JPS6028475B2 - correlation tracking device - Google Patents

Description

[Detailed Description of the Invention] This invention calculates the correlation coefficient between a previously stored reference screen and a newly input screen in images supplied from a television camera or an infrared imaging device, and detects the maximum correlation point. The present invention relates to a device for tracking a desired target or scene within the field of view of an imaging device.

Conventionally, this type of device has a width of m pixels (m pixels) as shown in FIG.
Horizontal and vertical correlated screens 1 and 2 are provided in a cross shape with
nVSI) are cut out and stored in memory in advance as reference screens 3 and 4.

Next, the upper reference screens 3 and 4 were scanned horizontally or vertically on the correlated screens 1 and 2 to obtain the correlation coefficient. If we call the area of the reference screen above the tracking window,
In conventional devices of this type, the size of the tracking window is manually set by a person observing the image displayed on a monitor TV and determining the size of the target or scene to be tracked and the detail of the image. Was.

However, in order to obtain the required tracking accuracy, the follow-up method had to manually set several different tracking window sizes and select the size that would result in higher tracking accuracy, which had the disadvantage of being time consuming.

It was also difficult to set the size of the tracking window to an optimal value. In particular, when used outdoors, it is common for the targets and scenes to be tracked to be different, and each time tracking is started (hereinafter referred to as lock-on for convenience), the size of the tracking window must be manually resized. This type of device is complicated to operate.

In order to eliminate these drawbacks, this invention calculates the curvature near the peak value of the correlation coefficient, changes the size of the tracking window so that this value is maximized, and automatically sets the tracking window to its optimal value. This is what I did.

A detailed explanation will be given below with reference to the drawings.

Generally, correlation tracking accuracy depends on the shape of the correlation coefficient near its peak value, and the sharper the correlation coefficient, the higher the tracking accuracy.

That is, the greater the curvature near the peak value of the correlation coefficient, the higher the tracking accuracy. By the way, the curvature depends on the fineness of the image of the target or scene to be tracked, the SNR of the video signal, and the size of the tracking window, so if the fineness of the image and the SNR of the video signal are given, the curvature is determined by the size of the tracking window. There will be a dependency only on

FIG. 2 is a diagram illustrating the operating principle of the correlation tracking device according to the present invention.

In Fig. 2, 1 and 2 are horizontal and vertical correlated screens;
3 and 4 are horizontal and vertical reference screens, and the areas 3 and 4 are called the tracking windows mentioned above. Now, as an initial state, the length of the horizontal and vertical tracking windows is set to m pixels, elbow pixels and nV pixels, and then automatically increased by K pixels (K≧1) on the left and right sides of the tracking window in each direction. =ni=m+2k (i is the number of trials: iZO)
, nv=nJ=m0ak (i is the number of trials: i≧0), and the correlation coefficients C(x) and C(y) at that time are determined.

Here, when determining the curvature of the correlation coefficients C(x) and C(y), the same processing is performed in both the horizontal and vertical directions, so the explanation will be given below taking the horizontal direction as an example. By the way, in this type of device, the maximum correlation value of the correlation coefficient C(x) and its coordinates are detected and used as a tracking point, so it is only necessary to consider the curvature near the peak value of the correlation coefficient.

Therefore, as shown in Figure 3, the coordinate where the correlation coefficient C(x) is maximum is ×p, and the correlation value at that time is C(Xp)
Let X1 and Xu be the coordinates that give the doubled (K<1) correlation value.

Then, the curvature R near the peak value of the correlation coefficient C(x)
Since it is given by the second derivative of C(x) at X=XP, it can be approximately given by the formula [1] with the proportionality constant a (a>0).

a {1) R=The above curvature R is calculated from the correlation coefficient shown in Fig. 2, and the size ni of the tracking window that maximizes the curvature R is the size mH of the tracking window at lock-on. 'Set this.

Note that when there are multiple maximum values of the curvature R, the smallest value is set among the sizes of the multiple tracking windows in order to ensure a wide tracking range.

FIG. 4 is a block diagram of an embodiment of the correlation tracking device according to the present invention.

In FIG. 4, an imaging device 6 mounted on a servo stand 5
After converting the video signal supplied from V into a digital signal through the analog-to-digital conversion circuit 7, the area division circuit 8 divides and extracts horizontal and vertical correlated screens.

The correlated screen is stored in advance in the memory 10th and 10V using switches 9th and 9V for each correlated screen.

Tracking window setting circuit 11 at the time of lock-on
The size of the tracking window is automatically changed using 12V and 11V, and the video signal within that window is input to the reference signal input terminal of the correlator 12V and 12V.

On the other hand, the video signal in the screen to be correlated is
The correlation coefficient is calculated by inputting it to the 2V correlated signal input terminal. The maximum value detection circuit 13 and 13 calculate the calculated correlation coefficient.
Enter the maximum value in V and detect the coordinates that give the maximum value,
The curvature near the maximum value is detected through the curvature detection circuit 14V and 14V.

The above operation is performed by automatically changing the size of the tracking window, its maximum value is detected, and the size of the tracking window at that time is determined.Tracking window setting circuit 11 and 1 1
This is fed back to V and set, and the video in this tracking window is locked on as a reference screen. Note that after lock-on, the size of the tracking window is kept constant and the correlator 12
The correlation coefficient is calculated using the maximum value detection circuit 13 and 12V, and the maximum value and the coordinates giving the maximum value are detected by the maximum value detection circuit 13 and 13V. By moving the correlated screen, a desired target or scene within the field of view of the imaging device 6 can be tracked by an electronic loop.

In addition, the servo mount 5 can be
By outputting an error message to the camera and controlling the attitude of the imaging device 6, a desired target or scene can be kept stationary on the screen and tracked. According to the automatic setting of the tracking window size described above, the curvature of the correlation coefficient is maximized.
In other words, it is possible to realize a case where the correlation coefficient becomes sharp, and the tracking accuracy can be increased. Although the above description has been made of a cross-shaped case in which tracking windows are set in the horizontal and vertical directions to perform real-time processing and the correlation coefficients are determined and tracked independently of each other, this invention is not limited to this. It can also be used when the correlation coefficient screen is determined by scanning the correlated screen in a two-dimensional direction on the correlated screen and using the inside of one rectangular tracking window as the reference screen.

As described above, in the correlation tracking device according to the present invention, the curvature in the vicinity of the peak value of the correlation coefficient is determined at the time of lock-on, and the size of the tracking window is automatically changed to the optimum value so that this value becomes the maximum. This has the advantage of being able to lock on easily and quickly depending on the target or scene to be tracked, improving the operability of the device and increasing tracking accuracy.

[Brief explanation of the drawing]

FIG. 1 is a layout diagram of each screen when determining a correlation coefficient using a cross-shaped correlated screen and a reference screen, FIG. 2 is a diagram of the operating principle of the correlation tracking device according to the present invention, and FIG. FIG. 4, which is an explanatory diagram of the curvature detection method of the correlation coefficient, is a configuration diagram of an embodiment of the correlation tracking device according to the present invention. In the figure, 1 and 2 are horizontal and vertical correlated screens, 3 and 4 are horizontal and vertical reference screens, 5 is a servo mount, 6 is an imaging device, 7 is an analog-to-digital conversion circuit, 8 is an area division circuit, 9th, 9V is horizontal and vertical switch, 10th, 1
0V is horizontal and vertical memory, 11th, 11V is horizontal and vertical memory, 11th, 11V is horizontal and vertical tracking window setting circuit, 12th, 12V is horizontal and vertical correlator, 13th, 13V is horizontal and vertical maximum value detection circuit, 14th, 14V is horizontal and vertical curvature detection circuit, 15
is a tracking coordinate signal output circuit. It should be noted that the same or corresponding parts in the figures are indicated by the same reference numerals. Figure 1 Figure 3 Figure 2 Figure 4

Claims (1)

[Claims] 1. Find the correlation coefficient between the screen obtained by analog-to-digital conversion of the video signal supplied from the imaging device and the previously stored reference screen, detect the maximum correlation point, and select the desired point within the field of view of the imaging device. In a correlation tracking device that tracks a target or scene, before starting tracking, the size of the tracking window that captures the image of the reference screen is sequentially changed to obtain the correlation coefficient, and the curvature near the peak value of the correlation coefficient is detected. A correlation tracking device characterized in that a size of a tracking window that maximizes the detected curvature is determined, an image within this tracking window is stored as a reference screen, and tracking is started.