JPH06187449A - Moving object tracking device - Google Patents

Abstract

PURPOSE:To track an object moving in a wide space by moving an image pickup part and to secure the correspondence between the object to be tracked in the viewfield of the unmoved image pickup part and the object to be tracked in the viewfield of the moving image pickup part. CONSTITUTION:An image pickup operation is started by the instruction of a control part 7, and a driving part 6 is instructed to stand still. The part 7 also instructs an animation image memory part 2 to store the animation images sent from an image pickup part 1. When a fixed time passed, the part 7 instructs the part 2 to stop the fetching of the animation images. Then a motion vector extracting part 3 extracts a moving area by the instruction of the part 7 based on the animation images stored in the part 2 and then calculates the motion vector of the moving area. When the processing of the part 3 is over, the part 7 actuates a moving object extracting part 4 to extract the moving object as a set of moving areas. Then a single moving object to be tracked is decided out of those extracted moving objects by an external instruction or based on the feature of the moving object.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a moving body tracking device for tracking a moving body using an image picked up by a TV camera or the like. With this device, it is possible to monitor the moving body and analyze the behavior.

[0002]

2. Description of the Related Art Conventionally, as a moving object tracking device, various devices having a fixed visual field of an image pickup unit (eg, Ueda et al .: Contour tracking method based on elastic contour model and energy minimization principle, IEICE Transactions, Vol. .J75-D-II, pp.111-120,19
92) has been proposed.

[0003]

The conventional apparatus having a fixed field of view of the image pickup section has a problem that the tracking space is narrow and sufficient tracking cannot be performed.

[0004]

In order to solve the above-mentioned problems, the present invention is an apparatus for tracking a moving object using a picked-up image, and an image pickup section having a predetermined field of view such as a TV camera, and the image pickup thereof. A means for storing an image obtained by the means, a means for calculating a motion vector from the stored moving image, a means for extracting a region of a moving object from the calculated motion vector, and the extracted moving object and the previous one. A means for associating the moving object extracted in the field of view and a means for moving the imaging means according to the motion vector of the associated moving object.

[0005]

According to the moving body tracking device of the present invention, a moving vector is calculated using a moving image obtained from the image pickup means, the moving body is extracted based on the moving vector, and the moving image pickup means is moved based on the moving vector of the focused moving body. Since the visual field is moved by driving and the moving object in the moved visual field is associated with the moving object in the visual field before being moved, the visual field can be moved following the movement of the moving object. Therefore, it becomes possible to track an animal body in a wide space.

[0006]

Embodiments of the present invention will now be described in detail with reference to the drawings. FIG. 1 is a diagram showing the configuration of a moving body tracking device according to an embodiment of the present invention. In the figure, 1 is an image pickup unit, 2 is a moving image memory unit, 3 is a motion vector extraction unit, 4 is a moving object extraction unit, 5 is a moving object visual field correspondence unit, 6 is a driving unit, and 7 is a control unit. .

The image pickup unit 1 is an image pickup device such as a TV camera, and converts an image in the field of view of the camera into a digital image and sends it to the moving image memory unit 2. The moving image memory unit 2 includes the image pickup unit 1
It is a storage means for accumulating a series of images (moving images) continuously photographed.

The motion vector extraction unit 3 calculates a motion vector using the moving image stored in the moving image memory unit 2. Extraction of a motion vector is expressed by I (x, y, t) = I (x, y, t) = I for a certain short time dt when the density value at a pixel (x, y) in a moving image is expressed as I (x, y, t) A motion vector (dx / dt, dy / dt) is obtained by searching for a pixel (x + dt, y + dt) satisfying the relationship of (x + dx, y + dy, t + dt).
Is to calculate. Although many methods have been proposed as motion vector extraction methods, here, a region of pixels whose density value has changed due to the motion of an object (called a motion region) is extracted, and from the temporal displacement of the center of gravity of the motion region, A method of extracting a motion vector for each moving area is used. For this specific method, see “S.Suzuki and J.Sklansky: Extracting
and tracking nonrigid moving objects by temporal e
dges, Image Recognition and Understanding Symposium, pp.I-121 to I-12
Since it is known by "8,1992", a more detailed explanation is omitted here.

The moving body extraction unit 4 is a motion vector extraction unit 3
The motion area obtained by
Based on features such as size when integrated, moving objects are extracted as a set of moving regions. This specific method is also described in “S.Suzuki and J.Sklansky: Extrac
ting and tracking nonrigid moving objects by tempo
ral edges, Image Recognition and Understanding Symposium, pp.I-121
~ I-128, 1992 ", detailed description thereof is omitted here.

The moving body visual field correspondence unit 5 correlates the moving body extracted by the moving body extracting unit 4 with the moving body extracted in the previous visual field. This association method will be described when explaining the operation of the entire apparatus.

The drive unit 6 moves the image pickup unit 1 in parallel according to an instruction from the control unit 7, and can be realized by a robot arm, for example. The control unit 7 is a unit that controls the entire apparatus so as to track the target moving object.

These operations are as follows. First, imaging is started according to an instruction from the control unit 7. Next, the control unit 7 instructs the drive unit 6 to stop. Then, the moving image memory unit 2 is instructed to store the moving image sent from the image pickup unit 1. After a certain period of time, the control unit 7 controls the moving image memory unit 2
Instruct to stop the acquisition of moving images. Next, according to an instruction from the control unit 7, the motion vector extraction unit 3 uses the moving image stored in the moving image memory unit 2 to extract a moving region and calculate the motion vector thereof. Further, after the processing of the motion vector extraction unit 3 is completed, the control unit 7 operates the moving body extraction unit 4 to extract the moving body as a set of moving regions. From the extracted moving bodies, one moving body to be tracked is determined according to an external instruction or the characteristics of the moving body.

FIG. 2 shows an example of a moving body to be extracted. That is, as shown in FIG. 2 (a), if the moving body 10 is a rectangle whose field of view moves from left to right, the moving body 1
0 is extracted as a set of two moving areas 11 and 12, as shown in FIG.

Maximum area in the moving area of the moving object
The coordinates of the center of gravity of the moving region of (x₀, Y ₀). The movement
X-direction component per unit time of the vector₀,unit
V is the y-direction component per unit time₀Express. The control unit 7
The imaging unit 1 is attached to the moving unit 6 in the x direction.₀× t₀, V in y direction₀
× t₀Just instruct to move. t₀Is a moving image memo
After the moving image capturing unit 2 captures the moving image, the movement of the image capturing unit 1
Is the estimated time to complete.

Next, the control unit 7 instructs the drive unit 6 to stop. Then, the moving image memory unit 2 is instructed to store the moving image sent from the image pickup unit 1. After a certain period of time, the control unit 7 instructs the moving image memory unit 2 to stop capturing the moving image. Next, according to an instruction from the control unit 7, the motion vector extraction unit 3 uses the moving image stored in the moving image memory unit 2 to extract a moving region and calculate a motion vector. Further, after the processing of the motion vector extraction unit 3 is completed, the control unit 7 operates the moving object extraction unit 4 to extract the moving objects as a set of moving regions.

Next, according to an instruction from the control unit 7, the visual field of the moving object
The inter-correspondence unit 5 operates in the rectangular range shown in FIG.
In the area, the motion characteristics and the image density characteristics are similar, and their center of gravity
(X ₁, Y₁) To (x₀, Y₀Is closest to
The animal body including the animal is the animal body to be tracked.

In FIG.
FIG. 3 is a diagram showing a search range of a moving area, which is a search range of a moving object.
Is the seat of the center of gravity of the moving area of the moving body in the previous visual field.
Mark (x₀, Y₀) And the previous view to the current view
Amount of movement u₀× t₀And v₀× t ₀Shows that it is limited by
You Where α and β are the ranges for allowing the error
It is a constant and is given in advance. Of the moving body
The unit of the motion vector of the largest moving area in the moving area
U component in the x direction per time₁, Y per unit time
V component₁Express. The control unit 7 includes the drive unit 6 and the imaging unit 1
U in the x direction₁× t₁, V in y direction₁× t₁Just move
To instruct. t₁Is the moving picture memory unit 2
From capturing the image until the movement of the image pickup unit 1 is completed
Estimated time.

Thereafter, the processing is repeated in the same manner until the control section 7 gives an instruction to finish. Although the present invention has been specifically described based on the embodiments, the present invention is not limited to the above embodiments, and it goes without saying that various modifications can be made without departing from the gist of the present invention. .

[0019]

As described above, according to the present invention,
An object moving in a wide space can be tracked by moving the image pickup unit and associating the tracking moving object in the field of view before moving with the tracking moving object in the field of view after moving.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a moving body tracking device according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating a moving object extracted as a set of moving regions.

FIG. 3 is a diagram showing a search range of a moving region in a moving object visual field correspondence unit.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Imaging part 2 Moving image memory part 3 Motion vector extraction part 4 Moving body extraction part 5 Moving object visual field matching part 6 Driving part 7 Control part

Claims (1)

[Claims] 1. A moving object tracking device for tracking an moving object using a picked-up image, an image pickup means having a predetermined visual field, a means for storing an image obtained by the image pickup means, and a stored moving image. And a means for calculating a motion vector from the calculated motion vector, a means for extracting a region of the moving object from the calculated motion vector, and a means for associating the extracted moving object with the moving object extracted in the previous visual field. And a means for moving the image pickup means according to a motion vector of the moving body.