JP2019068339A - Image processing apparatus, image processing method, and program - Google Patents

Abstract

To provide an image processing apparatus capable of appropriately counting the number of objects within a target range even in an imaging range wider than an angle of view.SOLUTION: An image processing apparatus includes: detection means for detecting an object from an image captured by imaging means; detection means for detecting that the object detected by the detection means has passed a detection line set within an angle of view; counting means for counting the number of objects detected by the detection means that have passed through the detection line among the objects detected by the detection means; and control means for controlling the imaging means so as to move the detection line within a target range in which the number of objects should be counted.SELECTED DRAWING: Figure 6

Description

  The present invention relates to an image processing apparatus, an image processing method, and a program.

Conventionally, there is known a method of counting the number of people in an image by analyzing an image captured by a surveillance camera.
For example, Patent Document 1 discloses a technique for detecting a human body in a captured image by image analysis. Specifically, in the technique of Patent Document 1, an edge background difference image is generated, an edge of only a moving object is detected, an interframe difference image between temporally consecutive images is generated, and an edge background difference image is generated. Labeling is performed by superposing inter-frame difference images. Furthermore, according to the labeling result, edge information of only the moving object is tracked to generate an outline image of the moving object, and the number of people is counted based on the outline image.
Moreover, in patent document 2, the image processing of the image | video from the camera for surveillance is carried out, and the technique which counts the number of persons who passed specific places, such as the entrance to each residence area, is disclosed. Specifically, in the technique of Patent Document 2, the detected human body is tracked, the difference between the number of authenticated people and the number of people who actually passed the specific location is detected, and it is confirmed whether co-delivery is occurring. Is going. Then, when an unauthenticated person who is not a resident who is recognized as a suspicious person occurs, a real time image is automatically transmitted to a previously registered distribution destination together with the image taken by the monitoring camera.

JP, 2015-70359, A JP 2008-217205 A

  However, in the prior art, the range for counting the number of people is limited within the imaging angle of view of the monitoring camera. For this reason, when counting the number of people for a wider range, it is necessary to divide the range to be counted into a range that falls within the angle of view and to capture the number, and add the number of people counted by image analysis for each image. In this way, when the imaging range is divided and imaged, the image of the person is cut off halfway along the divided boundary portion, or the person moves while moving the angle of view, so double counting or count omission occurs. Was likely to occur.

  The present invention has been made to solve the above problems, and an image processing apparatus capable of appropriately counting the number of objects within a target range even if the target range is wider than the angle of view. The purpose is to provide a processing method and program.

  In order to solve the above problems, according to an aspect of the present invention, a detection unit that detects an object from an image captured by an imaging unit, and the object detected by the detection unit are set within an angle of view. Detection means for detecting that the detection line has passed, counting means for counting the number of objects of the objects detected by the detection means that have been detected to have passed the detection line by the detection means; And a control unit configured to control the imaging unit so that the detection line moves in a target range in which the number of objects is to be counted.

  According to the present invention having the above configuration, the number of objects within the target range can be appropriately counted even if the target range is wider than the angle of view.

Block diagram showing the configuration of the network camera system according to the first embodiment Block diagram showing an example of hardware configuration of an information processing apparatus Block diagram showing an example of functional configuration of network camera Block diagram showing an example of functional configuration of information processing apparatus Block diagram showing a detailed configuration example of a camera management unit and an image analysis unit A flowchart showing an example of a processing procedure of the people counting process according to the first embodiment A figure explaining processing which determines operation of a network camera concerning Embodiment 1. A figure explaining processing which determines operation of a network camera concerning Embodiment 2. A figure explaining change processing of shape of a passage detection line concerning Embodiment 3 A figure showing an example of an image concerning a modification

  Hereinafter, an embodiment for carrying out the present invention will be described in detail with reference to the attached drawings. The embodiment described below is an example as a realization means of the present invention, and should be appropriately corrected or changed according to the configuration of the apparatus to which the present invention is applied and various conditions. It is not necessarily limited to the embodiment. Moreover, not all combinations of the features described in the present embodiment are essential to the solution means of the present invention. In addition, about the same structure, the same code | symbol is attached | subjected and demonstrated.

First Embodiment
A configuration example of a network camera system according to the present embodiment will be described with reference to FIG. In FIG. 1, a network camera 2 and an information processing apparatus 3 are connected via a network 1. A plurality of network cameras 2 and a plurality of information processing devices 3 may exist.
The network camera 2 may be, for example, a surveillance camera installed in a theater, a cinema, a stadium or the like. The network camera 2 captures an image of a subject according to control from the information processing device 3, and delivers the captured image to the information processing device 3 via the network 1. In the present embodiment, the network camera 2 can move the imaging angle of view with respect to the measurement target range where the number of people should be counted, and can capture an object within the measurement target range. An image captured by the network camera 2 includes a still image and a moving image.

The information processing device 3 controls the network camera 2 and controls the image quality and the like of the image (moving image) from the network camera 2 and analyzes the image distributed from the network camera 2. The information processing device 3 can be configured by, for example, a PC (Personal Computer).
The network 1 can be configured by, for example, the Internet, a wired LAN (Local Area Network), a wireless LAN, a WAN (Wide Area Network), an analog cable, or the like. The network 1 only needs to be able to communicate between the network camera 2 and the information processing apparatus 3, and the communication standard, scale, and configuration thereof are not limited.

(Hardware configuration)
The functions of the network camera 2 or the information processing apparatus 3 in FIG. 1 may be realized by the CPU 11 executing a program stored in a storage unit such as the ROM 12 described later. Note that at least a part of the functions of the network camera 2 or the information processing device 3 may be realized by hardware. In the case of hardware implementation, for example, by using a predetermined compiler, a dedicated circuit may be automatically generated on the FPGA from a program for realizing each step. FPGA is an abbreviation for Field Programmable Gate Array. Further, a Gate Array circuit may be formed in the same manner as an FPGA and realized as hardware. Also, it may be realized by an application specific integrated circuit (ASIC).

FIG. 2 shows an example of the hardware configuration of the information processing device 3. The information processing device 3 includes a CPU 11, a ROM 12, a RAM 13, an external memory 14, an input unit 15, an output unit 16, and a communication interface (I / F) 17. The CPU 11 to the communication I / F 17 are connected by a bus 18.
The CPU 11 controls each part constituting the network camera 2 or the information processing device 3 in accordance with the input signal or program. Specifically, the CPU 11 performs processing such as image analysis processing. Functional blocks in FIG. 3 and FIG. 4 described later illustrate functions executed by the CPU 11.
The RAM 13 stores temporary data and is used for work of the CPU 11. The ROM 12 stores programs for executing the functional units of the network camera 2 or the information processing apparatus 3 and various setting information. The external memory 14 is, for example, a removable memory card, and can be attached to a PC (personal computer) or the like to read out data.
Further, a predetermined area of the RAM 13 or the external memory 14 is used as a buffer of input / output data, temporary storage of data in processing executed by the CPU 11, or the like.
The input unit 15 includes, for example, a device such as a keyboard and a mouse for inputting an operation from the user. The output unit 16 includes, for example, a device such as a monitor that displays an image.
The communication I / F 17 performs communication via the above-described network 1.
Note that the network camera 2 may also have the hardware configuration shown in FIG. 2, and in that case, the network camera 2 further includes an imaging unit that images a subject.

Hereinafter, configurations of the network camera 2 and the information processing device 3 for counting the number of human bodies (people) present in an image will be described. In the present embodiment, an example of counting a human body in an image will be described, but in the present embodiment, any other object, for example, an animal or a vehicle can be counted, and an object to be counted is a moving object. And stationary.
FIG. 3 is a block diagram showing a functional configuration of the network camera 2. The network camera 2 includes an imaging unit 21, an imaging control unit 22, a processing unit 23, a communication unit 24, and a storage unit 25.
The imaging unit 21 images a subject and generates an image signal. The imaging unit 21 includes an imaging element such as a CMOS (Complementary MOS) image sensor. In addition, the imaging unit 21 includes a lens having a mechanism such as zoom, focus, and aperture, a pan head for changing an imaging direction by a pan / tilt mechanism, and a mechanism for image processing for changing luminance and color tone. Have.
The imaging control unit 22 controls zoom, focus, pan / tilt, and the like of the imaging unit 21 in accordance with a control command from the processing unit 23. Thereby, the object angle can be imaged by moving the angle of view of the imaging unit 21 with respect to the measurement target range. In the present embodiment, the imaging control unit 22 moves the angle of view of the network camera 2 by controlling operations such as pan / tilt, zoom and focus of the imaging unit 21. In each embodiment, the “angle of view” means a range (imaging range) to be imaged by the imaging unit 21.

The processing unit 23 receives a request command from the outside through the communication unit 24 described later, analyzes the received request command, and executes the request command. For example, if the request command is a command related to control of the network camera 2, the processing unit 23 converts the request command into a control command and sends the control command to the imaging control unit 22. The imaging control unit 22 executes a control command from the processing unit 23. If the request command is a command related to the inquiry about the setting state of the network camera 2, the processing unit 23 acquires the state of the corresponding setting from the storage unit 25 described later, and converts the response to the request into a response format. It transmits via the communication unit 24.
The communication unit 24 communicates with other devices. The communication unit 24 is connected to the network 1 and functions as an interface with the network 1.
The storage unit 25 stores model information, a name, a setting state, and the like of the network camera 2.

FIG. 4 is a block diagram showing a functional configuration of the information processing device 3. The information processing device 3 includes a communication unit 31, a camera management unit 32, an image analysis unit 33, a primary storage unit 34, a secondary storage unit 35, an input unit 36, a display unit 37, and a processing unit 38. And.
The communication unit 31 communicates with the network camera 2. The communication unit 31 is connected to the network 1 and functions as an interface with the network 1.
The camera management unit 32 acquires the state of the setting from the network camera 2 or receives the notification of the change of the state from the network camera 2. The camera management unit 32 grasps and manages the states of the camera platform and the imaging mechanism of the network camera 2 based on the acquired state or the received notification. In addition, the camera management unit 32 generates a request command for changing the camera platform of the network camera 2 or the imaging mechanism to a desired state or a request command for acquiring an image from the network camera 2, via the communication unit 31. To send.
The image analysis unit 33 acquires an image from the network camera 2, performs image analysis on the acquired image, detects an object (human body) to be measured in the image, and counts the detected human body. Do.

The primary storage unit 34 is a RAM (Random Access Memory) or the like that temporarily stores an image analysis program and data, and temporarily stores information of analysis processing executed by the image analysis unit 33.
The secondary storage unit 35 is a hard disk, an optical disk, a memory card, or the like, and stores a program for executing image analysis. Further, the secondary storage unit stores information such as the number of people output as a result of the image analysis.
The input unit 36 is a device such as a pointing device such as a mouse or a keyboard operated by the user for input.
The display unit 37 is configured by a display such as a CRT (Cathode Ray Tube) or an LCD (Liquid Crystal Display), and displays a GUI (Graphical User Interface) of an image analysis program. The display unit 37 also displays information according to the result of the image analysis.

The processing unit 38 expands the image analysis program read from the secondary storage unit 35 in the primary storage unit 34 and executes the program, and the instruction input to the input unit 36 or the state of the camera obtained from the camera management unit 32. The network camera 2 is controlled based on the camera management unit 32. In addition, the processing unit 38 supplies the image from the network camera 2 acquired from the camera management unit 32 to the image analysis unit 33 to execute the image analysis. Furthermore, the processing unit 38 causes the display unit 37 to display the result of the image analysis obtained from the image analysis unit 33 and stores the result in the secondary storage unit 35.
The processing unit 38 can be configured as a processor such as a central processing unit (CPU), for example. The processing unit 38, the camera management unit 32, and the image analysis unit 33 can be combined to form a single processor. Alternatively, the processing unit 38, the camera management unit 32, and the image analysis unit 33 may be configured by a plurality of CPUs, or may be configured to have a GPU (Graphics Processing Unit). Further, the processing unit 38 may execute the processing of the camera management unit 32 and the image analysis unit 33.

(Detailed Configuration of Camera Management Unit 32 and Image Analysis Unit 33)
FIG. 5 is a block diagram showing detailed functional configurations of the camera management unit 32 and the image analysis unit 33. As shown in FIG. The camera management unit 32 includes a range setting unit 41, an operation determination unit 42, and a camera control unit 43. The image analysis unit 33 further includes a line setting unit 44, a human body detection unit 45, a passage determination unit 46, and a number-of-people calculation unit 47.
Exchange of information between the camera management unit 32 and the image analysis unit 33 is performed via the processing unit 38.
The range setting unit 41 causes the display unit 37 to display a setting screen for setting a measurement target range (or a target range) which is a range to which the number of people is counted. The range setting unit 41 receives the range designated by the user via the input unit 36 on the setting screen, and stores the range in the primary storage unit 34. The range in which the number of persons is to be counted can be specified by, for example, the rotational position of pan and tilt of the network camera. For example, if two or more combinations of pan and tilt positions are set, the area surrounded by these positions can be set as a measurement target range in which imaging is performed by moving the angle of view of the imaging unit 21.

The operation determining unit 42 determines how to operate the network camera 2 with respect to the measurement target range set by the range setting unit 41. The operation determining unit 42 determines pan, tilt and zoom positions at the start of shooting, pan / tilt speed and direction during zoom, direction and zoom position, and pan and tilt positions at the end of shooting.
According to the imaging operation of the network camera 2 determined by the operation determination unit 42, the camera control unit 43 issues a command to control the network camera 2 in accordance with the progress of imaging while acquiring the state of the camera. Also, the camera control unit 43 acquires an image from the network camera 2 and stores the image in the primary storage unit 34.
The line setting unit 44 causes the display unit 37 to display a setting screen for setting a passage detection line (detection line) for counting the number of people who have passed within the angle of view of the image. The line setting unit 44 stores, in the primary storage unit 34, information indicating the passage detection line designated by the user via the input unit 36 according to the setting screen. The line setting unit 44 can set the position and length of the passage detection line, and the direction of passage to be detected. Further, the passage detection line may be not only a straight line but also a curved line. Also, a plurality of passage detection lines may be set.

The human body detection unit 45 reads an image acquired by the camera control unit 43 from the camera and stored in the primary storage unit 34, performs image processing, and detects a human body shown in the image. The human body detection unit 45 detects the number of human bodies in the image and the coordinates and size of each human body. The human body detection unit 45 can identify the same human body when detecting the human body continuously from a plurality of frame images in a moving image. That is, the human body detection unit 45 can detect and track the same person from the moving image.
The human body detection unit 45 does not necessarily have to detect a human body from the entire image. That is, the human body may be detected only in the vicinity of the passage detection line, as long as the passage determination unit 46 described later can determine the passage of the human body. Thereby, the processing load of the human body detection unit 45 can be reduced.
The passage determination unit 46 determines whether or not the human body detected by the human body detection unit 45 has passed through the passage detection line set by the line setting unit 44, and when the human body passes, the number calculation unit 47 described later. Notice.
The number-of-people calculating unit 47 counts the number of people who have passed the passage detection line by measuring the number of times of notification from the passage determining unit 46.

(Processing procedure of people count process)
FIG. 6 is a flowchart showing a series of processing (number counting process) executed by the processing unit 38, the camera management unit 32, and the image analysis unit 33.
First, in S1, the processing unit 38 resets the number of people counted by the number-of-people calculation unit 47 to zero. In S2, the processing unit 38 moves the pan / tilt of the network camera 2 to the imaging start position determined by the operation determination unit 42. In S3, the processing unit 38 moves at least one of pan and tilt of the network camera 2 at the speed determined by the operation determination unit 42.

In S4, the passage determination unit 46 determines whether the human body detected by the human body detection unit 45 has passed the passage detection line (detection line). In S5, when it is determined that the human body has passed through the passage detection line in S4, the number-of-people calculating unit 47 counts the number of people passing through. On the other hand, if it is determined in S4 that the human body has not passed the passage detection line, the process proceeds to S6 without performing the process of S5, and the number calculation unit 47 does not count the number of people.
In S6, the processing unit 38 determines whether the pan and tilt of the network camera 2 have reached the imaging end position determined by the operation determination unit 42. If the pan and tilt of the network camera 2 have not reached the imaging end position, the processing unit 38 repeats the processing of S4 and subsequent steps. If the pan / tilt of the network camera 2 has reached the imaging end process, the pan / tilt operation of the camera is stopped in S7, and the process is ended.

(Details of camera operation determination process)
Hereinafter, an example of processing in which the operation determining unit 42 determines the operation of the network camera 2 will be described with reference to FIGS. 7A to 7C. It is assumed that the target range for counting the number of people is set as shown by 61 in FIG. 7A by the range setting unit 41. In order to count the number of persons within the target range 61, a path for moving the imaging angle of view of the network camera 2 may be set such that the passage detection line passes through the entire target range 61. Therefore, the range setting unit 41 can set an area wider than the imaging angle of view of the imaging unit 21 as the measurement target range.

Further, the line setting unit 44 vertically draws the passage detection line from the upper end to the lower end of the angle of view of the network camera 2 to the left and right centers, for example. That is, the line setting unit 44 sets the passage detection line in the direction orthogonal to the direction in which the imaging angle of view of the imaging unit 21 moves.
Then, the operation determination unit 42 determines the imaging start position so that the left end of the passage detection line matches the left end of the target range 61, and the lower end of the passage detection line matches the lower end of the target range 61. As a result, as shown in FIG. 7B, the imaging angle of view indicated by the rectangular area at the imaging start position is 62, and the passage detection line is 63.

  Next, the operation determination unit 42 determines to move the pan of the network camera 2 to a position 64 where the passage detection line and the right end of the target range 61 coincide with each other. Thereafter, as shown in FIG. 7C, the operation determining unit 42 determines to move the imaging angle of view to the position 65 in the upward direction by the vertical angle of view. Then, the operation determination unit 42 again determines that only the pan is moved to the position 66 where the passage detection line coincides with the left end of the passage detection line 61. As described above, by repeating the movement of the pan and the movement of the tilt, the number of people in the entire target range 61 can be seamlessly counted.

(Effect of this embodiment)
As described above, in the present embodiment, the measurement target range is imaged while moving the imaging angle of view of the imaging unit with respect to the measurement target range. Then, in the present embodiment, an object to be measured is detected from an image from the imaging unit, and whether the detected object has passed a passage detection line set according to the measurement target range and the imaging angle of view of the imaging unit. And count the number of times the passing detection line has been crossed. That is, control is performed to move the imaging angle of view of the imaging unit so that the passage detection line set in the measurement target range moves. Thereby, double counting and count leakage can be reduced. That is, even if it is an imaging range (measurement object range) wider than the angle of view of an imaging part, the object of a measurement object can be counted appropriately.

The operation determining unit 42 and the line setting unit 44 automatically set the imaging operation of the imaging unit and the passing knowledge line in accordance with the measurement target range set by the range setting unit 41.
When the user designates the pan and tilt coordinates of the network camera 2 in the moving order and designates the passage detection line, the range setting unit 41 may not set the measurement target range. Further, the line setting unit 44 can automatically set the passage detection line in, for example, a direction orthogonal to the moving direction of the camera determined by the operation determining unit 42.
Further, the line setting unit 44 can set the passage detection line so as to count only the human body passing in the opposite direction with respect to the movement direction of the network camera 2. The line setting unit 44 may set the passage detection line so as not to detect passage while the network camera 2 is moving in parallel to the passage detection line. This configuration can reduce count errors.

In the description of FIG. 7 described above, the passage detection line is set in the direction perpendicular to the angle of view, and the pan of the network camera 2 is moved to count the number of persons. However, when the horizontal angle of view of the camera is longer than the vertical angle of view, setting the passage detection line in the horizontal direction with respect to the angle of view increases the number of people that can be counted at one time because the passage detection line becomes longer. . Depending on the arrangement of people to be counted, it may be possible to count with a smaller movement amount by setting the passage detection line in the horizontal direction with respect to the angle of view and moving the tilt of the network camera 2.
Further, in the present embodiment, while the pan and tilt of the network camera 2 are moved, the number of people passing through the passage detection line is counted. It may be a method. For example, a method such as zooming or rotation may be used, or imaging may be performed by attaching the network camera 2 to a moving object such as a car or a drone. Further, the same processing as described above may be performed on an image cut out while moving the cutout position from a large resolution or wide angle (wide area) image.
Further, by fixing the passage detection line with respect to the angle of view of the imaging unit 21, the processing load can be reduced.

Second Embodiment
In the first embodiment, the number of people detected exceeds the passage detection line is measured, but in the present embodiment, the number of people who have passed the passage detection line and the number of people counted by measuring the number of human bodies detected The method of combining will be described.
Hereinafter, a method of determining the operation method of the network camera 2 by the operation determination unit 42 in the present embodiment will be described with reference to FIG.
In the first embodiment, the angle of view of the network camera 2 is moved such that the passage detection line passes through the entire target range 61 where the number of people is counted. Then, the number-of-people calculation unit 47 counts only the number of the passage determination unit 46 that has determined that the passage detection line has passed.
In the present embodiment, when counting the number of people in the target range 61, the angle of view captured by the network camera 2 is only the minimum range in the target range 61. The number-of-people calculation unit 47 determines the number of human bodies within the measurement target range as the number determined by the passage determination unit 46 as having passed through the passage detection line and the number detected as a human body by the human body detection unit 45 Add as the number of
For example, the camera control unit 43 sets the initial position of the angle of view of the network camera 2 to 71 so that the left end of the angle of view and the left end of the target range 61 coincide with each other. Then, the camera control unit 43 moves the pan of the network camera 2 to the right, and moves it to a position 72 where the right end of the angle of view coincides with the right end of the target range 61. At this time, while the pan is moving, the number of human bodies passing through the passage detection line is counted as the number of people. In addition, the number of human bodies detected on the left side of the passage detection line when the angle of view is 71 and the number of human bodies detected on the right side of the passage detection line when the angle of view is 72 Do.

In the present embodiment, by combining the count of the number of people who have passed the passage detection line and the count of the number of people by measuring the number of detected human bodies, the moving amounts of pan and tilt of the network camera 2 are reduced compared to the first embodiment. Can count the number of people in the entire target range.
It is also effective when it is desired to count the number of persons in the image up to the limit position of the pan and tilt operating range of the network camera 2.

Embodiment 3
In the first and second embodiments, the passage detection line set by the line setting unit 44 is constant with respect to the angle of view of the image. However, in the present embodiment, the line setting unit 44 controls the pan and tilt of the network camera 2. The position and the shape of the passage detection line are changed according to the position.
Hereinafter, a specific example of the process in the present embodiment will be described with reference to FIG.
As shown in FIG. 9, consider a scene where there is a matrix in the direction of depth with respect to the angle of view of the camera, and the matrix is folded back and continued. When imaging is performed from the front or the side of the matrix, a person may overlap and a detection omission is likely to occur. Therefore, imaging may be performed from diagonally above the matrix. In such a case, pan and tilt of the camera are moved along the matrix as in the position 82 from the position 81 in FIG. 9, and the passage detection line can be efficiently counted by setting it in the direction orthogonal to the matrix. However, if there is a depth in the matrix, the width of the matrix on the image is different between the near side and the far side, so if you set the passage detection line with the width of the near side matrix, the human body of the next column folded back is also It may be counted incorrectly. Therefore, in this embodiment, when the angle of view of the network camera 2 is on the front side of the matrix, a long passage detection line is set as 81a, and when the angle of view of the network camera is on the back of the matrix, as 82a. Set a short passage detection line to That is, the line setting unit 44 changes the position or the shape of the passage detection line according to the movement of the angle of view of the imaging unit 21. Specifically, the length of the passage detection line is set according to, for example, the distance to the subject. By counting the number of people according to the passage detection line set in this way, the number of people can be counted for each row, so the influence of the movement of the row and the double count can be reduced.

In the present embodiment, by changing the position or the shape of the passage detection line according to the movement of the angle of view of the imaging unit 21, it is possible to accurately count the number of persons in the image.
In the above description, the length of the passage detection line is changed depending on the position of the pan and tilt of the network camera 2. However, not only the length of the passage detection line but also the position and the shape may be changed. In addition to the pan and tilt of the network camera 2, the position and the shape of the passage detection line may be changed according to the zoom.

Fourth Embodiment
In the present embodiment, the operation determination unit 42 automatically sets the moving speed and the moving direction of the network camera 2 in accordance with the imaging environment.
When the shutter speed of the network camera 2 is slow, if the moving speed of the angle of view by operating the pan, tilt and the like of the network camera 2 is fast, the subject on the image will be greatly blurred. When the blurring of the subject increases, the detection accuracy of the human body detected by the human body detection unit 45 decreases. Therefore, in the present embodiment, the angle of view is moved at a moving speed set in advance for each shutter speed. Here, the moving speed of the angle of view is a speed of moving by one angle of view per unit time, and also differs depending on the zoom position. Specifically, for example, when the shutter speed is 1/1000, movement is performed by one angle of view per second, and when the shutter speed is 1/2000, movement is performed by two angle of view per second. That is, the speed of movement of the angle of view by the imaging unit 21 is changed according to the shutter speed of the imaging unit 21.

In this embodiment, by moving the angle of view of the network camera 2 at a moving speed set in advance with respect to the shutter speed, a moving speed suitable for human body detection is realized even when the shutter speed changes. it can.
Note that the moving speed and moving direction such as pan and tilt of the camera can also be changed according to the detected condition of the human body.
For example, when the size of the human body detected by the human body detection unit 45 is smaller than a predetermined threshold or when the number is large, the load of the human body detection process may be large and it may take time. Therefore, it is conceivable to change the frame rate in accordance with the number of human bodies. However, if the frame rate is slow and the moving speed of the angle of view is fast, the moving distance of the human body in the previous and subsequent frames becomes large, and the accuracy of human tracking becomes low. In such a case, the moving speed of the angle of view may be reduced. That is, the speed of movement of the angle of view by the imaging unit 21 is changed according to at least one of the number and the size of the specific object detected by the human body detection unit 45.

  Furthermore, it is conceivable to determine the moving speed of the angle of view by detecting the moving speed of the human body. When tracking and detecting a human body, the amount of movement and the movement direction per unit time within the angle of view can be known from the coordinates in each frame. Therefore, if the direction in which the person is moving is detected and pan / tilt of the network camera is moved in the opposite direction, the number of people can be efficiently counted. If the angle of view must be moved in the same direction as the direction in which the person is moving, the angle of view can be moved faster than the speed at which the person is moving. That is, the moving direction and the moving speed of the person are detected, and the moving speed of the angle of view by the imaging unit 21 is changed according to the moving direction and the moving speed.

<Modification>
In the above-described embodiments, the image angle is obtained by moving the angle of view within the measurement target area by the pan / tilt function of the network camera 2, but the method for moving the angle of view within the measurement target area is limited to this. Absent.
For example, an area 91 cut out in a predetermined size from the omnidirectional image shown in FIG. 10 is sequentially changed in a predetermined direction (circumferential direction) 92 in the omnidirectional image, and sequentially supplied to the human body detection unit 45 as a cutout image. included.
An omnidirectional image obtained by a hemispherical mirror, a fisheye lens or the like is affected by image distortion, aberration, etc. even if a specific region is simply extracted. On the other hand, by performing coordinate conversion, distortion correction, and the like of the cutout image forming the angle of view, the image can be converted into an image of a rectangular area similar to an image of a normal camera.
The detection of the human body is performed on the cutout image on which the distortion correction and the like are sequentially performed while moving the region 91 cut out in a predetermined size from the omnidirectional image in the predetermined direction 92 on the circumference, and then the passage detection line Set By determining whether or not the passage determination unit 46 has passed the passage detection line set in the cutout image, the number of people can be counted.
In this case, the passage detection line may be fixed in the cutout image converted into the rectangular area. Or you may change according to the detected human body etc.
Thus, the accuracy of counting the number of persons included in the image can be improved by performing the same processing as in each embodiment on the image that is omnidirectional image output and distortion-corrected.

  REFERENCE SIGNS LIST 1 network 2 network camera 3 information processor 21 imaging unit 22 imaging control unit 32 camera management unit 33 image analysis unit 38 processing unit 41 range setting unit 44 ... line setting unit, 45 ... human body detection unit, 46 ... passage determination unit, 47 ... number of people calculation unit

Claims (16)

Detection means for detecting an object from an image captured by the imaging means;
Detection means for detecting that the object detected by the detection means has passed a detection line set within the angle of view;
Counting means for counting the number of objects detected by the detecting means that have passed through the detection line among the objects detected by the detecting means;
Control means for controlling the imaging means to move the detection line within a target range in which the number of objects should be counted;
An image processing apparatus comprising: The control means controls the operation of the imaging means to move the detection line in the target range by moving the angle of view of the imaging means.
An image processing apparatus according to claim 1, characterized in that. The detection means detects that, of the objects detected by the detection means, the object that has passed the detection line in the direction opposite to the direction of the operation of the imaging means has passed the detection line.
The image processing apparatus according to claim 1, wherein the image processing apparatus comprises: The detection means sets the detection line in a direction orthogonal to the direction of operation of the imaging means.
The image processing apparatus according to any one of claims 1 to 3, characterized in that: The control means controls the operation of the imaging means in accordance with the position or the shape of the detection line set by the detection means.
The image processing apparatus according to any one of claims 1 to 4, characterized in that: The control means sets an area having a predetermined width with respect to the angle of view of the imaging means as the target range, and controls the operation of the imaging means so that the detection line passes the entire target range. Set the route to
The image processing apparatus according to any one of claims 1 to 5, characterized in that: The counting means counts the number of the objects according to the number of the objects detected as having passed the detection line by the detection means and the number of the objects detected by the detection means.
The image processing apparatus according to any one of claims 1 to 6, characterized in that: The detection unit changes the position or the shape of the detection line according to the operation of the imaging unit.
The image processing apparatus according to any one of claims 1 to 7, characterized in that: The control unit controls the operation of the imaging unit according to at least one of the number and the size of the object detected by the detection unit.
An image processing apparatus according to any one of claims 1 to 8, characterized in that. The control means controls the speed of operation of the imaging means in accordance with the shutter speed of the imaging means.
The image processing apparatus according to any one of claims 1 to 9, characterized in that: The control means controls the operation of the imaging means in accordance with the moving direction of the object detected by the detection means.
The image processing apparatus according to any one of claims 1 to 10, characterized in that: The control means controls the operation of the imaging means in accordance with the moving speed of the object detected by the detection means.
The image processing apparatus according to any one of claims 1 to 11, characterized in that: The imaging means constructs the angle of view by cutting out an image of a predetermined size from an omnidirectional image including the target range,
The control means controls the imaging means to move the detection line in the target range by changing a position at which the imaging means cuts out the image of the predetermined size from the omnidirectional image.
An image processing apparatus according to claim 1, characterized in that. The detection means sets the detection line at a fixed angle with respect to the angle of view.
The image processing apparatus according to claim 13, characterized in that: Detecting an object from an image taken by the imaging means;
Detecting that the detected object has passed a detection line set in an angle of view;
Counting the number of objects detected to have passed through the detection line among the detected objects;
Controlling the imaging means to move the detection line within a target range in which the number of objects should be counted;
An image processing method comprising:   The program for functioning a computer as each means of the image processing apparatus of any one of Claims 1-14.

Images