JPS5894220A - Separating filter - Google Patents

Abstract

PURPOSE:To ensure the good filtering, by using a low-order transversal filter as a BPF which extracts a prescribed band component of a signal. CONSTITUTION:The signal supplied from an input terminal 10 is supplied to a series circuit of delaying circuits 31 and 32. The signal of the input side of the circuit 31 is added 33 to the signal of the output side of the circuit 32. A secondary transversal filter is formed with the circuits 31-33. The output signals of delaying circuits 11 and 12 are supplied to a transversal filter consisting of delaying circuits 34 and 35 and an adder circuit 36 or delaying circuits 37 and 38 and an adder circuit 39. The signals supplied from these transversal filters are supplied to a logical arithmetic circuit 15 as well as to inverters 14 and 18. Then the signal of the output side of the circuit 34 is supplied to an adder circuit 19 and a subtractor circuit 22. Thus the luminance signal is completely eliminated since the luminance components excepting that of a chroma signal band, are attenuated by the circuits 19 and 22. Therefore, the chroma signal is separated from the luminance signal.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a separation filter that separates a plurality of correlated signals based on their respective correlations.

For example, when separating a bright station signal and a chroma signal in a color video signal of the NTSC system, a method of separating using a bandpass filter has been widely used. However, in this method, for example, 3,58'
! Since all the brightness signals included in the J14z±5QQkHz band are also considered to be chroma signals, the bird area component of the brightness signal is mixed into the chroma signal in parts where the s-degree signal changes sharply, resulting in an image called cross color. deterioration occurs.

By the way, in conventional filter devices, a signal that changes in the time dimension is converted into a frequency dimension by Fourier transform, and the frequency component is filtered. In response to this, the inventor of the present application previously proposed a filter device that converts the signal into a pattern space that changes in the time dimension and performs filtering by deforming this pattern.

An example of such a filter device will be described below with reference to the drawings.

First, the pattern space will be explained. In other words, one frame of TV signal is placed horizontally with m and n i! The amplitude of each 1m tree is f (xl・y3 ). However, if 1×i(m, 1×j<n), then the TV 4g issue mentioned above is f(xl, Yj ) to m
It can be regarded as having been arranged in order of Xn=.

For example, F=(f, , t2-...fk) However, -fl = f (xt e yl) fk=f (
xm+Yn), it can be considered as a dimensional vector F for this l-frame television signal 1.

In this way, when signal levels at a plurality of time points are applied to each dimension and vectors are constructed, it is called a pattern space, which is a multidimensional space where K can be obtained.

Similarly, the signal levels ft-1, f of the three adjacent points
A three-dimensional pattern space can be constructed using tvft-z (2<t<k-1).

The m1 diagram is a perspective view of such a three-dimensional pattern, with each vector represented by an arbitrary point in the space encompassed by the maximum level of the signal.

In this three-dimensional pattern space, the line segment connecting the origin O and the point P with the largest vector is ft-1=(1=f1+
1.

In addition, the plane shown in FIG. 2 AK shows that ft = 1'', ftNfi +1. Furthermore, the plane shown in VC in FIG. 2 BK shows that ft = 1 to ft - ft +1, and these This shows that the signal changes in a stepwise manner.

On the other hand, the plane shown in FIG. 2CK shows that ft'=1"ft+1~fi, which indicates that the signal is changing rapidly.

When viewed from above, the three-dimensional pattern space 10-P line looks like FIG. 3. Here, the signals in each range change around the outer periphery as shown. In the figure, the first
The 0-P line in the figure is the origin, the person's plane in Figure 2 is the A axis, the B plane is the B axis, and the C plane is the C axis.

In this case, since the original signals have an extremely strong correlation between adjacent signals, a line from A to B as shown in FIG.
It is concentrated and distributed in the -d line range, and does not exist near the C-C' line. On the other hand, since there is no correlation between adjacent signals of stockpiles, etc., signals are uniformly distributed throughout as shown in No. 41i1BK.

That is, for example, in the case of a signal as shown in FIG. 5, when this is converted into the above-mentioned pattern space, it becomes as shown in FIG. 6.

Note that (0) in Figure 5 indicates the position of the signal.

In this case, the signal outside the diagonally shaded range in FIG. 6, for example, is deformed as shown by the arrow.
Ns can be removed.

This transformation can be done in five rows using, for example, the following five logical operations.

That is, fl' = MAX (MIN (ft 1 , ft
MIN (ft, ft hand 1).

MIN('t:s, ft +1) )...
・fl) a-MIN (MAX('t 1- ft
), MAX('t + ft + 1) - MAX
(ft-1r fl+lid)) - ・ ・ ・(
2) - A deformed performance is performed on the burial muzzle. Here, MAX indicates that the maximum value *1 in the parentheses below is to be extracted, and MIN indicates that the smallest value in the parentheses below is to be extracted.

By performing this logical operation, for example, in the noise N1 part of FIG. 5, f &'Ahlα(MIN (f 1 e f @
) e MIN (f g * f 9 ) e MIN
(r7.fe))=MAX(fm e fs w'v)=f7. In addition, in the noise N2 part, f 11'=MAX (MIN (fl6. fll
), MIN (fll, fim).

MIN(flL'll)) = MAX(fl(1, flg, fH))"fil, and the noise N1 e N! is removed. Even if the noise Ns K is present, it is removed by the same 11 Km as the noise N1. .

On the other hand, for example, in the fI signal, f, I−
第'u(MIN(f4.f, ), MIN(fI,f
, ).

MIN(f4.fm)) = MAX(f4.fs, f4) = fi, and in the f- signal, f@'xMAX(MIN(f6.fI), MIN(
f6. f7).

MIN(f,,f, )) W MAX(f@, f@, fH)=f
As a result, the original signal is extracted as is.

In this way, in the above-described filter device, noise can be removed without degrading the original signal.

Further, Figures 7 to 9 show specific circuit configurations.

First, Fig. 7 shows the overall configuration, with input terminals (delay circuit + 2 having a delay time such that the signal supplied to ll reaches the period of the highest frequency of the original signal), (supply to 31 series circuits). Furthermore, the signal from the input terminal (1) and the output signals from the delay circuits (2) and (3) are
Alternatively, (2) K equivalent logical operation circuits (4) K are supplied. Then, the signal processed by the logical operation is sent to the output terminal (5)
It is taken out.

The logic operation circuit (4) is configured as follows.
, FIG. 8 corresponds to the above equation (1). In this figure, pnp type transistors (51m), (52m
), (51b), (52b), (51c), (52
Three sets of circuits according to c) and K are provided. These collectors are commonly grounded, and two emitters are connected to each other, and each of these connection points is connected to a resistor (53M).
).

(53b), (53e) ik: Through power terminal (b)
connected to.

* Terminals (55a) (ssb), (ssc) to which signals ft-1-ft and 't+1 are supplied are provided, and the terminal (55m) is connected to transistors (51m), (52c),
Terminal (55b) is transistor (52m), (51b)
, the terminal (S5c) is a transistor! (52b).

(51c) are connected to their respective bases. Furthermore, the connection points of the entrants of the transistors (51m) to (52c) are npn type transistors/disp (ssa) and (56
b) + (86c), and the collectors of these transistors (56m) to (56C) are connected to the power supply terminal (
b) and the emitters are connected to each other,
This connection point is grounded through a resistor 67). The emitter connection points of the transistors (56m) to (56c) are connected to the output terminal tAK.

According to this circuit, transistors (51a), (52m
), the smaller signal from the terminals (55m) and (55b) is taken out, and the transistors (51b) and (52b
), the smaller of the signals from the terminals (55b) and (55c) is taken out, and the transistors (51C) and (52e)
The smaller signal from the terminal (55c) and (55) is taken out.
6c) The maximum signal from the transistors (51m) to (52c) is extracted and output to the output terminal 51.

Moreover, FIG. 9 corresponds to the above equation (2)K, and
PNP type transistors (51a) to (52c) in FIG.
) are respectively npn type transistors (511') ~
(52C') K substituted, npn type transistor (
5651) to (56c) are respectively replaced with pnp type transistors (56m') to (56c') by K, and the power supply terminal (b) and ground are reversed. According to this circuit, the larger one of transistors (511) to (52C') is taken out, and the transistor (56m')
~(The smallest of them is output in 56C.

Furthermore, in the KNTSC color video signal, the phase of the subcarrier of the chroma signal is inverted every horizontal period. Therefore, instead of the above signal 't1+'t+'1+IK, the signal fl, the signal ft-H before 1 horizontal period 0, and the signal f after 1 horizontal period 10H are used to form a pattern space like h rW]. do. Here, the luminance signal component (◯) and the chroma signal component (x) are expressed separately.

In this way, these signals are [Fig.

As shown in FIG. 11, the luminance signal component is distributed between the A-axis and the B-axis, and the chroma signal component is distributed near the C-axis. Therefore, in order to extract the chroma signal from the above-mentioned signal, it is sufficient to deform it in the C-axis direction on the pattern space.

Figure 12 shows a circuit for extracting chroma signals in this manner. In the figure, a signal from an input terminal OG is supplied to a series circuit of delay circuits Oυ, 02 each having a delay time corresponding to one horizontal period. In addition, input terminal 0
The signals from I are supplied to a logic operation circuit αS through a bandpass filter 0 and an inverter 0 corresponding to the band on which the chroma signal is superimposed. Also, the signal from the delay circuit αυ is passed through the same bandpass filter αe & Wa as described above.
The signal is supplied to the arithmetic operation circuit 05. Further, the signal from the delay circuit QB is supplied to the logic operation circuit (11) through the same bandpass filter 07) and inverter 0s as described above.

Here, logic operation circuit a! As 9, a circuit equivalent to the logic operation circuit (4) in FIG. 7 described above is used.

Therefore, the scanning lines 11jlk, l, as shown in FIG.
When the signal m is supplied, the signals T, j, and H are supplied to the logic operation circuit aCJ as shown in FIG. Further, at the time when the chroma signal 9 is output to the scanning line, the signal t is supplied to the logical operation circuit αl at 51C as shown in FIG. 13CK. Further, at the time of extracting the chroma signal of the scanning line exposure, the signals of V, t, and m are supplied to the logic operation circuit 05, as shown in FIG. 13DK.

By supplying such a signal, the logic operation circuit 0S outputs a signal J1.K as shown in FIG. 13EK. ', love' is taken out.

This signal is supplied to the adder circuit QIK, and the original signal from the delay circuit aυ is supplied to the bandpass filter αJ.

(The chroma signal from which the luminance signal has been removed is supplied to the adder circuit CIIK through the delay circuit (2) corresponding to the delay characteristic of ILOη, etc., and the luminance signal component is removed.Then, the chroma signal from which the luminance signal has been removed is taken out to the output terminal C!υ.

Furthermore, the chroma signal from the addition circuit α bell is sent to the subtraction circuit (2).
At the same time, the signal from the delay circuit (2) is also supplied to the subtraction circuit @ to remove the chroma signal component. Then, the luminance signal from which the chroma signal component has been removed is taken out to the output terminal (to).

Therefore, according to this circuit, it is clear from FIG.
Furthermore, it is possible to separate the 4fI signal and the chroma signal without image deterioration due to cross color equal pressure.

By the way, in the above-mentioned filter device, the filtering is not ideal, and there is a risk that low-frequency components, especially DC components, are not removed and may leak to the output terminal t2v of the black 1 (No. 11).For this reason, in the above-mentioned circuit, the bandpass Filters (13, αe, αη) are provided to remove low frequency components.

However, in this case, when a signal is supplied to a bandpass filter, the phase is generally distorted around the cutoff frequency. For this reason, if the band of the bandpass filter is narrowed, the phase of the chroma signal will be distorted, the frequency characteristics will deteriorate, and image quality deterioration such as ringing will occur. On the other hand, if the band of the bandpass filter is made wider than necessary, cross color occurs due to leakage of low frequency luminance signal components.

Therefore, in the past, a band pass filter of 3.58 MHz±500 kHz was used (.However, this was not always satisfactory.

More resistors and funds? The bandpass filter according to the above was not easy to manufacture, and the adjustment to obtain the desired frequency characteristics was also complicated.

In view of these points, the present invention has been developed to perform 5K filtering with a simple configuration.

That is, in the present invention, in a separation filter that filters a plurality of correlated signals based on their respective correlations, a low-order transfer ζ-sal filter or the like is used as a bandpass filter for extracting a predetermined band component of the signal. It was established.

An embodiment of the present invention will be described below with reference to the drawings.

In FIG. 14, the signal from the input terminal (Ll) is shown.

3.58MJ-1g color [7 waves J of 8 transmissions [corresponding 1
dQnl! 1) is supplied to the delay circuit C1υ and the series circuit (to). The signals on the input side of the delay circuit c31) and on the output 11 side of the delay circuit 6υ are weighted -, -, -7, respectively, and added between the adder circuits.

Here, these delay circuits 6υ, 03 and an adder circuit.

A second-order transversal filter is formed. The frequency characteristic curve of this transversal filter is as follows: (Milo) However, TD = 140n16C, 3.58 MHz 4I: mljS as the apex
The result will be as shown in the figure.

11fjllK and the output 1H of the extension circuit aυ and a are respectively i11Igl path circle, connection and addition circuit (to),
Or a slow bat! l road 0. (2) and addition-path (
2) is supplied to a transpernal filter consisting of:

and S11 from these transpersal filters
The α-stage arithmetic circuit and inverter were presented as a gift.

Furthermore, the signal on the output side of the delay circuit (b) is supplied to the addition circuit 01 and the subtraction circuit (2).

In this circuit, since the frequency characteristics of the transversal filter are as gentle as described above, there is no possibility of phase distortion occurring near the cutoff frequency. Furthermore, since the luminance signal component outside the chroma signal band is appropriately attenuated, the luminance signal component can be completely removed by using a substantially ideal filter device.

In this way, the luminance signal and chroma signal are separated, but F! According to AK, a low-order transversal filter is used as a bandpass filter, and this is used in combination with the above-mentioned approximately ideal filter device.
Therefore, there is no risk of image quality deterioration due to phase distortion of the chroma signal, and generation of four scalars due to leakage of luminance signal components can also be prevented. Furthermore, the transpersal filter does not require any adjustment (it can easily be provided with desired frequency characteristics, and its configuration is extremely simple).

Furthermore, the present invention can be applied not only to the filter device K11jI using the above-mentioned logic operation circuit, but also to a separation filter using a so-called comb filter.

In the 16th E, the signal from the input terminal (1) is supplied to the subtractor circuit (Kl)K, and is also supplied to the subtractor circuit (41)K through the delay circuit (6) corresponding to one horizontal period to form a comb filter. , the black 1 signal is separated. This claw w 140 naec delay circuit (c),
The signal is taken out to the output terminal 3υ through a transversal filter consisting of a circle and an adder circuit (c). In addition, the signal from the input terminal (14) is supplied to the power reduction circuit (14) through a delay circuit (14), and the chroma signal from the adder (3) is supplied to a subtracter (4) to separate the luminance signal. This luminance signal is taken out at the output terminal.

In this circuit as well, since the comb filter is a substantially ideal filter device, it is possible to obtain a separation filter that is free from the risk of image quality deterioration or cross color as described above.

Note that the present invention can also be applied to filter devices and comb filters using other types of logical operation circuits.

[Brief explanation of the drawing]

1 to 13 are diagrams for explaining a filter device previously proposed by the inventor of the present application, FIG. 14 is a configuration diagram of an example of the present invention, FIG. 15 is a diagram for explaining the same, and FIG. The figure is a configuration diagram of another example. OI is an input terminal, αυ, a2 is a delay circuit, α$ is a logic operation circuit, Qυ, Cj3 is an output terminal, C11), cw
, u, ca, C12) (to) is a delay circuit that constitutes a transversal filter, and (to) is an addition circuit that constitutes a transversal filter. Fig. 4 8 Fig. 7 Fig. 8 Fig. 9r21 Fig. 12 Fig. 13 D E j-■-111--

Claims (1)

[Claims] In a separation filter that filters multiple correlated signals based on their respective correlations, a low-order transversal filter is used as a bandpass filter to extract a predetermined band component of the signal. Separation filter.