JP2004199623A - Movable body counting method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a movable body counting method capable of counting the number of movable bodies bidirectionally passing in an observation area. <P>SOLUTION: The observation area is set so that the same movable body can be detected by at least two continued images, and images P1-P4 are continuously acquired in this state. Movable bodies A and B bidirectionally moving in the observation area are detected from the acquired images, and the matching of the movable body of the previous image with the movable body of the subsequent image is performed for every advancing direction. When the movable body of the previous image cannot be matched with the movable body of the subsequent image, this movable body is considered to move out of the observation area, and its number is counted. <P>COPYRIGHT: (C)2004,JPO&NCIPI

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a moving object counting method used for counting a moving object such as a person passing through an observation area such as a road or a passage.
[0002]
[Prior art]
As a method of counting moving objects as described above, a method of extracting the moving objects and associating the moving objects between successive frame images to count the number of moving objects that have passed through the observation area is known. (See Patent Document 1).
[0003]
[Patent Document 1]
Claim 1 of JP-A-8-185521
In this counting method, the moving objects detected in each of the successive images are associated with each other, and if the moving object detected in the previous image cannot be associated with the subsequent image, the moving object is observed. The number of moving objects is counted by counting the number of moving objects as if they were out of the area.
[0004]
[Problems to be solved by the invention]
However, in the conventional moving object counting method described in Patent Literature 1, since the direction of the moving object passing through the observation area is predetermined in one direction, the moving object passes through the observation area in both directions. In such a case, there was a problem that it could not be used.
[0005]
The present invention has been made to solve such a problem, and an object of the present invention is to provide a moving object counting method capable of counting the number of moving objects that pass through an observation area in both directions.
[0006]
[Means for Solving the Problems]
The above-mentioned subject is a step of continuously acquiring images in a state where an observation region is set so that the same moving object can be detected in at least two consecutive images, and moving the observation region bidirectionally from the acquired image. Detecting a moving object, associating the moving object detected in the previous image with the moving object detected in the subsequent image for each traveling direction, and And a step of counting the number of moving objects when the moving object cannot be associated with the moving object.
[0007]
In this counting method, an observation area is set so that at least two consecutive images can detect the same moving object, and images are continuously acquired in this state. Therefore, when the moving object passes through the observation area in both directions. In any case, in any direction, after being continuously detected in at least two images, it is not detected when it moves out of the observation region. Therefore, from the acquired image, the moving body that moves the observation area in both directions is detected, and the moving body of the previous image and the moving body of the subsequent image are associated with each other in the traveling direction, and the moving of the previous image is performed. If the body cannot be associated with the moving body in the subsequent image, it can be considered that this moving body has moved out of the observation area, and the number is counted. Thus, the number of moving objects that pass through the observation region in both directions can be counted at high speed by simple processing.
[0008]
In the moving object counting method, when the moving object is a person, for example, the number of people passing on a road or a passage can be counted.
[0009]
Also, in the moving object counting method, when the moving object of the previous image can be associated with the moving object of the subsequent image, the position information of the moving object is updated, and the position of the moving object in the previous image is detected. For a moving object that could not be associated with the moving object, the position information and the traveling direction may be determined.
[0010]
In this counting method, if the moving object in the previous image can be associated with the moving object in the subsequent image, this moving object can be considered to be passing through the observation area, and thus the position information of the moving object should be updated. Thereby, the position of the moving body can be easily grasped. In addition, a moving object that is detected in a later image and cannot be associated with the moving object in the previous image can be regarded as a moving object that has newly moved to the observation area, so that by determining the position information and the traveling direction, The moving object can be easily associated with the next image. In this manner, the position information and the traveling direction are determined for the moving object newly moving to the observation region, and the position information is updated for the moving object that has already passed through the observation region, so that the moving object is executed in accordance with the image update. And the associated moving objects can be smoothly performed.
[0011]
In this case, by assuming that the position information of the moving object is the coordinates on the image, accurate position information can be obtained reliably and easily. In addition, the traveling direction of the moving object detected in the subsequent image and not associated with the moving object of the previous image is detected in any one of the observation regions divided in advance into two regions. Also in the case where the configuration is determined depending on whether the moving direction has been determined, the traveling direction can be determined reliably and easily.
[0012]
BEST MODE FOR CARRYING OUT THE INVENTION
Next, an embodiment of the present invention will be described. In this embodiment, the moving object to be counted is assumed to be a person, and the traveling direction of the person passing through the observation area is described as a vertical direction.
[0013]
FIG. 1 is a block diagram showing an example of a moving object counting device for implementing the moving object counting method according to this embodiment.
[0014]
In FIG. 1, reference numeral 1 denotes an image pickup device such as a CCD camera that picks up an image of a predetermined target area and outputs the signal, and 2 denotes an image storage unit that temporarily stores a signal output from the image pickup device 1 as image data. It is.
[0015]
The image storage unit 2 has two built-in frame buffers capable of sequentially storing continuous images, and is designed to store signals output from the imaging device 1 in the frame buffer alternately.
[0016]
Reference numeral 3 denotes a moving object detection unit that detects a moving object using an image temporarily stored in the image storage unit 2. As a method for detecting a moving object, a method disclosed in JP-A-2001-222719 is used. In this method, an edge image is created by calculating a time difference between images temporarily stored in the image storage unit 2, and a plurality of similar circular templates of different sizes prepared in advance are obtained. This is a method of extracting a head region by sequentially moving on the edge image and voting. Alternatively, as another method, a person area may be detected only by a motion difference of an image temporarily stored in the image storage unit 2 without extracting a head area.
[0017]
Reference numeral 4 denotes direction-dependent moving object counting means. The direction-dependent moving object counting means 4 determines the traveling direction of the newly detected head and associates the moving object detected in each of the successive images with the traveling direction. A moving object information storage unit 6 for temporarily storing information about the moving object such as the position, size, and speed of the moving object; a moving object counting unit 7 including a counter 71 for counting the number of moving objects; When the body is no longer detected, the mobile body information reset section 8 returns the mobile body information temporarily stored in the mobile body information storage section 6 to an initial state. Hereinafter, the direction-specific moving object counting means 4 will be described.
[0018]
FIG. 2 is a diagram illustrating an installation state of the imaging apparatus 1. The imaging apparatus 1 is installed directly below the ceiling so that the same moving object can be detected in at least two consecutive frames of images. For example, when the maximum speed of the moving objects A and B is 5 m / s and the total processing time of the moving object detection process and the moving object counting process is 200 msec per time, the length of the moving object in the traveling direction is 2 m or more. What is necessary is just to set the observation area so that the image is taken and to install the imaging device. By installing the imaging device 1 in this manner, the moving object moving in the observation region not only necessarily exists in the observation region over two or more consecutive frames, but also newly enters the observation region. If the moving object exists in the upper half of the image, it can be determined that it has entered from above, and conversely, if it exists in the lower half of the image, it can be determined that it has entered from below.
[0019]
When the moving object detecting unit 3 detects the moving object from the image and the same moving object detecting unit 5 determines that the moving object is a new moving object not detected in the previous image, The traveling direction of the moving object is determined. For the moving body for which the traveling direction has been determined, the position and size of the moving body are stored as moving body information in the moving body information storage unit 6 prepared for each traveling direction.
[0020]
FIG. 3 is a diagram illustrating an internal structure of the mobile information storage unit 6. The mobile unit information storage unit 6 has two types of storage tables M1 and M2 composed of a plurality of buffers, all of which have initial values set to -1. The two types are prepared for storing moving body information for each traveling direction. In the present embodiment, the description has been given assuming that three types of position, size, and speed can be stored as the moving object information. In FIG. 3, X and Y in the storage tables M1 and M2 indicate an X coordinate and a Y coordinate of an image, and represent position information. R represents size information. S represents speed information. In addition, the moving body information may be one kind, two kinds, four kinds or more, and the number of moving body information can be increased or decreased as needed.
[0021]
In the same moving object detecting unit 5, if a moving object is detected in the previous image, the next image is searched for a moving object corresponding to the moving object information temporarily stored in the moving object information storage unit 6. If found, the mobile unit information is updated by overwriting, and if not found, the mobile unit information is deleted, and the count counter 70 of the mobile unit counter 7 is incremented.
[0022]
In this embodiment, the search for the corresponding moving object is performed using two methods, a method using speed information and a method not using speed information. The speed information is a value represented by a difference between the Y coordinate of the moving object in the previous image and the Y coordinate of the moving object in the next image. When searching for the moving object of the next image corresponding to the moving object of the previous image having the speed information, by adding or subtracting the speed information to the position of the moving object of the previous image, The destination can be estimated. If a moving object exists at the estimated location of the next image, it is associated with them. However, in this embodiment, in consideration of the error of the speed information, if the error is within ± 10, the correspondence is performed.
[0023]
When the moving body of the previous image does not have speed information, comparing the moving body of the previous image with the moving body of the next image, the moving body of the next image is at least downstream in the traveling direction. Therefore, when the traveling direction is from top to bottom, the Y coordinate of the moving body of the next image is larger than the Y coordinate of the moving body of the previous image. Conversely, when the traveling direction is from top to bottom, the Y coordinate of the moving object of the next image is smaller than the Y coordinate of the moving object of the previous image.
[0024]
Further, since it can be assumed that the moving object moves almost linearly in a narrow range, the moving object of the previous image and the corresponding moving object of the next image should exist on substantially the same line. In consideration of this point, in this embodiment, when the difference between the X coordinate of the moving object of the previous image and the X coordinate of the moving object of the next image is within ± 10, the moving object Shall be treated as Then, when there are a plurality of moving objects of the next image on the same line downstream of the moving object of the previous image, the moving object at the most downstream position is associated. This association is performed on the assumption that the mobile object information stored in the mobile object information storage unit 6 is stored according to a certain rule. A description of this rule follows.
[0025]
After examining all the moving object information temporarily stored in the moving object information storage unit 6, all the temporarily stored moving object information is rearranged according to a certain rule and stored in order from the upper buffer. . The rearrangement method gives priority to those having speed information, and to those having no speed information, when the traveling direction is from top to bottom, the Y coordinates are arranged in descending order. In the case of from above, the Y coordinates are arranged in the order of smaller values. This makes it possible to start searching for a moving object having no speed information from a moving object existing downstream in the previous image. Can be prevented from being associated with each other.
[0026]
The moving object of the next image that is not associated with the moving object information of the previous image stored in the moving object information storage unit 6 is assumed to have newly entered the observation area at the time of capturing the next image. After the moving direction of the moving body is determined, the position and size of the moving body are stored as moving body information in the storage tables M1 and M2 of the moving body information storage unit 6 prepared for each moving direction. The above processing is sequentially repeated.
[0027]
When the moving object is no longer detected by the moving object detection unit 3, there is no moving object of the next image corresponding to the moving object information temporarily stored in the moving object information storage unit 6, and therefore, the moving object information reset unit In step 8, all the moving object information temporarily stored in the moving object information storage unit 6 is discarded, and this is performed for each traveling direction. The count counter 70 of the moving object counting unit 7 is discarded by the moving object information reset unit 8. Increment by number.
[0028]
Next, a description will be given of a method of counting the number of persons passing through the observation region in both directions using the moving object counting device of FIG.
[0029]
P1 to P5 in FIG. 4 are images of one frame (hereinafter, referred to as frame images) representing the head of the moving object (person) detected in time series, and are M1n and M2n (n = 1 to 5). Is a storage table of the mobile unit information storage unit 6. Assuming that the size of the frame image from P1 to P5 is 320 × 240 and the upper left is the origin, M1n (n = 1 to 5) contains the moving object information passing through the image from top to bottom, M2n (n = 1 to 5). In 5), information on a moving object passing from the bottom to the top of the image is stored.
[0030]
5 to 8 are flowcharts showing a counting process by the moving object counting device of FIG. 1, FIG. 5 shows a main routine, and FIG. 6 shows an identical moving object detection process (Step 3 in FIG. FIG. 7 shows a subroutine of the moving object information storing process (S4 in FIG. 5), and FIG. 8 shows a subroutine of the moving object information resetting process (S5 in FIG. 5). Actually, the moving object counting device includes a computer, and the computer executes the processes illustrated in FIGS. 5 to 8 according to a moving object counting program recorded on a recording medium such as a ROM or a hard disk.
[0031]
Hereinafter, a description will be given with reference to FIGS.
[0032]
When the frame image P1 in FIG. 4 is input to the direction-dependent moving object counting means 4, the same moving object detecting unit 5 determines whether or not a moving object exists (S1 in FIG. 5). In the frame image P1, since the number of moving objects is 0 (NO in S1), the process proceeds to the moving object information reset process in S5 in FIG. 5, and all the moving object information stored in the storage tables M11 and M21 is deleted. It is discarded (reset) (S5-1 in FIG. 8), and the number of moving objects is counted by the number of discarded moving object information (S5-2 in FIG. 8). However, since the storage tables M11 and M21 are in the initial state, the count value of the moving object remains 0. After the counting, the process returns to the main routine of FIG. 5 and ends the main routine.
[0033]
Next, when the frame image P2 of FIG. 4 is input to the moving object counting means 4 for each direction, the same moving object detecting unit 5 determines whether or not there is a moving object (S1 of FIG. 5). Since two moving objects A2 and B2 exist in the frame image P2 (the determination in S1 is YES), the same moving object detecting unit 5 determines whether the moving object is detected in the immediately preceding frame image (P1). Is determined based on whether or not mobile object information is stored in the storage tables M11 and M21 of the mobile object information storage unit 6 (S2 in FIG. 5).
[0034]
In M11 and M21, the moving object is not detected in the initial state, that is, the previous frame image P1 (the determination in S2 is NO). Therefore, the process proceeds to the moving object information storage process of S4 in FIG. ), It is determined whether or not there is a moving object that is not stored in the storage tables M11 and M21 among the moving objects detected in (S4-1 in FIG. 7).
[0035]
Since the information of the moving objects A2 and B2 in the frame image P2 has not been temporarily stored in the storage tables M11 and M21, the traveling directions of the moving objects A2 and B2 are determined (S4-2 in FIG. 7). The determination of the traveling direction is determined based on which of the observation regions divided into two regions in advance has detected the moving object. In this embodiment, the frame image is equally divided into two upper and lower parts, and moves from top to bottom when detected in the upper area, and moves from bottom to upper when detected in the lower area. Shall be.
[0036]
When the moving direction of each moving body is determined, the position and size information of the moving body are stored in the storage tables M11 and M21 for each moving direction (S4-3 in FIG. 7). The mobile unit information stored in the storage tables M11 and M21 is rearranged according to the above-described rules (S4-4 in FIG. 7). After the rearrangement, the process returns to the main routine of FIG. 5 and ends the main routine.
[0037]
Next, when the frame image P3 in FIG. 4 is input to the moving object counting means 4 for each direction, the same moving object detecting unit 5 determines whether or not a moving object exists (S1 in FIG. 5). In P3, since three moving objects A3, B3, and C3 exist (the determination of S1 is YES), whether the moving object is detected in the immediately preceding frame image P2 is moved to the storage tables M12 and M22. It is determined whether the body information is stored (S2 in FIG. 5). Since the moving object information is stored in each of the storage tables M12 and M22 (the determination in S2 is YES), the moving object associated with this moving object, that is, the moving object recognized as the same is stored in the frame image P3. (S3 in FIG. 5). The detection is performed according to the flowchart of FIG.
[0038]
First, it is determined whether the moving object No. 1 in the storage table M12 has speed information (S3-1 in FIG. 6). Since the moving object No. 1 in the storage table M12 does not have the speed information (the determination in S3-1 is NO), in the frame image P3, the moving object No. 1 in the storage table M12 is downstream in the traveling direction and on the same line as the moving object No. 1 in the storage table M12. It is determined whether or not there is a moving object (S3-4 in FIG. 6). Since the moving object A3 of the frame image P3 exists on the same line as the moving object A3 of the frame image P3 with respect to the moving object of No. 1 in the storage table M12 (the determination of S3-4 is YES), The moving body of the frame image P3 corresponding to the moving body is set to A3, and the moving body information in the storage table M12 is updated to the storage table M13 (S3-5 in FIG. 6).
[0039]
Next, it is determined whether moving object information other than the No. 1 moving object remains in the storage table M12 (S3-8 in FIG. 6), and if it remains (YES in S3-8), FIG. The processing of S3-1 to S3-8 of No. 6 is repeated. In this embodiment, since there is no remaining information (NO in S3-8), it is determined whether or not there is a traveling direction for which the moving object information has not been checked (S3-9 in FIG. 6). In this embodiment, since the other direction has not been checked (NO in S3-9), the process returns to S3-1 to determine whether the moving object No. 1 in the storage table M22 has speed information. .
[0040]
Since the moving object No. 1 in the storage table M22 does not have the speed information (the determination in S3-1 is NO), in the frame image P3, the moving object downstream of the Nol moving object in the storage table M22 is on the same line as the moving object. It is determined whether or not there is a moving object (S3-4 in FIG. 6). With respect to the moving object No. 1 in the storage table M22, the moving object B3 exists downstream in the traveling direction and exists on the same line (the determination in S3-4 is YES). Is set to B3, and the storage table M22 is updated to the storage table M23 (S3-5 in FIG. 6).
[0041]
Next, it is determined whether or not moving object information other than the No. 1 moving object remains in the storage table M22 (S3-8 in FIG. 6). If the moving object information remains (YES in S3-8), FIG. S3-1 to S3-8 are repeated. In this embodiment, since there is no remaining information (NO in S3-8), it is determined whether or not there is a traveling direction for which the moving object information has not been checked (S3-9 in FIG. 6). Since both directions have been checked (YES in S3-9), the process returns to the main routine of FIG. 5 and proceeds to the process of S4.
[0042]
Next, in S4, it is determined whether or not there is a moving object that is not stored in the immediately preceding storage table M12 and the storage table M22 among the moving objects detected in the frame image P3 (S4 in FIG. 7). -1). Since the information on the moving object C3 in the frame image P3 has not been temporarily stored in the storage table M12 or the storage table M22 (YES in S4-1), the traveling direction of the moving object C3 is determined (S4-2). The moving object C3 is in the lower area of the frame image P3, and therefore, the traveling direction is determined to be from bottom to top.
[0043]
Since the traveling direction of the moving object C3 has been determined, the position and size information of the moving object are stored in the corresponding storage table M23 (S4-3 in FIG. 7). Then, the mobile unit information stored in the storage tables M13 and M23 is rearranged according to the rules (S4-4 in FIG. 7). After the rearrangement, the process returns to the main routine of FIG. 5 and ends the main routine.
[0044]
Next, when the frame image P4 of FIG. 4 is input to the moving object counting means 4 for each direction, the same moving object detecting unit 5 determines whether or not there is a moving object (S1 of FIG. 5). In P4, since two moving bodies A4 and C4 exist (the determination in S1 is YES), whether the moving body is detected in the previous frame image P3 is stored in the storage tables M13 and M23. It is determined whether or not the information is stored (S2 in FIG. 5). Since one moving object information is stored in the storage table M13 and two moving object information is stored in the storage table M23 (YES in S2), the moving object associated with this moving object is stored in the frame image P4. It is detected (S3 in FIG. 5).
[0045]
First, it is determined whether the moving object No. 1 in the storage table M13 has speed information (S3-1 in FIG. 6). Since No. 1 in the storage table M13 has a value of “100” as the speed information (the determination in S3-1 is YES), the location where “100” is added to the Y coordinate of the moving body of No. 1 in the storage table M13 is It is determined whether or not the moving object of the frame image P4 exists (S3-2 in FIG. 6). Since the moving object A4 is within a predetermined range with respect to the moving object of No. 1 in the storage table M13 (the determination in S3-2 is YES), the moving object of the frame image P4 corresponding to the moving object of No. 1 in the storage table M13. Is set to A4, and the storage table M13 is updated to the storage table M14 (S3-3 in FIG. 6).
[0046]
Next, it is determined whether or not moving body information other than the moving body of No. 1 remains in the storage table M13 (S3-8 in FIG. 6). If there remains (YES in S3-8), the processing of S3-1 to S3-8 in FIG. 6 is repeated again. In this embodiment, since there is no remaining information (NO in S3-8), it is determined whether or not there is a traveling direction for which the moving object information has not been checked (S3-9 in FIG. 6). In this embodiment, since the other direction is not checked (NO in S3-9), the process returns to S3-1 to determine whether the moving object No. 1 in the storage table M23 has the speed information. .
[0047]
Since the moving object of Nol in the storage table M23 has a value of “100” as the speed information (the determination in S3-1 is YES), the moving object in the storage table M23 is obtained by subtracting “100” from the Y coordinate of Nol in the storage table M23. It is determined whether or not the moving object of the frame image P4 exists (S3-2 in FIG. 6). Since the moving object No. 1 in the storage table M23 has already escaped from the observation area in the frame image P4 (NO in S3-2), the information of No. 1 in the storage table M23 is deleted (S3-6 in FIG. 6). , The counter 70 is incremented (S3-7 in FIG. 6). As a result, the count value of the counter 70 is 1, which indicates that one person has passed the observation area. Then, it is determined whether moving body information remains in the storage table M23 (S3-8 in FIG. 6).
[0048]
In this embodiment, since the information on the No. 2 moving body remains (YES in S3-8), it is determined whether the No. 2 moving body has speed information (S3-1 in FIG. 6). . Since the moving body of No. 2 in the storage table M23 does not have the speed information (the determination in S3-1 is NO), in the frame image P4, the moving body is moved downstream with respect to the moving body of No. 2 in the traveling direction and on the same line. It is determined whether a body exists (S3-4 in FIG. 6).
[0049]
Since the moving object C4 of the frame image P4 is on the same line as the moving object of the frame image P4 with respect to the moving object of No. 2 in the storage table M23, the frame image P4 corresponding to the moving object of No. 2 in the storage table M23 is displayed. The moving object is C4, and the moving object information in the storage table M23 is updated to the storage table M24 (S3-5 in FIG. 6).
[0050]
Then, it is determined whether or not the moving body information still remains in the storage table M23 (S3-8 in FIG. 6). If not (NO in S3-8), the traveling direction in which the moving body information is not checked. It is determined whether or not remains (S3-9). Since both directions have been checked (YES in S3-9), the process returns to the main routine of FIG. 5 and proceeds to the process of S4.
[0051]
Next, in S4, it is determined whether or not there is a moving object that is not stored in the storage tables M13 and M23 among the moving objects detected in the frame image P4 (S4-1 in FIG. 7). Since there is no moving object that is not stored (NO in S4-1), the process proceeds to S4-4, and the moving object information stored in the storage tables M14 and M24 is rearranged according to the rules. After the rearrangement, the process returns to the main routine of FIG. 5 and ends the main routine.
[0052]
Next, when the frame image P5 of FIG. 4 is input to the direction-dependent moving object counting means 4, the same moving object detecting unit 5 determines whether or not there is a moving object (S1 of FIG. 5). In P5, since the number of moving objects is 0 (NO in S1), it is considered that all the moving objects stored in the storage tables M14 and M24 have passed the observation area. Therefore, the process proceeds to S5, where all the moving object information stored in the storage tables M14 and M24 is discarded (reset) (S5-1 in FIG. 8), and the number of moving objects is counted by the number of discarded moving object information. (S5-2 in FIG. 8), the process returns to the main routine. In this embodiment, since the number of discarded moving object information is two, the count value of the counter 70 is a total of three, that is, it indicates that three persons have passed the observation area.
[0053]
As described above, if the observation area is set so that the same moving object can be detected in at least two consecutive images, it is possible to count the moving objects passing through the observation area in both directions by a simple process for each traveling direction. it can. Further, by using the speed information of the moving object, the moving object counting can be more stably realized.
[0054]
In the present embodiment, the size of the moving object is not used as the moving object information. However, when performing the association, if the size is extremely different, it is determined that the association is distant and another association is performed. Is performed, a more accurate and stable moving object count can be performed.
[0055]
The embodiment of the present invention has been described above, but the present invention is not limited to the above embodiment. For example, in the above-described embodiment, the moving object passing in both directions through the observation area is counted by one counting counter without any direction. However, a counting counter is provided for each direction, and the moving object passing in each direction is counted. You may ask for the number.
[0056]
In addition, although the case where the observation region of the imaging device is set so that the same moving object is detected in two consecutive images is shown, the same moving object is detected in three or more consecutive images. The observation region may be set as described above. Also in this case, when the presence of a new moving body is detected, the moving body information including the position information and the traveling direction are determined, and the information is updated each time the same moving body is detected, and the same moving body is detected. When the moving object no longer exists, its number may be counted.
[0057]
The specific embodiments described above include inventions having the following configurations.
(1) continuously acquiring images in a state where an observation area is set so that the same moving object can be detected in at least two consecutive images;
Detecting a moving object that moves in both directions in the observation area from the acquired image;
Performing a correspondence between the moving body detected in the previous image and the moving body detected in the subsequent image for each traveling direction,
If the moving body of the previous image could not be associated with the moving body of the subsequent image, counting the moving body;
Including
The correspondence between the moving object of the previous image and the moving object of the subsequent image depends on whether or not the moving object of the subsequent image exists on a straight line or almost linearly in the traveling direction of the moving object of the previous image. A moving object counting method characterized by performing.
[0058]
In this counting method, when associating the moving object of the previous image with the moving object of the subsequent image, the moving object of the subsequent image is positioned on a straight line or almost linearly in the traveling direction of the moving object of the previous image. Since it is sufficient to determine whether or not there is, the association is simplified.
(2) continuously acquiring images in a state where an observation area is set so that the same moving object can be detected in at least two consecutive images;
Detecting a moving object that moves in both directions in the observation area from the acquired image;
Performing a correspondence between the moving body detected in the previous image and the moving body detected in the subsequent image for each traveling direction,
If the moving body of the previous image could not be associated with the moving body of the subsequent image, counting the moving body;
Including
The association between the moving object of the previous image and the moving object of the subsequent image is based on whether or not the moving object exists at a predetermined position of the image after being estimated based on the speed information of the moving object of the previous image. A moving object counting method,
[0059]
In this counting method, when associating the moving object of the previous image with the moving object of the subsequent image, the moving object is located at a predetermined position of the image after the estimation based on the speed information of the moving object of the previous image. It is only necessary to determine whether or not exists, so that the association is simplified.
(3) imaging means for continuously acquiring images in a state where an observation area is set so that the same moving object can be detected in at least two consecutive images;
A moving object detecting unit that detects a moving object that moves the observation region in both directions from the acquired image,
Associating means for associating the moving object detected in the previous image with the moving object detected in the subsequent image for each traveling direction;
Counting means for counting the number of moving objects if the moving object of the previous image could not be associated with the moving object of the subsequent image;
A counting device for a moving object, comprising:
[0060]
Even with this counting device, it is possible to count moving objects that move in the observation area in both directions.
[0061]
In this moving object counting device, the moving object may be a person.
[0062]
If the moving object of the previous image can be associated with the moving object of the subsequent image, the position information of the moving object is updated, and the moving object of the previous image is detected and cannot be associated with the moving object of the previous image. The moving body may further include a moving body information management means for determining position information and a traveling direction. In this case, the position information of the moving object may be coordinates on the image. In addition, the traveling direction of the moving object detected in the subsequent image and not associated with the moving object of the previous image is detected in any one of the observation regions divided in advance into two regions. Alternatively, the configuration may be determined depending on whether the operation has been performed.
(5) continuously receiving an image obtained with an observation area set so that the same moving object can be detected in at least two consecutive images;
Detecting, from the received image, a moving object that bidirectionally moves the observation area;
Performing a correspondence between the moving body detected in the previous image and the moving body detected in the subsequent image for each traveling direction,
Counting the number of moving objects if the moving object in the previous image could not be associated with the moving object in the subsequent image;
Computer-readable recording medium on which a moving object counting program for causing a computer to execute the program is recorded.
[0063]
According to this recording medium, it is possible to cause a computer to execute a process of counting moving objects that pass through the observation region in both directions.
[0064]
In this recording medium, the moving object may be a person.
[0065]
If the moving object of the previous image can be associated with the moving object of the subsequent image, the position information of the moving object is updated, and the moving object of the previous image is detected and cannot be associated with the moving object of the previous image. For the moving object, the computer may further execute a step of managing the moving object for determining the position information and the traveling direction.
[0066]
In this case, the position information of the moving object may be coordinates on the image. In addition, the traveling direction of the moving object detected in the subsequent image and not associated with the moving object of the previous image is detected in any one of the observation regions divided in advance into two regions. Alternatively, the configuration may be determined depending on whether the operation has been performed.
(6) continuously receiving an image acquired with an observation area set so that the same moving object can be detected in at least two consecutive images;
Detecting, from the received image, a moving object that bidirectionally moves the observation area;
Performing a correspondence between the moving body detected in the previous image and the moving body detected in the subsequent image for each traveling direction,
Counting the number of moving objects if the moving object in the previous image could not be associated with the moving object in the subsequent image;
Moving object counting program for causing a computer to execute.
[0067]
According to this program, it is possible to cause the computer to execute a process of counting moving objects that pass through the observation region in both directions.
[0068]
In this program, the moving object may be a person.
[0069]
If the moving object of the previous image can be associated with the moving object of the subsequent image, the position information of the moving object is updated, and the moving object of the previous image is detected and cannot be associated with the moving object of the previous image. For the moving object, the computer may further execute a step of managing the moving object for determining the position information and the traveling direction.
[0070]
In this case, the position information of the moving object may be coordinates on the image. In addition, the traveling direction of the moving object detected in the subsequent image and not associated with the moving object of the previous image is detected in any one of the observation regions divided in advance into two regions. Alternatively, the configuration may be determined depending on whether the operation has been performed.
[0071]
【The invention's effect】
According to the moving object counting method of the present invention, as described above, in a state where an observation area is set so that the same moving object can be detected in at least two consecutive images, a step of continuously acquiring images; Detecting a moving object that bidirectionally moves the observation area from the obtained image, and performing a correspondence between the moving object detected in the previous image and the moving object detected in the subsequent image for each traveling direction. And counting the number of moving objects in the case where the moving object in the previous image could not be associated with the moving object in the subsequent image. Can be counted by simple processing, and the number of moving bodies passing in both directions can be counted at a high speed because the processing is simple.
[0072]
Further, when the moving object is a person in the method of counting moving objects, for example, it is possible to count the number of people passing on a road or a passage, and it is necessary to accurately grasp the moving state of the person. , It is of great practical value.
[0073]
Further, in the moving object counting method, when the moving object of the previous image can be associated with the moving object of the subsequent image, the position information of the moving object is updated, and the position information of the previous image is detected in the subsequent image. For a moving object that could not be associated with the moving object, if the position information and the traveling direction are determined, it is easy and easy to grasp the position of the moving object and associate the moving object between images. It can be done smoothly.
[0074]
In this case, by assuming that the position information of the moving object is coordinates on the image, accurate position information can be obtained reliably and easily. In addition, the traveling direction of the moving object detected in the subsequent image and not associated with the moving object of the previous image is detected in any one of the observation regions divided in advance into two regions. In the case where the configuration is determined depending on whether the operation has been performed, the traveling direction can be reliably and easily determined.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a moving object counting apparatus in which a counting method according to an embodiment of the present invention is performed.
FIG. 2 is an explanatory diagram of a setting state of an observation region by an imaging device.
FIG. 3 is a diagram showing a storage table of a mobile information storage unit.
FIG. 4 is a diagram showing a transition of a storage table of a mobile information storage unit when a mobile object counting method is performed.
FIG. 5 is a flowchart showing the operation of the moving object counting device of FIG. 1;
FIG. 6 is a flowchart showing a subroutine of the same moving object detection process in the flowchart of FIG. 5;
FIG. 7 is a flowchart showing a subroutine of a moving object information storage process in the flowchart of FIG. 5;
FIG. 8 is a flowchart showing a subroutine of a moving object information reset process in the flowchart of FIG. 5;
[Explanation of symbols]
1 ... Imaging device
3 ・ ・ ・ ・ ・ Moving object detection unit
4 ・ ・ ・ ・ ・ ・ ・ Moving object counting unit for each direction
5 ... Same moving object detection unit
6: Mobile information storage unit
7 ···· Mobile object counting unit
8 ····· Moving object information reset unit
A, B, A2 to A4, B2 to B4, C3, C4 ... Moving object (person)
M1, M11 to M15, M2, M21 to M25 ... storage table

Claims (5)

Continuously acquiring images in a state where an observation region is set so that the same moving object can be detected in at least two consecutive images;
Detecting a moving object that moves in both directions in the observation area from the acquired image;
Performing a correspondence between the moving body detected in the previous image and the moving body detected in the subsequent image for each traveling direction,
If the moving body of the previous image could not be associated with the moving body of the subsequent image, counting the moving body;
A moving object counting method, comprising: The counting method of a moving object according to claim 1, wherein the moving object is a person. If the moving object of the previous image can be associated with the moving object of the subsequent image, the position information of the moving object is updated, and the moving object detected in the subsequent image and not associated with the moving object of the previous image can be updated. The moving object counting method according to claim 1, wherein the position information and the traveling direction are determined for the body. 4. The moving object counting method according to claim 3, wherein the position information of the moving object is coordinates on an image. The moving direction of the moving object detected in the subsequent image and not associated with the moving object of the previous image is determined in which of the observation regions of the observation region divided in advance into two regions. The moving object counting method according to claim 3, which is determined by:

Images