Human flow determination method, human flow determination system, and human flow determination program

Abstract

To provide a technique for determining a human flow based on passing speeds of passers-by and an index relating to a passer-by density in a target space.SOLUTION: A human flow determination method includes: an image acquisition step of acquiring a space image which is a dynamic image or a plurality of static images capturing a target space including passers-by; a passer-by density calculation step of analyzing the space image acquired in the image acquisition step to calculate an index relating to a passer-by density in the target space; a passer-by identification step of analyzing the space image acquired in the image acquisition step to identify the passers-by; a movement track analysis step of analyzing a movement track of the passers-by identified in the passer-by identification step; a passing speed calculation step of calculating passing speeds of the passers-by from the movement track of the passers-by analyzed in the movement track analysis step; and a human flow determination step of determining whether or not a human flow in the target space satisfies a predetermined condition based on the calculated index relating to the passer-by density and the passing speed.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a people flow determination method. [Background technology]

[0002] There is technology that can predict how crowded a station will become before it gets too crowded by using information that affects the increase or decrease in the number of station users, such as train operation information, in addition to the number of people entering through the ticket gates, to estimate the number of users that will flow into the station when an irregular situation such as a train delay occurs (for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2016-150664 Summary of the Invention [Problem to be solved by the invention]

[0004] In the case of Patent Document 1, predictions are made using the number of users passing through a specified location and information that affects the increase or decrease in the number of station users, so it is possible to make some predictions before the station becomes overcrowded, but users tend to accumulate and become overcrowded gradually, and it was difficult to predict people flow taking this trend into account, so there was room for improvement.

[0005] The present invention has been made in view of the above-mentioned problems, and relates to a technology for determining people flow based on an index relating to the speed and density of pedestrians passing through a target space. [Means for solving the problem]

[0006] The present invention relates to a pedestrian flow determination method including an image acquisition step of acquiring a spatial image, which is a moving image or a plurality of still images of a target space including passersby; a pedestrian density calculation step of analyzing the spatial image acquired by the image acquisition step and calculating an index related to the density of passersby in the target space; a pedestrian identification step of analyzing the spatial image acquired by the image acquisition step and identifying the passersby; a movement trajectory analysis step of analyzing the movement trajectories of the passersby identified by the passerby identification step; a movement speed calculation step of calculating the movement speed of the passersby from the movement trajectories of the passersby analyzed by the movement trajectory analysis step; and a pedestrian flow determination step of determining whether the pedestrian flow in the target space satisfies a predetermined condition based on the calculated index related to the pedestrian density and the movement speed.

[0007] The present invention also relates to a pedestrian flow determination system including: an image acquisition means for acquiring a spatial image, which is a moving image or a plurality of still images of a target space including passersby; a pedestrian density calculation means for analyzing the spatial image acquired by the image acquisition means and calculating an index related to the density of passersby in the target space; a pedestrian identification means for analyzing the spatial image acquired by the image acquisition means and identifying the passersby; a movement trajectory analysis means for analyzing the movement trajectories of the passersby identified by the passerby identification means; a movement speed calculation means for calculating the movement speed of the passersby from the movement trajectories of the passersby analyzed by the movement trajectory analysis means; and a pedestrian flow determination means for determining whether the pedestrian flow in the target space satisfies predetermined conditions based on the calculated index related to the pedestrian density and the movement speed.

[0008] The present invention also relates to a pedestrian flow determination program, which is a computer program to be executed by a computer device, and includes: an image input process for inputting a spatial image, which is a moving image or a plurality of still images of a target space including passersby, to the computer device; a pedestrian density calculation process for analyzing the spatial image input by the image input process and calculating the density of passersby in the target space; a pedestrian identification process for analyzing the spatial image input by the image input process and identifying the passersby; a movement trajectory analysis process for analyzing the movement trajectories of the passersby identified by the passerby identification process; a movement speed calculation process for calculating the movement speed of the passersby from the movement trajectory of the passersby analyzed by the movement trajectory analysis process; and a pedestrian flow determination process for determining whether the pedestrian flow in the target space satisfies predetermined conditions based on the calculated pedestrian density and movement speed. [Effects of the Invention]

[0009] The technology provided by the present invention determines the flow of people based on indicators related to the speed of pedestrians passing through the target space and the density of pedestrians, making it possible to determine the flow of people in a way that allows for the understanding of gradually changing pedestrian congestion and congestion. [Brief explanation of the drawings]

[0010] [Figure 1] 1A is an image of an image of an escalator, which is an example of a target space, and FIG. 1B is an image of an image of an escalator, which is an example of a target space. [Figure 2] 1A is an image of an image of an escalator, which is an example of a target space, and FIG. 1B is an image of an image of an escalator, which is an example of a target space. [Figure 3] 1 is a flowchart of a people flow determination method. [Figure 4] is a flowchart of the analysis process. [Figure 5] 1 is a conceptual diagram of a method for identifying a passerby. [Figure 6] FIG. 10 is a conceptual diagram of a passerby information table. [Figure 7] is an example of a determination condition. [Figure 8] is an example of a determination condition. [Figure 9] 1 is a block diagram of a people flow determination system. [Figure 10] is an example of a target space and a neighboring space. [Figure 11] 1 is a conceptual diagram showing the relationship between the head and foot positions of a pedestrian. [Figure 12] 1 is a conceptual diagram of a method for calculating the coordinates of the foot position of a pedestrian. DETAILED DESCRIPTION OF THE INVENTION

[0011] Hereinafter, an embodiment of the present invention will be described. FIG. 1 shows an overview of a target space for determining people flow using the people flow determination method (hereinafter, sometimes referred to as "this method"). Fig. 1(a) is a spatial image of the target space S captured at a predetermined timing (t), and Fig. 1(b) is a spatial image of the target space S captured at the timing (t+1) following the predetermined timing. Fig. 2(a) is a spatial image of the target space S captured at a predetermined timing (t) in a more mixed state than Fig. 1(a), and Fig. 2(b) is a spatial image of the target space S captured at the timing (t+1) following the predetermined timing. Since the target space S is an escalator 10, it is assumed that passersby M move in a fixed direction, and that the escalator 10 moves in the direction of the arrows shown from Fig. 1(a) to Fig. 2(b). In addition, an imaging device 300 (not shown) is provided at a position where it can image an area including the target space S.

[0012] Next, the processing content of this method will be described. The method includes an image acquisition step, a pedestrian density calculation step, a pedestrian identification step, a movement trajectory analysis step, a passing speed calculation step, and a people flow determination step. It may also include a notification step, an object identification step, and a people flow prediction step.

[0013] The "image acquisition step" is a step of acquiring a space image, which is a moving image or a plurality of still images of a target space including passersby. A "passerby" is a person present in the target space, regardless of age, sex, walking style, or whether they are walking, running, standing, sitting, or any other movement state. The "target space" is a space where people flow is desired to be determined, and is also a space where passersby exist, and specifically, it applies to spaces such as inside train stations, airports, shopping malls, and specific spaces inside train stations, airports, and shopping malls, such as stairs, escalators, passageways, elevators, entrances, etc. "People flow" means the flow of people, and in this embodiment, it includes not only when, from where to where, and how long it took people to move, but also how many people are present in the target space and the degree of congestion in the target space. A "spatial image" is a moving image or multiple still images captured of a target space including pedestrians. In this method, the movement trajectories of pedestrians are analyzed to calculate indices related to pedestrian density and traffic speed. Since the movement trajectories of pedestrians cannot be analyzed with only one image (one still image) taken at a single point in time, it is necessary to acquire a moving image or multiple still images. Furthermore, the spatial image is an image captured by an imaging device 300 (camera). The imaging device 300 may be a general RGB camera or a monochrome camera, and there are no restrictions on the performance or specifications of the imaging device 300.

[0014] The "passenger density calculation step" is a step of analyzing the spatial image acquired in the image acquisition step and calculating an index related to the pedestrian density in the target space. A "pedestrian density index" is a value that represents the number of pedestrians present in a target space. Generally, when there are many pedestrians in the same target space, the pedestrian density is higher than when there are few pedestrians in the same target space. Furthermore, even if the number of pedestrians present is the same depending on their physical size, the pedestrian density is higher when large pedestrians are present than when small pedestrians (e.g., infants) are present. Therefore, a pedestrian density index is a value that represents the number of pedestrians present in a target space, such as a value based on the floor area of ​​the target space and the number of pedestrians in the target space, a value based on the sum of the floor area of ​​the target space and the equivalent of the ground contact area of ​​pedestrians in the target space (total pedestrians present), a value based on the floor area of ​​the target space and the equivalent of the maximum horizontal cross-sectional area of ​​pedestrians (total pedestrians present), or the ratio of the total volume of pedestrians (total pedestrians present) to the volume of the target space. The term "calculating" includes not only calculating the index numerically, but also calculating the index in color or in graph form.

[0015] The "passerby identification step" is a step of analyzing the spatial image acquired in the image acquisition step and identifying the passerby. "Identifying passersby" refers to identifying passersby, which is necessary when calculating the passing speed, because the spatial image contains objects other than passersby that exist in the target space, such as pillars, signs, and guide boards. Note that it is not necessary to identify individual passersby; it is sufficient to identify passersby (people). Any method for identifying passersby may be used, and in this embodiment, passersby are identified using a trained model obtained by machine learning.

[0016] The "movement trajectory analysis step" is a step of analyzing the movement trajectory of the identified passerby. "Movement trajectory" refers to tracking a pedestrian's entry into a target space and their movement within the target space. Here, "tracking" refers to, when analyzing spatial images of the target space captured at predetermined intervals, identifying the position of the pedestrian in the spatial image captured at the predetermined timing, and then, when analyzing the spatial image captured at the next predetermined timing, identifying how far the same pedestrian has moved from their previous position. The pedestrian's position can be any part of the body, but it is preferable that it be the same part before and after the movement, and it is also preferable that the pedestrian's position be at their feet.

[0017] The "passenger speed calculation step" is a step of calculating the passing speed of the pedestrian from the analyzed movement trajectory of the pedestrian. "Pedestrian passing speed" refers to the passing or moving speed of pedestrians in the target space, regardless of the type of movement, such as walking, running, etc. In this embodiment, "pedestrian passing speed" includes both the passing speed of a single pedestrian and a value calculated from the passing speeds of multiple pedestrians in the target space (a total value, an average value, a value classified by a predetermined threshold (e.g., fast, normal, slow) etc.).

[0018] The "people flow determination step" is a step of determining whether the people flow in the target space satisfies a predetermined condition based on the calculated pedestrian density index and the passing speed. The "predetermined condition" in "people flow satisfies the predetermined condition" refers to a condition for determining or estimating the state of people flow, such as the state of people flow at a predetermined time, the state of people flow a predetermined period after the predetermined time (future people flow state), the state of people flow a predetermined period before the predetermined time (past people flow state), the state of people flow in a predetermined space related to the target space, etc. Also, if there is processing to be performed based on the determination result, the condition may be whether or not to execute the processing based on the determination result.

[0019] The "alert process" is a process of issuing an alert depending on the determination result of the people flow determination process. "Alert process" depending on the people flow state means varying the alert mode (alert content) depending on the people flow state, such as when there is a possibility of danger depending on the people flow state (when the state is congested or when it is estimated that it will become congested), when there is a low possibility of danger depending on the people flow state (when the state is empty or when it is estimated that it will become empty), or when it is possible to attract people depending on the people flow state (when the state is empty or when it is estimated that it will become empty).

[0020] The "object identification step" is a step of analyzing the acquired spatial image and identifying an object other than a passerby that is equal to or larger than a predetermined size. An "object of a predetermined size or larger" is an object that affects the flow of people in the target space, makes it difficult for pedestrians to walk, occupies space for pedestrians, and is not present in the target space in its initial state (at the start of business, operation, or operation). Therefore, objects that are present in the target space even in its initial state (e.g., pillars, signboards, advertisements) are not included in the objects identified in the object identification process. Furthermore, because the size that affects pedestrian flow varies depending on the area and shape of the target space, it is preferable to set the "predetermined size" according to the target space. Furthermore, regardless of the area and shape of the target space, the "predetermined size" may be a size that obstructs pedestrian passage or obstructs the flow of pedestrians when multiple objects accumulate, such as the size of a beverage container (e.g., a 120ml beverage bottle, a 160ml beverage can, or a 350ml PET beverage bottle).

[0021] The "people flow prediction step" is a step of predicting people flow in the adjacent space based on the determination result in the people flow determination step. An "adjacent space" is a space adjacent to the target space, and may be not only a space adjacent to the target space, but also a space that partially overlaps with the target space or a space separated from the target space by a certain distance, and is a space in which pedestrians entering the target space exist, and / or a space into which pedestrians exiting the target space enter.

[0022] FIG. 3 shows the main flow chart of the method. Step S100 is an image acquisition process, which corresponds to the image acquisition step described above. Step S200 is a process of analyzing the acquired spatial image, and corresponds to the pedestrian density calculation step, the pedestrian identification step, the movement trajectory analysis step, and the traffic speed calculation step. If the object identification step is included, it is included in step S200. Step S300 is a people flow determination process, which corresponds to the people flow determination step described above.

[0023] FIG. 4 shows a detailed flowchart of the process of analyzing the acquired spatial image in step S200. Step S205 is a process for calculating an index relating to pedestrian density, which corresponds to a pedestrian density calculation step. A method for calculating an index relating to pedestrian density will be described.

[0024] FIG. 5 shows a passerby M present in a target space S. In this embodiment, the passerby M is identified by identifying the head of the passerby M present in the target space S. As shown in FIG. 5, a region 501 is defined as a region for identifying the head of the passerby M captured in the spatial image. The region 501 for identifying the head of the passerby M is determined taking into consideration the average height of an adult. Furthermore, when the target space S is an escalator as in this embodiment, the distance (height) between the image capture device 300 and the feet of the passerby M varies depending on the standing position of the passerby M, so it is preferable to determine the region 501 for identifying the head taking into consideration the position of the passerby M. Furthermore, a passerby M farther from the image capture device 300 appears smaller than a passerby M closer to the image capture device 300, so it is preferable to determine the region 501 for identifying the head taking into consideration the position from the image capture device 300. In this way, the number of passersby M present in the target space S is calculated from the identified head, and an index related to passerby density is calculated from the calculated number of people and the size of the target space S, for example, the floor area of ​​the target space S. In this embodiment, the value shown in formula (1) is calculated as the index related to passerby density. Number of pedestrians / floor area of ​​target space = index related to pedestrian density ··· Equation (1)

[0025] Step S210 is a process for identifying a passerby M. As in step S205, the passerby M is identified by identifying the head of the passerby M present in the target space. As shown in FIG. 5, a region 501 for identifying the head of the passerby M captured in the spatial image is defined as the region 501. The region 501 for identifying the head of the passerby M is determined taking into consideration the average height of an adult. Furthermore, because a passerby M farther from the image capture device 300 appears smaller than a passerby M closer to the image capture device 300, it is preferable to determine the region 501 for identifying the head by taking into consideration the position of the passerby M from the image capture device 300. Furthermore, a predetermined position of the identified head, for example, the center position of the top edge, is calculated as the passerby coordinates of the passerby M. Note that this process is the same as the process for identifying the passerby M performed in step S205, and therefore the results identified in step S205 may be used. In step S215, it is determined whether the passerby M identified in step S210 is a new passerby. Here, a new passerby refers to a passerby M who was not present in the spatial image captured at a predetermined time (e.g., t-1, t-2) before the spatial image used in the current processing (t) was captured. Specifically, the passerby information table is referenced to determine whether there is a passerby M whose passerby information (identification information ID) stores passerby coordinates within a predetermined range from the passerby coordinates calculated for the current spatial image. If there is no passerby M, the passerby is a new passerby, and the process proceeds to step S220. If there is a passerby M, the passerby is not a new passerby, and the process proceeds to step S225. Note that the area of ​​the passerby information table from the previous time (t-1) is referenced to determine whether passerby coordinates within the predetermined range are stored, and the area from the time before last (t-2) is also referenced. This is because, assuming that a person may not have been identified as a passerby the previous time due to the influence of the captured image, it is possible to more reliably analyze the movement trajectory of a passerby by not limiting it to the previous area. It is preferable to determine the range of the passerby information table to be referenced, taking into consideration the processing performance and desired performance of the arithmetic processing device. FIG. 6 shows an example of the passerby information table.

[0026] The passerby information table stores identification information (passerby ID) assigned to passerby M and the passerby coordinates of the passerby M, in association with each timing at which a spatial image was captured. In the table, "xxx" represents the passerby coordinate value, and "-" represents NULL, indicating that no data exists. In this embodiment, the data is stored in association with the image capture time. FIG. 6 shows that passerby M1 was present in the target space S from t-2 to t+1, and passerby M2 was present in the target space S from t-1 to t+2. For ease of explanation, areas shaded in the passerby information table indicate the same coordinate values. Therefore, passerby M+n was present in the same position in the target space S from t+4 onward. If the spatial image is a still image, the coordinate values ​​of passersby are stored for each imaging timing, for example, 20 spatial images per 3 seconds if imaging is performed every 3 seconds, or 10 spatial images per 6 seconds if imaging is performed every 6 seconds. If imaging is performed every 3 seconds, the difference between "t" and "t+1" in the passerby information table is 3 seconds, and the difference between "t" and "t+2" is 6 seconds. If imaging is performed every 6 seconds, the difference between "t" and "t+1" in the passerby information table is 6 seconds, and the difference between "t" and "t+2" is 12 seconds. If the spatial image is a moving image, the coordinate values ​​of passersby are stored for each predetermined period, for example, 20 times per minute if pedestrian coordinate values ​​are calculated every 3 seconds, or 10 times per minute if pedestrian coordinate values ​​are calculated every 6 seconds. As for the traveling speed, the travel speed of pedestrian M calculated in the current analysis process is stored in association with the pedestrian ID, as will be described in detail later. The pedestrian speed is updated to the value calculated in the latest analysis process and stored, so one value is stored for each pedestrian ID.

[0027] Step S220 is a process performed when a new pedestrian is determined in step S215, and is a process of adding pedestrian information. The pedestrian information includes at least identification information (pedestrian ID), pedestrian coordinate information, and traveling speed information. The identification information (passer-by ID) is unique information that is added when a passer-by first appears in the target space image. In this embodiment, it is denoted as M1, M2, M3, . . . The passerby coordinate information is the value of the passerby coordinate identified in step 210 . In step S215, since it is determined that there is no passerby M in the passerby information whose passerby coordinates are stored within a predetermined range from the passerby coordinates calculated for the spatial image at this time (t), new identification information (passerby ID) is added to the passerby M identified at this time (t). In addition, the passerby coordinate values ​​calculated as passerby coordinate information are stored. In addition, since this is a new passerby, 0 is stored as the passing speed.

[0028] Step S225 is a process performed when it is determined in step S215 that the passerby M is not a new passerby. In this step, passerby information of the passerby M identified in step S210 is extracted from the passerby M present in the spatial image captured at the previous predetermined timing (t-1). Specifically, among the passerby coordinates of the previous time (t-1) in the passerby information table, passerby coordinates that fall within a predetermined range of the passerby coordinates calculated for the current spatial image (t) are identified, passerby information storing the passerby coordinates is extracted, identification information (passerby ID) included in the passerby information is assigned to the passerby M, and the passerby information table is updated. In this way, the passerby M present in the spatial image captured at the previous time (t-1) and the passerby M present in the spatial image captured at the current time (t) can be managed as the same person, making it possible to analyze the movement trajectory of the passerby M.

[0029] Step S230 is a process of analyzing the movement trajectory of the passerby M from the previous time (t-1) to the current time (t) based on the data stored in the passerby information table updated in step S225. In this embodiment, the movement speed of the passerby M after entering the target space S is calculated and used for people flow determination, so the movement trajectory of the passerby M is analyzed.

[0030] Step S235 is a process for calculating the traveling speed of the pedestrian M moving through the target space S based on the analysis results of step S230. In this embodiment, the position of the pedestrian is identified in a spatial image captured at a predetermined timing, and when analyzing a spatial image captured at the next predetermined timing, the movement trajectory of the pedestrian M is analyzed by tracking the pedestrian M moving through the target space while identifying how far the same pedestrian M has moved from the previous position. The traveling speed is calculated based on the analysis results of this movement trajectory. Specifically, the previous time and position coordinates of the pedestrian M, and the current time and position coordinates are extracted from the pedestrian information table, and the time and distance required for the movement of the pedestrian M are calculated. By analyzing the movement trajectory, the distance traveled (traveled) by the pedestrian M is identified, and the time required for the movement (travel) is identified based on the time of image capture, so that the traveling speed can be calculated. Then, the traveling speed in the pedestrian information table of FIG. 6 is updated to the calculated value.

[0031] As described above, the process in Fig. 5 is performed for each spatial image. Furthermore, if multiple passersby are identified in the spatial image in step S210, the processes in steps S210 to S235 are performed for each of the multiple passersby.

[0032] Step S245 is a process of identifying an object other than a pedestrian that exists in the target space S and that is equal to or larger than a predetermined size, and determining whether it exists within a predetermined range for a predetermined period of time. The term "object" refers to portable objects such as bags and suitcases carried by station users or deliveries by vendors. It does not include objects fixedly installed in the target space, such as pillars and walls, or movable objects in the target space, such as the handrails and steps of escalators 10. Furthermore, since the purpose of identifying objects using this method is to determine whether an object's presence in the same location for a predetermined period of time affects pedestrian flow, in this embodiment, the term "predetermined size" refers to a size that obstructs passage, such as an object the size of a beverage container (e.g., a 120 ml beverage bottle, a 160 ml beverage can, or a 350 ml PET bottle). Note that objects are not limited to cylindrical or prismatic shapes, and may be approximately rectangular, spherical, or other shapes. The object to be identified is one whose shape is anticipated in advance. For example, if it is a beverage container, training data learned from images of various types of beverage containers can be stored in memory unit 190, and the acquired spatial image can be analyzed and identified. However, identification is difficult when an unexpected object is present. Therefore, an image of the target space S at a predetermined time (e.g., at the start of business hours) is captured (referred to as the "baseline image"), and any objects not captured in the baseline image are identified as objects. The term "within a predetermined range" refers to the range within which an object is determined to exist in the same position (same coordinates), including the range within which the object's location changes due to the vibration of the escalator's movement and the presence of pedestrians. Therefore, when determining whether an object exists in the same position, any object within the predetermined range is determined to exist in the same position, even if the object's coordinates are not exactly the same. In this embodiment, since the target space is an escalator, the predetermined range is set to one step (approximately 40 centimeters). The term "predetermined period" refers to the period within which an object is determined to exist within the predetermined range, and is the period during which it can be estimated that the object is remaining in the same position (abandoned). For example, the period is the period required for pedestrians to travel the entire length of the target space in the direction of travel at a normal speed. This period is the period during which an object may exist in the target space under normal conditions and when pedestrians are moving the object.

[0033] In step S250, if there is an object identified in step S245, the position coordinates of the object are calculated.

[0034] Step S255 is a process for determining whether an object was previously located at the same coordinate position as the coordinate position calculated in step S250. Specifically, the previous (t-1) portion of the object information table is referenced to determine whether the same coordinate position as the coordinate position identified this time (t) exists. If the coordinate position identified this time (t) exists in the previous (t-1) time, it is not a new object, and the process proceeds to step S265. On the other hand, if the coordinate position identified this time (t) does not exist in the previous (t-1) time, it is a new object, and the process proceeds to step S260. The "object information table" is a table that associates an object ID assigned to each object, a calculated coordinate position, and information indicating the time the image was captured when an object is identified in the target space S, and is stored in the storage unit 190.

[0035] Step S260 is a process for adding object information to the object information table when it is determined to be a new object in step S255. Since step S260 is a process for when it is determined to be a new object in step S255, a new object ID is added to the object information table, and information indicating the coordinate position and the time when the image was captured is stored in association with the new object ID.

[0036] Step S265 is a process for updating the information in the object information table if it is determined in step S255 that the object is not a new object. The information indicating the current time (t) is updated in association with the object ID already stored in the information table for the identified object. This process records in the object information table that the identified object exists at the same coordinate position both last time (t-1) and this time (t).

[0037] As described above, the process in Fig. 4 is performed for each spatial image. If multiple objects are identified in the spatial image by the process in step S245, the processes in steps S245 to S265 are performed for each of the multiple objects. The processes in steps S245 to S265 correspond to the object identification process and the object identification step.

[0038] In step S300, it is determined whether the pedestrian flow in the target space satisfies a predetermined condition based on the calculated pedestrian density index and passing speed. The "predetermined condition" is a determination condition for determining the pedestrian flow state, and may be a combination of a pedestrian density index and passing speed, or a combination of a pedestrian density index and passing speed plus whether an object has been present within a predetermined range for a predetermined period of time. Figure 7 shows four patterns of the predetermined condition in this embodiment, Condition 1 to Condition 4.

[0039] "Pedestrian density (p)" in Fig. 7 is the index calculated in step S205 of Fig. 4, and is the pedestrian density at this time (t). In this embodiment, the index related to the pedestrian density is the pedestrian density calculated by equation (1). "p0" is a preset "index for standard pedestrian density" that serves as the standard for the number of pedestrians in the target space. The pedestrian density at which people can walk comfortably and safely is 4 people / m. 2 Therefore, in this embodiment, taking into consideration false detection due to noise, etc., "p0" is set to 5 people / m 2 The index for standard pedestrian density, the value indicating normal pedestrian density, is the value beyond which an abnormality is judged, and is also the limit value indicating normal pedestrian density. "p1" is a predetermined threshold value smaller than "p0", and by providing multiple judgment conditions that do not reach "p0" but are close to "p0" (close to the value that is judged to be abnormal), the state of the target space can be judged in detail. In this embodiment, the threshold is 4.5 people / m 2 Let's say. Also, the "traffic speed (v)" in FIG. 7 is the value calculated in step S235 of FIG. 4, and is the traffic speed (v) calculated when the spatial image of this time (t) was acquired. "v0" is the "standard passing speed" which is a preset reference value for the passing speed within the target space, a passing speed at which one can walk comfortably and safely, a value indicating a normal state, and in this embodiment, since the target space is the escalator 10, it is the moving speed of the escalator 10, and is set in the range of 30 to 45 m / min. The standard passing speed, the passing speed at which one can walk safely, and the value indicating a normal state are values ​​beyond which an abnormality is judged, and are also the limit value indicating a passing speed at which one can walk safely and a normal state. "v1" is a predetermined threshold value greater than "v0" (a faster value). By setting multiple determination conditions that do not reach "v0" but are close to "v0" (close to the value that is determined to be abnormal), the state of pedestrians in the target space can be determined in detail. In this embodiment, the range is 35 to 50 m / min. Alternatively, "v1" may be set to the value set for "v0" + 5 m / min. Note that the travel speed may be the travel speed (v) of a specific pedestrian among those identified using the spatial image of this time (t), or the average travel speed of all identified pedestrians. In this embodiment, the average travel speed of all identified pedestrians is used.

[0040] "Condition 1" is the case where the value of pedestrian density p is greater than "p0" and the value of passing speed v is less (slower) than "v0". In this case, the number of pedestrians present in the target space is greater than the normal value, and the passing speed of the pedestrians is slower than the moving speed of the escalator 10, causing pedestrians to accumulate, and it is determined that this is an abnormal state in which there is a possibility of an accident such as a fall. Note that if the value of pedestrian density p is simply greater than "p0" (if the passing speed is faster than v0), it is not determined to be an abnormality because it is highly likely that a state in which pedestrians are close to each other by chance has occurred. Also, if the value of passing speed v is simply less (slower) than "v0" (if the pedestrian density is less than p0), it is highly likely that a state in which the passing speed is slow by chance has occurred, and it is not determined to be an abnormality. In this way, the state of the target space is determined using both the pedestrian density and the traffic speed, so that the determination result can be adapted to the actual state of the target space.

[0041] "Condition 2" is when the value of pedestrian density p is greater than "p1" and the value of traffic speed v is smaller (slower) than "v1." In this case, there are no more pedestrians in the target space than the normal value, but the value is close to the abnormal value, and the traffic speed is faster than the normal value but close to the abnormal value, so it can be inferred that if this state continues, there is a possibility of transitioning to an abnormal state, and the judgment result is that there is a possibility of an abnormality.

[0042] "Condition 3" is when the value of pedestrian density p is greater than "p1" or the value of traffic speed v is smaller (slower) than "v1," and an object of a predetermined size or larger has been identified for a predetermined period of time within a predetermined range. In this case, there are no pedestrians in the target space with a normal value or larger (abnormal value), but the value is close to the abnormal value, or the traffic speed is faster than the normal value but close to the abnormal value, and an object of a predetermined size or larger has been present for a predetermined period of time within a predetermined range. Therefore, when the pedestrian density or traffic speed is close to the abnormal value, it can be inferred that the presence of an object of a predetermined size or larger is causing a stagnation in the flow of people, and if this state continues, it can be inferred that there is a possibility of transitioning to an abnormal state, and therefore the judgment result is that there is a possibility of an abnormality.

[0043] Condition 4 is when the value of pedestrian density p is greater than "p1," the value of traffic speed v is less (slower) than "v1," and an object of a predetermined size or larger is present within a predetermined range for a predetermined period of time. In this case, there are no pedestrians in the target space whose numbers are greater than the normal value (abnormal value), but are close to the abnormal value; the traffic speed is faster than the normal value but close to the abnormal value; and an object of a predetermined size or larger is present within a predetermined range for a predetermined period of time. Therefore, the pedestrian density and traffic speed values ​​that are close to the abnormal value indicate that the presence of an object of a predetermined size or larger is causing a disruption in the flow of people. If this state continues, it can be inferred that there is a possibility of transitioning to an abnormal state. Furthermore, because it can be inferred that there is a higher possibility of transitioning to an abnormal state than in Conditions 2 and 3, the result indicates a higher possibility of an abnormality (e.g., "high possibility of an abnormality") than the "possibility of an abnormality" judgment results for Conditions 2 and 3.

[0044] As the number of pedestrians gradually increases or gradually stagnates, the state of the target space becomes a dangerous walking state. Therefore, as shown in "Condition 2" to "Condition 4," the state of the target space can be determined using a threshold value "p1" that is smaller than the normal value for pedestrian density, a threshold value "v1" that is faster than the normal value for passing speed, and information on whether an object of a specified size or larger is present within a specified range for a specified period of time, thereby making it possible to make a judgment that is appropriate to the change in the state of the target space.

[0045] 7 are merely examples, and other conditions may be set as long as they include an index indicating pedestrian density and a traffic speed. In addition, all of conditions 1 to 4 may be set, or specific conditions may be set.

[0046] Next, the notification mode based on the determination result will be described. The notification device may be any device, such as a display device or speaker installed in a facility, for example, a station premises, or a mobile terminal carried by a facility employee, such as a station staff member. It is preferable to vary the notification mode depending on the notification result. For example, when an abnormal condition occurs under Condition 1, a notification is issued instructing pedestrians not to enter the target space. Under Conditions 2 to 4, although an abnormal condition has not yet occurred, a notification is issued stating that the target space is congested and recommending an alternative route because an abnormal condition is likely to occur in the future. Condition 4 is closer to an abnormal condition than Conditions 2 and 3, so in addition to the above notification, it is preferable to issue a notification that entry into the target space may be restricted if the number of pedestrians increases. Furthermore, when an object larger than a predetermined size is present for a predetermined period of time, as under Conditions 3 and 4, it is preferable to issue a notification that an object obstructing passage is present and to instruct station staff to remove the object obstructing passage. In this way, by issuing a notification according to the determination result, it is possible to maintain the safety of the target space.

[0047] <People flow detection system> The following describes a people flow determination system 200 (also referred to as "this system") that realizes this method. FIG. 9 shows a block diagram of the people flow determination system 200. The people flow determination system 200 includes an image acquisition unit 110, a pedestrian density calculation unit 120, a passerby identification unit 130, a movement trajectory analysis unit 140, a passage speed calculation unit 150, a people flow determination unit 160, an object identification unit 170, a people flow prediction unit 175, a notification unit 180, a display unit 185, a memory unit 190, and an imaging device 300. The image acquisition unit 110, the passerby density calculation unit 120, the passerby identification unit 130, the movement trajectory analysis unit 140, the passage speed calculation unit 150, the people flow determination unit 160, and the object identification unit 170 are provided in an arithmetic processing device 100 provided in the system.

[0048] The image acquisition unit 110 (image acquisition means) is a means for acquiring a space image, which is a moving image or a plurality of still images of a target space including passersby, and is provided in the arithmetic processing device 100. The pedestrian density calculation unit 120 (pedestrian density calculation means) is a means for analyzing the space image acquired by the image acquisition unit 110 and calculating an index relating to the pedestrian density in the target space, and is provided in the arithmetic processing device 100. The passerby identification unit 130 (passerby identification means) is means for analyzing the spatial image acquired by the image acquisition unit 110 and identifying passersby, and is provided in the arithmetic processing device 100. The movement trajectory analysis unit 140 (movement trajectory analysis means) is means for analyzing the movement trajectory of the passerby identified by the passerby identification unit 130, and is provided in the arithmetic processing device 100. The traveling speed calculation unit 150 (traveling speed calculation means) is means for calculating the traveling speed of a pedestrian from the movement trajectory of the pedestrian analyzed by the movement trajectory analysis unit 140, and is provided in the arithmetic processing device 100. The people flow determination unit 160 (people flow determination means) is a means for determining whether the people flow in the target space satisfies a predetermined condition based on the calculated index relating to the passerby density and the passing speed, and is provided in the arithmetic processing device 100. The object identifying section 170 (object identifying means) is means for analyzing the spatial image acquired by the image acquiring section 110 and identifying an object other than a passerby that is equal to or larger than a predetermined size, and is provided in the arithmetic processing device 100. The people flow prediction unit 175 (people flow prediction means) is a means for predicting people flow in the adjacent space based on the determination result by the people flow determination unit 160, and is provided in the arithmetic processing device 100. The notification unit 180 (notification means) is means for making a notification in accordance with the determination result of the people flow determination unit 160, and is provided in the arithmetic processing device 100. In addition to the above, this system is equipped with a display unit 185 that displays captured images and information necessary for processing, a memory unit 190 that stores programs and specified data necessary for processing, and input devices such as a keyboard and a pointing device (not shown).

[0049] <Computer Program> A computer program (hereinafter also referred to as the program) for causing the processing device 100 to execute the above-described people flow determination method is installed in the processing device 100 of the people flow determination system 200. In this embodiment, the computer program is software that executes the people flow determination method. This program includes image acquisition processing, pedestrian density calculation processing, pedestrian identification processing, movement trajectory analysis processing, traffic speed calculation processing, and people flow determination processing. It may also include object identification processing, people flow prediction processing, and notification processing. The image acquisition process is a process of acquiring a space image, which is a moving image or a plurality of still images, of a target space including passersby. The pedestrian density calculation process is a process of analyzing the spatial image acquired by the image acquisition process and calculating an index related to the pedestrian density in the target space. The passerby identification process is a process of analyzing the spatial image acquired by the image acquisition process and identifying the passerby. The movement trajectory analysis process is a process of analyzing the movement trajectory of the passerby identified by the passerby identification process. The travel speed calculation process is a process for calculating the travel speed of a pedestrian from the movement trajectory of the pedestrian analyzed by the movement trajectory analysis process. The people flow determination process is a process of determining whether the people flow in the target space satisfies a predetermined condition based on the calculated pedestrian density index and the passing speed. The object identification process is a process of analyzing the spatial image acquired by the image acquisition process and identifying an object other than a passerby that is equal to or larger than a predetermined size. The people flow prediction process is a process for predicting people flow in an adjacent space based on the determination result of the people flow determination process. The notification process is a process of issuing a notification in accordance with the determination result of the people flow determination process.

[0050] <Modification> The present invention is not limited to the above-described embodiment, and various modifications and improvements are possible.

[0051] In the above-described embodiment, four determination patterns are shown in Fig. 7, but the present invention is not limited to this. For example, "Condition 5" shown in Fig. 8 may be included. "Condition 5" is a case where a judgment is made using pedestrian density and traffic speed calculated at different times. In FIG. 8, "p(t)" is the pedestrian density calculated based on the spatial image at time t, and "p(tn)" is the pedestrian density calculated based on the spatial image at time tn. Also, in FIG. 8, "v(t)" is the traffic speed calculated based on the spatial image at time t, and "v(tn)" is the traffic speed calculated based on the spatial image at time tn. If p(t) is greater than p(tn) and v(t) is slower than v(tn), it is judged that there is a possibility of an abnormality. This is because it can be estimated that the number of pedestrians in the target space is increasing, so p(t) is greater than p(tn) and v(t) is slower than v(tn). This also makes it possible to judge the state of the target space. In Figure 8, the judgment is made by comparing the indices related to traffic speed and pedestrian density between time t and time tn, but it may also be determined that an abnormality exists if, for example, the traffic speed slows down multiple times and the indices related to pedestrian density increase.

[0052] In this embodiment, the index indicating the pedestrian density is calculated based on Equation (1), but this is not limited to this. For example, the acquired spatial image may be divided into a predetermined number of parts, and the proportion of images corresponding to pedestrians among the predetermined number of parts may be calculated and used as the index. In this way, the index indicating the pedestrian density in this embodiment may be calculated without identifying each individual pedestrian.

[0053] In the passerby information table shown in FIG. 6, all of the identified passerby information is stored in the storage unit 190, but this is not limited to this. For example, if the coordinate position of a passerby stored at the previous timing is outside the target space, the information of that passerby may be deleted. In the present embodiment, the target space is the escalator 10, so the direction of movement of passersby is constant. Therefore, since the coordinate position of a passerby stored at the previous timing is no longer used in processing after it is outside the target space, deleting the information of that passerby makes it possible to reduce the amount of data stored in the storage unit 190.

[0054] In this embodiment, one target space is provided, and the flow of people in the target space is determined based on an index related to pedestrian density and a passing speed, but the present invention is not limited to this. As shown in FIG. 10, a target space C may be provided adjacent to a target space A for which people flow is to be determined, and people flow in the target space C may be predicted based on the determination results for the target space A. For example, as shown in FIG. 10, the target space A is a space in which an escalator is installed, and the target space C is a space in which an automatic ticket gate 30 is installed. The people flow is determined for the target space A shown in FIG. 10 using the method described above. Furthermore, the target space C adjacent to the target space A is assumed to receive a large number of pedestrians from the target space A. In this case, it is expected that the people flow state of the target space C will be similar to that of the target space A, but with a slight delay. Therefore, the people flow state of the target space C can be predicted based on the determination results for the target space A. Furthermore, by predicting the people flow in the target space C before the people flow in the target space C actually reaches a dangerous state such as congestion, it becomes possible to take measures in response to an increase in the number of pedestrians.

[0055] Also, as shown in FIG. 10, target space A and target space B are spaces in which escalators are installed, and target space C is a space in which an automatic ticket gate 30 is installed. Target space A and target space B for determining people flow, and target space C adjacent to each of target space A and target space B, may be provided, and the people flow in target space C may be predicted based on the determination results of target space A and target space B. In this case, weighting may be applied to either the determination result of target space A or the determination result of target space B based on the positional relationship between target space A, target space B, and target space C. For example, if the number of pedestrians in target space B under normal circumstances is greater than that of target space A, the determination result of target space A may be weighted (given priority) and used to make a prediction for target space C.

[0056] In this embodiment, the target space is the escalator 10, and therefore the standard travel speed "v0" is determined as the speed of the escalator 10 in the people flow determination process, but this is not limited to this. For example, if the target space is a moving walkway, it is preferable to set the travel speed of the moving walkway to the standard normal speed "v0" as in this embodiment. Furthermore, if the target space is a space where the floor (ground) on which pedestrians pass does not move, such as a staircase or a corridor, it is preferable to set the travel speed of an average adult (average travel speed) to the standard travel speed "v0."

[0057] In this embodiment, the target space is the escalator 10, and therefore, under normal conditions, movement is possible within the range of 30 to 45 m / min, which is the movement speed of the escalator. Therefore, a judgment condition may be set that judges an abnormality to exist regardless of the value of the pedestrian density p when the movement speed is less than a predetermined value relative to the standard travel speed (for example, when "v0" is 30 m / min, if "v" is judged to be less than 28 m / min).

[0058] In this embodiment, the target space is defined as a range for determining whether an object of a predetermined size or larger exists within a predetermined range for a predetermined period of time, but this is not limited to this. It may also be determined whether an object of a predetermined size or larger exists in a specific range of the target space, in a position where an object is likely to remain, or in a position where a remaining object is likely to obstruct passage. In this embodiment, the target space is an escalator, and since it is conceivable that an object will remain near the exit of the escalator, it may be determined that an abnormality has occurred if an object of a predetermined size or larger exists in a predetermined area near the exit for a predetermined period of time.

[0059] In this embodiment, the traveling speed of the pedestrian M is calculated from the previous time and coordinate position, and the current time and coordinate position, and the traveling speed of the pedestrian M in the pedestrian information table is updated each time the traveling speed is calculated. This is because if the traveling speed were calculated from the coordinate position of the pedestrian M when it first existed in the target space S and the image capture time at that time, and the coordinate position of the pedestrian M when it last existed in the target space S and the image capture time at that time, the travel distance would be determined from the coordinate position of the pedestrian M when it first existed and the coordinate position of the pedestrian M when it last existed, even though the pedestrian M does not necessarily travel in a straight line within the target space S, and this could differ from the actual travel distance. As a result, a value different from the actual traveling speed of the pedestrian M could be calculated. However, in the present embodiment, when the target space S is defined as the range within the target space S in which the pedestrian M moves in one direction in a straight line, the travel distance of the pedestrian M within the target space S is determined in advance. Therefore, if it is determined in step S210 that the coordinate position of the passerby M is outside the target space S, the image capturing time when the passerby M first existed in the target space S and the image capturing time when the passerby M last existed in the target space S may be identified from the passerby information table, and the passing speed of the passerby M may be calculated. By performing such processing, it is not necessary to calculate the passing speed of the passerby M every time a spatial image is acquired in the analysis processing of step S200, and therefore it is possible to reduce the processing load.

[0060] In the present embodiment, the traveling speed of the passerby M is calculated from the previous time and coordinate position of the passerby M and the current time and coordinate position, and the traveling speed of the passerby M in the passerby information table is updated each time the traveling speed is calculated. However, this is not limited to this. For example, from the time when the passerby M first entered the target space S until the coordinate position of the passerby M is determined to be outside the target space S in step S210, the travel distance may be calculated from the previous coordinate position and the current coordinate position, and the travel distance may be added each time step S210 is executed to calculate the travel distance of the passerby M from the first time to the last time the passerby M entered the target space S. The traveling speed of the passerby M may then be calculated based on the calculated travel distance, the image capture time when the passerby M first entered the target space S, and the image capture time when the passerby M last entered the target space S. This process eliminates the need to calculate the traveling speed of the passerby M each time a spatial image is acquired in the analysis process of step S200, thereby reducing the processing load.

[0061] In this embodiment, as shown in Fig. 5, when identifying a passerby from a captured image by machine learning, the head (face) is identified using training data, but in addition to this, after identifying the head, a process may be performed to identify the foot position of the passerby (person) and calculate the coordinates of the foot position. This will be explained using Fig. 11 and Fig. 12. Fig. 11 is a conceptual diagram showing the relationship between the head and foot positions of passerby M. Fig. 12 is a diagram explaining the correction process performed when determining the foot coordinates of passerby M. In the case of Fig. 12, it is assumed that the imaging device 300 is installed in a high position on a ceiling, wall, pillar, or the like near a line connecting feature points F1 and F2, which will be described later. In FIG. 11, the area analyzed as the head of pedestrian M when identifying the head of pedestrian M is shown as area 501. Area 501 for identifying the head of pedestrian M is determined taking into consideration the average height of adults. Furthermore, as shown in FIG. 12, a pedestrian M who is farther from image capture device 300 appears smaller than a pedestrian M who is closer to image capture device 300, and therefore area 501 for identifying the head is determined taking into consideration the position from image capture device 300. If the head can be identified in area 501 for identifying the head, a virtual line 502 is drawn perpendicularly from area 501 toward the bottom, and the position where the virtual line intersects with the bottom is identified as foot position 503 of pedestrian M. In this manner, the head and foot positions of pedestrian M are identified. Next, the coordinates of the foot positions are calculated from the identified foot positions. The method for calculating the coordinates of the foot positions will be described with reference to FIG. In this embodiment, the range captured by the imaging device 300 is fixed. For the sake of explanation, in FIG. 12 , it is assumed that the floor is composed of square tiles. Here, predetermined feature points, for example, F1 to F4 shown in FIG. 12 , are set in the spatial image. In FIG. 12 , the four corners of the floor of the target space S are set as feature points, and the feature points are marked by, for example, affixing circle-shaped stickers. However, the feature points are not limited to being located on the floor, but may also be located at a predetermined height from the floor, such as on a wall or pillar. The coordinates of the feature points are stored in advance, and the foot position coordinates of the passerby M are calculated based on the relative positional relationship between the coordinates of the feature points and the foot position 503 of the passerby M. By calculating the foot position coordinates in this manner, it is possible to calculate the foot position coordinates of the passerby M present in the spatial image captured by the imaging device 300. The calculated foot position coordinates may then be stored as pedestrian coordinates in the pedestrian information table.

[0062] The above embodiment encompasses the following technical ideas. (1) A pedestrian flow determination method including: an image acquisition step of acquiring a spatial image, which is a moving image or a plurality of still images of a target space including passersby; a pedestrian density calculation step of analyzing the spatial image acquired by the image acquisition step and calculating an index related to the density of passersby in the target space; a pedestrian identification step of analyzing the spatial image acquired by the image acquisition step and identifying the passersby; a movement trajectory analysis step of analyzing the movement trajectories of the passersby identified by the passerby identification step; a movement speed calculation step of calculating the movement speed of the passersby from the movement trajectories of the passersby analyzed by the movement trajectory analysis step; and a pedestrian flow determination step of determining whether the pedestrian flow in the target space satisfies a predetermined condition based on the calculated index related to the pedestrian density and the movement speed. (2) The people flow determination method according to (1), characterized in that it includes an alarm process for issuing an alarm in accordance with the determination result of the people flow determination process. (3) A pedestrian flow determination method according to (1) or (2), characterized in that a standard traffic speed, which is a reference value for the traffic speed within the target space, and an index relating to a standard pedestrian density, which is a reference for the number of pedestrians within the target space, are set in advance, and the pedestrian flow determination process determines whether the pedestrian flow in the target space satisfies predetermined conditions based on the standard traffic speed and the standard pedestrian density in addition to the pedestrian density and the traffic speed. (4) The pedestrian flow determination method described in (3) is characterized in that the pedestrian flow determination process determines that a predetermined condition is met when the traffic speed is slower than the standard traffic speed and the pedestrian density is higher than the standard pedestrian density. (5) The method for determining people flow described in (1) or (2) further includes an object identification process for analyzing the spatial image acquired by the image acquisition process and identifying objects other than the passersby that are equal to or larger than a predetermined size, and the people flow determination process determines whether the people flow in the target space satisfies predetermined conditions based on the identification results of the object identification process in addition to the passerby density and the passing speed. (6) The people flow determination method described in (5), characterized in that the people flow determination process determines that the specified condition is met when the object is identified within a specified range for a specified period of time. (7) A people flow determination method as described in (1) or (2), characterized in that an adjacent space is provided adjacent to the target space, and further includes a people flow prediction step of predicting people flow in the adjacent space based on the determination result by the people flow determination step. (8) A pedestrian flow determination system including: an image acquisition means for acquiring a spatial image, which is a moving image or a plurality of still images of a target space including passersby; a pedestrian density calculation means for analyzing the spatial image acquired by the image acquisition means and calculating an index related to the density of passersby in the target space; a pedestrian identification means for analyzing the spatial image acquired by the image acquisition means and identifying the passersby; a movement trajectory analysis means for analyzing the movement trajectories of the passersby identified by the passerby identification means; a movement speed calculation means for calculating the movement speed of the passersby from the movement trajectory of the passersby analyzed by the movement trajectory analysis means; and a pedestrian flow determination means for determining whether the pedestrian flow in the target space satisfies a predetermined condition based on the calculated index related to the pedestrian density and the movement speed. (9) A computer program to be executed by a computer device, the program including: an image input process for inputting a spatial image, which is a moving image or a plurality of still images of a target space including passersby, into the computer device; a pedestrian density calculation process for analyzing the spatial image input by the image input process and calculating the density of passersby in the target space; a pedestrian identification process for analyzing the spatial image input by the image input process and identifying the passersby; a movement trajectory analysis process for analyzing the movement trajectories of the passersby identified by the passerby identification process; a movement speed calculation process for calculating the movement speed of the passersby from the movement trajectory of the passersby analyzed by the movement trajectory analysis process; and a pedestrian flow determination process for determining whether the pedestrian flow in the target space satisfies predetermined conditions based on the calculated pedestrian density and the movement speed. [Explanation of symbols]

[0063] 10 Escalator 20 objects 30 Automatic ticket gates 100 Processing unit 110 Portrait Acquisition Department 120 Passenger Density Calculation Department 130 Passenger Special Department 140 Movement Trajectory Analysis Department 150 km / h speed calculation department 160 People Flow Determination Department 170 Specific parts of an object 175 People Flow Forecasting Department 180 News Department 185 indicates the section 190 Memory Department 200 Abortion Judgment System 300 imaging device 501 Leading Industries 502 Pseudo-Imagination Line 503 Foot position

Claims

1. an image acquisition step of acquiring a spatial image, which is a moving image or a plurality of still images of a target space including passersby; a pedestrian density calculation step of analyzing the spatial image acquired by the image acquisition step and calculating an index related to the pedestrian density in the target space; a passerby identification step of analyzing the spatial image acquired by the image acquisition step and identifying the passerby; a movement trajectory analysis step of analyzing the movement trajectory of the passerby identified by the passerby identification step; a travel speed calculation step of calculating a travel speed of the pedestrian based on the movement trajectory of the pedestrian analyzed in the movement trajectory analysis step; and a people flow determination step of determining whether the people flow in the target space satisfies a predetermined condition based on the calculated index related to the pedestrian density and the passing speed.

2. The people flow determination method according to claim 1, further comprising a notification step of issuing a notification in accordance with the determination result of the people flow determination step.

3. A standard traffic speed that is a reference value of the traffic speed in the target space; an index relating to a standard pedestrian density that is a standard for the number of pedestrians in the target space is set in advance; The pedestrian flow determination method according to claim 1 or 2, characterized in that the pedestrian flow determination process determines whether the pedestrian flow in the target space satisfies predetermined conditions based on the pedestrian density and the pedestrian speed, as well as the standard pedestrian speed and the standard pedestrian density.

4. The pedestrian flow determination method according to claim 3, characterized in that the pedestrian flow determination step determines that a predetermined condition is met when the traffic speed is slower than the standard traffic speed and the pedestrian density is greater than the standard pedestrian density.

5. an object identifying step of analyzing the aerial image acquired by the image acquiring step and identifying an object other than the passerby that is equal to or larger than a predetermined size; The pedestrian flow determination method according to claim 1 or 2, characterized in that the pedestrian flow determination step determines whether the pedestrian flow in the target space satisfies predetermined conditions based on the identification results of the object identification step in addition to the pedestrian density and the traffic speed.

6. 6. The people flow determination method according to claim 5, wherein the people flow determination step determines that the predetermined condition is met when the object is identified within a predetermined range for a predetermined period of time.

7. An adjacent space is provided adjacent to the target space, The people flow determination method according to claim 1 or 2, further comprising a people flow prediction step of predicting people flow in the adjacent space based on the determination result in the people flow determination step.

8. an image acquisition means for acquiring a spatial image, which is a moving image or a plurality of still images, of a target space including passersby; a pedestrian density calculation means for analyzing the spatial image acquired by the image acquisition means and calculating an index related to the pedestrian density in the target space; a passerby identification means for analyzing the spatial image acquired by the image acquisition means and identifying the passerby; a movement trajectory analysis means for analyzing the movement trajectory of the passerby identified by the passerby identification means; a travel speed calculation means for calculating a travel speed of the pedestrian based on the travel trajectory of the pedestrian analyzed by the travel trajectory analysis means; and a people flow determination means for determining whether the people flow in the target space satisfies a predetermined condition based on the calculated index relating to the pedestrian density and the passing speed.

9. A computer program to be executed by a computer device, an image input process for inputting a spatial image, which is a moving image or a plurality of still images of a target space including passersby, into the computer device; a pedestrian density calculation process that analyzes the spatial image input by the image input process and calculates the pedestrian density in the target space; a passerby identification process for analyzing the spatial image input by the image input process and identifying the passerby; a movement trajectory analysis process for analyzing the movement trajectory of the passerby identified by the passerby identification process; a travel speed calculation process for calculating a travel speed of the pedestrian based on the movement trajectory of the pedestrian analyzed by the movement trajectory analysis process; a people flow determination process that determines whether the people flow in the target space satisfies predetermined conditions based on the calculated pedestrian density and the passing speed.

Images