a plurality of parking determination areas by moving the imaging position of the parking space on the frame image in the height direction of the vehicle.

8. The process involves inputting a frame image of a target area where the parking space is defined as the contact surface of the vehicle, and The steps include detecting a vehicle captured in the frame image and setting the region on the frame image in which the detected vehicle is captured as the vehicle region, The steps include setting a plurality of parking determination regions on the frame image for the parking space based on the imaging position of the parking space on the frame image, The steps include: calculating the occupancy rate of the parking determination area in the vehicle area for each of the multiple parking determination areas set in the step of setting the parking determination area; and calculating the degree of occupancy of the parking determination area in the vehicle area from the occupancy rate; If the calculated occupancy level is higher than a threshold, the computer is instructed to perform the following steps: determine that the vehicle is parked in the parking space; An image processing program that, in the step of setting the parking determination area, causes a computer to set up a plurality of parking determination areas by moving the imaging position of the parking space on the frame image in the height direction of the vehicle.

Images