JP3440916B2 - Automatic tracking device, automatic tracking method, and recording medium recording automatic tracking program - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an automatic tracking technique for tracking an object in an image input from a surveillance camera or the like by an image pickup device having a turning function, and particularly to an automatic tracking device that is robust to the movement of the object. The present invention relates to an automatic tracking method and a recording medium having an automatic tracking program recorded therein.

[0002]

2. Description of the Related Art Conventionally, this type of automatic tracking system extracts an object of interest from an image taken by a surveillance camera or the like installed in a turning device, and turns the extracted object so that it is always contained in the image. Tracking is performed by controlling the pan / tilt angle of the device. An example of a conventional automatic tracking system is described in Japanese Patent Laid-Open No. 10-210343. The automatic tracking system described in this publication extracts an object to be tracked from an output image of an imaging device, converts the amount of displacement of the object from the image center into an angular velocity command of a pan / tilt turning device, and extracts the object. It is configured by means for converting a size shift between the size of the object on the image and the set target size into a zoom operation command.

[0003]

However, since the tracking device according to the conventional technique performs control using only the position information of the object on the screen in tracking in pan / tilt turning, tracking of a moving object is performed. There is a problem that it cannot be performed accurately. For example, as shown in FIG. 11, consider a case where an object is located at the edge of the screen and heads toward the center of the screen. With the control parameter set to operate the pan of the camera so that the object at the edge of the screen is captured in the center of the screen, the object is positioned at the edge of the screen as shown in FIG. When moving, if an attempt is made to perform the same control as in the case of FIG. 11, the object cannot be captured in the center of the screen, and if the movement of the object is large, it may deviate from the angle of view. That is, the method of tracking an object only on the position on the screen has a problem that stable tracking cannot be performed depending on the size and direction of movement of the object.

Further, in the zoom control method for making the size of the object on the image constant, the zoom control amount of the stationary object and the moving object are the same, but as shown in FIG. Since the apparent amount of movement on the image increases with zooming in on a moving object, it is easy to deviate from the angle of view for an object with a high moving speed, and tracking of a moving object is also possible when zoom control is used. There is a problem that it cannot be performed stably.

The present invention has been made in view of the above circumstances, and an automatic tracking device capable of performing zoom-up even with a moving object without impairing the stability of tracking and more accurately tracking. An object of the present invention is to provide an automatic tracking method and a recording medium recording an automatic tracking program.

[0006]

According to a first aspect of the present invention, there is provided an automatic tracking device for tracking an object using a captured image, wherein the automatic tracking device needs to track the object. Image input means including a camera for obtaining an image of a region, turning means for turning the image input means in two axial directions based on the input pan / tilt angle control amount, and an image obtained by the image input means Tracking object extraction means for extracting an object to be tracked from, and the position of the tracking point of the tracking target object at different times are converted into positions on absolute coordinates using the pan / tilt angle of the turning means. And position estimation means for estimating the moving position of the object from the position of the tracking point of the past object Means and past things And a turning control means for controlling the pan / tilt angle of the turning means based on the estimated moving position and the moving speed. And

According to a second aspect of the present invention, there is provided an automatic tracking device for tracking an object using a picked-up image, wherein the automatic tracking device acquires an image of a region in which the object needs to be tracked. The image input means provided with a camera and the image input means based on the input pan / tilt angle control amount
The turning means that turns in the axial direction, the tracking object extraction means that extracts the object to be tracked from the image acquired by the image input means, and the position of the tracking point of the object to be tracked at different times are the turning. The pan / tilt angle of the camera and the zoom factor of the image input device are used to convert the positions to absolute coordinates, which are stored in time series, and the positions of the stored tracking points are converted to camera coordinates again and output. Position storage means, position estimation means for estimating the moving position of the object from the position of the tracking point of the past object, speed calculation means for calculating the moving speed of the object from the position of the tracking point of the past object, and said estimation Pan of the turning means based on the moving position and the moving speed
Rotation control means for controlling the tilt angle, object movement area calculation means for calculating the area of the area swept on the camera coordinates by movement of the tracking object, and zoom magnification of the image input means from the movement area area of the tracking object And a zoom control means for controlling the zoom control.

According to a third aspect of the present invention, there is provided an automatic tracking method for tracking an object using a captured image, wherein the automatic tracking method determines the position of the tracking point of the tracking target object at different times. , The position of the absolute coordinate is converted using the pan / tilt angle of the image input means and stored in time series, and the positions of the plurality of tracking points stored are converted into camera coordinates again and output. The moving position of the object is estimated from the position of the tracking point of the object and the moving speed of the object is calculated, and the pan / tilt angle of the image input means is controlled from the estimated position and the moving speed.

According to a fourth aspect of the present invention, there is provided an automatic tracking method for tracking an object using a picked-up image, wherein the automatic tracking method determines the position of the tracking point of the object to be tracked at different times. , Converts to positions on absolute coordinates using the pan / tilt angle and zoom magnification of the image input means and stores them in time series, and converts the stored positions of a plurality of tracking points into camera coordinates again and outputs. Then, the moving position of the object is estimated from the position of the tracking point of the past tracking object and the moving speed of the object is calculated, and the pan / tilt angle of the image input means is controlled based on the estimated position and the moving speed, and the tracking object It is characterized in that the area of the area swept on the camera coordinates is calculated by the movement, and the zoom magnification is controlled from the area of the moving area of the tracking object.

According to a fifth aspect of the present invention, there is provided a computer-readable recording medium recording an automatic tracking program for performing tracking using a captured image, wherein the automatic tracking program is a tracking target at different times. The position of the tracking point of the object is converted into a position on the absolute coordinate by using the pan / tilt angle of the image input means and stored in time series, and the positions of the plurality of stored tracking points are again set to the camera coordinates. Converting and outputting, estimating the moving position of the object from the position of the tracking point of the past object and calculating the moving speed of the object, and controlling the pan / tilt angle of the image input means from this estimated position and moving speed. It is characterized by making a computer do.

According to a sixth aspect of the present invention, there is provided a computer-readable recording medium in which an automatic tracking program for performing tracking using a captured image is recorded, the automatic tracking program being a tracking target at different times. The position of the tracking point of the object is converted into a position on the absolute coordinates by using the pan / tilt angle and the zoom magnification of the image input means and stored in time series, and the positions of the plurality of stored tracking points are stored in the camera. The converted coordinates are output again, the moving position of the object is estimated from the position of the tracking point of the tracking object in the past, and the moving speed of the object is calculated. From the estimated position and moving speed, the pan / tilt angle of the image input means is calculated. It is characterized by causing the computer to calculate the area of the area to be swept on the camera coordinates by moving the tracking object and control the zoom magnification from the area of the moving area of the tracking object. .

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An automatic tracking device according to an embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of the first embodiment. In this figure, reference numeral 1 is an image input device composed of a CCD camera or the like, and the captured image is a moving image.
Reference numeral 3 is a turning device that turns the image input device 1 on two axes, that is, vertical and horizontal. Reference numeral 4 is an image output device that displays the image acquired by the image input device 1. Code 5
Is a tracking object extraction means for extracting a region of an object which is a tracking target from the image acquired by the image input apparatus 1. Code 7
Calculates the position of the tracking point from the tracking target object region extracted by the tracking object extraction means 5, and further acquires the pan / tilt angle from the turning device 3 to convert the position of the tracking point into a position on absolute coordinates. It is a position storage unit that stores the data afterwards, converts the stored positions of the tracking points at a plurality of absolute coordinates into the position on the camera coordinates again using the pan / tilt angle acquired from the turning device 3, and outputs the converted position. Reference numeral 8 is a speed calculation means for calculating the moving speed of the tracking point from the positions of the plurality of tracking points output from the position storage means 7. Reference numeral 9 is position estimation means for predicting the tracking point position of the tracking object at the next time from the positions of the plurality of tracking points output from the position storage means 7. Reference numeral 10 is a turning device control means 10 which determines the pan / tilt control amount of the turning device 3 from the tracking point position and the moving speed which are outputs of the speed calculation means 8 and the position estimation means 9.

The overall operation of this embodiment will be described in detail with reference to FIG. First, the tracking object extraction means 5 extracts the area of the target object to be tracked from the image captured by the image input device 1. An existing method using background subtraction and template matching will be described as a method for determining and extracting a tracking object. First, an intruding object is detected using the existing background subtraction method while the pan / tilt turning device 3 is stopped. In the background subtraction method, an image that does not include an object to be detected is stored in advance as a background image f (x, y), and a difference image h (x of the input image g (x, y) and the brightness value of the background image h (x , Y) = | g (x, y) −f (x, y) | Further, by binarizing the gradation value of h (x, y) with a certain threshold value, it is possible to detect the intruding object area when an object not present in the background image intrudes.

Here, it is assumed that the detected intruding object is an object to be tracked and tracking is performed. The area of this intruding object is bounded by a circumscribed rectangle, and the upper left coordinates are represented by (x1, y1) and the lower right coordinates are represented by (x2, y2). This (x1, y
1), the image of the object area represented by (x2, y2) is saved as an initial template t (x, y). Based on this template, template matching is used to extract the tracking target object from the input image that changes due to pan / tilt rotation.

Template matching is the input image g
This is a method of searching the position having the highest degree of similarity with the template image t (x, y) from (x, y). The template matching calculation method is "Introduction to computer image processing"
Supervised by Hideyuki Tamura, Research Institute Publishing (ISBN 4-7952-6
304-3) Details on p149. Since the size and the shape of the object on the image change as the tracking object moves or changes its direction, the tracking object in the input image and the template image may not match, which may make it difficult to extract the object.
Therefore, in order to respond to the change in size, prepare a plurality of templates that are enlarged and reduced in size in advance, perform template matching using these multiple templates, and select the template size and position that maximize the similarity. To decide. In order to deal with the shape change of the object, the image of the template size is cut out from the position determined to have the highest similarity in the input image and saved as a new template. This makes it possible to extract a tracking object whose size and shape gradually change in the input image.

An existing motion vector may be used for object extraction. In this method, motion in the image is represented by a vector, clustering is performed on blocks that include vectors in similar directions, and the largest block excluding the large background area caused by camera movement is regarded as the moving object area. Extract as a moving object. The circumscribed rectangle of the extracted area is obtained, and its upper left coordinates (x1, y1) and lower right coordinates (x
2, y2). The motion vector calculation method is described in “Image Processing and Recognition” by Takeshi Akuin and Tomoharu Nagao, Shokodo P.
164 (ISBN 4-7856-9043-7).

The position storage means 7 first extracts a point representing the area of the object extracted by the tracking object extraction means 5 as a tracking point. As a tracking point extraction method, when a tracking object was extracted using template matching, the upper left coordinates (x1, y1) and the lower right coordinates (x2, y2) of the template rectangle at the extracted positions were averaged. Center pixel ((x1
+ X2) / 2, (y1 + y2) / 2) are calculated as tracking points. In applications where only people are tracked,
Sometimes it is more convenient to track the head than to track the center of the person area. In this case, the tracking point is not the center pixel of the template, but the upper part of the template, for example, the position of 80% height ((x1 + x2) / 2, (4xy1 + y2)
/ 5), may be selected as the tracking point. When the area of the tracking object is extracted by the motion vector, the center of gravity of the tracking object area may be selected as the tracking point, or the center of the circumscribed rectangle of the object may be selected.

Next, the position storage means 7 acquires the pan angle corresponding to the turning angle in the left and right direction and the tilt angle corresponding to the turning angle in the up and down direction from the turning device 3, and the image input device 1 acquired in advance. From the angle of view (viewing angle) of, the position of the extracted tracking point is once converted into a position on absolute coordinates. As a method for converting a tracking point into a position on absolute coordinates, polar coordinates represented by pan / tilt angles are used as absolute coordinates, as shown in FIG.

Here, the point Oc at the center of the camera image (the origin of the camera coordinates) can be expressed by equation (1) when expressed by the pan / tilt angle. Similarly, the point P on the image can be expressed by the pan / tilt angle as the expression (2).

[Equation 1]

[Equation 2]

However, f is the focal length of the lens system [mm]
Where x and y are positions [mm] on the camera image plane (CCD element plane), and the position on the image plane is calculated from the size of the camera image sensor CCD and the number of pixels. Further, if x and y << f, it can be expressed by a linear approximation as in the equation (3).

[Equation 3]

By using this conversion, as shown in FIG. 4, the position of an arbitrary point on the camera coordinates can be expressed in polar coordinates using the pan / tilt angle, and the position of the tracking point of the object in absolute coordinates can be expressed. Can be grasped.

Next, the positions of the tracking points converted on the absolute coordinates are stored in time series. Since the memory to be stored is limited, only the positions of N (N ≧ 2) tracking points are stored. When memorizing the position of a new tracking point, the position of the oldest tracking point is erased and the position of a new tracking point is added.

Finally, as shown in FIG. 5, the positions of the N tracking points stored in this memory are again calculated by using the equation (4) according to the pan / tilt angle from the turning device 3. It is converted to the position on the camera coordinates and output. That is, the stored positions of the N tracking points are virtually reproduced on the line of sight of the camera, and the positions of the tracking points converted into the camera coordinates must be within the camera angle of view. There is no.

[Equation 4]

The speed calculating means 8 calculates the moving speed V from the N tracking point positions output from the position storing means 7. As shown in FIG. 6, assuming that the movement of the object is linear on the camera coordinates, the moving speed V uses the current tracking point position G0 and the tracking point position G1 one unit time before, and (5)
It is calculated by the formula.

[Equation 5]

The position estimating means 9 estimates the tracking point position Obj at the next time from the N tracking point positions output from the position storing means 7. The estimated position of the tracking point can be calculated from the past N positions by extrapolation using a polynomial or extrapolation using a rational function. These are described in "Numerical Recipe in Sea" Technical Review, William H. et al. Press, S
aul A. Teukolsky, William
T. Vetterling, Brian P.
By Flannery, Katsuichi Dankei, Haruhiko Okumura, Toshiro Sato,
Translated by Makoto Kobayashi P. 96 (ISBN4-87408-56
Details on 0-1). If N = 2, the current tracking point position is G0, and the tracking point position one unit time ago is G1 (6)
It is calculated by the formula.

[Equation 6]

The turning device control means 10 comprises a speed calculation means 8
An appropriate pan / tilt control value is calculated so that the tracking object (tracking point) can be captured in the center of the image from the moving speed calculated in step 1 and the position of the tracking point at the next time estimated by the position estimating means 9. The deviation (Δx, Δy) between the estimated position of the tracking point at the next time and the center of the image is calculated, and feedback control is performed so that the deviation becomes zero. As the pan / tilt control amount, as shown in FIG. 7, the pan movement angular velocity is calculated from the x coordinate shift Δx, and the tilt movement angular velocity is calculated from the y coordinate shift Δy. At this time, if the deviation (Δx, Δy) is smaller than the minimum movement amount of the pan / tilt on the image, the object cannot be captured in the center of the image, causing an overshoot (overshoot), which makes the control unstable ( Will vibrate).

Therefore, the minimum movement amount (Pmin, Tmin) on the image is calculated in advance from the angle of view of the image input apparatus 1 and the minimum pan and tilt control amounts, and the deviation (Δx, Δy) is panned. -If the tilt movement amount is smaller than the minimum movement amount, do not perform pan or tilt operation. Further, the pan / tilt control amount is changed based on the moving speed (vector) of the object obtained from the speed calculating means 8 depending on whether the object is moving in the image center direction or moving away from the image center. be able to. An object moving away from the center of the image may deviate from the angle of view, so it is necessary to perform quick pan / tilt control before the object deviates from the angle of view.

However, an object coming toward the center of the screen is less likely to deviate from the angle of view, so control is performed as gently as possible. In other words, it is not necessary to move the tracking point to the center of the image with one pan / tilt control, and it is sufficient to perform gentle control so as to gradually approach the center of the image with several times of control. This makes it possible to provide a tracking image that does not cause any discomfort to the observer without causing a sudden movement of the object in the image.

As described above, the positions of a plurality of tracking points of the object to be tracked at different times are converted into positions on the absolute coordinates by using the pan / tilt angles and stored, and the positions of these tracking points are stored. Once again, the position of the tracking point at the next time is estimated by converting the current pan / tilt angle to the camera coordinates,
Since the moving speed can be calculated and the pan / tilt control is performed using these values, the tracking can be smoothly performed without losing the object according to the position / speed of the object.

Next, a second embodiment of the present invention will be described in detail with reference to the drawings. FIG. 2 is a block diagram showing the configuration of the second embodiment. In this figure, the same parts as those of the device shown in FIG. 1 are designated by the same reference numerals, and the description thereof will be omitted. Reference numeral 2 denotes a zoom drive device including an electric zoom lens attached to the image input device 1,
The angle of view of the image input device 1 can be controlled by operating the zoom lens. By the turning device 3, it is possible to turn with the image input device 1 on two axes, that is, vertical and horizontal. Reference numeral 6 is an angle-of-view reference unit that refers to angle-of-view information of the image input device 1 that is set in advance by the zoom value. Reference numeral 11 is an object moving area calculating means for calculating the area on the screen swept by moving the object in a unit time, based on the area of the object area extracted by the tracking object extracting means 5 and the moving speed of the output of the speed calculating means 8. Is. Reference numeral 12 is
It is a zoom control unit that determines the zoom operation amount from the object movement region calculation unit 11. The device shown in this figure is different from the device shown in FIG. 1 in that a zoom drive device 2, an angle-of-view reference device 6, an object movement region calculation device 11, and a zoom control device 12 mounted on the image input device 1 are further provided. That is the point.

Next, the overall operation of this embodiment will be described in detail with reference to FIG. First, the angle-of-view reference unit 6
Acquires a zoom value from the zoom drive device 2 and outputs angle-of-view information corresponding to the obtained zoom value. The zoom value is a value corresponding to the focus value f of the zoom lens of the zoom driving device 2, and the combination of the angle of view of the image input device 1 corresponding to the zoom value is stored in a memory or the like in advance. Since the number of combinations of the zoom value and the angle of view to be stored in advance is limited due to the limitation of the memory or the like, only the main combinations may be stored and the rest may be approximated by interpolation.

Next, the position storage means 7 extracts a point representative of the area of the object extracted by the tracking object extraction means 5 as a tracking point, acquires the pan / tilt angle from the turning device 3, and obtains the angle-of-view reference means. 6, the angle of view of the image input device 1 is acquired, the extracted tracking point position is once converted into an absolute coordinate position and stored in time series, and the plurality of stored tracking point positions are stored in the turning device 3 The pan / tilt angle from is converted again to the tracking position in camera coordinates.

As shown in FIG. 8, the object moving area calculating means 11 calculates the area S swept on the screen by the movement of the object per unit time. When the object area is simplified and approximated to a rectangle, the object has a width W and a height H, and the moving velocity vector is V.
Assuming that (Vx, Vy), the object moving area S can be obtained by the equation (7).

[Equation 7]

The zoom control means 12 controls the zoom operation based on the moving area area calculated by the object moving area calculating means 11. As a zoom control method, it is determined in advance how much the object should be zoomed in on the image, and the object area at this time is registered as the set area S0 so that the moving area area S of the tracking object becomes S0. Performs feedback control of zoom. That is, a control signal is sent to the zoom operating device 2 so as to zoom out when S-S0> 0 and to zoom in when S-S0 <0. However, S ≒ S
In order to prevent the zoom control from becoming unstable due to a slight change in the object region area S at 0, as shown in FIG. 9, when the deviation from the set area is equal to or less than the threshold value TH, the zoom control is performed. If not performed, the acquired image will be stable and the electric zoom device will not be uselessly driven.

Further, when the object to be tracked is extracted by the template matching in the tracking object extracting means 5, it is better to limit the zoom speed according to the template update speed. This is because it is difficult to extract an object when the amount of change in zoom is larger than the allowable range of change in size (Zoom_Max shown in FIG. 9) in template matching.

With respect to zoom-in, as shown in FIG. 10, when the zoom-in possible area is set on the screen in advance and the object tracking point (Gx, Gy) is located within the area ((( Zoom-in control is performed only when (8) is satisfied. This is because, if the object is far from the center of the screen, zooming in may cause the tracking object to be out of the angle of view.

[Equation 8]

As described above, the positions of the tracking points of the tracking object at different times are converted into positions on the absolute coordinates by using the pan / tilt / zoom values and stored, and the positions of these tracking points are panned again.・ By converting from the tilt angle to the current camera coordinates, the position of the tracking point at the next time can be estimated, the moving speed can be calculated, and the area of the moving area of the object can be calculated from this and the object area. Since the pan / tilt and zoom controls are performed using these values, the pan / tilt control according to the position / speed of the tracking target object and the zoom ratio according to the position / speed / size of the tracking target object. By adjusting, it is possible to perform tracking with zooming up without losing sight of the object.

It should be noted that a program for realizing the configuration shown in FIGS. 1 and 2 is recorded in a computer-readable recording medium, and the program recorded in the recording medium is read by a computer system and executed to automatically execute the program. Tracking processing may be performed. The “computer system” mentioned here includes an OS and hardware such as peripheral devices. Also, "computer system"
Includes the homepage providing environment (or display environment) if the WWW system is used. The "computer-readable recording medium" means a portable medium such as a floppy (registered trademark) disk, a magneto-optical disk, a ROM, a CD-ROM, or a storage device such as a hard disk built in a computer system. . Further, the "computer-readable recording medium" is a volatile memory (RAM) inside a computer system which serves as a server or a client when a program is transmitted via a network such as the Internet or a communication line such as a telephone line. In addition, those that hold the program for a certain period of time are also included.

Further, the above program may be transmitted from a computer system in which the program is stored in a storage device or the like to another computer system via a transmission medium or by a transmission wave in the transmission medium. Here, the "transmission medium" for transmitting the program refers to a medium having a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line. Further, the program may be a program for realizing some of the functions described above. further,
It may be a so-called difference file (difference program) that can realize the above-mentioned functions in combination with a program already recorded in the computer system.

[0040]

As described above, according to the present invention, by converting the object positions at different times into positions on the absolute coordinates and storing the positions, the moving speed of the object to be tracked is tracked even with an arbitrary camera line of sight. And the predicted position of the object is calculated, and the pan / tilt is controlled according to the direction of the position / moving speed, so pan / tilt tracking can be performed according to the position and moving speed of the object. Even if the direction of the object is reversed, it is possible to obtain the effect that the tracking can be performed smoothly and accurately without being lost.

Further, according to the present invention, by converting the object positions at different times into positions on the absolute coordinates and storing them,
Since the zoom is controlled according to the area of the moving region swept by the object to be tracked even with the arbitrary line of sight of the camera, tracking can be performed by zooming up according to the moving speed of the object. In tracking, it is possible to obtain an effect that the tracking is performed with the zoom-up suppressed so as not to be overlooked, and the tracking can be automatically performed with the zoom-up even for an object having a small movement.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a first exemplary embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of a second exemplary embodiment of the present invention.

FIG. 3 is an explanatory diagram showing an operation of the first exemplary embodiment of the present invention.

FIG. 4 is an explanatory diagram showing an operation of the first exemplary embodiment of the present invention.

FIG. 5 is an explanatory diagram showing an operation of the first exemplary embodiment of the present invention.

FIG. 6 is an explanatory diagram showing an operation of the first exemplary embodiment of the present invention.

FIG. 7 is an explanatory diagram showing an operation of the first exemplary embodiment of the present invention.

FIG. 8 is an explanatory diagram showing an operation of the second exemplary embodiment of the present invention.

FIG. 9 is an explanatory diagram showing an operation of the second exemplary embodiment of the present invention.

FIG. 10 is an explanatory diagram showing operations according to the second embodiment of the present invention.

FIG. 11 is an explanatory diagram showing a tracking operation according to a conventional method.

FIG. 12 is an explanatory diagram showing a tracking operation by a conventional method.

FIG. 13 is an explanatory diagram showing a tracking operation according to a conventional method.

[Explanation of symbols]

1 ... Image input device, 2 ... Zoom drive, 3 ... turning device, 4 ... Image output device, 5 ... Tracking object extraction means, 6 ... View angle reference means, 7 ... position storage means, 8: speed calculation means, 9 ... Position estimation means, 10 ... turning device control means, 11 ... Object movement region calculation means, 12 ... Zoom control means.

Claims (6)

(57) [Claims] 1. An automatic tracking device for tracking an object using a captured image, wherein the automatic tracking device includes an image input means including a camera for acquiring an image of a region in which the object needs to be tracked. A turning means for turning the image input means in two axial directions based on the input pan / tilt angle control amount; and a tracking object extraction for extracting an object to be tracked from the image acquired by the image input means. Means and the position of the tracking point of the tracking target object at different times are converted into positions on the absolute coordinates by using the pan / tilt angle of the turning means and stored in time series. Position storage means that converts the position of the tracking point into camera coordinates again and outputs it, position estimation means that estimates the moving position of the object from the position of the tracking point of the past object, and object from the position of the tracking point of the past object Moving speed A speed calculation means for calculating, automatic tracking apparatus being characterized in that and a turning control means for controlling the pan and tilt angles of the turning unit based on the moving speed and the estimated travel position. 2. An automatic tracking device for tracking an object using a captured image, wherein the automatic tracking device includes an image input means including a camera for acquiring an image of a region in which the object needs to be tracked. A turning means for turning the image input means in two axial directions based on the input pan / tilt angle control amount; and a tracking object extraction for extracting an object to be tracked from the image acquired by the image input means. Means and the position of the tracking point of the tracking target object at different times are converted into positions on absolute coordinates using the pan / tilt angle of the turning means and the zoom magnification of the image input means, and stored in time series. , A position storage unit that converts the stored positions of a plurality of tracking points into camera coordinates again and outputs the position, a position estimation unit that estimates the moving position of the object from the positions of the past tracking points of the object, and a past object Addition Speed calculation means for calculating the moving speed of the object from the position of the point, turning control means for controlling the pan / tilt angle of the turning means based on the estimated moving position and the moving speed, and camera coordinates by moving the tracking object. An automatic tracking device, comprising: an object moving area calculating means for calculating an area of an area to be swept above; and a zoom control means for controlling a zoom magnification of the image input means based on a moving area area of the tracking object. . 3. An automatic tracking method for tracking an object using a captured image, wherein the automatic tracking method determines the position of a tracking point of an object to be tracked at different times by using a pan / scan of an image input means. The positions of the tracking points of the past object are converted and output again by converting the stored positions of the tracking points into camera coordinates and converting the positions into absolute coordinates using the tilt angle and storing them in time series. An automatic tracking method characterized in that the moving position of an object is estimated from the calculated position and the moving speed of the object is calculated, and the pan / tilt angle of the image input means is controlled from the estimated position and moving speed. 4. An automatic tracking method for tracking an object using a captured image, wherein the automatic tracking method determines the position of a tracking point of an object to be tracked at different times by the pan / scan of the image input means. It is converted to a position on absolute coordinates using the tilt angle and zoom magnification and stored in time series, and the positions of the stored multiple tracking points are converted to camera coordinates again and output, and the past tracking object The moving position of the object is estimated from the position of the tracking point and the moving speed of the object is calculated.The pan / tilt angle of the image input means is controlled based on the estimated position and moving speed, and the tracking object sweeps on the camera coordinates. An automatic tracking method, wherein the area of a region is calculated and the zoom magnification is controlled based on the moving region area of a tracking object. 5. A computer-readable recording medium in which an automatic tracking program for performing tracking using a captured image is recorded, wherein the automatic tracking program determines the positions of tracking points of an object to be tracked at different times. , The position of the absolute coordinate is converted using the pan / tilt angle of the image input means and stored in time series, and the positions of the plurality of tracking points stored are converted into camera coordinates again and output. The moving position of the object is estimated from the position of the tracking point of the object and the moving speed of the object is calculated, and the computer is made to control the pan / tilt angle of the image input means from the estimated position and the moving speed. A recording medium on which the characteristic automatic tracking program is recorded. 6. A computer-readable recording medium in which an automatic tracking program for performing tracking using a captured image is recorded, wherein the automatic tracking program determines the positions of tracking points of an object to be tracked at different times. , Converts to positions on absolute coordinates using the pan / tilt angle and zoom magnification of the image input means and stores them in time series, and converts the stored positions of a plurality of tracking points into camera coordinates again and outputs. Then, the moving position of the object is estimated from the position of the tracking point of the past tracking object and the moving speed of the object is calculated, and the pan / tilt angle of the image input means is controlled based on the estimated position and the moving speed, and the tracking object An automatic tracking program characterized by causing a computer to calculate the area of a region to be swept on camera coordinates by movement and control the zoom magnification from the moving region area of the tracking object. Recording the recording medium.