JPH06350397A - Interpolation circuit and electronic zoom equipment provided with it - Google Patents

Abstract

PURPOSE:To provide an excellent frequency characteristic and to reduce an interpolation arithmetic operation quantity by using an interpolation function generated by connecting plural quadratic functions in the interpolation circuit to output by converting a sampling interval to provide an output. CONSTITUTION:A control signal generator 45 calculates an address signal of picture memories 41, 43 and an interpolation coefficient of both a vertical interpolation filter 42 and a horizontal interpolation filter 44 based on data representing a zoom position and a zoom magnification outputted from a microcomputer 5 to provide an output. An interpolation relation used for the arithmetic operation of the interpolation coefficient uses functions formed by connecting quadratic functions to satisfy f(0)=1, f(n)=0, SIGMAf(n-a)=1, where n is an integer other than 0 and SIGMA is a sum with respect to the (n). Thus, the deterioration of frequency characteristic is less and the interpolation coefficient is obtained with less arithmetic operation quantity.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an interpolation circuit suitable for an electronic zoom device such as a camera-integrated VTR.

[0002]

2. Description of the Related Art In a consumer-use VTR with a built-in camera, an electronic zoom function using digital interpolation processing has been realized as a part of higher functionality. FIG. 7 is an explanatory diagram of such an electronic zoom function, and shows a case of magnifying 4 times horizontally and vertically.

FIG. 7 (a) is an explanatory view of an operation for vertically enlarging by 4 times, in which a hatched portion of an original image is cut out and adjacent pixels Y1, Y2, Y3 and Y4 in the vertical direction are cut out. By interpolating 3 pixels each between the pixels, the pixel is expanded four times in the vertical direction.

The image interpolated in the vertical direction is next shown in FIG.
As shown in (b), as in the vertical direction, three pixels are generated by interpolation between adjacent pixels of the pixels X1, X2, X3, and X4 in the horizontal direction.

FIG. 8 is an explanatory diagram of the interpolation operation using the interpolation function f (x) = sinx / x. Here, x is a distance in the pixel direction, and is a pixel (P (m-1), P (m), P (m +
1), P (m + 2), etc.) is set to 1. Then, the pixel Q (n) generated by interpolation is an infinite number of pixels P (m) and P (m) existing before and after in the horizontal direction.
+1), P (m-1), P (m + 2), P (m-2),
It is created based on P (m + 3), ... (Only 4 pixels are shown). This interpolation algorithm is Q [n] = f (−a) · P [m], where a is the distance between the pixel Q (n) generated by interpolation and the pixel P (m) before it. + F (-a + 1) · P
[M + 1] + f (-a-1) * P [m-1] + f (-a
+2) ・ P [m + 2] + f (-a-2) ・ P [m-2]
+ F (-a + 3) * P [m + 3] + ... In addition, Q [n] and P [m-2] to P [m +
3] represents the levels of the pixels Q (n) and P (m−2) to P (m + 3).

[0006]

As described above, it is ideal to use f (x) = sinx / x as an interpolation function and to interpolate based on an infinite number of pixels, but in reality, simplification is required. Therefore, as an interpolation function, Nearest-Neighbor [f (x) = {1
(0 ≦ | x | ≦ 0.5), 0 (0.5 <| x
|)}], Bi-Linear [f (x) = {1- | x
| (0≤ | x | ≤1), 0 (1 <| x |)}], C
sub-Convolution [f (x) = {1-
| X | + | x | (1- | x |) ² (0 ≦ | x | ≦
1), (1- | x |) (2- | x |) ² (1 <| x |
≦ 2), 0 (2 <| x |)}] and the like before and after the pixel Q (n) 4
Pixels P (m-1), P (m), P (m + 1), P
Interpolation is performed based on (m + 2).

Therefore, Nearest-Neighb
In the case of or, the frequency characteristic of the image signal after interpolation is as shown in FIG. 9, and the gain of the harmonic is larger than −30 dB with respect to the gain of the DC component, so that the problem that the false signal cannot be suppressed sufficiently There was a point.

Similarly, in the case of Bi-Linear, as shown in FIG.
However, there is a problem that the frequency near ½ of the sampling frequency fs, that is, the level of the high frequency component of the baseband image signal decreases.

Further, Cubic-Convoluti
On is as shown in FIG. 11, and the frequency characteristic is good, but there is a problem that the amount of calculation of the interpolation coefficient increases.

The present invention has been made in order to solve such a problem, and provides an interpolation circuit having a good frequency characteristic and a small amount of interpolation calculation, and an electronic zoom apparatus including the same. The purpose is to

[0011]

In order to solve the above-mentioned problems, the present invention provides a plurality of secondary circuits in an interpolation circuit which inputs digital data having a predetermined sampling interval, converts the sampling interval and outputs the digital data. The interpolation function created by connecting the functions is used. As the interpolation function, for example, f (x) = {1- | x | ² (0 ≦ | x |
≤1), (1- | x |) (2- | x |) (1 <| x |
≦ 2), 0 (2 <| x |)} (provided that the predetermined sampling interval = 1).

The interpolation circuit of the present invention can be used in an electronic zoom device such as a camera-integrated VTR.

[0013]

According to the present invention, since the interpolation function created by connecting the quadratic functions is used, the Nearest-Neig
It is possible to obtain better frequency characteristics than those of hbor and Bi-Linear. Also, Cubic-Convo
The interpolation coefficient can be obtained with a smaller amount of calculation than the solution. For example, if the above-mentioned f (x) is used as the interpolation function, only one multiplication is required to calculate the interpolation coefficient.

[0014]

[F (x) = {1- | x | ² (0 ≦ | x | ≦ 1), (1
− | X |) (2- | x |) (1 <| x | ≦ 2), 0 (2
<| X |)}]

 It was adopted.

FIG. 2 is a diagram showing frequency characteristics of an image signal interpolated by this interpolation function. This frequency characteristic is characterized by a high gain near fs / 4. This means that the deterioration of the frequency characteristic of the original image signal can be suppressed and the contour of the image is emphasized. Further, since the gain of the component of fs or more is almost -30 dB or less, it is possible to suppress the false signal.

FIG. 3 is a block diagram showing an example of the structure of the interpolation filter. This horizontal interpolation filter has four latches 11, 12, 13, 14 connected in cascade, and P [m + which is the output of these latches 11, 12, 13, 14.
2], P [m + 1], P [m], P [m-1] and the interpolation coefficients f (-a + 2), f created by the control signal generator 45.
The sum of the outputs of the multipliers 15, 16, 17, 18 and the multipliers 15, 16, 17, 18 for calculating the product of (-a + 1), f (-a), f (-a-1) and Adder 1 to calculate
It is composed of 6. Then, the latch 11 stores the level of the pixel read from the image memory.

Next, the interpolation operation will be described with reference to the operation timing chart of FIG. Here, the operation when three pixels are created between each of the pixels P (1), P (2), P (3), and P (4) before interpolation will be described. Also,
At the time when the system clock CLK rises at time t0, as shown in FIG.
2, 13 and 14 are the level P of the pixel before interpolation, respectively.
[3], P [2], P [1], P

It is assumed that [0] is stored.

First, at time t0, the system clock CLK
Rises, the pixel levels P [3], P [2], P [1], P output from the latches 11, 12, 13, 14 are output.

[0] is supplied to multipliers 15, 16, 17, and 18, respectively, where f (-a + 2), f (-a + 1), and f
The product of (-a) and f (-a-1) is calculated, but at this point a = 0.00, therefore

Q [n] = f (2) · P [3] + f (1) · P [2] + f (0) · P [1] + f (−1) · P

[0] = 0 × P [3] + 0 × P [2] + 1 × P [1] + 0 × P [0] = P [1], and the level P [1] of the point P (1) before interpolation is It is output as is.

Next, at time t1, the system clock CLK
Rises, the outputs of the D-latches 11 and 14 are the same as at t0, but since a = 0.25, Q [n] = f (1.75) .P [3] + f (0. 75) ・ P [2] + f (-0.25) ・ P [1] + f (-1.25) ・ P

[0] = (− 3/16) × P [3] + (7/16) × P [2] + (15/16) × P [1] + (− 3/16) × P [0] Become.

Further, at time t2, the system clock is C
When LK rises, a = 0.50, so Q [n] = f (1.50) · P [3] + f (0.50) · P [2] + f (−0.50) · P [1] + f (-1.50) · P

[0] = (-1/4) * P [3] + (3/4) * P [2] + (3/4) * P [1] + (-1/4) * P

It becomes [0].

Similarly, at time t3, the system clock becomes C
When LK rises, a = 0.75, so Q [n] = f (1.25) · P [3] + f (0.25) · P [2] + f (−0.75) · P [1] + f (-1.75) · P

[0] = (− 3/16) × P [3] + (15/16) × P [2] + (7/16) × P [1] + (− 3/16) × P [0] Become.

After outputting the outputs of the latches 11 and 14 to the multipliers 15 to 18 at time t3, these latches 11 and 14 are output.
The pixel level P [4] is read from the image memory and supplied to the latch 11 while the enable signal EN to be supplied to As a result, when the system clock CLK rises at time t4, the latch 11 is set to P
[4] is stored, and the pixel levels P [3], P [2], P [1] transferred from the latches 11, 12, 13 at the previous stage are stored in the latches 12, 13, 14.

After that, similarly to the time t0 to t3, the same operation is periodically repeated every four cycles of the system clock CLK, so that the pixels P (2), P before the interpolation are processed.
(3) Interpolate every three pixels between the pixels.

Here, the interpolation function of this embodiment is Cubic.
-The reason why the amount of calculation is smaller than that of Convolution will be described.

Substituting -a-1, -a, -a + 1, -a + 2 into the function of Cubic-Convolution and calculating the respective values is as follows. However, b = 1-a here. f (-a-1) =-(ab) b f (-a) = (ab) b + b f (-a + 1) = (ab) a + a f (-a + 2) =-(ab) a

On the other hand, in the function of this embodiment, f (-a-1) =-ab f (-a) = ab + b f (-a + 1) = ab + a f (-a + 2) =-ab

Comparing these, the interpolation function of the present embodiment does not require the operations of (ab) × b and (ab) × a which are required in Cubic-Convolution, so the amount of operation is small accordingly. I understand.

FIG. 5 is a block diagram showing an example of a circuit for calculating an interpolation coefficient. Thus, it can be realized with a simple configuration of three adders, one multiplier and one inverter. Therefore, since the calculation time is short, it is also suitable for moving image processing.

FIG. 6 is a block diagram showing an example of the configuration of the electronic zoom device according to the embodiment of the present invention.

In this figure, the image of the subject is the CCD 1
Is converted into an image signal by the CDS & AGC circuit 2, sampling and gain adjustment are performed, converted into a digital image signal by the A / D converter 3, and supplied to the electronic zoom device 4.

The electronic zoom device 4 includes image memories 41, 4
3, a vertical interpolation filter 42, a horizontal interpolation filter 44, and a control signal generator 45.

The image memory 41 stores one frame of the image signal supplied from the A / D converter 3 in accordance with the write address signal output from the control signal generator 45, and the zoom output from the control signal generator 45. The stored image signal is read according to the read address signal corresponding to the position and the zoom magnification, and is supplied to the vertical interpolation filter 42.

The vertical interpolation filter 42 is constructed as shown in FIG. 3, and creates an interpolation pixel in the vertical direction based on the pixel output from the image memory 41 and the interpolation coefficient output from the control signal generator 45. Output to the image memory 43.

The image memory 43 stores the image signal supplied from the vertical interpolation filter 43 according to the write address signal output from the control signal generator 45, and corresponds to the zoom position and zoom magnification output from the control signal generator 45. According to the read address signal, the image signal is read and supplied to the horizontal interpolation filter 42.

The horizontal interpolation filter 44 is constructed as shown in FIG. 3, and creates horizontal interpolation pixels based on the pixels output from the image memory 43 and the interpolation coefficients output by the control signal generator 45. Output to the D / A converter 6.

The control signal generator 45 calculates the address signals of the image memories 41 and 43, the interpolation coefficients of the vertical interpolation filter 42 and the horizontal interpolation filter 44 based on the data indicating the zoom position and the zoom magnification output from the microcomputer 5. Calculate and output.

In the above embodiment, the case where the magnification to be enlarged is an integer has been described, but the present invention can be enlarged to a magnification other than an integer.

The interpolation function is not limited to the above embodiment, f (0) = 1, f (n) = 0 [where n is an integer other than 0], and Σf (n−a). ) = 1 (where Σ is the sum of n), it may be an interpolation function that connects quadratic functions.

Furthermore, the present invention can be applied not only to image signals but also to conversion of sampling frequencies of digital signals in general.

[0041]

As described above in detail, according to the present invention, since the interpolation function created by connecting the quadratic functions is used, Nearest-Neighbor and Bi
-The deterioration of the frequency characteristic of the signal due to the interpolation is smaller than that of Linear. Also, Cubic-Convolutio
The interpolation coefficient can be obtained with a calculation amount smaller than n.

If the function described in claim 2 is adopted as the interpolation function, only one multiplication is required to calculate the interpolation coefficient, so that the circuit scale in the case of hardware can be small. Further, since the gain is applied to the high frequency portion, the contour of the image is emphasized in the interpolation of the image signal.

Therefore, if this interpolation circuit is applied to an electronic zoom device of a camera-integrated VTR, an electronic zoom device having a good frequency characteristic and a small circuit scale can be realized.

[Brief description of drawings]

FIG. 1 is a diagram showing an example of an interpolation function according to an embodiment of the present invention.

FIG. 2 is a diagram showing frequency characteristics of an image signal interpolated by the interpolation function of FIG.

FIG. 3 is a block diagram showing an example of a configuration of an interpolation filter.

FIG. 4 is a timing chart showing the operation of the interpolation filter of FIG.

FIG. 5 is a block diagram showing an example of a circuit for calculating an interpolation coefficient.

FIG. 6 is a block diagram showing a configuration of an electronic zoom device according to an embodiment of the present invention.

FIG. 7 is an explanatory diagram of an electronic zoom function.

FIG. 8 is an explanatory diagram of an interpolation operation using an interpolation function f (x) = sinx / x.

FIG. 9 is a diagram showing frequency characteristics of an image signal after interpolation in the case of Nearest-Neighbor.

FIG. 10 is a diagram showing frequency characteristics of an image signal after interpolation in the case of Bi-Linear.

FIG. 11 is a diagram showing frequency characteristics of an image signal after interpolation in the case of Cubic-Convolution.

[Explanation of symbols]

4 ... Electronic zoom device, 11, 12, 13, 14 ... Latch, 15, 16, 17, 18 ... Multiplier, 19 ... Adder

Claims (3)

[Claims] 1. An interpolating function created by connecting a plurality of quadratic functions in an interpolating circuit for inputting digital data having a predetermined sampling interval, converting the sampling interval and outputting the digital data. Interpolation circuit. 2. An interpolation function f (x) = {1- | x |
² (0 ≦ | x | ≦ 1), (1- | x |) (2- | x
|) (1 <| x | ≤2), 0 (2 <| x |)}
The interpolation circuit according to claim 1, wherein (where the predetermined sampling interval is 1) is adopted. 3. An electronic zoom device comprising the interpolation circuit according to claim 1.