JPH11155099A - Electronic zoom processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To correctly obtain a transformation rate by setting a transformation picture element rate by means of plural parameters, executing division between the respective picture elements of interpolating data generated by means of picture element transformation and executing approximating calculation through the use of a quadratic function which is calculated by front and rear specified numbers. SOLUTION: Eight-division is executed between the respective picture elements and an interpolating value is calculated by a quadratic curve obtained by three points. At the time of a reduction, input data is written in dual port RAM 4 by 19.6 MHz while executing reduction and interpolation so as to be read by 13.5 MHz. At the time of a magnification, input data is written in dual port RAM 4 by 19.6 MHz to be read by 13.5 MHz while executing magnification and interpolation. Input data is transformed from 886 fh into 710 fh in the picture element by a reducing and interpolating circuit 2, delayed for 1H in a dual port RAM 4 after that and, then, transformed in a frequency so as to be outputted. In the meantime, when the picture element is transformed from 665 fh to 710 fh, frequency transformation is executed in the dual port RAM 4 and, after that, magnifying processing is executed.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an imaging device such as a video camera and an electronic still camera, an editing device for editing a video signal obtained by imaging with the imaging device, and a display for outputting an image represented by the video signal. The present invention relates to an electronic zoom processing device suitable for use in a device or a printing device.

[0002]

2. Description of the Related Art Conventionally, an electronic zoom function for performing a zooming process employs, for example, an interpolation method based on a previous value hold of a pixel or a linear (primary) interpolation in response to a demand for simplification of a hardware configuration. Was output as an enlarged image.

In general, in such interpolation processing, higher-order interpolation processing of higher than quadratic interpolation can obtain better image quality. However, due to the above-mentioned demand, a relatively small amount of calculation can be applied to various patterns. In this case, an image having a small image quality is obtained by the primary interpolation which performs the interpolation processing safely.

An example of the prior art will be described in detail with reference to the drawings. FIG. 8 is a conceptual diagram for explaining the operation of the conventional device.

Referring to FIG. 8, the data to be obtained is calculated by linear interpolation of two points. The distance between the two points is divided into n, and the data to be obtained is x =
m, the coordinates of the two points are (x1, y1), (x
2, y2), the value y _m of y of the data to be obtained can be calculated as follows.

[0006]

(Equation 1)

[0007]

In this case, the interpolation data is based on linear approximation, and accurate interpolation cannot be performed. or,
Since the zoom magnification is determined by the number of divisions n, if a fine conversion ratio is to be set, the number of divisions between two points is increased, and the circuit scale becomes large. Conversely, if the number of divisions is reduced, a fine conversion ratio cannot be set.

[0008]

SUMMARY OF THE INVENTION The present invention has been made to solve such a problem. The present invention solves this problem by dividing each pixel into eight and calculating an interpolation value from a quadratic curve obtained from three points. Do it exactly. Further, the electronic zoom processing device can realize a finer conversion ratio even when a plurality of parameters are provided and the division between two points is fixed at 8.

In an electronic zoom processing device required for frequency conversion of a digital circuit, interpolation is calculated from a quadratic curve of three points, and the two points are divided into eight to set a plurality of parameters, thereby achieving finer details. Pixel conversion can be realized.

[0010]

DESCRIPTION OF THE PREFERRED EMBODIMENTS First, an outline of an embodiment of the present invention will be described.

In an electronic zoom processing device required for frequency conversion of a digital circuit or the like, two points of data before conversion are divided into eight in the time axis direction (hereinafter referred to as X axis), and the converted X axis is determined by a parameter A. Is sampled.

Thus, the zoom magnification can be obtained at 8 / A. However, in this case, detailed settings other than the magnification of 8 / A cannot be made. For example, horizontal 886 pixels → 7
For accurate conversion to 10 pixels, 886 × 8 = 710 ×
A that is A cannot be found.

Therefore, three parameters (A, B, C)
And A + B + C = 710, the following is obtained.

[0014]

[Table 1]

Here, the values of A, B, and C are set so that the values on the X axis before and after conversion are the same. Thus, the converted X-axis values are, for example, 9, 9, 10, 9, 9,
11. . . . Will be chosen like this. Also, 9,10,1
1 can also be changed by a parameter.

FIG. 2 is a conceptual diagram of the interpolation according to the present invention, showing a method of calculating a quadratic curve from three points, dividing the two points into eight, and calculating data to be interpolated from the quadratic curve. .

As for the interpolation, as shown in FIG. 2, first, the two points before conversion are divided into eight, and the number of apparent data is increased eight times.

Then, three points are defined as (x ₁ , y ₁ ), (x ₂ ,
y ₂ ), (x ₃ , y ₃ ), and a quadratic function f connecting them
(X) = ax ² + bx + c is obtained.

At this time, the coefficients a, b, and c divide the distance between the two points into eight, and can be calculated as follows by setting x ₁ = 0, x ₂ = 8, and x ₃ = 16.

[0020]

(Equation 2)

The pixel data after interpolation is obtained by substituting the value of the X axis calculated from the above parameters into the above equation f (x).

Next, embodiments of the present invention will be described in more detail.

FIG. 1 is a block diagram of the frequency conversion. FIG.
, 1 is a data input terminal, 2 is a reduction interpolation circuit, 3
Is an address generation circuit, 4 is a dual port RAM, 5 is an expansion interpolation circuit, and 6 is a data output terminal.

In the case of reduction, input data is written into the dual port ram 4 at 19.6 MHz and read at 13.5 MHz while performing reduction interpolation.

In the case of enlargement, the input data is 19.6M
Write to the dual port ram 4 at 13.5 Hz and read at 13.5 MHz while performing magnification interpolation.

In FIG. 1, the input data is obtained by converting the pixels from 886 fh to 710 fh by the reduction interpolation circuit 2, delaying the pixel by 1H in the dual port ram 4,
Perform frequency conversion and output.

On the other hand, when converting a pixel from 665fh to 710fh, after performing frequency conversion by the dual port ram 4, enlargement interpolation processing is performed.

Next, the processing of reduction interpolation and enlargement interpolation will be described.

FIG. 3 shows interpolation from 886 pixels to 710 pixels. First, a quadratic curve equation passing through three points is calculated.
The data to be interpolated is divided into 8 by dividing
Read every unit. This value to be read can take values of 11 or 9 for fine adjustment. All of these values can be changed as parameters.

In the case of the reduction interpolation, first, from FIG. 3, the interval between input pixels is divided into eight in the X-axis (time axis) direction, and a quadratic curve equation passing through three points is obtained. In FIG. 3, the reduction conversion from 886 to 710 is performed by dividing into eight.

The data to be read is calculated by inputting the value of X into the quadratic curve. In this case, the size is always reduced to 10
Although reading is performed every other time, a fine adjustment term is provided in order to perform accurate conversion from 886 to 710, and when a certain condition is satisfied, the reading interval (X-axis value) is changed to 11 or 9.

One condition is when the following equation is satisfied:
The values of A, B, and C are given by parameters, for example. Specifically, a counter is provided, and when the value satisfies A, B, C, a carry is issued, the carry is monitored, and the X-axis value is adjusted and determined.

709 = A + B + C 858 × 8 = (10 × A) + (11 × B) + (9 × C) FIG. 4 shows the interpolation from 665 pixels to 710 pixels. First,
Compute the quadratic curve equation passing through the three points. The distance between the two points is divided into eight, and the data to be interpolated is read out every seven units. This value to be read can also take the value of 8 or 9 for fine adjustment. All of these values can be changed as parameters.

In the case of enlargement interpolation as well, FIG.
After calculating the quadratic curve equation passing through the point, the value of X is input and calculated. In this case, it is always read out every 7th place,
In order to perform an accurate conversion from 65 to 710, similarly, if a fine adjustment term is provided and a certain condition is satisfied, the reading interval is set to 8 or 9
Change to

In FIG. 4, the enlargement conversion from 665 to 710 is performed by dividing into eight.

One condition is when the following equation is satisfied:
The values of A, B, and C are given by parameters, for example, and are the same as in the method of reduction.

[0037] 710 = A + B + C 886 × 8 = (7 × A) + (8 × B) + (9 × C) On the other hand, the quadratic curve passing through three points 3 points _{(x 1, y 1),} (x
₂ , y ₂ ) and (x ₃ , y ₃ ), the following can be written.

[0038] than ^{_{y 1 = ax 1 2 + bx}} 1 + c y 2 = ax 2 2 + bx 2 + c y 3 = ax 3 2 + bx 3 + c above three equations, and _{x 1 = 0, x 2 =} 8, x 3 = 16 When you put it,

[0039]

(Equation 3)

Here, A = y ₁ -2y ₂ + y ₃ , B = 4y ₂
Placing the _{-y 3 -3y 1, f (x} ) = A (x 2/128)
+ B (x / 16) + c. From FIG. 2, x
Since it is assumed that up to ^{0~8, x 2/128, x} /
16 is as shown in the following table. By shifting and adding the values of A and B, f (x) can be obtained without using a multiplier.

[0041]

[Table 2]

At this time, the values of A and B are signed data taking a positive number and a negative number. Here, in order to simplify the circuit, the signed data is offset to unsigned data, shifted, and returned again with the offset (FIG. 5). In this case, a positive number is correctly converted, but a negative number is, for example, -1 to 1 /
Even if 2 (1 bit shift), the error does not become 0 and remains at -1. However, in order to simplify the circuit, it is rounded to 0 and processed.

From the above, an example of parameters given at the time of enlargement (665 → 710) and at the time of reduction (886 → 710) is as follows.

[0044]

[Table 3]

FIG. 6A and FIG. 7A are block diagrams thereof.

FIG. 6A is a block diagram of the reduced interpolation according to the present invention, and FIG. 6B is a diagram for explaining the operation. FIG.
In (a), 10 is a first counter, 11 is a second counter, 12 is a selector, 13 is a bit converter, 14
Is a function generator, 15 is a function calculator, and 16 is a dual port ram.

Zoom, z1, z2 for determining the zoom magnification
And the parameters c1 and c2 that determine the timing of using this, x is calculated in the time axis direction, and an interpolation value is obtained based on the value.

The parameters selected by the zoom magnification are input to the first counter 10, the second counter 11, and the selector 12. In addition to the parameter input, the selector 12 outputs the output of the first counter 10 and the second counter 1
1 is input, and the variable Z is input by three inputs.
One of OOM, Z1 and Z2 is selected and bit converter 13
Is input to

The bit converter 13 calculates a division difference based on a variable selected for eight pixel divisions, and outputs the calculated difference to the function calculator 15 as an interpolation coefficient.

On the other hand, Y, B-
Inputs such as Y and R-Y are input to a function generator 14, which sequentially generates a function based on the three input points and outputs the function to a function calculator 15.

The function calculator 15 performs a function calculation using the generated function and the complement coefficient, and writes the reduced complement data into the dual port ram 16 at 19.6 MHz, for example.
By reading at 3.5 MHz, a reduced interpolation output is obtained.

FIG. 7A is a block diagram of the enlarged interpolation of the present invention, and FIG. 7B is a diagram for explaining the operation. FIG.
In FIG. 6A, each block is the same as in FIG. 6A, and a description thereof will be omitted.

In FIG. 7A, in the case of enlargement interpolation,
For example, the input is written to the dual port ram 16 at 13.5 MHz, and the data read at 19.6 MHz is input to the function generator 14, and the output of the function calculation 15 is used as the enlarged interpolation output. Is the same as that described for the reduced interpolation described above.

In the case of enlargement, processing is performed after reading from the dual port ram 16.

[0055]

As described above, according to the present invention, an arbitrary pixel conversion ratio can be realized by setting the conversion pixel ratio using a plurality of parameters. In addition, the interpolation data required at that time can be accurately and easily made by setting the number of divisions between pixels to eight and obtaining a quadratic curve from three points.

[Brief description of the drawings]

FIG. 1 is a frequency conversion block diagram of the present invention.

FIG. 2 is a diagram illustrating the concept of interpolation according to the present invention.

FIG. 3 is a diagram showing interpolation from 886 pixels to 710 pixels in the present invention.

FIG. 4 is a diagram showing interpolation from 665 pixels to 710 pixels in the present invention.

FIG. 5 is a diagram illustrating shift operation (division) of signed data in the present invention.

FIG. 6A is a block diagram of a reduction interpolation according to the present invention, and FIG. 6B is a diagram illustrating an operation.

FIG. 7A is a block diagram of the enlargement interpolation of the present invention, and FIG. 7B is a diagram for explaining the operation.

FIG. 8 is a conceptual diagram of conventional interpolation.

[Explanation of symbols]

 1 data input terminal 2 reduction interpolation circuit 3 address generation circuit 4 dual port ram 5 enlargement interpolation circuit 6 data output terminal

Claims (1)

[Claims] In an electronic zoom processing device that performs frequency conversion by a digital circuit, the ratio of converted pixels is set by a plurality of parameters, and each pixel of interpolation data generated by pixel conversion is divided into eight parts. An electronic zoom processing device wherein an approximate calculation is performed using a quadratic function calculated from three points.