JP2014106879A - Human distribution status estimation system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a system enabling quick estimation of a human distribution status within a facility from human count information of a limited zone of the facility.SOLUTION: A human distribution status estimation system 100 performs: generating a route information map representing connection relationship between zones R1-R5 and a distribution regulation table for defining a distribution regulation of the number of people moving from a certain zone to another zone for the individual zones R1-R5; estimating respective human distribution at individual mid-points P1-P4 between the zones R1-R5 at a clock-time Tthat has elapsed from a certain clock-time Tby a first period of time by using respective pieces of human count information of the zones R1-R5 at the clock-time Tobtained from monitoring cameras C1-C5, on the basis of the route information map and the distribution regulation table; and displaying the respective pieces of human count information of the zones R1-R5 at the clock-time Tobtained from the monitoring cameras C1-C5 and the respective pieces of human count information at the individual mid-points P1-P4 between the zones R1-R5 at the clock-time Tthat is estimated as described above together on a monitor 20.

Description

  The present invention relates to a human distribution status estimation system, and more particularly to a system capable of estimating a human distribution status in a facility from information on the number of people in a limited area in the facility.

  A large number of surveillance cameras are installed in large facilities such as stations, and security guards residing in the management room within the facility can monitor the distribution of people in the facility from images taken by multiple surveillance cameras. Monitoring. However, in many cases, the number of surveillance cameras is insufficient to cover the entire area in the facility, and there will be a large number of blind spot areas that are not photographed by the surveillance cameras. As a result, the security guards residing in the management room cannot fully grasp the distribution of people in the facility.

  As a method for dealing with the above problem, for example, it is conceivable to estimate the distribution of people in the blind spot area from images taken by a plurality of surveillance cameras. Patent Document 1 describes a technique for extracting a region where a moving object exists from an image photographed by a video camera or the like, calculating a motion vector thereof, and predicting the moving direction of the moving object using the motion vector. ing. Therefore, for example, by extracting a region where a person exists from an image taken by a surveillance camera and calculating a motion vector, and predicting the moving direction of each person using the motion vector, the distribution of people in the blind spot region Can be estimated.

JP 11-328407 A

  However, in the method of calculating the motion vector of each person in the image and predicting the movement direction as described above, when there are a large number of persons in the image, the calculation of the motion vector of each person and the movement direction Since the amount of computation required for the prediction of the system becomes enormous, there is a problem that the processing speed of the system decreases.

  The present invention has been made to solve such a problem, and an object thereof is to provide a system capable of quickly estimating the distribution of people in a facility from information on the number of people in a limited area in the facility. To do.

  In order to solve the above-described problems, a human distribution situation estimation system according to the present invention includes a route information defining unit that defines route information representing a connection relationship between a plurality of predetermined regions, and a certain predetermined number of predetermined regions. A distribution rule defining means for defining a distribution rule for the number of persons moving from one area to another predetermined area; a number information acquiring means for acquiring each number of persons information in a plurality of predetermined areas; and a first number acquired by the number information acquiring means. The number of persons on each route between the plurality of predetermined areas at the second time after the first predetermined time has elapsed from the first time based on the route information and the distribution rule from the number of persons information on the plurality of predetermined areas at the time of First person distribution estimating means for estimating the distribution, each piece of person information of a plurality of predetermined areas at the second time acquired by the person number information acquiring means, and the first person distribution estimating means estimated by the first person number distribution estimating means. Characterized in that it comprises a number distribution display means for displaying together with the number distribution on the path between the plurality of predetermined regions in time.

  Based on the route information and the distribution rule, the plurality of information at the fourth time after the second predetermined time from the third time, based on the route information and the distribution rule, from the number of people information of the plurality of predetermined areas at the third time acquired by the number of people information acquisition means. A second number distribution estimation unit that estimates each number distribution of the predetermined area, a number of persons information of the plurality of predetermined areas at a fourth time estimated by the second number distribution estimation unit, and a number information acquisition unit You may further provide the distribution rule correction | amendment means which corrects a distribution rule by comparing with each number of persons information of the several predetermined area | region in 4th time.

  According to the human distribution status estimation system according to the present invention, it is possible to quickly estimate the human distribution status in a facility from the number of people in a limited area in the facility.

It is a figure which shows the structure of the floor in the station premises to which the person distribution condition estimation system which concerns on embodiment of this invention is applied. It is a figure which shows the structure of the person distribution condition estimation system which concerns on embodiment of this invention. It is a flowchart which shows the detail of the setup process which the person distribution condition estimation system which concerns on embodiment of this invention performs. It is a figure which shows an example of the route information map in the person distribution condition estimation system which concerns on embodiment of this invention. It is a figure which shows an example of the initial state of the distribution rule table in the person distribution condition estimation system which concerns on embodiment of this invention. It is a figure which shows an example of each number of persons information of area | region R1-R5 in the time (tau) _n in the person distribution condition estimation system which concerns on embodiment of this invention. It is a figure which shows an example of the estimated value of each number of persons information of area | region R1-R5 in the time tau _{n + 1} in the person distribution condition estimation system which concerns on embodiment of this invention. It is a figure which shows an example of the actual value of each number of persons information of area | region R1-R5 in time (tau) _{n + 1} in the person distribution condition estimation system which concerns on embodiment of this invention. It is a figure which shows an example of the correction process of the distribution rule table in the person distribution condition estimation system which concerns on embodiment of this invention. It is a figure which shows an example of the final state of the distribution rule table in the person distribution condition estimation system which concerns on embodiment of this invention. It is a flowchart which shows the detail of the distribution estimation process which the person's distribution condition estimation system which concerns on embodiment of this invention performs. It is a diagram illustrating an example of each number information of the region R1~R5 at time T _n in the distribution condition estimation system of a person according to an embodiment of the present invention. It is a figure which shows an example of the estimated value of each number of persons information of the intermediate points P1-P4 in the time Tn _{+ 1} in the person distribution condition estimation system which concerns on embodiment of this invention. It is a figure which shows an example of each number of persons information of area | region R1-R5 in time Tn _{+ 1} in the person distribution condition estimation system which concerns on embodiment of this invention. It is a figure which shows an example of the monitor display in the person distribution condition estimation system which concerns on embodiment of this invention.

Embodiments of the present invention will be described below with reference to the accompanying drawings.
Embodiment.
A configuration of a human distribution situation estimation system 100 in a station premises according to an embodiment of the present invention will be described with reference to FIGS.

  On the floor F in the station shown in FIG. 1, two stairs A1 and A2 and one ticket gate B1 are provided, and two stores K1 and K2 are installed. Five monitoring cameras C1 to C5 are installed on the ceiling surface of the floor F, and each of the monitoring cameras C1 to C5 can capture a moving image of a predetermined area indicated by R1 to R5 in the drawing. .

  As shown in FIG. 2, each of the monitoring cameras C1 to C5 is connected to the distribution situation estimation computer 10, and the distribution situation estimation computer 10 includes each region R1 to R5 photographed by the monitoring cameras C1 to C5. A moving image is input.

  The distribution status estimation computer 10 has a function of sequentially estimating the distribution status of people in the floor F from the moving images of the regions R1 to R5 photographed by the monitoring cameras C1 to C5 and displaying them on the monitor 20. , Route information definition means 11, distribution rule definition means 12, number information acquisition means 13, first number distribution estimation means 14, second number distribution estimation means 15, distribution rule correction means 16, and number distribution display And means 17.

  The route information defining unit 11 creates a route information map representing the connection relationship between the areas R1 to R5 in the floor F. Details of the route information map will be described later.

  The distribution rule definition means 12 creates a distribution rule table that defines the distribution rules for the number of people who move from one area to another for each of the areas R1 to R5. Details of the distribution rule table will also be described later.

  The number-of-persons information acquisition means 13 cuts out still images from the moving images of the regions R1 to R5 photographed by the monitoring cameras C1 to C5, and applies image processing to the still images, so that the regions R1 to R5 are included. Information on the number of persons existing (number information) is acquired.

The first number distribution estimation unit 14 calculates a certain time T _n based on the route information map and the distribution rule table from each number information of the areas R1 to R5 at a certain time T _n acquired by the number information acquisition unit 13. To the number of persons on each route between the regions R1 to R5 at time Tn _{+ 1} after the first predetermined time has elapsed.

The second number distribution estimating unit 15, from the number information of the region R1~R5 at certain time tau _n acquired by the count information acquiring unit 13, based on the route information map and distribution rule table, certain time tau _n To estimate the number of persons in the regions R1 to R5 at time τ _{n + 1} after the second predetermined time has elapsed.

The distribution rule correcting unit 16 includes the information on the number of people in the regions R1 to R5 at the time τ _{n + 1} estimated by the second number distribution estimation unit 15 and the regions R1 to R5 at the time τ _{n + 1} acquired by the number of people information acquiring unit 13. The distribution rule table is corrected by comparing each number of persons information.

The number distribution display means 17 includes information on the number of people in the areas R1 to R5 at time Tn _{+ 1} acquired by the number information acquisition means 13 and the areas R1 to R5 at time Tn _{+ 1} estimated by the first number distribution estimation means 14. And the distribution of the number of people on each route are displayed on the monitor 20 together.

  Next, details of processing executed by the human distribution situation estimation system 100 according to this embodiment will be described with reference to FIGS. First, after describing the “setup process” executed at the time of setting up the distribution situation estimation computer 10, the distribution situation estimation computer 10 uses the moving images of the respective regions R1 to R5 captured by the monitoring cameras C1 to C5 to create the inside of the floor F. A “distribution estimation process” for sequentially estimating a person's distribution status and displaying it on the monitor 20 will be described.

(1. Setup process)

  Processing executed at the time of setting up the distribution state estimation computer 10 will be described with reference to the flowchart of FIG.

  First, in step S11, the route information defining means 11 of the distribution state estimation computer 10 determines the shape of the floor F, the installation positions of the monitoring cameras C1 to C5, and the store K1, which becomes an obstacle when a person moves in the floor F, Based on the installation position of K2, etc., a route information map representing the connection relationship between the regions R1 to R5 as shown in FIG. 4 is created and stored in a memory (not shown). In other words, the route information map connects regions in the floor F where people can move directly.

  In the example of FIG. 4, the person who can move directly in the area R1 is the area R2 or R4, and the person who can move directly in the area R2 (ticket gate B) is outside the area R1 or the floor F, and the area R3 (step) A person in A1) can move directly in the area R4 or outside the floor F, a person in the area R4 can move directly in the area R1, R3 or R5, and a person in the area R5 (step A2) can move directly. Possible is outside the region R4 or floor F. An intermediate point P1 is defined on the route between the regions R1 and R2, an intermediate point P2 is defined on the route between the regions R1 and R4, and an intermediate point P2 is defined on the route between the regions R3 and R4. A point P3 is defined, and an intermediate point P4 is defined on the route between the regions R4 and R5.

  Next, in step S12, the distribution rule defining means 12 creates a distribution rule table as shown in FIG. 5 in which the distribution rules for the number of persons moving from one area to another area are defined for each of the areas R1 to R5. . This distribution rule table defines an estimated value as to at what rate when people in the regions R1 to R5 move toward the adjacent region at a certain time.

  In the initial state shown in FIG. 5, it is defined that people in each of the regions R1 to R5 move equally to other adjacent regions, for example, when there are 10 people in the region R1 at a certain time. It is estimated that 50% (5 people) of the people in the region R1 move toward the region R2, and the remaining 50% (5 people) move toward the region R4. However, as a matter of course, the actual distribution ratio is not uniform, and the initial state of the distribution rule table does not reflect the actual movement of the person. Therefore, in the following steps S13 to S18, the distribution rule table is corrected so that the distribution rule reflects the reality.

First, in step S13, the number-of-persons information acquisition unit 13 cuts out a still image at a certain time τ _n from the moving images of the regions R1 to R5 photographed by the monitoring cameras C1 to C5, and performs image processing on the still images. Is obtained in the format as shown in FIG. 6 for each person number information in the regions R1 to R5 at time τ _n . The details of the image processing at this time are included in the image by, for example, extracting a contour for each person by performing contour extraction on a still image and then performing region division for each person. The number of persons can be determined.

Next, in step S14, the second number distribution estimation unit 15 calculates the route information map (FIG. 4) and the distribution rule table from the number information of the regions R1 to R5 at the time τ _n acquired by the number information acquisition unit 13. (Figure 5) on the basis, from the time tau _n second predetermined time estimate each number information of the region R1~R5 at time tau _{n + 1} has elapsed (20 seconds in this embodiment) in a format as shown in FIG. 7 To do.

For example, when estimating the number of people information in the region R1 at time τ _{n + 1,} the regions adjacent to the region R1 from the route information map of FIG. 4 are R2 and R4, and the region R2 at time τ _n from the number of people information in FIG. The number of people in the region R4 is 30, and the number of people in the region R4 is 30. From the distribution rule table of FIG. 5, 50% of the people in the region R2 move toward the region R1, and 33% of the people in the region R4 Since it is defined to move toward the region R1, the number of people in the region R1 at time τ _{n + 1} is estimated to be (20 × 0.5) + (30 × 0.33) ≈20 people. The number of people entering the floor F from the areas R2 (ticket gate B), R3 (staircase A1), R5 (staircase A2) is separately determined from the human detection sensors installed on the ticket gate B and the stairs A1 and A2. Shall be acquired.

Next, in step S15, the number-of-persons information acquisition means 13 cuts out a still image at time τ _{n + 1} from the moving images of the regions R1 to R5 photographed by the monitoring cameras C1 to C5, and extracts the regions R1 to R5 at time τ _{n + 1} . Each person number information is acquired in a format as shown in FIG.

Next, in step S16, the distribution rule correcting unit 16 acquires the individual number information (FIG. 7) of the regions R1 to R5 at the time τ _{n + 1} estimated by the second number distribution estimating unit 15 and the number information acquiring unit 13. The distribution rule table is corrected by comparing the number of persons information (FIG. 8) in the areas R1 to R5 at the time τ _{n + 1} . Further, the distribution rule table before correction is stored in the memory.

  As for details of the distribution rule table correction method, for example, focusing on the number of people in the region R1 in FIGS. 7 and 8, the estimated value in FIG. 7 is 20 people, but the actual value in FIG. is there. This means that more people than the number estimated based on the distribution rule table in FIG. 5 have actually moved from the regions R2 and R4 to the region R1. Therefore, by dividing the difference (8 people) by subtracting the estimated value (20 people) from the actual value (28 people) in the region R1 by the number of people in the source regions R2 and R4 (20 people and 30 people) The distribution rules of the source areas R2 and R4 are respectively corrected. As a result, as shown in FIG. 9, the ratio of the number of persons moving from R2 to R1 is corrected from 50% to 65%, and the ratio of the number of persons moving from R4 to R1 is corrected from 33% to 50%. The distribution rule of each area is corrected by performing the same process for other areas.

  Next, in step S17, it is determined whether or not the above steps S13 to S16 are repeated a predetermined number of times (for example, about 10,000 times). If the number is less than the predetermined number, the process returns to step S13 again and is repeated a predetermined number of times. If yes, go to Step 18.

  In step S18, the second distribution rule correcting means 15 selects the distribution rule having the highest appearance frequency as the distribution rule for each of the regions R1 to R5 based on the plurality of distribution rule tables stored in the memory for each loop in step S16. As a final distribution rule, the contents of the distribution rule table are stored in the memory in a final state as shown in FIG.

  The route information map and the distribution rule table are completed and stored in the memory by the setup process in steps S11 to S18, and the preparation for executing the distribution estimation process described below is completed.

(2. Distribution estimation process)
FIG. 11 is a flowchart of processing in which the distribution status estimation computer 10 sequentially estimates the distribution status of people in the floor F from the moving images of the regions R1 to R5 photographed by the monitoring cameras C1 to C5 and displays them on the monitor 20. Will be described with reference to FIG.

  First, in step S21, the route information definition unit 11 of the distribution state estimation computer 10 reads the route information map (FIG. 4) created during the previous setup process from the memory.

  Next, in step S22, the distribution rule definition means 12 reads the final distribution rule table (FIG. 10) created during execution of the previous setup process from the memory.

Next, in step S23, count information acquiring unit 13 cuts out a still image at a certain time point _{T n} from the moving image of each region R1~R5 taken by surveillance cameras C1 to C5, region at time _{T n} R1 to Each number information of R5 is acquired in a format as shown in FIG.

Next, in step S24, the first number distribution estimating means 14, from the number information of the region R1~R5 at time T _n, which is acquired by the count information acquiring unit 13, the route information map (FIG. 4) and the distribution rule table based on (FIG. 10), from the time T _n first predetermined time each number information of waypoints P1~P4 at time T _{n + 1} has elapsed (10 seconds in this embodiment) in the format as shown in FIG. 13 presume.

For example, when estimating the number information of the intermediate point P1 at time T _{n + 1,} the regions at both ends of the intermediate point P1 are R1 and R2 from the route information map of FIG. 4, and the region at the time Tn from the number information of FIG. The number of people in R1 is 10 and the number of people in region R2 is 20, and from the distribution rule table of FIG. 10, 80% of people in region R1 move toward region R2 and 20% of people in region R2 Is defined as moving toward the region R1, the number of people at the intermediate point P1 at time T _{n + 1} is estimated to be (10 × 0.8) + (20 × 0.2) = 12 people.

Next, in step S25, the number-of-people information acquisition unit 13 cuts out a still image at time Tn _{+ 1} from the moving images of the regions R1 to R5 captured by the monitoring cameras C1 to C5, and extracts the regions R1 to R5 at time Tn _{+ 1} . Each person number information is acquired in a format as shown in FIG.

Next, in step S <b> 26, the number distribution display unit 17 determines the number of people information in the regions R <b _{> 1 to} R <b _{> 5} at the time T _{n + 1} acquired by the number information acquisition unit 13 and the time T estimated by the first number distribution estimation unit 14. _The information on the number of persons at the intermediate points P1 to P4 at _{n + 1} is displayed on the monitor 20 in the format as shown in FIG.

By repeatedly executing the above steps S23 to S26, the times T ₀ , T ₁ , T ₂ ,. . . Each of count information, the time _{T 1, T 2, T 3} , regions R1~R5 in. . . Each person number information of the intermediate points P1 to P4 is sequentially estimated, and each person number information of the areas R1 to R5 and the intermediate points P1 to P4 is sequentially displayed on the monitor 20. Each number information waypoint P1~P4 at time _{T n + 1} on the basis of the route information map and distribution rule table from the number information of the area R1~R5 first number distribution estimating means 14 in the step S24 at time _{T n} Since the process of estimating is very simple, it is possible to quickly estimate the distribution of people in the floor F.

As described above, in the person distribution situation estimation system 100 in the station premises according to this embodiment, the route information map indicating the connection relationship between the regions R1 to R5, and the regions R1 to R5 from other regions. create a distribution rule table that defines the distribution rules number of people moving into the area, from the people count information area R1~R5 at certain time T _n obtained from the monitoring camera C1 to C5, the route information map and distribution rules based on the table, to estimate the respective number distribution of each waypoint P1~P4 between regions R1~R5 at time _{T n + 1} from the time _{T n} has elapsed the first predetermined time, the time _{T n + 1} obtained from the monitoring camera C1~C5 each number information of the area R1~R5 in, and the number distribution of each waypoint P1~P4 between regions R1~R5 at time _{T n + 1} estimated by the To be displayed on the monitor. Thereby, it is possible to quickly estimate the distribution of people in the floor F from the information on the number of people in the limited areas R1 to R5 in the floor F in the station premises.

100 person distribution status estimation system, 11 route information definition means, 12 distribution rule definition means, 13 number information acquisition means, 14 first number distribution estimation means, 15 second number distribution estimation means, 16 distribution rule correction means, 17 number of persons Distribution display means, T _n first time, T _{n + 1} , second time, τ _n third time, τ _{n + 1} fourth time.

Claims (2)

Route information defining means for defining route information representing a connection relationship between a plurality of predetermined areas;
A distribution rule defining means for defining a distribution rule for the number of persons moving from a predetermined area to another predetermined area for the plurality of predetermined areas;
Number-of-persons information acquisition means for acquiring each number of people information in the plurality of predetermined areas;
Based on the route information and the distribution rule, the second predetermined time elapses from the first time based on the route information and the distribution rule from the number information of the plurality of predetermined areas at the first time acquired by the number information acquisition means. First number distribution estimation means for estimating each number distribution on each route between the plurality of predetermined areas at the time of
Between the number information of the plurality of predetermined areas at the second time acquired by the number information acquisition means and the plurality of predetermined areas at the second time estimated by the first number distribution estimation means A person distribution status estimation system comprising: a person distribution display means for displaying each person distribution on each route together. Based on the route information and the distribution rule, the fourth predetermined time elapses from the third time based on the route information and the distribution rule from the number information of the plurality of predetermined areas acquired at the third time acquired by the number information acquisition means. Second number distribution estimation means for estimating each number distribution of the plurality of predetermined areas at the time of
Each of the plurality of predetermined areas at the fourth time estimated by the second number distribution estimation means and each of the plurality of predetermined areas at the fourth time acquired by the number information acquisition means The human distribution situation estimation system according to claim 1, further comprising distribution rule correction means for correcting the distribution rule by comparing with the number information.

Images