JPH0898079A - Image signal processing unit - Google Patents

Abstract

PURPOSE: To avoid mis-recognition of an object by detecting a moving object in image data based on number of picture elements in which a change in luminance of Y, R, B of image data and delayed image data is a threshold level or over in the automatic tracking of a video camera. CONSTITUTION: While the field angle of a television camera 1 is not changed, an input image signal S2 is A/D-converted, and a processing circuit 7 processes image data Y, R, B periodically and stores them to a memory 10. A characteristic picture element extract means of a tracking processing circuit 11 extracts picture elements in which a change in the luminance is a prescribed threshold level or over with respect to picture elements of optional characteristic luminance and chrominance information between the image data and image data delayed by a prescribed time. A moving body detection means of the tracking circuit 11 counts the number of extracted picture elements to detect a moving amount from the number of changed picture elements thereby detecting the moving body in the image data. Thus, the moving body is detected based on luminance and chrominance information to avoid mis-recognition of the object in the case of automatic tracing.

Description

Detailed Description of the Invention

[0001]

[Table of Contents] The present invention will be described in the following order. Field of Industrial Application Conventional Technology Problem to be Solved by the Invention Means for Solving the Problem (FIGS. 1, 6 and 10) Action (FIGS. 1, 6 and 10) Embodiment (1) First Video camera system of the embodiment (FIGS. 1 to 5) (2) Video camera system of the second embodiment (FIGS. 6 to 9) (3) Video camera system of the third embodiment (FIGS. 10 to 13) ( 4) Other Examples Effect of the invention

[0002]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image signal processing apparatus, and can be applied to, for example, a video camera system for tracking a target by moving a television camera in accordance with the movement of a target object.

[0003]

2. Description of the Related Art Conventionally, as a video camera system, there is a video camera system which performs automatic tracking from a difference between frames of an image relating to luminance. In this video camera system, a difference regarding luminance is calculated for all pixels of image data of two consecutive frames, and a pixel whose difference value exceeds a certain threshold is counted as a pixel changed from the previous image. The moving object is detected by obtaining the center of gravity of the pixel set of. Then, the pan motor and tilt motor of the video camera are controlled according to the difference in the distance from the center of the position of the moving object on the screen detected for tracking the subject.

[0004]

However, in the conventional automatic tracking method based on the detection of a moving object such as the video camera system having such a configuration, the difference regarding the brightness is calculated for all the pixels of the two image data, and the difference is calculated. Since we are paying attention to the value, even if the object has a completely different color and brightness from the subject,
If there is a brightness change in the pixel, there is a problem that the pixel is counted as a moving point, that is, a subject. In addition, when moving objects are detected while the camera block or video camera body is being moved to track an object, there is a drawback in that motion detection is performed on random pixels in the screen.

Further, conventionally, as one of the functions of the video camera system, there is an autofocus for automatically focusing an image. This controls the focus lens so as to increase the high frequency component of the spatial frequency related to the luminance within the area set in the center of the screen. The "mountain climbing method" and "wow ring method" are generally used that determine the direction in which the lens is moved so that the evaluation value in the detection frame in the center of the screen becomes high by moving the focus lens.

However, like the video camera system having such a configuration, the conventional autofocus function, which focuses only on the area fixed at the center of the screen, matches the background in the center when the target object deviates from the center of the screen. I got impatient,
There was a drawback that the original purpose of focusing on the subject could not be achieved.

The present invention has been made in consideration of the above points, and an object of the present invention is to propose an image signal processing apparatus capable of effectively eliminating erroneous recognition when detecting a moving object in an image and improving its usefulness. It is a thing.

[0008]

In order to solve such a problem, in the present invention, the input image signal (S2) is converted into a digital signal, and the image memory (10) is periodically used as image data (Y, R, B). ) Image storage control means (9), image data (Y, R, B) stored in the image memory (10) and a delayed image delayed by a predetermined time from the image data (Y, R, B). Characteristic pixel extraction means (11) for extracting pixels having a luminance change amount of a predetermined threshold value or more between pixels of arbitrary characteristic luminance and color between the data (Y, R, B), and the characteristic thereof. The number of pixels extracted by the pixel extracting means (11) is counted to obtain image data (Y, R,
The moving body detecting means (11) for detecting the moving body in B) is provided.

In the present invention, the image signal (S2)
Is input from the television camera (1, 2, 3, 4), and the feature pixel extraction means (11) selects an arbitrary feature when the angle of view of the television camera (1, 2, 3, 4) is unchanged. With respect to the pixels of specific luminance and color, the pixels whose change in luminance is equal to or larger than a predetermined threshold value are extracted.

Further, in the present invention, image data (Y,
Image data (Y, R, B) is stored using the image storage control means (9) that periodically stores R, B) in the image memory (10).
The feature amount setting means (11) for setting the feature amount relating to the brightness and color information of the target object in B), and the feature model determining means (11) for determining the feature model indicating the feature of the target object from the set feature amount. And a pixel determination means (11) for detecting a pixel that is a candidate for a target object from the image data (Y, R, B) using the determined feature model, and horizontal and vertical directions of the detected pixel set. Coordinate provisional determining means (11) for determining the center of gravity of the target object and detecting the position of the target object in the image data (Y, R, B), and moving body detecting means for detecting the moving body in the image data (Y, R, B). The position verification means (11) for verifying the certainty of the position of the detected target object using (11), and the television camera according to the position of the target object verified by the position verification means (11). (1, 2,
The camera control means (12, 13) for controlling (3, 4) is provided to track the target object in the image data (Y, R, B).

In the present invention, the image data (Y,
An image storage control means (9) for periodically storing R, B) in the image memory (10) is used to store the entire initial image at startup in the image memory (10) with respect to luminance and color information. Means (9) and image data (Y, R, B)
A moving point detecting means (11) for detecting a moving body in the inside of the initial image stored in the image memory (10), and a moving point detecting means (11) for detecting a pixel whose luminance change is equal to or more than a predetermined threshold value. , A change area detecting means (11) for detecting a changed area from the detected pixel from the initial image, and when the change area detecting means (11) detects no change area from the initial image, An initial image resetting means (11) for resetting the initial image at regular intervals is provided.

Further, in the present invention, the changing area detecting means (11) for detecting the changing area from the initial image is used,
Recording control means (15) that enters an image recording state only when a changed area is detected from the initial image and frame display means (14) that superimposes a frame display surrounding the changed area on the video signal. I decided to set and.

Further, in the present invention, image data (Y,
Correlation detecting means (11) for detecting the correlation between two consecutive image data (Y, R, B) by using the moving body detecting means (11) for detecting the moving body in R, B), Scene change detecting means (1) for detecting scene change in the image signal (S2) input by the correlation detecting means (11).
1) and are provided.

In the present invention, the image data (Y,
In the target object area setting means (11) for setting an area including the target object extracted from R, B) and the target object area set by the target object area setting means (11), image data (Y, Lens control means (14, 15, 16) for controlling the focus lens (2) of the television camera (1) so as to increase the spatial high frequency component of R, B).
Was set up.

[0015]

By verifying the positioning by detecting the moving object in the image, it is possible to avoid erroneously recognizing the case where the automatic tracking is performed. In addition, by updating the image data to be compared only when a moving object is detected, a large amount of difficult movement occurs when updating each frame.

Further, by comparing from the initially set initial image, recording is performed only when there is a change, and by detecting the position of the change, it is possible to perform auto panning or tilting on the object. Also, since the moving object is detected by image processing, the system does not need to be so large as compared with the conventional system. Further, since the scene change detection information source performs only from the image information, the scene change can be surely detected.

By controlling the focus lens so that the spatial high-frequency component of the image data relating to the region tracked to the target object becomes large, an image focused on the target object can be obtained no matter where the target object is on the screen. Is obtained.

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described in detail below with reference to the drawings.

(1) Video Camera System of First Embodiment FIG. 1 shows, as a whole, an embodiment of a video camera system having a subject tracking function using motion detection according to the present invention. In the following, the luminance is Y, the color difference RY signal is R, and the color difference BY signal is B, and the value of the i-th pixel in the vertical direction and the j-th pixel in the horizontal direction in the image in the memory described later is Y, and Y is Y. _ij and R are R _ij , and B is B _ij . In this system, the image pickup light coming from the subject passes through the lens 2 and the iris 3 of the lens block 1 and is received by the image pickup device 4, which is, for example, a CCD, and as a result, the image pickup output signal S1 representing the subject image is separated / demultiplexed. It is given to the automatic gain adjustment circuit section (AGC) 5.

The signal separation / automatic gain adjustment circuit section 5 samples and holds the image pickup output signal S1 and controls the gain so as to have a predetermined gain by a control signal from an auto iris (AE) system (not shown). The image pickup output signal S2 thus obtained is supplied to the digital camera processing circuit 7 via the analog / digital conversion circuit unit 6. The digital camera processing circuit 7 forms a luminance signal Y, a chroma signal C, and color difference signals R and B based on the digitized image pickup output signal, and the luminance signal Y and the chroma signal C.
Is transmitted as a video signal S3 via the digital / analog conversion circuit section 8.

In addition to this, the digital camera processing circuit 7 supplies the luminance Y and the color difference signals R and B as the subject tracking detection image signal S4 to the tracking control circuit unit 9, and the tracking control circuit unit 9 outputs the subject tracking detection signal. A tracking control signal S9 for the panning drive motor 12 and the tilting drive motor 13 provided for the lens block 1 is generated based on the image S4. The tracking control circuit unit 9 sends the color-difference signals R and B together with the luminance signal Y to the image memory 10 formed of, for example, a frame memory, and the write permission signal S sent from the tracking signal processing circuit 11 of the microprocessor configuration.
After the arrival of 5, Y, R, and B are stored independently for each pixel.

When the subject tracking detection image signal S4 has been stored in the image memory 10, a read permission signal S6 is sent to the tracking signal processing circuit 11. Tracking signal processing circuit 1
1 confirms the read permission signal S6 and then supplies an arbitrary memory address signal S7 to the image memory 10 to supply data at a predetermined position in the image memory 10 to the tracking signal processing circuit 11 as a data signal S8. . After the subject position calculation processing including the moving body detection is completed in the tracking signal processing circuit 11, the write permission signal S5 is generated again, and the image memory 10 confirms this, and then the image memory 10 rewrites the memory.

Next, a series of processing performed inside the tracking signal processing circuit 11 will be described. FIG. 2 shows a subject recognition processing procedure. In this step SP0, the region D0 centered on the center of the screen is set because the subject is initially in the center of the screen.
This is processing for obtaining subject information from (for example, a block of 5 × 5 pixels). Further, the next step SP1 is a process of determining a feature model including allowable errors in the quadratic approximation curves for Y and R, and Y and B based on the subject information obtained in step SP0.

Further continuing step SP2 is step SP
It is a determination process of each pixel that determines whether or not each pixel in the screen is a candidate point of the subject using the feature model obtained in 1. Further, step SP3 is a process for temporarily determining the coordinates of the subject from the pixels determined as part of the subject in step SP2. Furthermore, the next step SP4 is a motion verification process for detecting the motion in the image and verifying the authenticity of the object coordinates provisionally determined in step SP3.

An example of each step is shown below. In the subject information acquisition process of step SP0, (Y _ij ,
R _ij and B _ij ) have noise, and
Since it is only a representative point of the subject, in order to have a certain degree of width in the subject information (Y _ij , HR _ij , H
B _ij ), (Y _ij , LR _ij , LB _ij ) as subject information. Here, the subject information HR _ij , HB _ij , LR _ij , LB
_ij has the allowable error of 10%,

[Equation 1]

[Equation 2]

[Equation 3]

[Equation 4] Given in.

The characteristic model determination process of step SP1 is as follows.
Subject information (Y _ij , HR _ij , HB _ij ), (Y _ij , LR _ij , LB) obtained by the subject information acquisition processing of step SP0
_ij ), a feature model based on a quadratic approximation function of R and B for Y is determined. At this time, the intercepts R _low , R _high , B _low , and B _high on the Y axis are set as constants so that the feature model that is quadratic-approximated with respect to an arbitrary subject has a certain degree of constant shape. The following coefficient A by multiplication
_{0 to} A ₄ is determined, and R and B feature models HF related to Y
_r (Y), HF _b (Y), LF _r (Y), LF _b (Y)
Is

[Equation 5]

[Equation 6]

[Equation 7]

[Equation 8] To decide.

Whether each pixel determination process of step SP2 applies to the feature model obtained by the feature model determination process of step SP1 for each pixel, that is, whether both pixels R and B have the feature model HF of R and B with respect to Y
_r (Y), HF _b (Y), LF _r (Y), LF _b (Y)
As a result, it is determined whether to enter the hatching portion shown in FIG. That is,

[Equation 9] And the following formula

[Equation 10] If so, the pixel is determined to be a part of the subject.

In the coordinate provisional determination process of step SP3, the coordinates of the subject are determined by obtaining the horizontal and vertical center of gravity of a plurality of pixels determined to be a part of the subject in each pixel determination process of step SP2. Where step S
If there is no pixel that is determined to be a subject in each pixel determination process of P2, it is determined that there is no subject in the screen.

Next, in step SP4, a panning drive motor 12 for moving the lens block 1 of the camera,
It is detected whether the tilting drive motor 13 is moving. As a result, when the panning drive motor 12 and the tilting drive motor 13 are moving, that is, when the angle of view is changing, the subsequent motion verification using the moving body detection is not performed, and the coordinates are temporarily calculated in step SP3. The determined coordinates are determined as the subject coordinates. In the subsequent step SP5, it is detected whether or not the coordinates calculated in the coordinate provisional determination processing of step SP3 have moved by, for example, 5 pixels or more from the coordinates one frame before. This allows
Only when the coordinates have moved significantly compared to the past coordinates, the subsequent motion verification is performed, and when the coordinates have not changed, the panning drive motor 12 and the tilting drive motor 13 are performed.
As in the case where is not moving, the coordinates calculated in the temporary coordinate determining process of step SP3 are determined as the subject coordinates without performing the motion test.

Further, in the next step SP6, since it is judged that the subject coordinates in the current frame have moved more than the position in the previous frame, the amount of movement of the subject from the previous frame is naturally detected near the coordinates. If it should be, the motion extraction is performed in the vicinity of the coordinates in the previous frame and the newly calculated coordinates. When the amount of movement is not detected, the movement of the subject coordinates is not due to the movement of the subject, but is determined as an erroneous recognition in the subject recognition processing of each pixel determination processing and the temporary coordinate determination processing of steps SP2 and SP3, and is newly calculated The coordinate set is invalid and the coordinate in the previous frame is retained. If the amount of movement is detected, it means that the subject has moved to the newly calculated coordinate by the motion test, and thus the coordinate calculated in the current frame is adopted as the subject position.

As a first method of extracting the motion amount in step SP6, as shown in FIG. 4, for example, in step SP0 in the neighborhood area D1 surrounding the calculated coordinates and the coordinates calculated in the previous frame, The neighborhood of R _org and B _org which are R and B of the subject learned when the subject is first set in the subject information acquisition process, that is,

[Equation 11] And the following formula

[Equation 12] The difference regarding the luminance is calculated for the pixels that satisfy the condition. When the difference value exceeds a certain threshold value, it is determined that the motion amount is detected for the pixel. In this way, when the number of pixels N _move1 determined to have moved in the area D1 exceeds a certain ratio P ₁ of the total number of pixels N _D1 in D1, it is determined that the subject has moved in D1.

As a second method of extracting the motion amount,
For example, as shown in FIG. 5, for pixels in the region D1,
An evaluation data image in which a pixel satisfying the condition of each pixel determination process in step SP2 is set to a value "1" and a pixel not satisfying the condition is set to a value "0" is provided for the previous frame and the current frame, and is common between the two. It is determined that the pixels that do not move have moved. After that, when the number of pixels N _move2 determined to have moved in the same manner as described above exceeds a certain ratio P2 of the total number of pixels N _D1 in D1, it is determined that the subject has moved in D1.

The processing of step SP7 is the evaluation data of the previous frame, that is, in the case of the first example described above, the luminance data for the pixels satisfying the expressions (11) and (12) as shown in FIG. In the case of the above example, the evaluation data image of the previous frame in which the point satisfying the condition of each pixel determination process in step SP2 is set to the value "1" and the other pixel is set to the value "0" as shown in FIG. Update.

According to the above configuration, by performing the positioning verification by detecting the moving object in the image, it is possible to avoid the erroneous recognition of the object to some extent when the automatic object tracking is performed, and to detect the moving image of the image data to be compared. It is possible to realize a television camera system in which a large amount of difficult movement occurs when updating each frame by updating only every time.

(2) Video camera system of the second embodiment As a video camera system for monitoring to which the video camera system using the moving object detection of the first embodiment described above is applied, the moving object is detected from the initial image and displayed on the screen. An example of recording only when there is a change will be described. The configuration of this video camera system is shown in FIG. 6 in which the same parts as in FIG. FIG. 7 shows a procedure for detecting a moving object from the initial image. Step SP10 is a process for storing the luminance data of the initial image. From the next frame, the luminance changes as compared with the initial image luminance data D _org1. Performs motion detection.

The next step SP11 is a process for extracting a moving point. A difference is taken for each pixel between the initial image luminance data D _org1 (FIG. 8A) and the current frame image luminance data D _now1, and when this difference exceeds a certain threshold value α ₁ , the initial state is set for this pixel. It is judged that there has been a change from. In this way, the moving point extraction is performed for all the pixels, and the number N _{c of} pixels determined to have changed as in step SP12
When a certain threshold value α ₂ is exceeded, it is determined that the screen has changed from the initial state, such as an intruder entering the screen (FIG. 8B), and the recording state of step SP13 is entered.
After that, the frame display processing of step SP14 is performed, and
As shown in (C), the upper end of the pixel determined as the moving point,
A framed signal S10 that surrounds the bottom, right, and left pixels
(FIG. 6) is generated and a frame is placed on the video signal.

Further, in step SP12, when the number of pixels N _c judged to have changed does not exceed the threshold value α ₂ , it is judged that the screen has not changed from the initial state,
At step SP15, the recording standby state is entered, and the frame indicating the intruder or the change point is erased. Next, in step SP16, it is determined whether or not the state of the number of pixels N _c <threshold value α ₂ determined to have changed continues for a certain period of time, and if so, the process proceeds to step SP17 to set the initial image brightness data. D _org1
Update.

As an application example of the video camera system of FIG. 6, FIG. 9 shows a processing procedure for detecting scene change in an image signal. In this scene change detection processing procedure, the step SP20 determines the brightness data D of the image of the previous frame.
_This is the process of storing _org2 . At the following step SP21, it obtains a difference for each pixel between the luminance data D _org2 image of the previous frame and the luminance data D _Now2 the image of the current frame, when the threshold is exceeded alpha ₃ there is the difference, the pixel It is judged that there is a change from the previous frame.

In this way, moving point extraction is performed for all pixels,
When the number N _{d of} pixels determined to have a change in luminance (that is, moved) in step SP22 exceeds a certain threshold value α ₄ , it is determined that the correlation between both screens has significantly decreased, and in step SP23. Generates a scene change detection signal. After this, regardless of the presence or absence of scene change detection,
In step SP24, the luminance data D of the image of the current frame
The _now2 returns to step SP21 after replacing the brightness data D _org2 of the previous frame image.

According to the above configuration, as an application to the security system, by comparing from the initial setting image, recording is performed only when an intruder encounters, and the position of the intruder is detected. It is possible to perform auto panning and tilting on an object. Also, since the moving object is detected by image processing, the system does not need to be so large as compared with the conventional system. Further, since the scene change detection information source is generated only from the image information, the scene change can be detected without depending on the recorded tape.

Further, according to the above configuration, like a surveillance camera applied to a conventional security system,
It is possible to solve the problem that shooting must be performed continuously only at a fixed angle and a fixed panning, and a significantly smaller surveillance camera can be realized as compared with moving object detection using an infrared sensor.

Further, according to the above-mentioned configuration, as in the case of scene change detection in the conventionally conceived system, the time code recorded on the video tape is read, and a comparison is made in the case where the continuity is found. do it,
Even when the video tape is not time-coded or has a different time code standard, the scene change can be effectively detected, and the usefulness can be significantly improved.

(3) Video camera system of the third embodiment of the autofocus function focusing on the luminance signal in the area where the target object enters, which is an application of the video camera system using the subject extraction of the first embodiment described above. An example will be described. The structure of this video camera system is shown in FIG. 10 in which the same parts as in FIG.

In FIG. 10, the video signal detector 16 sets the detection area from the target object area coordinate signal S9 sent from the tracking signal processing circuit 11, and the digital signal sent from the digital camera processing circuit 7 with respect to this detection area. Detection for focusing is performed based on the Y signal. The video signal processing circuit 17 determines how to move the focus motor 18 based on the received detection output, and controls the focus motor 18. The series of processes will be described below.

FIG. 11 shows the state of each pixel which is judged as a candidate point in step SP2 of the above-mentioned subject recognition processing procedure. Pixels that are determined to be candidate points in the figure are shaded, and points that are not are shown in white. In addition, the subject coordinates determined in step SP8 are (Xt, Y
t), the upper left coordinate of the frame is (X1, Y1), and the lower right coordinate is (X2, Y2) as the initial frame coordinates of the subject area. Further, among the pixels in the subject area, the number of pixels determined to be the candidate points of the subject, that is, the number of pixels included in the shaded area in FIG. 11 is N _t .

In this state, first move X1 one screen to the left of the screen.
Shift by minutes, then N_tIs X₁Large due to shifting
If it is tight, change to the X1 offset position. Reverse
N _tIf is not increased, X1 is returned to the original position.

Next, Y1 is similarly shifted upward by one pixel, and if N _t at that time is increased, it is changed to the shifted position of Y1, and if not increased, it is returned to the original position. Similar processing is continued to shift X2 to the right and then to Y2. This series of routines is continued until N _t does not increase. The tracking signal processing circuit 11 sends the final (X1, Y1) and (X2, Y2) coordinates to the video signal detector 16 as the coordinates of the subject area.

The video signal detector 16 detects the Y signal for the received subject area. With respect to the Y signal sent from the digital camera processing circuit 7, a high-pass filter is applied in the horizontal direction for one line, and the absolute value thereof is taken for rectification (FIG. 12). The maximum value of this data is the evaluation value of this line. Similarly, evaluation values are obtained for all lines, and the maximum value among them is communicated to the video signal processing circuit 17 as a detection output.

The video signal processing circuit 17 controls the focus motor 18 based on the detection output from the video signal detector 16. There are various methods in this algorithm, but this time, the "Woobling method" will be explained. That is, the focus lens 2 is moved so that the detection output value is maximized to determine whether or not the focus is achieved. That is,
As shown in FIG. 13, the apex of the evaluation curve with the horizontal axis representing the lens position and the vertical axis representing the detection output is the in-focus position.

First, the focus lens 2 is made minute by 15 [Hz].
By moving the lens 2 by a certain degree, it is determined which position on the evaluation curve the lens is currently on the basis of the detection output at the time of the maximum and the minimum of the minute position variation of the lens, and the direction in which the lens 2 is moved is determined. That is, in FIG. 13, it can be seen that if the lens is moved in the positive direction, the focus is achieved. When it is determined that the evaluation value has converged to the maximum value, that is, it has come near the apex of the evaluation curve, it is determined that the focus is achieved.

Through the series of operations described above, the focus motor 18 is controlled to perform the auto focus tracking the subject. If it is determined in step SP8 that there is no subject or the subject is unknown, the process returns to normal autofocus with the detection frame fixed.

(4) Other Embodiments In the above embodiments, the present invention is applied to a video camera system and a security system for automatically tracking a target, but the present invention is not limited to this. Instead, such as a TV conference conference system or multimedia system that targets a subject with a television camera, or auto focus, auto iris, auto zoom, and image stabilization that lock the subject from its position in the image. It is suitable for wide application to television cameras.

[0053]

As described above, according to the present invention, erroneous recognition of the target object can be avoided to some extent when the automatic tracking is performed by performing the positioning verification by detecting the moving object in the image. By updating the data only when a moving object is detected, the amount of motion that is difficult to come out becomes large when updating each frame, thus effectively eliminating false recognition when detecting a moving object in an image and useful. It is possible to realize an image signal processing device capable of improving the property.

By entering the recording state only when a change from the initial setting image is detected, an intruder finds that the surveillance camera of the conventional security system must continuously record images. It is possible to record only when the time or the situation changes. In addition, the initial setting image is not fixed and is updated at regular intervals while motion is not detected, so that the screen may change when the camera is installed outdoors, such as when it changes from daytime to evening, and changes slowly. It will not be misjudged that Furthermore, by using the moving body detection, scene change detection can be performed with a simple system.

By tracking the subject and performing autofocus using a detection frame according to the size of the subject, the photographer can obtain a video image in which the subject is focused regardless of where the subject is on the screen. You can

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a video camera system having a subject tracking function using moving body detection as an embodiment of the present invention.

FIG. 2 is a flowchart showing a subject recognition processing procedure for performing verification by detecting a moving object.

FIG. 3 is a schematic diagram for explaining a subject feature model setting process in the subject recognition processing procedure.

FIG. 4 is a schematic diagram for explaining a method of extracting a motion amount based on a difference between frames as a motion test.

FIG. 5 is a schematic diagram for explaining a method of extracting a motion amount based on a difference between frames as a motion test.

FIG. 6 is a block diagram showing a configuration of a video camera system having a moving body detection from an initial image as an application of the video camera system of FIG.

FIG. 7 is a flow chart showing a moving body detection processing procedure from an initial image.

FIG. 8 is a schematic diagram for explaining change point extraction from an initial image and frame display.

FIG. 9 is a flow chart showing a scene change detection processing procedure.

10 is a block diagram showing a configuration of a video camera system that autofocuses focusing on a region tracked by a target object as an application of the video camera system of FIG.

FIG. 11 is a schematic diagram used for explaining whether each pixel is a candidate point of a subject.

FIG. 12 is a schematic diagram used for explaining a processing process for calculating a detection output from a Y signal in a subject region.

FIG. 13 is a characteristic curve diagram showing an evaluation curve showing the relationship between the lens position and the evaluation value.

[Explanation of symbols]

1 ... Lens block, 2 ... Lens, 3 ... Iris, 4 ... Imaging element, 5 ... Signal separation / automatic gain adjustment circuit (AGC), 6 ... Analog / digital conversion circuit, 7 ... Digital camera processing circuit, 8 ... Digital / analog conversion circuit section, 9 ... Tracking control circuit section, 10
…… Image memory, 11 …… Tracking signal processing circuit, 12 ……
Panning drive motor, 13 ... Chilling drive motor, 14 ... Frame setting circuit, 15 ... Recording control circuit, 16 ... Video signal detector, 17 ... Video signal processing circuit, 18 ... Focus motor.

Claims (7)

[Claims] 1. An input image signal is converted into a digital signal,
The image storage control means for periodically storing the image data in the image memory, the image data stored in the image memory, and the delayed image data delayed by a predetermined time from the image data Regarding the luminance and color pixels, a characteristic pixel extraction unit that extracts pixels whose change in luminance is equal to or greater than a predetermined threshold value, and the number of the pixels extracted by the characteristic pixel extraction unit is counted, and An image signal processing apparatus, comprising: a moving body detecting means for detecting a moving body. 2. The image signal is input from a television camera, and the characteristic pixel extracting means relates to the pixels of the arbitrary characteristic luminance and color when the angle of view of the television camera is unchanged. The image signal processing apparatus according to claim 1, wherein the pixels having a change amount of luminance equal to or larger than a predetermined threshold value are extracted. 3. A feature quantity setting means for setting a feature quantity relating to luminance and color information of a target object in the image data by using the image storage control means for periodically storing the image data in the image memory. A feature model determining unit that determines a feature model indicating the feature of the target object from the set feature amount, and a pixel that is a candidate for the target object from the image data using the determined feature model. A pixel determination unit that detects the horizontal and vertical center of gravity of the detected pixel set, coordinate temporary determination unit that detects the position of the target object in the image data, and a moving object in the image data. Using the moving body detecting means for detecting, the position verification means for verifying the certainty of the position of the detected target object, and the target object verified by the position verification means Depending on the location, comprising a camera control means for controlling the television camera, the image signal processing apparatus according to claim 2, characterized in that tracking the target object in the image data. 4. An initial image storage means for storing the entire initial image at the time of activation with respect to luminance and color information in the image memory by using the image storage control means for periodically storing the image data in the image memory. A moving point detecting means for detecting a pixel whose luminance change is equal to or more than a predetermined threshold value from the initial image stored in the image memory by using the moving body detecting means for detecting a moving body in the image data; From the detected pixels, a change area detecting unit that detects a changed area from the initial image, and when the change area from the initial image is not detected by the change area detecting unit, the initial image is displayed. The image signal processing apparatus according to claim 1, further comprising an initial image resetting unit that resets the image every fixed time. 5. A recording control means for entering an image recording state only when a changed area is detected from the initial image by using the changed area detecting means for detecting a changed area from the initial image, The image signal processing apparatus according to claim 4, further comprising a frame display unit that superimposes a frame display surrounding the changed area on a video signal. 6. A correlation detecting means for detecting a correlation between two continuous image data by using a moving body detecting means for detecting a moving body in the image data, and the input by the correlation detecting means. 2. The image signal processing apparatus according to claim 1, further comprising a scene change detecting unit that detects a change in the image signal. 7. A target object area setting means for setting an area including the target object extracted from the image data, and a space of the image data in the target object area set by the target object area setting means. The image signal processing apparatus according to claim 2, further comprising a lens control unit that controls a focus lens of the television camera so as to increase a high frequency component.