JP5069442B2 - Human backflow detection system - Google Patents

Description

  The present invention relates to a system for detecting a backflow at a moving place of a person.

In the case of monitoring the movement of a person, a method is conceivable in which a person monitors the image captured by a television camera.
When a person monitors, there is a risk that the tension will continue for a long time and oversight may occur.
In addition, it is conceivable to recognize a person from image data captured by a television camera and to deal with this. In this case, for example, it is conceivable that a shadow on the floor surface of a person is also recognized as a person.

  The problem to be solved is to accurately recognize a person from image data obtained by imaging a human flow, particularly a reverse flow from above.

According to a first aspect of the present invention, there is provided a human backflow detection system for detecting a person based on distance information from image data of an image pickup means provided by a plurality of cameras provided above a person's moving location. Movement of the person detected by the person detecting means , with one side of a virtual boundary line drawn perpendicularly to a predetermined direction with respect to the photographing range of the means and the imaging means as a management area and the other side as a non-management area The tracking is completed when the detection area representing the entire screen is exited, and at that point, the movement of the person is confirmed, and the entrance position is determined depending on whether the person is detected from the controlled area or the unmanaged area. which confirmed, with finalizing the contestants position by either exiting from any area of the management area or unmanaged zone backflow by detecting a person moving in the direction set in advance and judging to have occurred A.

  According to a second aspect of the present invention, there is provided a human backflow detection system according to the first aspect, further comprising a backflow person counting means for counting the number of persons detected by the person detection means and moving in the backflow direction.

  A human backflow detection system according to a third aspect is the one according to the first or second aspect, wherein the human detection means passes through a virtual line on a preset image from a designated direction or is preset. It is characterized in that a reverse flow is determined when moving from a predetermined area on the image to another predetermined area.

According to the human backflow detection system according to claim 1, since a person is detected using a plurality of cameras, a person below a certain distance can be automatically detected without being recognized as a person. When the person detection means tracks the movement of the person and exits the detection area representing the entire screen, the tracking is completed, and at that point, the action of the person is confirmed, and the person is detected from either the controlled area or the unmanaged area. The entrance position is determined depending on whether it is detected, and the U-turn can be determined so that the entry position is determined depending on whether it has exited from the management area or the non-control area, and the movement of the person in the preset direction is detected. Can determine the occurrence of backflow.

  According to the human backflow detection system according to claim 2, in the system according to claim 1, it is possible to correctly count the persons who are detected by the person detection means and move in the backflow direction.

  According to the human backflow detection system according to claim 3, in the system according to claim 1 or 2, the backflow determination in the human detection means passes through a virtual line on the preset image from the specified direction. The determination of the backflow in the person detection means or the person detection means is performed by assuming that the determination is made to move from a predetermined area on the preset image to another predetermined area.

  FIG. 1 is a configuration diagram of one embodiment of a human backflow detection system according to the present invention. In FIG. 1, reference numeral 101 denotes a passage in which a person flows and the direction of passage is determined, and reference numeral 102 denotes an imaging means provided above a boundary 101 a as a predetermined portion of the passage 101, for example, a virtual gate. Cameras are used as a set. 105 detects a person based on distance information from the image data of the imaging means 102, tracks the movement of the detected person, determines the normal direction, and detects a person movement in a preset direction, thereby detecting backflow. A person detecting means 108 for determining that the person has occurred, 108 is a person counting means for counting the persons detected by the person detecting means 105, and 109 is a backflow person counting means for counting the persons detected by the person detecting means 105 and determined to be backflowed. is there. 114 receives the backflow determination of the person detection means 105 and notifies the security guard room or reaches a predetermined count of the person count means 108 and the backflow person count means 109 to an external recording section (not shown) via a communication means (not shown). On the other hand, it is a notification means for notifying the fact and the number.

FIG. 2 is a diagram for explaining the principle of obtaining distance information necessary for the first, second and third person detection means 105, 106 and 107 to detect a person. When a single TV camera is used, the image data cannot be distinguished from a person when the photographed subject is a shadow on the floor or noise. In the present invention, two or more cameras are used, the distance information of an object is obtained based on the principle of triangulation, and a person is detected when the distance indicated by the camera is appropriate as a human when measured from a television camera. It is what you did. As shown in FIG. 2, one point P (X, Y, Z) of a certain object is photographed by two cameras. An image pl (xl, yl) is imaged on the imaging surface of the left camera, and pr (xr, yr) is imaged on the imaging surface of the right camera. The position of the point P (X, Y, Z) is obtained by searching for a corresponding point between the two cameras. At this time, assuming that the focal length of both cameras is F and the distance between the two cameras is B, the coordinate points of the point P can be obtained by the following equations.
X = B (xl + xr) / 2d
Y = B (yl + yr) / 2d
Z = BF / d
Where d = xl−xr
When a plurality of cameras are used, the average value of each calculation result by an appropriate number of pairs selected from the two cameras is obtained as the position.
Each of the first, second, and third person detecting means 105, 106, and 107 measures the distance from the image data as described above, and has a predetermined range having distance information in a predetermined range from the distance. A closed area is detected and the area is determined to be a human image.

When the user passes through the passage 101, the person detected by the person detecting means 105 moves on a straight line and goes out of the shooting range of the camera, that is, the detection area. At this time, as shown in FIG. 3, assuming that the imaging screen is advanced from the left to the right (normal direction (forward direction)), a virtual line drawn, for example, at a right angle to the direction is regarded as a gate. The boundary line 301 is referred to as a management area, and the right side of the boundary line 301 is referred to as a non-management area. The person detection means 105 detects a person, tracks it with the movement of the person, completes the tracking when the person leaves the detection area, and at that point, confirms the action of the person and determines the entrance position and the entry position. Then, when a person whose entry position is a management area and whose entry position is a non-management area is detected, it is detected that the person is in the normal direction.
The imaging unit 102 captures a person passing through the boundary through the passage 101 and outputs the image data to the person detection unit 105. The person detection means 105 analyzes the input image data and detects a predetermined closed region as a person. The one closed area from the person detecting means 105 moves in the normal direction, and a predetermined place on each screen, that is, a predetermined entrance position and an exit position can be determined. The person counting means 108 is notified, and the person counting means 108 adds the number of people to the count value and counts it.

  The person detecting means 105 makes a reverse flow determination when a passing person flows backward through the passage 101. The person detected by the person detecting means 105 moves along the passage 101 and goes out of the shooting range of the camera, that is, the detection area. The person detection means 105 tracks the person and completes the tracking when the detection area representing the entire screen is exited. At that point, the action of the person is determined and the entrance position and the entry position are determined. Then, when a person whose entrance position is an unmanaged area and where the entry position is a managed area is detected or when a person passes through the boundary line 301, a reverse flow is determined (flows 302, 303, and 304 in FIG. 3). The one closed region from the person detecting means 105 moves in the reverse direction and flows backward at a predetermined place on the screen, that is, the boundary line 301 of the process, so that the backward movement is notified to the counter flow person counting means 109, The person counting means 109 counts the number of persons in addition to the count value. The person detection means 105 determines a U-turn when the person movement is an unmanaged area for both the entrance position and the entry position or the entry position and the entry position are the management area (flows 305 and 306 in FIG. 3). At this time, the boundary line 301 in FIG. 3 corresponds to the boundary.

In the first embodiment, a passage assuming a general gate has been described. However, the passage may be applied in place of a person's moving place using a man conveyor such as an escalator or a moving sidewalk. The case of an escalator is shown in FIG. The same imaging means 402, 403, 404 as in the first embodiment are installed at the boarding / exiting parts 401a, 401b and a plurality of appropriate intermediate parts in the escalator 401, and distance information from the image data of the imaging means 402, 403, 404 The person detection means 405, 406, which detects a person based on the movement, tracks the movement of the detected person, and determines that a backflow has occurred by detecting a normal direction and detecting a person movement in a preset direction. 407, receiving a backflow determination of the backflow person counting means 409 and the person detection means 405, 406, 407 detected by the person detection means 405, 406, 407 and notifying the guard room or receiving a backflow person When the predetermined count of the counting means 409 is reached, a notification means 41 for notifying the external recording unit of the fact and the number via a communication means (not shown). It includes a can detect the reverse flow in the same manner as in Example 1.
In the second embodiment, the imaging means 402, 403, and 404 are used at three locations. However, the imaging means 402, 403, and 404 are not limited to this, and may be appropriately installed at one or a plurality of locations depending on the operation location of backflow detection such as an escalator or a moving sidewalk. May be used.

It is a block diagram of one Example of the human backflow detection system which concerns on this invention. Example 1 It is a figure explaining the principle which obtains the distance information required in order to detect a person. It is a figure explaining a normal direction and a backflow. It is a block diagram of one Example of the human backflow detection system which concerns on this invention. (Example 2)

Explanation of symbols

  DESCRIPTION OF SYMBOLS 101 ... Escalator, 102, 103, 104 ... Imaging means, 105, 106, 107 ... Person detection means, 108 ... Person counting means, 108a ... Count determination part, 109, 110, 111 ... Fall determination means, 112 ... Danger range detection Means 114... Notification means.

Claims (3)

An imaging means provided above a person's moving place and comprising a plurality of cameras;
Person detecting means for detecting a person based on distance information from image data of the imaging means;
The movement of the person detected by the person detection means is tracked with one side of a virtual boundary line drawn perpendicularly to a predetermined direction in the imaging range of the imaging means as a management area and the other side as a non-management area. , Tracking is completed when the detection area representing the entire screen is exited, and at that point, the action of the person is confirmed, and the entrance position is determined depending on whether the person is detected from the management area or the non-control area, A human backflow detection system characterized in that a participation position is determined depending on whether a control area or a non-management area has exited , and that a backflow has occurred due to detection of a person movement in a preset direction.   The person backflow detection system according to claim 1, further comprising a backflow person counting means for counting the number of persons detected by the person detection means and moving in the backflow direction.   The person detection means determines that the current flows backward when passing through a virtual line on a preset image from a specified direction or when moving from a predetermined area on the preset image to another predetermined area. The human backflow detection system according to claim 1, wherein the system is a human backflow detection system.

Images