JPS5952361A - Arithmetic system for picture signal processing - Google Patents

Abstract

PURPOSE:To set the digit position of a multiplied and accumulated output at the input of each digital picture signal, by 2<n> multiplying said picture signal with an n-bit shift and at the same time 1/2<n> multiplying the coefficient corresponding to said shift. CONSTITUTION:A subtractor 1 calculates the difference between a sharp signal S0 and an unsharp signal U0, and a limiter 2 gives a left shift of 2 bits to satisfy (S0-U0)'=4.(S0-U0) and to be fed to a data selector 3. A controller 4 performs a control so that a selector 3 delivers (S0-U0)' in the timing of sharpness emphasizing processing and that a constant memory 5 delivers a coefficient K'i=1/4Ki for sharpness emphasizing processing respectively. In this timing a multiplier/accumulator 6 calculates K'i.(S0-U0)'. This result of calculation is added with the result of calculation which is set to the accumulator 6 in the previous timing. Thus the sharpness emphasizing processing is carried out.

Description

DETAILED DESCRIPTION OF THE INVENTION In image signal processing using an arithmetic method in image signal processing, particularly when emphasizing the sharpness of an image, sharpness enhancement is performed based on the difference between a sharp signal and an unsharp signal obtained from the image. A commonly used method is to form a signal and add this sharpness enhancement signal to the image signal. If the image is a color image, it is separated using a three-color filter, and each of the obtained three-color image signals is subjected to the above-mentioned sharpening process. The input image signals of three colors are ll (i=1゜2.3
), the sharp signal is represented by 8i (i = 1.2.3), and the unsharp signal is represented by Ui (i = 1, 2e3), then the image signal after sharpening processing is Ii'(i' = 1.2). .3
) is Ii'= It +Ki・(Si−Ui)...
...... It is expressed as (1). In addition, Ki
is a constant parameter that determines the degree of sharpness enhancement.

However, in the sharpness enhancement method using the above equation (1), it is necessary to form sharpness enhancement signals (8i - Ui) for three colors, which has the disadvantage of complicating the processing equipment. Therefore, as a method for simplifying the emphasis method expressed by the above equation (1), a method has been widely used in which the sharpness emphasis signal is represented by one of the three colors. In other words, the sharp signal of one selected color (
If the sharpness emphasis signal is (So-Uo) using the unsharp signal (UO) and the unsharp signal (UO), the image signal Ii' after the simplified sharpening process is Ii'=Ii +Ki・(So -Uo) ・・
・・・・・・・・・・・・(2)

By the way, when the above-described image sharpness enhancement processing is performed using an analog circuit, there are disadvantages such as instability of the analog circuit, remote control, and difficulty in automatic operation using a computer or the like. Further, the sharpness enhancement processing described above may be performed using a digital circuit (for example, Japanese Patent Laid-Open No. 55-142
(described in Publication No. 342),
The disadvantage is that the coefficient, which is the degree of emphasis, cannot be varied. Furthermore, in image processing, it is necessary to multiply color correction signals and other image signals by coefficients, and since the digit positions of these image signals are different, it is necessary to efficiently align the digits of the multiplication results. There are also demands that it not be.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide an arithmetic method for image signal processing that does not have the above-mentioned drawbacks and can also satisfy the requirements for digit alignment.

This invention will be explained below.

The present invention includes a data selector that inputs various digital image signals having different digit positions, such as an image sharpness emphasis signal and a single color correction signal, and a constant memory that stores coefficients to be multiplied by each digital image signal. A timing controller that controls the output timing of the data selector and constant memory, and a multiplication controller that multiplies and accumulates the digital image signals and coefficients output from the data selector and constant memory.
In order to fix the digit position of the accumulated value in the multiplier/accumulator, the size of each coefficient in the constant memory is changed according to the magnification due to the pit shift of each digital image signal. .

That is, the present invention relates to a method of multiplying various digital image signals with different digit positions by coefficients in image signal processing and cumulatively outputting the result. By setting the coefficient to 172n times the value, the digit positions of the multiplication cumulative output can be aligned with respect to the input of each digital image signal. In this way, in the method of this invention, the weights of the two bits of the multiplier are changed so that one is light and the other is heavy, and this is canceled out in the output so that the decimal point is always at the same position, and the digit of the multiplication accumulation is The matching is done automatically.

When the image signal processing circuit is configured with a digital circuit, it is desirable to process masking and selective color correction operations in a time-series manner in one multiplier in terms of cost and compactness. is also used for such processing and the results are accumulated.

In this case, if the digits are aligned on the output side, a shift circuit whose shift width changes depending on the processing must be provided between the multiplier and the accumulator. Since the shift width of the shift circuit is fixed, there is no need for any special circuit elements and it can be done by wiring.For devices that have an integrated structure of a multiplier and an accumulator, it is possible to perform digit alignment on the output side. Therefore, the effects of this invention are enormous.

Next, the present invention will be explained with reference to a specific circuit.

Multiplication and addition are performed in sharpness enhancement processing, but these two
These calculations are also common to other color correction processes. To explain the configuration of the attached diagram, the subtracter 1 receives the sharp signal SO.
and the unsharp signal UO (8o - Uo), and the limiter 2 calculates the difference signal (5o - Uo) within a predetermined range.
o).

The output of the limiter 2 is shifted to the left by 2 bits (4 times) to become (So - UO)' = 4.(So - Uo), and is input to the data selector 3. Further, the timing controller 4 outputs the selection signal 8BL and the address signal ADH so that the data selector 3 outputs (So-UO)' and each constant memo at the timing of sharpness enhancement processing. Then, at this timing, the multiplier/accumulator 6 calculates Ki/.(So-Uo)', and at a previous timing, the multiplier/accumulator 6 calculates Ki/.
By adding this signal to the original image signal set in the accumulator 6, the sharpness enhancement process of equation (2) above is performed. In this way, the image signal f subjected to the sharpness enhancement process is output from the multiplier/accumulator 6.

Here, the constant input to the multiplier is -1 to 127
/128 in steps of 1/128, and in sharpness enhancement processing, it is desirable to take a value of about 0 to 4 as a coefficient, and a step of 1/32 is sufficient. Also, the difference signal between the sharp signal and the unsharp signal (S-
The absolute value of U) is small compared to the values of the sharp signal S and the unsharp signal U, and is mostly the sharp signal S.
and is within 1/4 of the maximum value of the unsharp signal U. Therefore, even if the value exceeding this value is limited, the sharpness enhancement processing will not be adversely affected.

From the above, in this invention, for example, sharpness enhancement processing (UO)
After passing through a limiter, it is shifted to the left by 2 bits (x4), and these two values are input to a multiplier and multiplied.

The timing controller 4 also sends a selection signal S so that the color correction signal and its course keeping Kd are outputted from the data selector 3 and the constant memory 5 at different timings.
EL and an address signal ADH are output, and the multiplier/accumulator 6 performs color correction processing at the same decimal point position as the sharpness enhancement processing described above. The same applies to the input of other signals by the data selector 3, and multiplication and accumulation in the multiplier/accumulator 6 are always performed at the same decimal point position.

As described above, according to the calculation method of the present invention, it is possible to perform sharpness enhancement processing of a digital image and also process other signals, without increasing the bit width of the input (7) of the multiplier. By widening the range of values that the coefficients can take, the decimal point positions can always be aligned, so there is no need to use a shift circuit to align the digits.

Note that in the above, the sharpness emphasis signal (So −Uo) is
We have described an example of shifting the coefficient to the left (by 4) and multiplying the corresponding coefficient by 1/4, but if the range of the coefficient is 2 or less, shift it by 1 bit and prepare a coefficient multiplied by 172 in the coefficient memory. Alternatively, any bit shift may be performed.

[Brief explanation of drawings]

The attached figure is a block diagram showing an example of a device to which the method of this invention is applied. 1... Subtractor, 2... Limiter, 3...-Data selector, 4... Timing controller, 5... Constant memory, 6... Multiplier/accumulator. Applicant's agent Yuzo Yasugata (1)

Claims (1)

[Claims] a data selector for inputting various digital image signals having different digit positions such as an image sharpness emphasis signal and a two-color correction signal; a constant memory for storing coefficients to be multiplied by each of the digital image signals; and the data selector. and a timing controller that controls the output timing of the constant memory, and a multiplication controller that multiplies and accumulates the digital image signal and coefficient output from the data selector and the constant memory.
and an accumulator, and in order to fix the digit position of the accumulated value in the multiplier/accumulator, the size of the valley coefficient of the constant memory is changed according to the magnification by bit shifting of each digital image signal. An arithmetic method in image signal processing characterized by the following.