JP5862410B2 - Number of people counting server, number of people counting device, and number of people counting system - Google Patents

Description

  The present invention relates to a person counting server, a passing person counting apparatus, and a passing person counting system that count the number of photographed persons, and in particular, face position information indicating a face position and person position information indicating a position of the entire person. The present invention relates to a number counting server connected to a processing device that calculates

  A conventional pedestrian detection system includes a video sensor, a sensor unit that detects the figure and face of a person from the output video of the video sensor, detects and measures the number and attributes of the person, and an aggregation server that counts the measurement results of the sensor unit. (Refer nonpatent literature 1). This pedestrian detection system is configured to detect a person's figure and face from an image and measure age and sex from a face image. Non-patent document 2 describes that “the contour shape of a pedestrian is expressed using edge information of an image”. Note that Patent Document 1 describes an image processing technique for evaluating a result by applying a plurality of different conditions.

Akitoshi Tsukamoto, Satoshi Sugano, Yoshinori Okuma “Advertising Effect Measurement Support System by Video Recognition“ RESCAT ”” OKI Technical Review October 2011 / No. 218 Vol. 78 No. 1 Daisuke Takeuchi, Kensuke Marutani “Development of Pedestrian Recognition Technology in Safe Driving Support System” Panasonic Technical Journal Vol.54 No.4 Jan.2009

JP 2000-222576 A (paragraph 0030)

However, while the technique for measuring the number of persons based on face images is more accurate than when measuring the number of human figures and contours, it is necessary that at least a person's face be reflected in the video. . For this reason, the technique of measuring the number of passing pedestrians using this technique cannot measure the number of passing persons in the direction of turning away from the camera, for example. Therefore, if the number of passing pedestrians is to be measured in both directions, it is necessary to prepare two measurement systems together with a technique for measuring the number of pedestrians based on the figure and contour shape of the person.
However, using both the technology to measure the number of facial images and the technology to measure the number of pedestrians passing based on the number of human figures and contours (whole person), the number of people moving away can be measured. Accuracy cannot be improved.

  The present invention has been made to solve such a problem, and an object of the present invention is to improve the measurement accuracy for measuring the number of passing people based on the total number of people.

  In order to solve the above problems, the first means of the present invention provides a plurality of pairs of face position information indicating the position of the face of the imaged person and person position information indicating the overall position of the imaged person. A person counting server (103) that receives and counts the number of persons, a filter unit (232) for selecting a path of the person position information, and a path of the selected person position information selected by the filter unit; A totaling unit (233) that counts the number of (person trajectories), and the filter unit is an approaching person selecting unit (selecting an approaching person trajectory approaching the person in the trajectory of the person position information) 232b) and a parameter adjustment unit (232a) that automatically adjusts the filtering parameters so that the number of approaching person trajectories is approximately equal to the number of trajectories of the face position information. However, the symbols in parentheses are examples.

  The second means of the present invention includes a trajectory information calculation unit that sequentially calculates face position trajectory information indicating the movement trajectory of the face of the imaged person and person trajectory information indicating the entire movement trajectory of the imaged person. A person count server that is communicably connected to a video processing device (for example, the processing device 102) and that counts the number of persons, and a plurality of face position trajectory information and person trajectory information that are sequentially calculated; A receiving unit that receives continuous information, a filter unit that uses a filtering parameter to select among the plurality of received human trajectory information, and a totaling unit that counts the number of selected human trajectory information selected by the filter unit The filter unit includes an approaching person trajectory information that selects approaching person trajectory information representing an approaching person trajectory approaching the person in the person trajectory information, and the approaching person trajectory information. As the number of substantially equal to the number of the face position locus information, characterized by comprising a parameter adjusting unit for automatically adjusting the filtering parameters.

  According to a third aspect of the present invention, the position information indicating the position of the face of the person captured by the image capturing apparatus and the position information indicating the position of the entire captured person are sequentially calculated. A processing unit that performs selection, a filter unit that selectively selects a trajectory of the person tracking position information, and a totaling unit that counts the number of trajectories of the selected person position information selected by the filter unit, and counts the number of persons The filtering unit includes an approaching person selection unit that selects an approaching person trajectory that the person approaches in the trajectory of the person position information, and the number of approaching person trajectories is the number of approaching person trajectories. And a parameter adjusting unit that automatically adjusts the filtering parameter so as to be approximately equal to the number of trajectories of the face position information.

  Here, the filtering parameter is a tracking time length of the approaching person locus, and the filter unit selects the approaching person locus whose tracking time length is equal to or greater than a threshold.

  Further, the filtering parameter includes a moving direction of the person, and the filter unit excludes the person trajectory information moving in a specific direction before performing the selection of the approaching person trajectory according to the tracking time length. Features.

  Further, the initial value of the threshold value of the tracking time length is set to a value shorter than the initial value at the normal time when the number of persons is larger than the normal time.

  ADVANTAGE OF THE INVENTION According to this invention, the measurement precision which measures the number of passing persons based on the number of the whole person can be improved.

1 is a configuration diagram of a traffic number counting system according to a first embodiment of the present invention. It is a figure which shows an example of the captured image which the imaging device of 1st Embodiment images. It is a hardware block diagram of the processing apparatus of 1st Embodiment. It is a hardware block diagram of the totaling server of 1st Embodiment. It is a functional block diagram of the processing apparatus of 1st Embodiment. It is a main functional block diagram of the totaling server of 1st Embodiment. It is a flowchart which shows the operation | movement content of the processing apparatus of 1st Embodiment. It is a flowchart which shows the operation | movement content of the total server of this embodiment. It is a flowchart which shows the operation | movement content of the total server of the 2nd Embodiment of this invention.

Next, embodiments of the present invention will be described in detail with reference to the drawings as appropriate.
Hereinafter, an embodiment of the present invention (hereinafter referred to as “the present embodiment”) will be described in detail with reference to the drawings. Each figure is only schematically showing the present invention. Therefore, the present invention is not limited to the illustrated example. Moreover, in each figure, the same code | symbol is attached | subjected about the common component and the same component, and those overlapping description is abbreviate | omitted.

(First embodiment)
FIG. 1 is a configuration diagram of a traffic number counting system according to the first embodiment of this invention.
In FIG. 1, the passing number counting system 100 includes an imaging device (camera) 101, a processing device 102, and a counting server 103 as a number counting server, and the counting server 103 includes a data storage device 104. Here, the processing device 102 and the aggregation server 103 are communicably connected via a network (LAN (Local Area Network) or WAN (Wide Area Network)).
Here, in the present embodiment, the processing apparatus 102 and the aggregation server 103 are connected via a network, but the functions of the processing apparatus and the aggregation server can be integrated to constitute a “passenger count apparatus”.

The imaging device 101 is a camera that sequentially captures a human image and outputs a video signal. For example, a commercially available small camera that can be connected to a personal computer with a USB interface can be used as it is.
The processing device 102 processes the captured image captured by the imaging device 101, sequentially detects attributes such as a face position, a person position, and sex, and outputs position information and a moving direction. The counting server 103 counts the number of passing people based on the position information detected by the processing device 102.

FIG. 2 is a diagram illustrating an example of a captured image captured by the imaging apparatus 101 according to the present embodiment.
FIG. 2 shows a captured image of a person A 201 that faces the door 203 toward the imaging apparatus 101 (front) and a person B 202 that moves away from the imaging apparatus 101.

FIG. 3 is a hardware configuration diagram of the processing apparatus 102 according to the present embodiment.
The processing device 102 includes an input / output unit 121 that inputs and outputs data, a ROM (Read Only Memory) 122 that stores fixed data and the like, and a CPU (Central Processor Unit) 123 that performs control and arithmetic processing. A rewritable memory RAM (Random Access Memory) 124 that temporarily stores processing data and the like, and a communication unit (communication) that receives an imaging signal from the imaging device 101 and transmits data (position information) to the aggregation server 103 via the network Interface) 125 and an external storage unit 126 such as a hard disk drive device, which are connected to each other via a bus line.

FIG. 4 is a hardware configuration diagram of the aggregation server 103 according to the present embodiment. The aggregation server 103 is constituted by, for example, a PC, and an input / output unit 131, a ROM 132, a CPU 133, a RAM 134, a communication unit 135, an external storage unit 136, and a recording medium driving unit 137 are connected by a bus line. The data storage device 104 is connected to the recording medium driving unit 137.
The input / output unit 131 is connected to input / output devices such as a monitor and a keyboard. The ROM 132 stores an initial boot program that is activated when the power is turned on.

  The RAM 134 is used as a working memory. The communication unit 135 is a communication interface that inputs information including person position information and face position information from the processing apparatus 102 via a network. The external storage unit 136 is, for example, a hard disk drive device. The recording medium drive unit 137 is an interface that drives the data storage device 104 for data storage.

FIG. 5 is a functional block diagram of the processing apparatus 102 according to the present embodiment.
As shown in FIG. 5, the CPU 123 as the control unit of the processing apparatus 102 (FIG. 3) executes a program stored in the ROM 122, thereby causing the video reception unit 221, the person position detection unit 222 a, and the person position comparison. Section 223a, person continuity determination section 224a, person position data file creation section 225a, person position data storage function 226a, face position detection section 222b, face position comparison section 223b, face continuity determination section 224b, face position data file creation section 225b, a face position data storage function 226b, and an output unit 227.

  The video receiver 221 receives an imaging signal (video signal) from the imaging device 101 (FIG. 1). The person position detection unit 222a detects the position of the person using the output signal of the video reception unit 221. The person may be detected by using, for example, the outline shape of the pedestrian from the edge information of the image. The person position data storage function 226a stores the person position data detected by the person position detection unit 222a as a file in the external storage unit 126 (FIG. 3). The person position comparison unit 223a determines whether or not the newly detected person is substantially the same as the position of the person stored as a file. The person continuity determination unit 224a determines that the tracking is being continued when the person position comparison unit 223a determines that the person is located at substantially the same position. On the other hand, if the person continuity determination unit 224a is a person who has not been in the same position and is a newly detected person, the person continuity determination unit 224a determines that the person is a new character. The person position data file creation unit 225a stores the current position of the person who is being tracked as a tracking position in the file, and stores the time and position of the new character in the file.

  The functions of the face position detection unit 222b, the face position comparison unit 223b, the face continuity determination unit 224b, the face position data file creation unit 225b, and the face position data storage function 226b are the same as those of the person position detection function. Is omitted. The output unit 227 outputs person position information indicating the position of the person imaged by the imaging apparatus 101 and face position information indicating the position of the face.

FIG. 6 is a main functional block diagram of the aggregation server 103 of the present embodiment.
As shown in FIG. 6, the totaling server 103 as the number totaling server is configured to include an input unit 231, a filter unit 232, and a totaling unit 233, and the filter unit 232 includes a parameter adjustment unit 232 a and an approaching person. And a selector 232b. In addition, the code | symbol in () is demonstrated in 2nd Embodiment.

  The input unit 231 inputs the position of the face and the person and the trajectory (tracking result) from the processing device 102 via the communication unit 135 as a file. The filter unit 232 inputs the person position information, automatically adjusts a filtering parameter (for example, a tracking time length threshold), and filters the approaching person data using the automatically adjusted threshold. That is, the approaching person selection unit 232b selects approaching person data, and the parameter adjustment unit 232a automatically adjusts the threshold of the tracking time length so that the number of approaching person data and the number of face data approximate. To do.

  The totaling unit 233 inputs the person position information that has approached by filtering from the filter unit 232, and inputs the unfiltered face position information from the input unit 231. Then, the totaling unit 233 totals the number of approaching person position information and the number of face position information, and feeds back the totaled data to the parameter adjustment unit 232a. Thereby, the parameter adjusting unit 232a can automatically adjust the tracking time length threshold so that the number of person data approaching and the number of face data approximate.

FIG. 7 is a flowchart showing the operation content of the processing apparatus 102 of the present embodiment.
Hereinafter, the operation of the processing apparatus 102 will be described with reference to the functional block diagram of FIG. 5 and the flowchart of FIG.

  In step S <b> 301, the processing apparatus 102 determines whether the video reception unit 221 has received a video (image) from the imaging apparatus 101. While new images are being received sequentially (Yes in S301), the processing from step S302 is executed. On the other hand, if there is no video reception, this routine is stopped (No in S301).

Next, the face position detection unit 222b detects a face in step S302 and determines its position and attribute. The face can be detected based on, for example, whether there are eyes, nose or mouth inside the face outline.
Next, in step S303, the face position comparison unit 223b determines whether or not the detected face is the same as the face that has already been detected. For example, the face position comparison unit 223b determines whether or not a face has already been detected at substantially the same position in the video received immediately before being stored by the face position data storage function 226b.

When the face position comparing unit 223b determines that the face is already detected at the same position, the face continuity determining unit 224b determines that the face is being tracked and determines the current position. In step S304, the tracking position is recorded in the file in the external storage unit 126 via the face position data file creation unit 225b. If the attribute determination result has changed, the attribute of the file is updated as necessary. .
When it is determined by the face position comparison unit 223b that the face has not been previously located at the same position as a newly detected face, the face continuity determination unit 224b, in step S304, The tracking start position is stored in a file via the face position data file creation unit 225b.

  On the other hand, for the face that has not been tracked, the face continuity determination unit 224b determines the tracking start time and tracking start position in step S304, the tracking end time at which the position was last stored, and the tracking end position as face position data. It is stored in a file via the file creation unit 225b.

Next, in step S305, the person position detector 222a detects the position of the person.
Further, in step S306, the person position comparison unit 223a determines whether or not the detected person is the same as the already detected person. Specifically, the person position comparison unit 223a determines whether a person has already been detected at substantially the same position in the video received immediately before being stored in the file via the person position data storage function 226a. .

When the person position comparing unit 223a determines that the person has already been detected at the same position, the person continuity determining unit 224a determines that the person is continuing to be tracked, and the current position is determined in step S307. Is stored as a tracking position in a file via the person position data file creation unit 225a.
On the other hand, if it is determined that the person position comparison unit 223a is a person who was not previously in the same position and is a newly detected person (new character), in step S307, the person continuity determination unit 224a. Stores the time and position as a tracking start time and a tracking start position in a file via the person position data file creation unit 225a.

  Next, in step S307, the person continuity determination unit 224a sets the tracking start time, the tracking start position, and the time at which the position was last stored as the tracking end time for the person who is determined not to continue tracking. The position is stored as a tracking end position in a file via the person position data file creation unit 225a.

  If there is a face detected in step S302 whose position is not included in the position of the person detected in step S305, the person position comparison unit 223a determines the position of the undetected person. It is also possible to execute the processing in step S306 after estimating the position and adding the result to the detection result in step S305.

Next, the output unit 227 transmits the data file in which the tracking result created by the face position data file creation unit 225b and the person position data file creation unit 225a is recorded to the aggregation server 103 to the aggregation server 103 in step S308. .
Thereafter, the flow returns to the start position, and the next video signal (image) is sequentially received.

FIG. 8 is a flowchart showing the operation content of the aggregation server 103 of this embodiment.
The operation of the aggregation server 103 will be described below with reference to this flowchart and the functional block diagram of FIG.

First, in step S <b> 401, the input unit 231 confirms whether a new file has been transmitted from the processing apparatus 102 via the communication unit 135.
If a new file has arrived (Yes), the process proceeds to step S402, and the input unit 231 starts tracking data recorded in the file, that is, each detected and tracked face and each person. The time, the tracking end time, the tracking start position, and the tracking end position are stored in the data storage device 104 (FIGS. 4 and 6) in the aggregation server 103.

  That is, the input unit 231 sequentially receives the tracking start time, the tracking end time, the position at the start of tracking, and the position at the end of tracking as face / person pair information, and the information storage device 104 receives the information pair. Will be stored sequentially. That is, as a result, the data storage device 104 stores the locus of the position of the face and the person. Further, the processing device 102 has transmitted the tracking start time, the tracking end time, the position at the start of tracking, and the position at the end of tracking to the aggregation server 103. It can also be transmitted to the sequential counting server 103. Also by this, the data storage device 104 stores the locus of the face and the position of the person.

Next, whether the new file has arrived in step S401 (Yes) and the tracking data is stored in S402 or has not arrived (No), the input unit 231 receives the data from the data storage device 104 in step S403. The face that started tracking in the time zone specified by the user and the tracking result data of each person are read out.
In the following description, the number of face data read from the data storage device 104 is F, and the number of person data is H.

  Next, in step S <b> 404, the approaching person selection unit 232 b of the filter unit 232 checks the passing direction of each read data of the person. Specifically, the approaching person selection unit 232b examines the direction of travel by examining the relationship between the tracking start position and the tracking end position for each person's data. In the following description, as a result of the determination of the direction of travel, the tracking start position is behind the tracking end position and the number of data of the person determined to have approached the imaging apparatus 101 is represented by Hn. The number of persons to be played is represented by Hf.

Next, in step S405, the approaching person selection unit 232b automatically adjusts the filtering parameter, and performs filtering on each of the Hn pieces of person data determined to have approached in step S406.
This filtering can be used to determine whether the tracking start position and the tracking end position are at appropriate positions on the screen, and whether the time from the tracking start to the tracking end is an appropriate time length. it can. The parameter adjustment unit 232a adjusts the filtering parameter (for example, the threshold of the tracking time length) so that the number of data Hna of the narrowed down person is relatively close to the number of face data F as a result of filtering.

  For example, whether or not the time from the tracking start time to the tracking end time is an appropriate time length is determined by whether or not the tracking time length is longer than a threshold automatically adjusted by the parameter adjustment unit 232a. By this filtering, the tracking time is shortened as in the case where it is temporarily hidden by others, and the case where it is detected again (double count) is excluded.

  The threshold value is determined by setting the threshold value T to the initial value T0 in the process of step S305 in FIG. 7 when the number of personal data items whose tracking time length is equal to or greater than the threshold value T is Hn (T). The value of Hn (T) is lowered by increasing the value from. Further, by adding a temporary undetected person, Hn (T0) is always larger than the number F of face data. Therefore, the parameter adjustment unit 232a can determine the threshold value T at which the number Hn (T) of person data is close to the number F of face data. The comparison between the number Hn (T) of person data and the number F of face data is executed by the totaling unit 233, and the comparison information is fed back to the parameter adjusting unit 232a of the filter unit 232 for adjusting the threshold T. Used.

  The value of the threshold value T is not only automatically adjusted, but can also be adjusted in advance from a sample video and set as a fixed value. Further, the threshold value T need not be determined as a single value. For example, a plurality of threshold values T are obtained from a plurality of sample videos having different face detection numbers, and the threshold value T is applied depending on the size of the face data number F. It is also possible to select and use the value of.

  Note that such filtering can also be applied to each of Hf person data. In this case, the number of person data narrowed down by filtering is set to Hfa.

  Next, in step S407, the totaling unit 233 outputs Hna, Hfa person data by direction of travel, and F face data. These data can be output only as a number, output as a table, or displayed as a graph for each attribute if necessary.

  Hereinafter, the filtering performed by the filter unit 232 will be described with reference to another example other than the tracking time length threshold. First, the filter unit 232 may have a function of determining whether the tracking start position and the tracking end position are at appropriate positions on the screen. For example, when a passage and escalator that do not require measurement are connected to a passage where the number of passing people is to be measured, the person position detection unit 222a and the face position detection unit 222b may perform the passage that does not require measurement in advance before the position detection. Or mask the escalator area.

Further, the filter unit 232 can have a function of eliminating position information in which the movement direction (Δx / Δy) of the person is equal to or greater than a predetermined value in order to exclude a person who crosses the door 203 (FIG. 2). .
Further, in a time zone where there are many passing people, most people are excluded at a normal threshold (for example, 1 second). For this reason, a threshold value at the time of congestion that is smaller than the threshold value at the normal time (tracking time length shorter than the normal time) is set. In this case, the threshold value can be set to a fixed value, but the initial value T0 of the automatic threshold adjustment can be changed to a value smaller than the normal threshold value.

  As described above, according to the first embodiment of the present invention, it is possible to simultaneously measure the number of passing people and determine the attributes of passing people using the video of one imaging device (camera). The number of passing people can be shown by passing direction. Furthermore, by optimizing the person detection using the number of approaching human faces, it is possible to accurately detect the number of persons with a face detection accuracy even for backward-facing persons who cannot detect faces or horizontal persons. Data can be obtained. That is, even if the person is facing in any direction, the number data of the person can be obtained accurately with the accuracy of the number of faces.

  According to the present embodiment, the tracking time length threshold is automatically adjusted to about 1 second as a result. If the threshold value of the tracking time length is made longer, the tabulation server 103 of the present embodiment increases the human detection accuracy, but this time a situation where it cannot be detected occurs. For example, if a threshold is set to 10 seconds at a place where it normally passes in 2 to 3 seconds, a person cannot be detected. The tabulation server 103 according to the present embodiment uses the value that makes the number of persons close to the number of faces as a threshold for the tracking time length, thereby eliminating such an undetectable situation and improving the detection accuracy of the person. Automatic adjustment is made up to the last threshold (optimum threshold) that improves the detection accuracy.

(Second Embodiment)
Next, a second embodiment of the present invention will be described.
The second embodiment is characterized in that, in addition to the functions of the first embodiment, the contents of filtering performed in the aggregation server can be set in advance.
The overall configuration of the passable person counting system 100-1 of the second embodiment is different from the overall configuration of the first embodiment in that the counting server 103 is the counting server 103-1. That is, the filter unit 132-1 of the aggregation server 103-1 includes a parameter adjustment unit 232a-1 and an approaching person selection unit 132b-1. The summary server 103-1, the filter unit 132-1, the parameter adjustment unit 232a-1, and the approaching person selection unit 132b-1 are described in parentheses in FIG.

  Here, the operation of the aggregation server in the second embodiment will be described. Note that the operation of the processing apparatus 102 is the same as that of the first embodiment.

FIG. 9 is a flowchart showing the operation content of the aggregation server 103-1 of this embodiment.
The operation content of the aggregation server 103-1 of this embodiment will be described along the flowchart of FIG.

The processing from step S501 to S504 in FIG. 9 is the same as the processing from step S401 to S404 in the flowchart of FIG. 8 of the first embodiment.
In step S505, the filter unit 232-1 performs filtering on each of the Hn pieces of person data.
At this time, a plurality of determination conditions to be applied to each person data are set in advance, and filtering is performed by determining whether or not these conditions are met.

  In the following description, the number of data of persons who respectively match these conditions J1, J2 to JN by filtering on Hn pieces of person data is represented as Hna (k) (where k = 1 to N).

Next, in step S506, the counting unit 233 selects an optimum filtering result. This can be done, for example, by adopting the filtering result that has the smallest number of person data with the number of face data F or more.
The number of person data of the selected filtering result is HNa. Also, the selected filtering can be applied to each of Hf person data. In this case, the number of person data narrowed down is HFa.

  Next, in step S507, the counting unit 233 outputs HNa pieces, HFa pieces of person data according to the direction of travel, and F pieces of face data. These data can be output only as a number, output as a table, or displayed as a graph for each attribute if necessary.

  As described above, the passing number counting system 100-1 according to the second embodiment of the present invention performs filtering in the counting server when measuring the passing number for each passing direction using a single imaging device. Can be set in advance.

  In the first embodiment and the second embodiment, as a method of determining the passing direction, it is determined whether or not the imaging apparatus 101 is approached from the relationship between the tracking start position and the tracking end position of each person. However, if the tracking start position is to the right of the tracking end position based on the installation position of the imaging device and the movement direction of the person at the time of measurement, it can also be determined that the movement has been made from right to left.

100 Passenger Counting System 101 Imaging Device 102 Processing Device (Video Processing Device)
103 Counting server (Number of people counting server)
104 Data storage device (storage unit)
121,131 Input / output unit 122,132 ROM
123, 133 CPU (control unit)
124,134 RAM
125, 135 Communication unit 126, 136 External storage unit 137 Recording medium drive unit 201 Person A
202 Person B
203 Door 221 Video reception unit 222a Human position detection unit 223a Human position comparison unit 224a Human continuity determination unit 225a Human position data file creation unit 226a Human position data storage function 222b Face position detection unit 223b Face position comparison unit 224b Face continuity determination Unit 225b face position data file creation unit 226b face position data storage function 227 output unit 231 input unit 232 filter unit 232a parameter adjustment unit 232b approaching person selection unit 233 tabulation unit

Claims (7)

A number counting server that receives a plurality of pairs of face position information indicating the position of the face of a captured person and person position information indicating the position of the entire captured person and counts the number of persons. ,
A filter unit for selecting a trajectory of the person position information;
A totaling unit that counts the number of trajectories of the selected person position information selected by the filter unit;
The filter unit is
An approaching person selection unit that selects an approaching person trajectory that the person approaches in the trajectory of the person position information;
And a parameter adjusting unit that automatically adjusts a filtering parameter so that the number of approaching person trajectories is substantially equal to the number of trajectories of the face position information. It is possible to communicate with a video processing apparatus including a trajectory information calculation unit that sequentially calculates face position trajectory information indicating the movement trajectory of the imaged person's face and person trajectory information indicating the entire movement trajectory of the imaged person. A number of people counting server connected to count the number of persons,
A receiving unit that receives a plurality of pieces of continuous information in which the face position locus information and the person locus information are sequentially calculated;
A filter unit for selecting using a filtering parameter from the received plurality of person trajectory information;
A totaling unit that counts the number of selected person trajectory information selected by the filter unit;
The filter unit is
An approaching person selecting unit that selects approaching person trajectory information representing an approaching person trajectory approaching the person in the person trajectory information;
And a parameter adjusting unit that automatically adjusts the filtering parameter so that the number of approaching person trajectory information is substantially equal to the number of face position trajectory information. The filtering parameter is a tracking time length of the approaching person trajectory,
The number-of-persons aggregation server according to claim 1, wherein the filter unit selects the approaching person locus whose tracking time length is equal to or greater than a threshold value. The filtering parameter is the moving direction of the person and the tracking time length of the approaching person trajectory ,
The filter unit, number aggregation server of claim 2, wherein after excluding the person locus information to move to a specific direction, to make a selection of the approaching person locus by the tracking time length.   The initial value of the threshold of the tracking time length is set to a value shorter than the initial value at the normal time when the number of the persons is larger than the normal value. Number counting server. Is connected to the imaging device, a processing unit for sequentially calculating a person position information indicating the position of the entire person is the imaging face position information indicating the position of the face of the person captured by the imaging device, the human object position A traffic number counting device for counting the number of persons, comprising: a filter unit for selecting a locus of position information; and a counting unit for counting the number of tracks of selected person position information selected by the filter unit. ,
The filter unit is
An approaching person selection unit that selects an approaching person trajectory that the person approaches in the trajectory of the person position information;
And a parameter adjusting unit that automatically adjusts a filtering parameter so that the number of approaching person tracks is substantially equal to the number of tracks of the face position information. An imaging apparatus, a video processing apparatus that sequentially calculates a pair of face position information indicating the position of the face of a person imaged by the imaging apparatus and person position information indicating the position of the entire imaged person, and the video processing A number-of-passenger counting system that includes a plurality of information counting servers that receive a plurality of information pairs calculated by the device and count the number of the persons,
A filter unit to sift the trajectory of the human object position location information,
A totaling unit that counts the number of trajectories of the selected person position information selected by the filter unit;
The filter unit is
An approaching person selection unit that selects an approaching person trajectory that the person approaches in the trajectory of the person position information;
And a parameter adjusting unit that automatically adjusts a filtering parameter so that the number of approaching person tracks is substantially equal to the number of tracks of the face position information.

Images