JP2010262527A - Passing person counting device, passing person counting method and passing person counting program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To accurately measure the number of passing persons even regardless of the height thereof. <P>SOLUTION: A highest point HMA of a target image TG1 is selected by a highest point selection part. Slice images parallel to a horizontal plane in a closed area HR1 are extracted and held by a holding part for each predetermined distance from the selected highest point HMA to a predetermined height (e.g. a floor face of height 0) to a vertical lower direction. Based on an area of the target image in the slice image lower by a predetermined distance (e.g. 150 mm) from the highest point HMA, a passing person recognition part recognizes whether the target image indicates a passing person. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a passer counting device, a passer counting method, and a passer counting program for counting passers.

  Conventionally, entrance / exit management devices for monitoring the passage of people at the entrances and the like of buildings have been proposed. For example, Patent Document 1 discloses an entrance / exit management device that detects that an unauthorized person enters with an authorized person and allows the authorized person to pass through an entrance more smoothly. .

  In the entrance / exit management device of Patent Document 1, a passage permission information input unit for inputting passage permission information for permitting passage through an entrance, an opening / closing detection unit for detecting an opening / closing state of an opening / closing unit such as a door, and the opening / closing unit A stereoscopic image capturing unit that captures a stereoscopic image of an object passing through the object from above, a storage unit, a passage permission information input unit, an open / close detection unit, a stereoscopic image capturing unit, and a processing unit connected to the storage unit. The storage unit includes a head identification distance from a preset reference point that identifies the head of the object in the image captured from the stereoscopic image capturing unit, and a torso based on the reference point or the head The torso specific distance, the reference size of the head and torso, and the number of permitted passages associated with the passage permission information, and the processing unit is input from the passage permission information input unit. The function of specifying the number of permitted passages stored in the storage unit based on the information and processing the image input from the stereoscopic image capturing unit when the open / close means open state signal is input from the open / close detection unit A function for specifying as a target image, a function for specifying all objects in the processing target image, a function for specifying the vertices of each object, the specified vertices and the head specific distance stored in the storage unit A function for identifying the head based on the above, a function for identifying the trunk based on the apex or the head and the trunk specific distance stored in the storage, and the identified head and storage And a function of determining whether or not the specified head is within the reference size, and the torso stored in the specified torso and storage unit Contrast with the standard size of A function for determining whether the determined torso is within the reference size, a function for outputting an abnormal signal when at least one of the head or torso exceeds the reference size, and the specified head When both the body part and the corresponding body part are within the standard size, the function of counting one passing person with the set of head and body part, the total number of passing persons and the storage part A function of outputting an abnormal signal when the total number of passing persons is larger than the number of permitted passing persons is compared with the stored permitted number of passing persons.

JP 2006-241682 A

  However, in the access control apparatus described in Patent Document 1 as described above, determination is made using one predetermined reference (head specific distance, trunk specific distance) for specifying the head and trunk. Therefore, for example, when a person with a high height and a person with a low height pass together, it is difficult to accurately measure the number of these passers-by.

  In view of such a problem, the present invention aims to accurately measure the number of passers-by.

(1)
A passer-by counting apparatus according to a first aspect of the present invention is an imaging device that captures a 3D image to acquire 3D image data, a recording unit that records the initial 3D image data, and a recording unit that records the 3D image data. An extraction unit that extracts one or more differences by comparing the three-dimensional image data in the initial state with the three-dimensional image data acquired by the imaging device, and a predetermined closed region including the one or more differences The closed region determination unit to be determined, the highest point selection unit for selecting the highest point of the 3D image data for each determined closed region, and vertically downward from each highest point selected by the highest point selection unit for each closed region And a holding unit that extracts and holds a plurality of slice images parallel to the horizontal plane at predetermined intervals, and a slice at a position lower than the highest point by a predetermined interval among the plurality of slice images held by the holding unit Place in the image A passer-by recognition unit that recognizes for each closed region whether or not the determination target image is an image showing a passer-by, based on the area of the determination target image that is a closed region, and passer-by recognition for each closed region And a measuring unit that totals the number of passers-by recognized by the unit.

  In the passer-by counting apparatus according to the first invention, the three-dimensional image data is acquired by the imaging device, and the initial three-dimensional image data is recorded in the recording unit. The three-dimensional image data in the initial state recorded in the recording unit and the three-dimensional image data acquired by the imaging device are compared by the extraction unit, and one or more differences are extracted. A predetermined closed region including one or a plurality of different points is determined by the closed region determining unit, and the highest point of the three-dimensional image data for each closed region is selected by the highest point selecting unit.

  For each closed region, a plurality of slice images parallel to the horizontal plane are extracted and held by the holding unit at predetermined intervals from each highest point vertically downward. Based on the area of the judgment target image in the slice image at a position lower than the highest point by a predetermined interval among the plurality of slice images held by the holding unit, whether or not the judgment target image is an image indicating a passerby Is recognized by the passer-by recognition unit for each closed region. The number of passers-by recognized by the passer-by recognition unit is summed up by the measurement unit for each closed region.

  In this case, since the area of the determination target image in the slice image is determined based on the highest position in each closed region, the number of passers-by can be accurately determined even for a tall person or a short person. It can be measured.

(2)
The passer-by identification unit includes one determination target image recognized as an image indicating a passer-by and an other determination target image different from the one determination target image among the plurality of slice images held by the holding unit. In a slice image, when the distance between the center of gravity of one judgment target image and the center of gravity of another judgment target image is calculated and it is determined that the distance is within a predetermined range, the other judgment target image indicates a passerby. You may recognize that it is not an image.

  In this case, since the passer-by recognition unit can recognize that the image is not an image showing the passer-by, even if the passer-by has an accessory (luggage), the number of passers-by can be accurately measured. it can.

(3)
The passer-by identification unit includes one determination target image recognized as an image indicating a passer-by and an other determination target image different from the one determination target image among the plurality of slice images held by the holding unit. In a slice image, when the distance between the center of gravity of one determination target image and the center of gravity of another determination target image is calculated and it is determined that the distance is not within the predetermined range, the position of the slice image in the three-dimensional image data Whether or not the other determination target image is an image indicating a passerby may be recognized based on the area of another determination target image in the slice image at a position lower than the predetermined interval.

  In this case, the passer-by identification unit can recognize not only one passer but also other passers-by, so even if there are other passers other than the one passer-by, the number of passers can be accurately measured. can do.

(4)
The passer-by counting method according to the second invention includes an imaging step of capturing a 3D image to acquire 3D image data, a recording step of recording the initial 3D image data captured by the imaging step, An extraction step of extracting one or more differences by comparing the recorded three-dimensional image data in the initial state and the acquired three-dimensional image data, and a predetermined closed region including the one or more differences The closed region determination step to be determined, the highest point selection step for selecting the highest point of the three-dimensional image data for each determined closed region, and vertically below each highest point selected by the highest point selection step for each closed region And a holding step for extracting and holding a plurality of slice images parallel to the horizontal plane at predetermined intervals, and a highest point among the plurality of slice images held by the holding step. A pass for recognizing for each closed region whether or not the determination target image is an image showing a passerby based on the area of the determination target image that is a predetermined closed region in the slice image at a position lower by a predetermined interval. A person recognition step and a measurement step of summing up the number of passers-by recognized by the passer-by recognition step for each closed region.

  In the passer-by counting method according to the second invention, the three-dimensional image data is acquired by the imaging step, and the initial three-dimensional image data is recorded by the recording step. The recorded three-dimensional image data in the initial state and the acquired three-dimensional image data are compared in the extraction step to extract one or a plurality of differences. A predetermined closed region including one or a plurality of different points is determined by the closed region determining step, and the highest point of the three-dimensional image data for each closed region is selected by the highest point selecting step.

  For each closed region, a plurality of slice images parallel to the horizontal plane are extracted and held at a predetermined interval from each highest point vertically downward. Based on the area of the determination target image in the slice image at a position lower than the highest point by a predetermined interval among the plurality of held slice images, whether or not the determination target image is an image indicating a passerby is closed. Each area is recognized by a passer-by recognition step. The number of recognized passersby is summed by the measurement step for each closed area.

  In this case, since the area of the determination target image in the slice image is determined based on the highest position in each closed region, the number of passers-by can be accurately determined even for a tall person or a short person. It can be measured.

(5)
A passer counting program according to a third aspect of the present invention is an imaging process for capturing a 3D image to acquire 3D image data, a recording process for recording the initial 3D image data captured by the imaging process, An extraction step of extracting one or more differences by comparing the recorded three-dimensional image data in the initial state and the acquired three-dimensional image data, and a predetermined closed region including the one or more differences The closed region determining step to be determined, the highest point selecting step for selecting the highest point of the three-dimensional image data for each determined closed region, and vertically below each highest point selected by the highest point selecting step for each closed region Of a plurality of slice images parallel to the horizontal plane at predetermined intervals, and a plurality of slice images held by the holding step at a position lower than the highest point by a predetermined interval. A passer-by recognition process for recognizing, for each closed region, whether or not the determination target image is an image indicating a passer-by based on the area of the determination target image that is a predetermined closed region in the slice image; And a measuring step of totaling the number of passers-by recognized by the passer-by recognition step.

  In the passer-by counting program according to the third aspect of the invention, the three-dimensional image data is acquired by the imaging process, and the initial three-dimensional image data is recorded by the recording process. The recorded three-dimensional image data in the initial state and the acquired three-dimensional image data are compared by an extraction process to extract one or a plurality of differences. A predetermined closed region including one or a plurality of differences is determined by the closed region determining step, and the highest point of the three-dimensional image data for each closed region is selected by the highest point selecting step.

  For each closed region, a plurality of slice images parallel to the horizontal plane are extracted and held at predetermined intervals from each highest point downward in the vertical direction. Based on the area of the determination target image in the slice image at a position lower than the highest point by a predetermined interval among the plurality of held slice images, whether or not the determination target image is an image indicating a passerby is closed. Each area is recognized by a passer-by recognition process. The number of recognized passersby is summed up by the measuring step for each closed region.

  In this case, since the area of the determination target image in the slice image is determined based on the highest position in each closed region, the number of passers-by can be accurately determined even for a tall person or a short person. It can be measured.

It is a schematic diagram which shows the installation state of the passer-by counter according to the first embodiment. It is a block diagram which shows the example of the hardware constitutions of a passer-by counter. It is a block diagram which shows the example of a functional structure of a passer-by counter. It is a schematic diagram which shows progress until a closed region is determined by the closed region determination part. It is a schematic diagram for demonstrating the measuring method of the number of passers by a measurement part. It is a schematic diagram for demonstrating the method to measure the passerby in the closed area | region of FIG.4 (d). It is a schematic diagram for demonstrating the method to measure the passerby in the closed area | region of FIG.4 (d). It is a schematic diagram for demonstrating the method to measure the passerby in the closed area | region of FIG.4 (d). It is a schematic diagram for demonstrating the method to measure the passerby in the closed area | region of FIG.4 (d). It is a flowchart which shows the main flow of a passer counting method. It is a flowchart which shows the counting process in the main flow of FIG.

  Hereinafter, a passer counting device, a passer counting method, and a passer counting program according to an embodiment of the present invention will be described in detail.

(First embodiment)
FIG. 1 is a schematic diagram showing an installation state of a passer counting device 100 according to the present embodiment, FIG. 2 is a block diagram showing an example of a hardware configuration of the passer counting device 100, and FIG. 3 is a block diagram illustrating an example of a functional configuration of the apparatus 100. FIG.

  As shown in FIG. 1, the passer-by counting device 100 is installed above a building 200 such as a building so that the blind spot is as small as possible. When a passerby enters from the entrance / exit 201 of the building 200, the passer-by counting device 100 performs imaging in the imaging range SH. Then, the number of passers by is measured by the passer counting device 100.

  As shown in FIG. 2, a passer-by counting device 100 includes a central processing unit (hereinafter referred to as a CPU) 1, a read-only memory (hereinafter referred to as a ROM) 2 that stores a program, and an arithmetic processing work area. A random access memory (hereinafter referred to as RAM) 3, a storage unit 4 that stores various data files, an output unit 5 that outputs count results, and an imaging device 7 are included. In this embodiment, a well-known TOF (Time of Flight) type three-dimensional distance measuring camera is used as the imaging device 7, but stereo vision for measuring distance using two cameras, or a specified projection The pattern may be photographed with a single camera to measure the distance.

  The CPU 1 is communicably connected to the ROM 2, RAM 3, storage unit 4, output unit 5, and imaging device 7, and each of these components is controlled by the CPU 1.

  As shown in FIG. 3, the passer-by counting apparatus 100 includes a recording unit 10, an extraction unit 11, a closed region determination unit 12, a highest point selection unit 13, a slice image holding unit 14, and passer-by recognition as functional components. The unit 15 and the measuring unit 16 are configured.

  When the CPU 1 executes the passer counting program of the present embodiment stored in the ROM 2, the recording unit 10, the extraction unit 11, the closed region determination unit 12, the highest point selection unit 13, the slice image holding unit 14, The passer-by identification unit 15 and the measurement unit 16 are functionally realized.

  The passer-by counting program can also be executed on a personal computer that includes the CPU 1, ROM 2, and RAM 3. In this case, a passer-counting program is stored in the RAM 3 of the personal computer from a portable recording medium such as a recording medium BT1 such as a CD-ROM or DVD-ROM or a semiconductor memory BT2 in which the program is recorded. Installed. Further, if the personal computer is configured to be connectable to a network, the program can be downloaded from the server via the network.

  In addition, all or part of the recording unit 10, the extraction unit 11, the closed region determination unit 12, the highest point selection unit 13, the slice image holding unit 14, the passer-by recognition unit 15, and the measurement unit 16 may be ASIC (Application Specific Integrated). Circuit) or the like, and may be realized by firmware or hardware.

  Hereinafter, the passer-by measurement method will be described in detail with reference to the drawings for each function of each component.

  FIG. 4 is a schematic diagram showing a process until the closed region is determined by the closed region determining unit 12, and FIG. 5 is a schematic diagram for explaining a method of measuring the number of passers by the measuring unit 16.

  As shown in FIG. 4A, first, an initial three-dimensional image is taken by the imaging device 7, and initial three-dimensional image data SS is acquired. The initial three-dimensional image data SS includes a background image HG such as a flower. The recording unit 10 records the initial three-dimensional image data SS. Note that the three-dimensional image data including the height from the floor is calculated by the CPU 1 using the distance and angle between the imaging device 7 and the object to be measured, which are measured by the imaging device 7. Since the calculation method is a known geometric operation, it is omitted.

  Next, as shown in FIG. 4B, the current three-dimensional image is taken by the imaging device 7, and the current three-dimensional image data ST is acquired. The three-dimensional image data ST includes a background image HG and a plurality of determination target images TG1 to TG4 that are images that should be determined whether or not the image indicates a passerby. FIG. 4B illustrates four determination target images TG1 to TG4.

  Next, as illustrated in FIG. 4C, the extraction unit 11 extracts one or a plurality of differences (differences) by comparing the initial three-dimensional image data SS and the current three-dimensional image data ST. The

  The difference 3D image data SB generated by extracting the differences includes the determination target images TG1 to TG4 as the differences. In addition, the hatched area in FIG. 4C and FIG. 4D described later indicates an area where there is no difference (difference) between the three-dimensional image data SS in the initial state and the current three-dimensional image data ST. .

  Next, as shown in FIG. 4D, the closed region determination unit 12 determines a closed region HR1, which is a rectangular, circumscribed rectangle so as to include the determination target image TG1 as a difference. A rectangular region circumscribing so as to include the determination target image TG2 that is a difference is determined, and a closed region HR2 that is a predetermined closed region is determined and circumscribed to include the determination target image TG3 that is a difference. A closed region HR3 that is a predetermined closed region is determined, and a rectangular closed region HR4 that is circumscribed so as to include the determination target image TG4 that is a difference is determined. Although the closed regions HR1 to HR4 are not rectangular, the following processing can be performed on the closed regions cut out on the outer periphery of the determination target images TG1 to TG4. However, in order to speed up the processing. In this embodiment, processing is performed on a rectangular closed region. The three-dimensional image data including the closed regions HR1 to HR4 is referred to as closed region three-dimensional image data SH.

  Thus, first, the closed regions HR1 to HR4 of the determination target images TG1 to TG4, which are images to be determined whether or not the user is a passerby, are determined. Below, the case where a passerby in closed area HR1 is measured is demonstrated, and the method of measuring a passerby in closed areas HR2-HR4 is explained in detail in below-mentioned 2nd-4th embodiment, respectively.

  Next, the height (height from the floor surface) of the closed region three-dimensional image data SH of the closed region HR1 shown in FIG. FIG. 5B shows the height data of the AA cross section of FIG. 5A, and the point with the maximum height selected by the highest point selection unit 13 (hereinafter referred to as the highest point). ) HMA is schematically shown. In addition, if HMA is by a passerby, it will substantially correspond to the height of the top of the head.

  Subsequently, as shown in FIG. 5 (c), the closed region HR1 is set at a predetermined interval D1 from the selected highest point HMA to a predetermined height (for example, a floor surface having a height of 0) vertically downward. Slice images parallel to the horizontal plane are extracted and held by the slice image holding unit 14. In this embodiment, by setting the predetermined distance D1 to 150 mm or less, which is about half of the standard height of the head of an adult, only the head can be extracted without erroneously detecting the torso in the processing described later. I have to. In FIG. 5C, 14 slice images SG1 to SG14 are extracted.

  In FIG. 5D, the slice image at the height of number 1 in FIG. 5C is SG1, and similarly, the slice image at the height of number 2 is SG2, and the slice image at the height of number 3 is the same. SG3, a slice image at the height of number 4 is SG4, and a slice image at the height of numbers 5 to 14 is SG5 to SG14. Since the slice images SG5 to SG14 are the same, the illustrations are the same.

  Here, based on the area of the determination target image in the slice image at a position lower than the highest point HMA by a predetermined distance D2, the passer recognition unit 15 determines whether or not the determination target image is an image indicating a passerby. Be recognized. In the present embodiment, the predetermined distance D2 is set to 150 mm, which is about half of the adult head standard height, so that only the head can be extracted without erroneously detecting the body part.

  That is, in FIG. 5D, when the area of the determination target image TG1 in the slice image SG3 at a position 150 mm lower than the highest point HMA, that is, the height of number 2, is within a predetermined range, the determination target image The passer-by recognition unit 15 recognizes that TG1 is an image showing a passer-by.

  Then, the number of passers-by recognized by the passer-by recognition unit 15 is measured by the measuring unit 16 for each of the closed regions HR1 to HR4 (the closed regions HR2 to HR4 will be described later), and these are totaled. Thereby, the number of passers-by is accurately measured. In addition, since there is no other determination target image other than the determination target image TG1 in each slice image, it is determined that there are no other passersby in the closed region HR1. Note that the slice image is extracted as described above in order to reduce the processing and increase the speed, so if there is no problem in the processing speed, the area is extracted continuously from the highest point in order to enhance the detection system. You may do it.

(Second Embodiment)
FIG. 6 is a schematic diagram for explaining a method of measuring a passerby in the closed region HR2 of FIG. In the following description, the same description as that of the first embodiment is omitted.

  The height of the closed region three-dimensional image data SH of the closed region HR2 shown in FIG. The determination target image TG2 includes determination target images TG21 and TG22 described later. FIG. 6B shows the height data of the BB cross section of FIG. 6A, and schematically shows the highest point HMB selected by the highest point selection unit 13. In addition, if HMB is by a passerby, it will correspond substantially to the height of the top of the head.

  Subsequently, as shown in FIG. 6C, from the selected highest point HMB to a predetermined height (for example, a floor surface having a height of 0) vertically downward at every predetermined interval D1, the closed region HR2 Slice images parallel to the horizontal plane are extracted and held by the slice image holding unit 14. In FIG. 6C, 14 slice images SG1 to SG14 are extracted.

  Similar to the first embodiment, in FIG. 6D, the determination target image TG21 in the slice image SG3 at a position 150 mm lower than the highest point HMB by a predetermined distance D2, that is, at a height of number 2. Is within the predetermined range, the passer-by recognition unit 15 recognizes that the determination target image TG21 is an image showing a passer-by. Hereinafter, the next process when the determination target image TG21 is recognized as the passerby image will be described.

  Next, the passer-by recognition unit 15 selects a slice image in which another determination target image exists in addition to the determination target image TG21, that is, a slice image SG4 having the determination target image TG22 that has newly appeared.

  Then, in this slice image SG4, a distance d1 between the center of gravity G1 of the determination target image TG21 and the center of gravity G2 of the determination target image TG22 is calculated by the passer-by recognition unit 15.

  When the calculated distance d1 is within the predetermined range, the passerby recognition unit 15 recognizes that the determination target image TG22 is not an image showing a passerby, in other words, an image showing an accessory such as a luggage. Is done.

  That is, in this embodiment, when the distance between one determination target image and the other determination target images other than the determination target image is extremely small, it is usually difficult to think of passing in close proximity. It should not be recognized as a passer-by image. In addition, even when the distance between one determination target image and the other determination target images other than the determination target image is not extremely small, even when the area of the other determination target image is equal to or larger than a predetermined value, The other images to be judged are not recognized as passersby images.

  Thereby, even when the passer-by passes with accessories such as luggage, the number of passers-by is accurately measured. In addition, since there are no other determination target images other than the determination target images TG21 and TG22 in each slice image, it is determined that there are no other passersby and appendages in the closed region HR2.

(Third embodiment)
7 and 8 are schematic views for explaining a method of measuring passersby in the closed region HR3 of FIG. 4D. In the following description, the same description as that of the first embodiment is omitted.

  The height of the closed region three-dimensional image data SH of the closed region HR3 shown in FIG. Note that the determination target image TG3 includes determination target images TG31, TG32, and TG33 described later. FIG. 7B shows the height data of the CC cross section of FIG. 7A, and schematically shows the highest point HMC selected by the highest point selection unit 13. In addition, if HMC is by a passerby, it will correspond substantially to the height of the top of the head.

  Subsequently, as shown in FIG. 7C, from the selected highest point HMC to a predetermined height (for example, a floor surface having a height of 0) vertically downward at every predetermined interval D1, the closed region HR3. Slice images parallel to the horizontal plane are extracted and held by the slice image holding unit 14. In FIG. 7C, 14 slice images SG1 to SG14 are extracted.

  Similarly to the first embodiment, in FIG. 8D, the area of the determination target image TG31 in the slice image SG2 at a position lower than the highest point HMC by a predetermined distance D2 of 150 mm, that is, the height of the number 2 is shown. When it is within the predetermined range, the passerby recognition unit 15 recognizes that the determination target image TG31 is an image showing a passerby. Hereinafter, the next process when the determination target image TG31 is recognized as the passerby image will be described.

  Next, the passer-by recognition unit 15 selects a slice image in which other determination target images exist in addition to the determination target image TG31, that is, the slice image SG4. It should be noted that the highest point may be extracted again in the same manner as described above for the portion excluding the determination target image TG31 and the surrounding range, and a slice image may be generated again and the processing described later may be performed.

  Then, in this slice image SG4, a distance d2 between the center of gravity G3 of the determination target image TG31 and the center of gravity G4 of the determination target image TG32 is calculated by the passer-by recognition unit 15.

  Here, when the calculated distance d2 is not within the predetermined range, and a slice image (for example, a slice) at a position lower than the slice image (in this embodiment, the slice image SG4) by which the distance d2 is calculated by a predetermined interval. When the area of the determination target image TG32 in the image SG5) has a predetermined size, the determination unit TG32 recognizes that the determination target image TG32 is an image indicating another passerby.

  Subsequently, in the same manner as described above, the passer-by recognition unit 15 selects the slice image in which other determination target images exist in addition to the determination target images TG31 and TG32, that is, the slice image SG8 having the determination target image TG33 that has newly appeared. Is done.

  In this slice image SG8, the distance d3 between the center of gravity G3 of the determination target image TG31 and the center of gravity G5 of the determination target image TG33, and the distance d4 between the center of gravity G4 of the determination target image TG32 and the center of gravity G5 of the determination target image TG33. In the case where both are not within the predetermined range and in the slice image (for example, the slice image SG9) at a position lower than the slice image (in this embodiment, the slice image SG8) for which the distances d3 and d4 are calculated by a predetermined interval When the area of the determination target image TG33 has a predetermined size, the passerby recognition unit 15 recognizes that the determination target image TG33 is an image indicating another passerby.

  Thereby, in this embodiment, even when a plurality of passers-by (for example, father, mother and child) pass, the number of passers-by is accurately measured. In addition, since there are no other determination target images other than the determination target images TG31, TG32, and TG33 in each slice image, it is determined that there are no other passersby and appendages in the closed region HR3.

(Fourth embodiment)
FIG. 9 is a schematic diagram for explaining a method of measuring a passerby in the closed region HR4 of FIG. In the following description, the same description as that of the first embodiment is omitted.

  The height of the closed region three-dimensional image data SH of the closed region HR4 shown in FIG. The determination target image TG4 includes determination target images TG41 and TG42 described later. FIG. 9B shows the height data of the DD cross section of FIG. 9A, and schematically shows the highest point HMD selected by the highest point selection unit 13. In addition, if HMC is by a passerby, it will correspond substantially to the height of the top of the head.

  Subsequently, as shown in FIG. 9C, from the selected highest point HMD to a predetermined height (for example, a floor surface with a height of 0) vertically downward at every predetermined interval D1, the closed region HR4. Slice images parallel to the horizontal plane are extracted and held by the slice image holding unit 14. In FIG. 9C, 14 slice images SG1 to SG14 are extracted.

  In the present embodiment, in the slice image SG2 having the plurality of determination target images TG41 and TG42 that appear first, the passer-by recognition unit 15 determines the distance d5 between the center of gravity G6 of the determination target image TG41 and the center of gravity G7 of the determination target image TG42. Calculated.

  When the calculated distance d5 is within a predetermined range, the passer-by recognition unit 15 recognizes that the determination target image TG42 is not an image indicating a passer-by (passerby's head).

  Next, based on the area of the determination target image in the slice image at a position lower than the highest point HMD by a predetermined distance D2, the passer recognition unit 15 determines whether or not the determination target image is an image indicating a passerby. Be recognized.

That is, in FIG. 9D, the area of the determination target image TG41 in the slice image SG2 at a position 150 mm lower than the highest point HMD by a predetermined distance D2, that is, the height of number 2 in FIG. When it is within the predetermined range, the passerby recognition unit 15 recognizes that the determination target image TG41 is an image showing a passerby.
Thereby, in this embodiment, even when other judgment object images exist in the vicinity of a passer's head, the number of passers is measured correctly. In addition, since there are no other determination target images other than the determination target images TG41 and TG42 in each slice image, it is determined that there are no other passersby and accessories in the closed region HR4.

(Flow chart of passer counting method)
FIG. 10 is a flowchart showing the main flow of the passer-by counting method, and FIG. 11 is a flowchart showing the counting process in the main flow of FIG.

  As shown in FIG. 10, first, three-dimensional image data SS in an initial state is acquired by the imaging device 7 (step S1). Next, the current three-dimensional image data ST is acquired by the imaging device 7 (step S2).

  Then, the difference 3D image data SB is generated by extracting the difference (difference) between the 3D image data SS and the 3D image data ST by the extraction unit 11 (step S3).

  Subsequently, closed region three-dimensional image data SH including the closed regions HR1 to HR4 determined by the closed region determination unit 12 is generated (step S4).

  Next, in the counting process described in detail later with reference to FIG. 11, the number of passers-by is measured for each of the closed regions HR1 to HR4 by the measuring unit 16 (step S5). When the counting process is completed in the fully closed area (Yes in step S6), the number of passers-by measured in each closed area is totaled by the measuring unit 16, and the total result is output to the control system of the building 200, for example. (Step S7). On the other hand, if the counting process is not completed in the fully closed region, the process returns to step S5 and is repeated.

  If necessary, the 3D image data SS in the initial state may be periodically updated when there is no passerby. Thereby, it becomes possible to reduce the influence by the change of sunlight or illumination at the time of imaging by the imaging device 7.

  Hereinafter, the counting process in step S5 of FIG. 10 will be described. As shown in FIG. 11, first, the highest point is selected by the highest point selection unit 13 for a certain closed region (step S11). In selecting the highest point, it is also determined whether or not the closed region immediately below the highest point has a predetermined area in the slice image at a predetermined interval D1 that is vertically downward from the highest point. By doing so, it is possible to avoid erroneously recognizing noise as the highest point.

  Next, slice images parallel to the horizontal plane in one closed region are sliced images at predetermined intervals D1 from the selected highest point to a predetermined height (for example, a floor surface having a height of 0) vertically downward. Each is extracted and held by the holding unit 14 (step S12). A slice image from the highest point to the height of a child who can walk (for example, about 800 mm) may be extracted.

  Subsequently, the passer-by identification unit 15 determines whether or not the area of the determination target image in the slice image at a position lower than the highest point by, for example, 150 mm that is the predetermined interval D2 is within a predetermined range (step S13). . The predetermined range is set on the basis of, for example, the size of the cross section of the child's head. The predetermined range may be changed based on the place of traffic.

  When the area of the determination target image is within the predetermined range (Yes in step S13), the passer recognition unit 15 recognizes that the determination target image is an image indicating a passerby, and the number of the passersby is measured. 16 (step S14).

  Next, after sequentially checking other slice images (step S15), if there is no other image to be judged (No in step S16), the process returns to the main flow of FIG. On the other hand, when another determination target image exists in a certain slice image (Yes in step S16), is the distance between the center of gravity of the determination target image and the center of gravity of the other determination target image within a predetermined range? Whether or not it is determined by the passer-by recognition unit 15 (step S17).

  If the distance is within the predetermined range (Yes in step S17), it is recognized that the other determination target image is not an image indicating a passerby and is not counted as a passerby (step S18). Thereafter, the process returns to step S15.

  On the other hand, if the distance is outside the predetermined range (No in step S17), then the other determination target image in the slice image at a position lower than the slice image for which the distance is calculated by a predetermined interval. It is determined whether or not the area has a predetermined size (step S19).

  When the area of the other determination target image has a predetermined size (Yes in step S19), the other determination target image is an image indicating a passerby, and the number of passersby in the closed region is calculated. Addition is performed (step S20), and the process returns to step S15. On the other hand, when the area of the other determination target image does not have a predetermined size (No in step S19), the process returns to step S15.

  A preferred embodiment of the present invention is as described above, but the present invention is not limited thereto. It will be understood that various other embodiments may be made without departing from the spirit and scope of the invention. Furthermore, in this embodiment, although the effect | action and effect by the structure of this invention are described, these effect | actions and effects are examples and do not limit this invention.

DESCRIPTION OF SYMBOLS 7 Imaging device 10 Recording part 11 Extraction part 12 Closed area determination part 13 Maximum point selection part 14 Slice image holding part 15 Passer recognition part 16 Measurement part 100 Passer counting device BT1, BT2 Recording medium d1-d5 Distance G1-G7 Center of gravity HMA, HMB, HMC, HMD Highest point HR1 to HR4 Closed region SB Difference 3D image data SH Closed region 3D image data SS Initial 3D image data ST Current 3D image data SG1 to SG14 Slice image TG1 to TG4 Judgment target image

Claims (5)

An imaging device for capturing 3D images and acquiring 3D image data;
A recording unit for recording the three-dimensional image data in an initial state;
An extraction unit for extracting one or more differences by comparing the three-dimensional image data in the initial state recorded in the recording unit with the three-dimensional image data acquired by the imaging device;
A closed region determining unit that determines a predetermined closed region including the one or more differences;
A highest point selection unit for selecting the highest point of the three-dimensional image data for each determined closed region;
A holding unit that extracts and holds a plurality of slice images parallel to the horizontal plane at predetermined intervals from the highest points selected by the highest point selection unit vertically downward for each of the closed regions;
Of the plurality of slice images held by the holding unit, the determination target image is based on the area of the determination target image that is a predetermined closed region in the slice image at a position lower than the highest point by a predetermined interval. A passer-by recognition unit that recognizes for each closed region whether or not the image shows a passer-by;
A passer counting device comprising: a measuring unit that sums up the number of passers-by recognized by the passer-by-passenger for each closed region. The passer-by identification unit
Among the plurality of slice images held by the holding unit, in the slice image in which one judgment target image recognized as an image indicating a passerby and another judgment target image different from the one judgment target image exist, When calculating the distance between the center of gravity of the one determination target image and the center of gravity of the other determination target image and determining that the distance is within a predetermined range, the other determination target image is an image indicating a passerby The passer-by counting apparatus according to claim 1, wherein the passer-by counter is recognized as not being. The passer-by identification unit
Among the plurality of slice images held by the holding unit, in the slice image in which one judgment target image recognized as an image indicating a passerby and another judgment target image different from the one judgment target image exist, When the distance between the center of gravity of the one determination target image and the center of gravity of the other determination target image is calculated and it is determined that the distance is not within a predetermined range, the position of the slice image in the three-dimensional image data is 2. The method of recognizing whether or not the other determination target image is an image showing a passerby based on an area of the other determination target image in a slice image at a position lower by a predetermined interval. The passer-by counting device according to claim 1 or 2. An imaging step of capturing a 3D image and acquiring 3D image data;
A recording step of recording the initial three-dimensional image data imaged by the imaging step;
An extraction step of comparing the recorded initial three-dimensional image data with the acquired three-dimensional image data to extract one or more differences;
A closed region determining step for determining a predetermined closed region including the one or more differences;
A highest point selecting step of selecting the highest point of the three-dimensional image data for each determined closed region;
A holding step for extracting and holding a plurality of slice images parallel to the horizontal plane at predetermined intervals from the highest points selected by the highest point selection step vertically downward for each closed region;
Based on the area of the determination target image that is a predetermined closed region in the slice image at a position lower than the highest point by a predetermined interval among the plurality of slice images held by the holding step, the determination target image is A passer-by recognition step for recognizing each closed region whether or not the image shows a passer-by;
A passer counting method comprising: a step of summing up the number of passers-by recognized by the passer-by-passenger step for each closed region. An imaging process of taking a 3D image and obtaining 3D image data;
A recording step of recording the initial three-dimensional image data imaged by the imaging step;
An extraction step of comparing the recorded initial three-dimensional image data and the acquired three-dimensional image data to extract one or more differences;
A closed region determining step of determining a predetermined closed region including the one or more differences;
A maximum point selection step of selecting the highest point of the three-dimensional image data for each determined closed region;
For each of the closed regions, a holding step of extracting and holding a plurality of slice images parallel to the horizontal plane at predetermined intervals from the highest points selected by the highest point selection step vertically downward;
Based on the area of the determination target image that is a predetermined closed region in the slice image at a position lower than the highest point by a predetermined interval among the plurality of slice images held in the holding step, the determination target image is A passer-by recognition process for recognizing for each closed region whether or not the image shows a passer-by;
A passer-counting program comprising: a counting step of totaling the number of passers-by recognized by the passer-by-passenger step for each closed region.