JPS6068792A - Y/c separating circuit - Google Patents

Abstract

PURPOSE:To obtain a Y/C separation circuit in which an optimum filter is decided definitely by applying weighting so as to minimize an error of a luminance signal and a chrominance carrier signal to a true value thereby separating the Y/C. CONSTITUTION:A digital composite NTSC color television signal sampled by a frequency of 4fsc is fed to an input terminal 1. The signal is fed to a series circuit comprising delay circuits 2-4 of one sample and a series circuit comprising delay circuits 9, 19 of one line respectively. Moreover, an output of the delay circuits 9, 19 is fed to a series circuit comprising delay circuits 10-13 and a series circuit comprising delay circuits 20-30 respectively. After each output of the delay circuits are extracted as shown in Fig. and weighted, addition/ subtraction is executed by an addition circuit or a subtraction circuit, the luminace signal Y is extracted from an output terminal 31 or the chrominance carrier signal C is extracted from an output terminal 32.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention This invention relates to a Y/C separation circuit that separates a luminance signal and a carrier color signal from a composite color television signal converted into a digital signal.

"Background technology and its problems" As a digital Y7C separation circuit, color subcarrier frequency fsc (3.58 MHz in NTSC system)
A bandpass filter with a passband centered at
Traditionally used. However, bandpass filters can only use one-dimensional information in the horizontal direction, resulting in incomplete separation. Similarly, the vertical one-dimensional Y/C separation circuit configured with a comb filter has incomplete Y/C separation, resulting in large errors with respect to the true values of the separated luminance signal and carrier color signal. There was a point. Furthermore, the horizontal Y
A two-dimensional Y/C separation circuit in which a /C separation circuit and a vertical Y/C separation circuit are connected in cascade merely performs processing that repeats one-dimensional processing, and does not perform separation based on original two-dimensional information. It was not possible to solve the incompleteness of processing, which is essentially one-dimensional.

Further, conventional filters for Y/C separation are designed in the frequency domain, and the filters are designed through repeated operations based on experience from the frequency characteristics of input and output signals. Unfortunately, the design method was based on so-called know-how, making it difficult to create a unified method.

``Object of the Invention'' This invention provides a fixed weighting coefficient that minimizes the error with respect to the true value of a separated luminance signal or a carrier color signal, which is identified in advance by the least squares method. If two-dimensional Y/C separation is performed using coefficients, conventional Y
It is an object of the present invention to provide a Y/C separation circuit in which imperfections in /C separation are eliminated.

Our goal is to provide a Y/C separation circuit in which the optimal fill can be uniquely determined once the size of the hardware that can actually be used is determined, that is, there is a unified design method. The purpose is to

"Summary of the Invention" The present invention is a non-separable two-dimensional Y/C separation circuit that performs vertical and horizontal separation processing at the same time. This invention is based on factors such as the size of the substrate and processing speed.
The processing range (called the order) is determined according to the scale of the software. MΔ(Moving Ave
Applying actual color television data to this model, the optimal weight coefficients are identified by the method of least squares. This optimal weighting performs Y/C separation by multiplying a predetermined number of pixel data corresponding to the order by a coefficient.

Furthermore, the present invention is a Y/C separation circuit in which the weight coefficients are approximated by integers and multipliers are omitted, if necessary.

"Example" When an NTSC composite color television signal is converted into a digital color television signal with a sampling pulse having a predetermined phase and a frequency of 4 fsc (fsc: subcarrier frequency), one pixel data and a two-dimensional array are converted. There is a strong correlation with each pixel data in the vicinity of the inside. Therefore, Y/C separation can be performed using a plurality of peripheral neighboring pixels in a two-dimensional array of pixel data to be subjected to Y/C separation. An embodiment of the present invention will be described in which Y,/C separation is performed using peripheral neighboring pixels in a two-dimensional array.

For each pixel, the luminance signal y(i,) in the first line sample I-t is divided into a plurality of peripheral neighboring pixels Zl+Z2+Z3+... in the two-dimensional array.
It can be considered as a linear combination with +Zn from its correlation power (yumi).In other words, weighting is given by the coefficients al+82+23+..., all, then the estimated luminance signal - y(1 , j) is Y(i, j) = at Zt + a2Z2-1-
It can be written as a3Z3 + .--+anZn. Which weights are the coefficients al, a2,
a3...all are identified in advance using the least squares method. In other words, it was filmed by a television camera.
Applying the actual data of a certain image to be visualized, the surrounding neighboring pixel data is multiplied by a weighted coefficient, and the combined estimated luminance signal and the luminance signal before color encoding are calculated as shown in the above formula. The error (Y-Y
) The weighting coefficients such that 2 is the highest are calculated using a computer.

For the picture that becomes f-yen for one sheet (Q, let the error be e.

data is obtained, and using these data, the weighting when the error variance is the smallest is the coefficient al l a2 + a3
...and calculated by computer.

Similarly, for example, using three pictures as data, find the coefficients to find the optimal weight profit for each picture, and calculate the weight of these three pictures by 11 digits (5
) The average value of the λ number (not shown below) is identified in advance in the Y/C separation circuit, and the Y/C separation circuit is realized by using hardware to which this weight (1j coefficient) is applied. .

The configuration of an embodiment of the present invention that performs the above processing using a 9-order MA model will be described with reference to the drawings. FIG. 1 is an enlarged view of a two-dimensional array of pixel blocks in which a composite color television signal of the NTSC system is ring-ringed with a Zamplin-Danoculus having a frequency of 4 fsc. Each pixel data lined up on the same line is arranged at an interval of Px (1/4j'sc when sampled with a sampling pulse of frequency 4fsc),
Data indicated by Z5 indicates a pixel take to be subjected to Y/C separation. The pixel data indicated by the white circular dot is Y/
The pixel data shown by the black circular dots is of the opposite phase, and the white triangular dots are of the same phase as the subcarrier in the carrier color signal in the pixel take Z5 that is the target of C separation. It has a phase shift of , and a black triangular phase shift of .

Zl-Z4 and Z6 to Z9 indicate neighboring and peripheral pixel takes used for pixel data Z5 to be subjected to Y/C separation. For pixel data Z5 to be subjected to Y/C separation, pixel takes Z4 and Z6 are respectively 2 on the same line.
Pixel takes separated by a sample are shown, and Z2 and Z8 show pixel takes one line before and after.

Z1 and Z3 indicate pixel data that is two samples apart from pixel data Z2 one line before, and Z7 and Z9 indicate pixel data that is two samples apart from pixel data Z5 one line later. These eight neighboring pixel data ZI-Z4 and Z6 to Z9 have subcarriers in their carrier color signals that are in the same phase or in opposite phase with respect to the pixel data Z5 to be subjected to Y/C separation. . These peripheral neighboring pixel data Zl-Z4 and Z6 to Z9 are Y/
Since there is a strong correlation with the pixel data Z5 to be subjected to C separation, the luminance signal Y in the pixel data to be subjected to Y/C separation can be expressed as a linear combination with peripheral neighboring pixels, as described above. In other words, the weighting is the coefficient a+, a
Let z, a3°..., a9 be 7, and let Y be the estimated luminance signal.

'l == al zl -1-az Z2 +a3
It can be expressed as z3 + ·−−·−10a9Z9. The weighting coefficients a] to a9 are calculated in advance using the method described above.

The inventor of the present application used the above-mentioned method to subtract the weighting coefficients, and in order to simplify the real coefficients and the software used, the coefficients obtained by approximating the real coefficients by integers were used. Shown in Figure 2. By multiplying this integer approximation value as a weighting factor by "-j" by one peripheral neighboring pixel data Zl to Z4, Z6 to Z9, and pixel data Z5 to be subjected to Y/C separation, and composing these values. , Y/C separation can be performed.

FIG. 3 shows the configuration of an embodiment of the present invention in which the above-described processing is performed, and an input terminal 1 is supplied with a digital composite NTSC color television signal sampled at a frequency of 41 sC.

In one embodiment of the present invention, a weight coefficient approximated by an integer is used to separate luminance signals according to the following equation.

Y=-T-g(Zl +2Z2 Z3 +224 +1
225 +2Z6Z7 +2ZgZ9 ) This human input signal is passed through a delay circuit 2゜3.4 with one sample each.
5 in series and one line of delay circuits 9 and 19 in series. When the input signal supplied through input terminal 1 is pixel take Z9, the delay circuit 2.3.4.5
Since the data is delayed by one sample each by the series connection of , pixel data Z8 is taken out from between the stages of delay circuits 3 and 4, and pixel data Z7 is taken out from delay circuit 5, respectively.

By inverting the pixel take Z9 supplied to the input terminal 1, the pixel data Z9 is multiplied by -1 and supplied to the addition circuit 7. On the other hand, pixel data Z8 taken out from between the stages of delay circuits 3 and 4 is supplied to a multiplication circuit 6, which multiplies the pixel data Z8 by 2 and supplies it to an addition circuit 7. The multiplication circuit 6 is constituted by a shift register, and the multiplication circuit 6 is configured by a shift register, and is
This is used to perform bit shift I. The output of adder circuit 7 is supplied to adder circuit 8 . The pixel data Z7 taken out from the delay circuit 5 is inverted, multiplied by -1, and then supplied to the adder circuit 8. The output of adder circuit 8 is supplied to adder circuit 27 .

By connecting the line delay circuits 9 and 19 in series, pixel data Z6 is extracted from between the stages of the delay circuits 9 and 19.

The output from the delay circuit 9 is a delay circuit 10 whose output is one sample.
11, 12, and 13 are each delayed by one sample, and the Y/C
Pixel data Z5 to be separated is taken out, and pixel data Z,+ is taken out from the delay circuit 13. Pixel data Z6 taken out from between the stages of delay circuits 9 and 19 is supplied to 1 multiplication circuit 16 composed of a shift register,
The pixel data Z6 is multiplied by 2 and supplied to the adder circuit 17. Y taken out from between stages of delay circuits 11 and 12
/' Pixel data Z5 to be subjected to C separation is multiplier circuit 1
4, the pixel data Z5 is multiplied by 12, and the pixel data Z5 is supplied to the adder circuit 17. The multiplication circuit 14 is composed of a shift register for shifting 3 bits to the left, a shift register for shifting 2 bits to the left, and an adder. The output of adder circuit 17 is supplied to adder circuit 18 .

The pixel data Z4 taken out from the delay circuit 13 is supplied to a multiplication circuit 15 composed of a shift register, the pixel data Z4 is multiplied by 2, and the pixel data Z4 is supplied to an addition circuit 18. The output of the adder circuit 18 is supplied to the adder circuit 27, and the output of the adder circuit 27 is supplied to the adder circuit 28.

(2) By connecting the line delay circuits 9 and 19 in series, pixel data Z3 is taken out from the delay circuit 19.

A delay circuit 20 in which the output from the delay circuit 19 is 1 time.
21. 22 and 23 are connected in series, and are delayed by one sample each, pixel data Z2 is taken out from between the stages of delay circuits 21 and 22, and pixel data Z is output from delay circuit 23.
l is taken out. The pixel data Z3 taken out from the delay circuit 19 is inverted, the pixel data Z3 is multiplied by -1, and the pixel data Z3 is supplied to the addition circuit 25. Delay circuit 21. 22
The pixel data z2 taken out from between the stages is supplied to a multiplication circuit 24 composed of a shift register, multiplied by 2, and supplied to an addition circuit 25. The output of the adder circuit 25 is
The signal is supplied to an adder circuit 26. Take IJ from delay circuit 23
” and fc tl-! ! The i-element data 1 is inverted, the pixel data Z1 is multiplied by -1, and the result is supplied to the adder circuit 26.

The output of adder circuit 26 is supplied to adder circuit 28 .

The output of the adder circuit 28 is supplied to a 1/16 multiplier circuit 29. The 1/16 multiplication circuit 29 is constituted by a shift register that performs 4 bits to the right. 1
/16 multiplication circuit 29 converts the original weighting coefficient to a value 16 times the weighting coefficient multiplied by each of the pixel data Z1 to Z9, thereby obtaining the luminance signal Y and outputting it to the output terminal 3'. 1 and is also supplied to the subtraction circuit 30. One of the subtraction circuits 30 is supplied with pixel data Z5 to be subjected to Y/C separation from between the stages of the delay circuits 11 and 12. The subtraction circuit 30 subtracts the luminance signal Y from the pixel data Z5 to form a carrier color (i-bow C), and the carrier color A, No. 1 C is taken out from the output terminal 32.

FIGS. 4 and 5 show 15th and 45th order MA models, respectively. These 15th-order and 45th-order "warming weight f" coefficients are calculated using the same method as described above. This weighting is achieved by implementing hardware based on coefficients, and Y/C using 15th and 45th order MA models, respectively.
One circuit can be constructed. The order of the MA model to be used is determined depending on the scale of the actual hardware.

"Application Example" The present invention can be applied not only when the sampling frequency is 4 fsc, but also when using an arbitrary sampling frequency. 1. In this invention, the luminance signal Y is a one-component combination of surrounding pixels,
The carrier color signal C may be a one-component combination of surrounding pixels.

Furthermore, the present invention is applicable not only to color television signals of the NTSC system but also to color television signals of the PAL system.

"Effects of the Invention" According to this invention, since a Y/C separation circuit based on two-dimensional information is realized, a one-dimensional Y/C separation Mlf circuit based on information only in the horizontal or vertical direction or a one-dimensional Y/C separation circuit based on information only in the horizontal or vertical direction is realized. , horizontal Y/C division 1 (compared to a two-dimensional Y/C division %1 circuit in which a circuit and a vertical Y/C separation circuit are connected in cascade,
ξ The error with respect to the true value of each Y/C separated luminance signal and carrier color signal is small, improving Y/C separation ability ξ
I can do that.

1. This is 1. According to υj, a fixed weight τJ coefficient that minimizes the error 4 to the true value of the separated J luminance signal or carrier color signal is identified in advance by the method of least squares, and this weight is determined by the coefficient 2. Since it is designed to perform a Y/C separation circuit of 300 degrees, once the size of the hardware is determined, the optimal filter can be uniquely determined, and a Y/C separation circuit for which there is a unified design method can be created. (1'4 can be made.

According to the present invention, a Y/C separation circuit with a simple circuit configuration can be realized by using coefficients for weighting approximated by integers.

[Brief explanation of the drawing]

FIG. 1 is a route map showing a two-dimensional redundant array of pixel data in an embodiment of the present invention, FIG. 2 is a route map showing an example of weighted coefficient values, and FIG. 3 is an embodiment of the present invention. The block diagram, FIGS. 4 and 5 are route maps showing one example and another example of a two-dimensional array of pixel data to which the present invention can be applied, respectively. 1... Input terminal, 9.19... L line delay circuit, 31... Luminance signal output terminal, 32...
Output terminal for carrier color signal. Agent Masatomo Sugiura 4th 0 △ 0 0 Δ ○ ○ △ O 5th 0 Δ0 Δ 0 0 ΔO△ 0 0 △0Δ 0 0 Δ○ △ 0 0 △ 0 Δ ○ Δ ○ △ 0 Δ ○

Claims (1)

Claims: A Y/C separation circuit for separating a luminance signal and a carrier color signal from a digital composite color television signal. The pixel data to be separated and a plurality of pixel data located at predetermined positions around the pixel data in the two-dimensional array are extracted, and the least square method is applied to each of the extracted pixel data. The predetermined weighting is multiplied by each of the coefficients and this multiplied lI! A Y/C separation circuit configured to obtain a composite signal of II elementary data as a separated output.