JP2001291184A - Traffic volume investigating system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To realize a system capable of automatically and correctly grasping a traffic volume of people or vehicles without manual intervention. SOLUTION: This traffic volume investigating system 10 is a system for automatically measuring the traffic volume of mobile bodies in an area to be observed and is provided with digital video cameras 14a-14p, an inspection object storage means 22 for storing the feature of the inspection objects, a mobile body detecting means 18 for detecting the appearance of the mobile body in images inputted from the cameras 14a-14p, an inspection object recognizing means 24 for deciding whether the detected mobile body coincides with the object to be inspected or not by referring to the storage means 22, an image fetching means 26 for fetching a still image when the mobile body coincides with the inspection object, an image registering means 28 for storing the still image into an image database 30 by a JPEG form and a calculating means 32 for counting image files in the image database 30 to calculate the traffic volume.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0101]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a system for automatically counting the traffic of a moving object such as a person or an automobile.
In particular, the present invention relates to a system that can grasp not only a mere amount of traffic but also an attribute configuration of a moving body that has passed.

[0092]

2. Description of the Related Art The traffic volume of people, automobiles, and the like in a specific place is the most basic data in marketing, and it is not an exaggeration to say that an effective strategy cannot be formulated without it. For this reason, a large number of personnel are arranged around a destination, and the number of passing people and vehicles is counted by a manual counter. In the case of manual traffic volume surveys, it is necessary to understand not only traffic volume but also gender composition, age composition, and vehicle type composition by using multiple counters according to the attributes of the survey target. There are advantages that can be.

[0093]

However, it is generally difficult to carry out a long-term survey at many observation points in a survey relying on human naval tactics, as described above, because of the high labor costs. At present, marketing has to be based on insufficient data. In addition, in a dangerous zone where humans cannot approach or in a severe environment, it is naturally impossible to investigate. Of course, automatic counting using a computer has also been attempted, but in the conventional method, shading is performed on a certain range of images captured as a sample, and the number of objects included in each cell is determined to calculate the total number. Since the estimation is performed, there is a disadvantage that an error easily occurs. Also, it is difficult to automatically recognize the attribute of the investigation target.

The present invention has been devised to solve the conventional problems described above, and requires no manual operation.
It is an object of the present invention to realize a system that can automatically and accurately grasp the traffic amount of a person, a car, and the like, and that can automatically analyze the attribute configuration.

[0056]

In order to achieve the above object, a traffic volume survey system according to the present invention is a system for automatically measuring the traffic volume of a specific moving object in a predetermined observation area. Imaging means for inputting an image in a region, a research object storage means for storing characteristics of a moving object to be investigated, and a moving object for detecting when a moving object appears in an image input from the imaging means A detection unit, a survey target confirming unit that refers to the survey target storage unit and determines whether the detected moving body matches the survey target; and, if the moving body matches the survey target, Image capturing means for capturing a still image, image registering means for storing the image in an image database in a predetermined file format, and counting image files stored in the image database. And a counting means for calculating the traffic amount. For example, a CCD camera corresponds to the imaging unit.

In this traffic volume survey system, the moving object detection means constantly monitors the image in the observation area, and when the appearance of the moving object is detected, it is determined whether or not the moving object matches the object to be investigated. The image registration means functions to take in a still image of the moving object only when the investigation target confirmation means makes a judgment and a positive judgment result is derived, so that the image file registered in the image database by the counting means By counting the number, it is possible to automatically and accurately grasp the traffic amount of the moving object.

A plurality of the image pickup means are desirably arranged on the front side or the back side in the traveling direction of the moving body. Further, the observation area is divided into a plurality of areas having a predetermined width set based on the width of the moving object to be investigated, and each of the imaging units includes the plurality of areas. In each of them, an area in charge of each is assigned in advance. In this case, the moving body detection means detects the appearance of the moving body in the shared area of each imaging means, and the investigation target confirmation means determines whether or not the moving body matches the investigation target for each of the imaging area sharing areas. It is determined, and the image capturing unit captures a still image of the moving object in units of the sharing area of each imaging unit. In this way, by dividing the observation area to be investigated into a plurality of areas, and arranging the imaging means in each of the sections to capture a still image of the moving body for each shared area, the moving body detection means, Confirmation means, image capture means,
There is an advantage that the processing in the image registration unit and the counting unit can be simplified and simplified.

It is desirable that the width of each assigned area be set smaller than the average width of the moving object to be investigated. For example, when the survey target is a human, a width of 30 cm is one standard. In this way, by arranging the plurality of imaging units on the front side or the back side in the traveling direction of the moving body, and allocating a sharing area smaller than the width of the moving body to each imaging unit, the plurality of moving bodies are arranged side by side in the horizontal direction. Even when moving in the observation area at the same time, each moving object is surely captured by the imaging means in charge of the area to which it belongs. In addition, since one moving object is photographed in the still image captured by the image capturing unit, the traffic amount of the moving object can be calculated by simply counting the number of image files in the counting unit. In addition, as a result of setting the width of each sharing area narrower than the width of the moving body, a part of the moving body naturally extends to another adjacent area, and the movement is captured by the moving body detecting means. If only the main part of the moving object (in humans, the “head” or “torso”) is set as the subject of investigation, the image of the same moving object may be erroneously captured in two or more shared areas. Can be prevented.

Alternatively, a plurality of the image pickup means may be arranged on the front side or the rear side in the traveling direction of the moving object, and the traveling direction regulating means is arranged along the traveling direction of each moving object to be investigated. May be divided into a plurality of areas, and each of the imaging units may be assigned in advance an area in charge of each of the plurality of areas.
Each region has a predetermined width set based on the width of the moving body. This width is desirably set to less than twice the average width of the moving body. Again,
The moving object detection means detects the presence of a moving object in the sharing area of each imaging means, and the investigation target confirmation means determines whether or not the moving object matches the investigation target for each sharing area of each imaging means, The image capturing unit captures a still image of the moving object in units of the sharing area of each imaging unit. As described above, by dividing the observation region into a plurality of sharing regions by the traveling direction restricting means and setting the width of each sharing region to be less than twice the average width of the moving object, each sharing region is set to two or more. The moving objects cannot travel side by side, and only one moving object always advances in one sharing area. As a result, if the image capturing means captures an image for each assigned area, only a still image of one moving object is automatically captured, and the counting means simply counts the number of image files. With this, the traffic amount of the moving object can be grasped.

[0010] In the above traffic amount survey system, a feature extracting means for subjecting each image file stored in the image database to necessary image processing to extract features of a photographed moving object; Comparing a feature pattern database storing a plurality of feature patterns corresponding to the attributes of a certain moving object, the features of the moving object extracted by the feature extracting means, and the feature patterns stored in the feature pattern database; It is desirable to provide attribute analysis means for estimating the attribute of the moving object. As a result, not only can the amount of traffic of the moving object be automatically grasped, but also its attribute configuration can be grasped to some extent, and there is an advantage that valuable data as basic data for marketing can be collected.

[0011]

FIG. 1 shows an embodiment of a traffic volume survey system 10 according to the present invention. For example, a plurality of digital video cameras 14a to 14c arranged on a passage 12 at an event venue such as a concert. 14p and each camera 14a ~
A personal computer (personal computer 16) for performing predetermined processing on image data input from 14p.

The cameras 14a to 14p are CCD cameras and are arranged at a height of 3 m or more from the floor. Each camera 14 and the personal computer 16 are connected via a cable and a dedicated interface board (stored in an expansion slot of the personal computer 16). Of course, each camera 14 and the personal computer 16 may be connected wirelessly by wireless. PC 16 above
In addition to the predetermined OS, a dedicated application program for realizing this system is installed in the built-in hard disk device.

FIG. 2 is a block diagram showing the functional configuration of the system 10. The moving object detecting means 18, the investigation target setting means 20, the investigation target storage means 22, and the investigation target confirmation means 24
Image capturing means 26, image registering means 28, captured image database 30, registered image counting means 32, feature extracting means
34, attribute analysis means 36, feature pattern database 38
And a result display means 40.

The moving object detecting means 18 and the investigation object confirming means 2
4, image capturing means 26, image registering means 28, registered image counting means 32, feature extracting means 34, attribute analyzing means 36,
Is executed by the CPU of the CPU executing predetermined processing according to the OS and the dedicated application program.
The captured image database 30 and the feature pattern database 38 are both stored in the hard disk device of the personal computer 16. Further, as the result display means 40,
The display device of the personal computer 16 corresponds to this.

The survey target setting means 20 is provided in the system
This is for inputting and setting the characteristics of the object whose traffic amount is to be investigated by 10, and specifically includes input devices such as a keyboard and a mouse of the personal computer 16. For example, when investigating the traffic volume of humans, the human vertical dimension range (70
Human characteristic data such as 〜220cm), Width range (25-60cm), outline line, Dimension range of main parts such as head and torso, color range of face part, moving speed range (4-8km / h) Is entered. These settings can be made on the personal computer 16
This is done by calling on the screen and inputting necessary information from the keyboard or mouse. Of course, it is also possible to register a template for each of several survey targets in advance and operate such that only necessary portions are corrected according to the characteristics of the survey. The data set here is stored in the investigation target storage means secured in the hard disk device or memory of the personal computer 16.
Stored in 22.

FIG. 3 shows an example of a screen display of the personal computer 16 at the time of conducting the survey, in which image data input from each of the cameras 14a to 14p is combined and displayed. In other words, this image is in the observation area (near entrance 42 of the event venue)
The images captured by a total of 16 cameras 14a to 14p installed in the camera are connected together to form the entire range of the observation area, and are divided into 16 small areas in total. Among them, the area A is the camera 14a, and the area B is the camera 14a.
b, the area C is the camera 14c, the area D is the camera 14d,
Area F is camera 14f, area G is camera 14g, area H
Is camera 14h, area I is camera 14i, area J is camera 14j, area K is camera 14k, and area L is camera 14l.
However, the area M is assigned to the camera 14m, the area N is assigned to the camera 14n, the area O is assigned to the camera 14o, and the area P is assigned to the camera 14p. The sharing area of each camera 14a to 14p is 3m in height.
Since it has a rectangular shape with a width of 30 cm, the area to be observed has a total area of 3 m long x 4 m wide 80 cm.

Each of the cameras 14a to 14p actually captures an image in a wider range than the above-mentioned sharing area, but a part of the area is to be monitored by the cameras 14a to 14p. Is preset on the program. For example, FIG. 4 shows an image of a video captured by the camera 14l, in which only the L region in the entire observation region is set as the observation sharing region.
Similarly, FIG. 5 shows the image of the image captured by the camera 14k, FIG. 6 shows the image of the image captured by the camera 14i, FIG. 7 shows the image of the image captured by the camera 14h, and FIG. The images of the video captured by each are shown. As is clear from FIGS. 4 to 8, the hardware performance of each camera 14 can cover a wider range than each assigned area, but the camera 14 is only responsible for one-sixteenth of the entire observation area. -ing

Next, a procedure for investigating the traffic volume of a person will be described with reference to the flowchart of FIG. First,
The characteristics of the person to be investigated are set in the investigation object storage means 22 in advance by the investigation object setting means 20. From each of the cameras 14a to 14p, the respective sharing area A
-P (30 frames per second) is always transmitted to the personal computer 16, and the personal computer 16 has a moving object detecting means.
18 checks whether there is a moving object entering each of the areas A to P (S10). Here, when the moving object detecting unit 18 detects the appearance of any moving object, the moving object waits until the entire moving object enters the sharing areas A to P, and the moving object is stored in the investigation target storage unit 22 by the inspection target confirmation unit 24. It is determined whether or not the condition is set in advance, that is, whether or not it has a human characteristic (S12). If an affirmative decision is made here, the image of the assigned area is instantaneously cut out by the image capturing means 26 (S14), and the image registration means
After being converted into a predetermined file format (for example, JPEG format) via the file 28, it is stored in the captured image database 30 (S16). Even when the passage of a person is detected in another plurality of shared areas, the same processing as described above is performed, and a plurality of images are captured at the same time.

As described above, the captured image database
Since an image of each passer-by is stored in 30 and one image file is generated for each person, the number of image files is counted by the image counting means 32 after the survey is completed. S18), the total traffic is calculated. This result is displayed on the display of the personal computer 16 (S
20). Alternatively, the data is printed out as aggregation data from a printer (not shown) connected to the personal computer 16.

The feature extracting means 34 converts the fetched image file into vector data and performs image processing such as contour extraction and color recognition, thereby extracting feature data of each passerby. Next, by matching the characteristic data with the characteristic pattern data stored in the characteristic pattern database 38 in the attribute analysis means 36, it is possible to determine the attributes such as the sex and age of each passer to some extent.

That is, in the feature pattern database 38, a number of feature patterns for gender identification are stored in advance. For example, “bald → male”, “skirt wear → female”, “long hair → female”, “height 180cm or more → male”, “lipstick applied → female”, “beard → male”,
"Wear a tie → male". Recently, long hair owners are not limited to women, but by combining and determining other characteristic data, it is possible to distinguish between men and women at a fairly high rate. The distinction between adults and children is
Judged primarily by height. Of course, there is a great difference between individuals in height, so it is not simply divisible as `` how many centimeters is more than an adult and how many centimeters are less than a child '', but the height data of each passer calculated by the feature extraction means is Attribute analysis means 36 can derive a certain degree of attribute distribution by comparing with statistical data indicating the correlation between gender and height.

The attribute analysis means 36 assigns an index of "indistinguishable" to an image file in which a moving object that cannot be identified is photographed. Later, an image file of "indistinguishable" may be extracted by using this index as a key, and sequentially displayed on the screen of the personal computer 16 for human identification.

Since the cameras 14a to 14p are arranged in front of the moving body (passenger) in the traveling direction, the cameras 14a to 14p can normally capture the face of a person, but some of them move in the opposite direction. There is also. In this case, the cameras 14a to 14p
Will capture the back of the human head, so that the object to be checked is not determined by the object to be checked 24 and no useless image is captured. In other words, the fact that this system 10 is installed at a concert venue to investigate the traffic of people is because the organizer wants to know the number of visitors. Therefore, it can be said that it is meaningless to count those who proceed in the opposite direction to go to the toilet or the stall after once sitting in the seat. On the other hand, if the purpose of the survey is to simply grasp the traffic volume at a specific place, the number of people who go back may be counted.

Each image file is stored in the captured image database 30.
When the registration is made within the registration, the registration date and time data is added, so that the attribute distribution of passers-by can be analyzed for each time zone. In addition, since it is easy to search for a person who has passed at a specific time later, it can be used for criminal investigation.

In the above description, the width of each of the sharing areas AP of the cameras 14a to 14p is set to 30 cm for the following reason. That is, the width of a human being to be surveyed is usually 30 cm or more. For this reason, even when a large number of people are moving side by side, it is unlikely that a plurality of people enter the sharing area. On the other hand, if the width of each sharing area is set to about 1 m, two people may enter the sharing area side by side. In this case, two people are included in the image file captured by the image capturing unit 26, and it becomes necessary to separate each person in the registered image counting unit 32 and the attribute analyzing unit 36, and the processing becomes complicated. It will be. Therefore, in this system 10, the width of each of the shared areas A to L assigned to one camera 14a to 14p can be barely stored by the main part of one person.
It is set to cm.

As described above, the width of a human is usually 30 cm.
As described above, a part of the body protrudes into the adjacent allotment area. In response to this, by setting the shapes and dimensions of the main parts of a human (for example, “head” and “torso”) in the investigation target storage unit 22 as the investigation target,
It is possible to suppress as much as possible an image relating to the same person from being acquired in two regions. In the example of FIG. 3, the main part of the person passes through the area L, but the left shoulder and the left arm of the person pass through the area M as well. In this case, the presence of a moving object in the area M is detected by the moving object detecting
Is determined to be out of the survey target, and no image is captured. Of course, if the person has advanced so that the center line of the person and the boundary line of the two regions just overlap, the image of the same person may be captured in two regions, and the image may be captured in either region. There is no denying the possibility that this is not the case, but these are only rare cases and some errors can be tolerated in this type of investigation.

The number of cameras 14 to be installed is determined based on a balance between the width of a person to be observed and the size of the observation area. That is, the sharing area of each camera 14 is set to 30 cm for the above reason, and the width of the monitoring target area is 4 m.
Although it was 80 cm, the number of 16 cameras was led, but the number of cameras 14 can be reduced if a narrower portion is to be observed. Conversely, if you want to investigate traffic in a wider area, more cameras
14 needs to be installed.

As described above, the width of each of the sharing areas A to P is
The reason why we set the camera as narrow as 30 cm and installed many cameras 14a to 14p to cover the entire width of the observation area is based on the assumption that the pedestrian to be surveyed will proceed naturally without any restrictions on the course. It is because it is. On the other hand, as shown in FIG. 10, a plurality of courses (sharing areas A to D) are formed by providing a traveling direction regulating means 44 in the passage 12 for regulating the progress of each person in the lateral direction. When the width of the camera 14a is set to be small enough to allow one person to pass, the sharing area of each of the cameras 14a to 14d can be set to 30 cm or more, and the number of cameras installed can be reduced accordingly. For example, at a ticket gate of a station, an automatic ticket gate corresponds to the traveling direction restricting means 44, and only one passerby passes through one course for a few seconds. In this case, the traffic amount can be counted very accurately only by installing one camera for each ticket gate.

In each of the above embodiments, each camera 14
The width of the shared areas A to P of a to p is set to be narrower than the width of the moving object, or the traveling direction restricting means 44 for dividing the course.
Is provided, each sharing area is configured so that only one person is included at a time.Therefore, the image capturing means 26 captures a still image for each sharing area, and the registered image counting means 32 There is an advantage that the traffic amount of a person can be grasped by simply counting the number of still images. However, in order to carry out the present invention, it is not always necessary to configure such that one sharing area includes only one person as described above. That is, even when a plurality of persons are included in one sharing area, the survey target confirmation unit 24 identifies each person, and the image capturing unit 26 captures a still image for each person, or registers a registered image. What is necessary is just to process in the counting means 32 so that each person may be counted separately.

In the above description, an example in which the present invention is used to investigate the traffic of a person has been described. However, the present invention can naturally be applied to the investigation of the traffic of an automobile. In this case, for example, the width of the shared area assigned to each camera 14 is set to be narrower (for example, 1 m50 cm) than the width of one car, and the survey target storage means 22 is adapted to the characteristics of the car. The set conditions (the width range of the vehicle body, the traffic speed range, etc.) are set in advance. The feature pattern database 38 stores feature patterns (vehicle-specific silhouettes, etc.) necessary for vehicle type identification.
Is registered in large numbers. It can also be applied to ecology research of rare animals such as Japanese serow.

[0031]

In the traffic volume survey system according to the present invention, the moving object detecting means constantly monitors the image of the observation area, and when the appearance of the moving object is detected, the survey object confirming means sets the object to be investigated. The image file registered by the counting means in the image database is determined because the image registration means functions to determine whether or not they match and to take in a still image of the moving object only when a positive determination result is obtained. By counting the number of objects, it becomes possible to automatically and accurately grasp the traffic amount of the moving object. For this reason, a sufficient range of investigations can be performed at low cost compared to the conventional traffic volume investigation that relies on manual labor. In addition, it is possible to conduct surveys in danger zones where humans cannot access.

In addition, with respect to the above traffic volume survey system,
Further, a feature extraction unit for performing predetermined image processing on each image file stored in the image database to extract a feature of the moving object, and a plurality of feature patterns corresponding to attributes of the moving object to be investigated are stored. By providing a feature pattern database and attribute analysis means for comparing the features of the moving object extracted by the feature extraction means with the feature patterns stored in the feature pattern database and estimating the attributes of the moving object, Not only can the amount of traffic of a moving object be grasped, but also its attribute configuration can be grasped, and there is an advantage that valuable data can be collected as basic data for marketing.

[Brief description of the drawings]

FIG. 1 is a conceptual diagram showing an implementation state of a traffic volume survey system according to the present invention.

FIG. 2 is a block diagram showing a functional configuration of the traffic amount inspection system.

FIG. 3 is an explanatory diagram showing a screen display example of a personal computer at the time of conducting a survey.

FIG. 4 is an explanatory diagram showing a shared area L of the camera 14l.

FIG. 5 is an explanatory diagram showing a sharing area K of a camera 14k.

FIG. 6 is an explanatory diagram showing a shared area I of a camera 14i.

FIG. 7 is an explanatory diagram showing a shared area H of the camera 14h.

FIG. 8 is an explanatory diagram showing a shared area C of the camera 14c.

FIG. 9 is a flowchart showing a processing procedure when checking a traffic volume.

FIG. 10 is a conceptual diagram showing an example in which a traveling direction restricting means for restricting each person from traveling in a lateral direction is provided in a passage.

[Explanation of symbols]

 10 Traffic volume survey system 14a-14p Camera 16 Personal computer 18 Moving object detection means 20 Survey target setting means 22 Survey target storage means 24 Survey target confirmation means 26 Image capture means 28 Image registration means 32 Registered image counting means 34 Feature extraction means 36 Attributes Analysis means 38 Feature pattern database 44 Travel direction regulation means

 ──────────────────────────────────────────────────続 き Continued on front page F term (reference) 3E038 AA01 AA07 BA11 CA03 CA07 CB04 CC01 EA02 GA01 JB02 JB10 5B057 AA16 AA19 CC03 DA06 DA13 DB02 DB06 DC03 DC16 DC25 DC33 DC36 5H180 AA01 AA21 CC04 DD02 EE07 5L0952 BA02 FA02 HA03 HA08 JA11

Claims (4)

[Claims] 1. A system for automatically measuring a traffic amount of a specific moving object in a predetermined observation area, wherein imaging means for inputting an image in the observation area and characteristics of the moving object to be investigated are stored. Investigation target storage means, when a moving object appears in the video input from the imaging means, a moving object detection means for detecting this, and referring to the investigation target storage means, the detected moving object is determined as the investigation target. An investigation target confirming means for determining whether or not they match, an image capturing means for capturing a still image of the moving object when the moving object matches the investigation object, and storing the image in an image database in a predetermined file format. A traffic amount survey system comprising: image registration means for storing; and counting means for counting the image files stored in the image database to calculate the traffic amount. 2. The imaging device according to claim 1, wherein a plurality of the imaging units are arranged on a front side or a back side in a traveling direction of the moving body, and the observation area has a predetermined width set based on a width of the moving body to be investigated. Each of the imaging means is assigned in advance a region in charge of each of the plurality of areas, and the moving object detection means is a shared area of each imaging means. In the detection of the appearance of the moving object, the investigation target confirmation means determines whether or not the moving object matches the investigation target for each assigned area of each imaging means, and the image capturing means The traffic amount survey system according to claim 1, wherein a still image of the moving object is captured in units of the sharing area. 3. A plurality of said image pickup means are arranged on a front side or a rear side in a traveling direction of said moving object, and said observation area is a traveling direction regulating means arranged along a traveling direction of each moving object to be investigated. Each area is provided with a predetermined width set on the basis of the width of the moving object, and each of the imaging means includes a plurality of areas in the plurality of areas. Are assigned in advance, the moving object detecting means detects the appearance of the moving object in the shared area of each imaging means, and the investigation target confirming means checks the moving object for each assigned area of each imaging means. The traffic amount survey system according to claim 1, wherein it is determined whether or not the traffic coincides with a survey target, and the image capturing unit captures a still image of the moving object in units of a shared area of each image capturing unit. . 4. A feature extracting means for subjecting each image file stored in the image database to a predetermined image processing to extract the feature of a photographed moving object, and a feature extracting means corresponding to the attribute of the moving object to be investigated. A feature pattern database storing a plurality of feature patterns, a feature of the moving object extracted by the feature extracting unit is compared with a feature pattern stored in the feature pattern database, and an attribute of the moving object is estimated. The traffic analysis system according to any one of claims 1 to 3, further comprising: attribute analysis means.