JPS5911305B2 - Automatic distortion removal device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an automatic distortion removal device that detects the difference between a reference signal portion included in an output signal of an acyclic digital filter and a standard signal set to the original waveform of this reference signal portion. By observing and successively modifying the weighting coefficient values, the reference signal portion is brought closer to the standard signal, and ghosts (
The purpose of the present invention is to provide a device that can automatically reduce and eliminate distortion such as echo distortion.

Various methods have been tried in the past as countermeasures against ghosts in video signals received by television receivers.

For example, research on wall materials of high-rise buildings that are sources of radio wave reflection, methods for reducing ghosts by optimizing the directivity of receiving antennas, methods for removing ghosts at the intermediate frequency stage of images from television receivers, and videos from television receivers. This is a method for removing ghosts in signal systems. Among the above ghost removal methods,
A ghost removal method in a video signal system has become seriously considered with the development of an analog shift register using a charge transfer device such as a charge coupled device (CCD).

This analog shift residence samples the analog signal at a frequency of F. sampled at unit time (1/FO
), and plays an extremely important role in constructing a digital filter. As a conventional ghost removal device using such a digital filter, there is the ``Ghost Killer Eye Movement Method'' which was announced at the National Conference of the Television Society in 1971. This method uses a weighting control circuit for the negative feedback loop,
A CCD transversal filter whose delay time is controlled by the output signal of this weighting control circuit is provided, and by performing negative feedback synthesis of the output signal of this transversal filter and a video signal having a ghost, ghosts are automatically removed. This is a method of erasing the data. However, the above method is only intended to eliminate delayed ghosts that arrive later than the desired wave and have the same information as the desired wave, although they have a different phase and amplitude from the desired wave. could not be deleted.

In other words, as is well known, most of the ghosts are delayed ghosts, but for example, if a ghost wave enters from the direction of the antenna's directional characteristics, or if it ) etc., when the desired wave propagates through a coaxial cable and enters the television receiver, if the television radio wave directly enters the television receiver, the desired wave is preceded by a certain amount of time. As a result, a leading ghost having the same information as the desired wave is generated. For this reason, the above method has the disadvantage that a sufficient ghost removal effect cannot be obtained. Furthermore, since the above system uses a cyclic digital filter, it is technically difficult to suppress the problematic oscillation phenomenon because it is a feedback type. On the other hand, as shown in FIG. 1, there has been a conventional 7'S ghost automatic removal device using a non-recursive type (feed-forward type) digital filter.

This conventional device has τ1, τ2,..., τ. n delay circuits 11, 12, . . . , 1n are cascade-connected and have a delay time of .
, and 31, 32, ..., 3. are added and synthesized in synthesis circuits 4 and 5 via the respective signals. Here, the input video signal from which ghosts are removed is assumed to have been transmitted using the vestigial sideband transmission method, similar to current television broadcast waves, so the video signal of the desired wave has a vector that is in phase with the carrier wave. component and orthogonal component. Therefore, although the ghost wave also has an in-phase component and a quadrature component, it has a different phase and amplitude from the in-phase component and quadrature component of the desired wave. Therefore, the coefficient units 21, 22, . . . , 2 are provided. represents the ghost in-phase component, and the coefficient multipliers 31, 3, . . . , 3. creates ghost orthogonal components from the in-phase components of the desired wave, respectively. The reason why the in-phase component of the desired wave is output in this way is to use a synchronous waveform circuit as a video detection circuit of a television receiver so that ghosts can be sufficiently reduced. The ghost orthogonal components synthesized by the synthesis circuit 5 are filtered in a conversion circuit 6 to obtain the original ghost orthogonal components, and then supplied to an addition circuit 7, where the ghost in-phase components from the synthesis circuit 4 are is added to the component.

After the output signal of this adder circuit 7 is subtracted from the input video signal having ghosts, a video signal from which ghosts have been removed is obtained. However, the conventional device shown in FIG. 1 was intended to remove delayed ghosts, and was unable to remove preceding ghosts.

Further, the specific setting method for each coefficient unit 21-12n, 31-3n was not specified. The present invention eliminates the above-mentioned drawbacks, and an embodiment thereof will be described below with reference to FIGS. 2 to 5.

FIG. 2 shows a block system diagram of an embodiment of the apparatus of the present invention.

A video signal input from the input terminal 8 is supplied to a digital filter section 9 and a timing signal generation section 10, respectively. The digital filter section 9 converts the video input signal into a unit time (
One sampling frequency F. An analog shift register 11 is equipped with a plurality of parallel output terminals that delay and output the reciprocal of the color burst signal, FO is three or four times the frequency of the color burst signal, and the signal from this parallel output terminal is , a weighting coefficient circuit that multiplies the weighting coefficient value set externally (
Such a multiplier can be realized using an operational amplifier, etc.)12
, an adder circuit 13 that adds each parallel output signal of the weighting coefficient circuit 12, and a signal from a predetermined one reference terminal 11a of the parallel output terminals of the analog shift register 11. An amplifier 14 that compensates for the amount of attenuation that occurs while the signal passes through it.
, a subtraction circuit 15 that subtracts the output signal of the amplifier 14 and the output signal of the adder circuit 13, and an amplifier 16 that amplifies the output signal of the subtraction circuit 15 to the level value of the video signal at the input terminal 8. ing. The delay time given to the video signal by the analog shift register 11 increases toward the right in FIG. 2, and the output terminal 11a obtains a delayed signal with a delay time approximately in the middle of the parallel output terminals. Selected as an output terminal.

Therefore, among the output signals supplied from the analog shift register 11 to the weighting coefficient circuit 12, for the output signal from the output terminal 11a, the analog shift register section 11b
Each of the plurality of output signals taken out from the analog shift register section 11c has a small delay time, while each of the plurality of output signals taken out from the analog shift register section 11c has a long delay time.

On the other hand, the timing signal generation section 10 includes a sync separation circuit 17 and a timing signal generator 18, and the sync separation circuit 17 reproduces and separates vertical and horizontal sync signals and color burst signals contained in the input video signal. ,
Based on these synchronization signals, a sampling signal of the digital filter unit 9, a timing signal for setting a weighting coefficient in a weighting coefficient control unit 21, which will be described later, and a reference signal processing unit 26
The timing signal generator 18 outputs a timing signal for reference signal processing in the reference signal processing section 10, a timing signal for synchronizing the arithmetic unit 19 with the input video signal, and the like.

The arithmetic unit 19 is a block equipped with arithmetic capabilities, such as a microcomputer, and includes a central processing unit (CPU) 2.
0, the difference data of the reference signal part from the reference signal processing unit 26, which will be described later, is read as needed, and a predetermined level change part (hereinafter referred to as slope part) of the standard signal that is preset as the standard of the reference signal part is read as needed. A difference zeta of a portion corresponding to is created, subtracted from the difference data of the reference signal portion, and this value is normalized by the amplitude value of the standard signal.

The arithmetic unit 19 is connected to a weighting coefficient memory 22 which will be described later.
reads the value already set in the weighting coefficient memory 22, adds the above-mentioned normalized zeta to it, corrects the value in a direction that eliminates distortion in the reference signal part, and rewrites the contents of the weighting coefficient memory 22. This is the role of the calculation unit 19, which mainly performs calculations to correct the weighting coefficient values. The weighting coefficient data calculated and corrected by the calculation unit 19 is set and stored in a weighting coefficient memory 22 that constitutes a part of the weighting coefficient control unit 21.

This weighting coefficient control unit 21 is a block that functions to control the value of the weighting coefficient set in the weighting coefficient circuit 12 in the digital filter unit 9. The zeta set in the weighting coefficient memory 22 is sequentially supplied to the DA conversion circuit 23, where it is converted into analog data and sent to the next analog shift register 24. The weighting coefficient values converted into analog by this analog shift register 24 are sent to the next analog memory 25 and set when they are all read. The analog memory 25 is an analog memory that can temporarily hold charge and is composed of elements such as field effect transistors. Therefore, the above-described weighting coefficient values are set in the analog memory 25 periodically at time intervals such that previously stored data will not be lost due to discharge. This operation of rewriting the weighting coefficient values in the analog memory 25 is performed during a time period such as a horizontal or vertical blanking period that does not affect the screen of the television receiver. The output video signal of the digital filter section 9 is supplied to the reference signal processing section 26.

The reference signal processing unit 26 detects the reference signal portion (a portion including the slope portion of the vertical synchronization signal) in the output video signal using the timing signal from the timing signal generation unit 10, and detects the reference signal portion (the portion including the slope portion of the vertical synchronization signal) in the output video signal, and a difference circuit 27 that takes the difference between
It is comprised of an AD conversion circuit 28 that converts the output signal of the difference circuit 27 into binary data (quantization), and a buffer memory 29 that stores the output data of this AD conversion circuit 28. Since the reference signal section is a part of the vertical synchronization signal, the data stored in the buffer memory 29 can be rewritten every reciprocal of the field frequency at the minimum, but the processing speed of the calculation section 19 at the next stage Since there is also a trade-off, the data set in the buffer memory 29 is controlled to be maintained as it is until the series of processing in the arithmetic unit 19 is completed. In this way, the digital filter section 9 successively corrects and sets the weighting coefficient values, and finally the video signal whose reference signal section is brought closer to the standard signal is output from the amplifier 16, and then passed through the low-pass filter 30. Used in digital filter section 9 in
] A video signal S from which distortion-corrected sampling frequency components and the like have been removed.

is output. Next, the principle of operation of the main parts of the device of the present invention, such as the digital filter section 9, will be explained in detail theoretically.

The operation of the digital filter section 9 is expressed by the following equation. y
(t) = x (t) - W (△n) Assuming that the attenuation amount is O, △ is the unit delay time of the analog shift register 11 (the reciprocal arrow W(△n) of the sampling frequency is x(
The weighted coefficient value multiplied by t-△n), x(t) at time t
The output video signal value of the reference terminal 11a (or the amplifier 14) at time t, y(t) is the output video signal value of the digital filter unit 9 at time t, and Nl and n2 are positive integers.

This formula is expressed as the value x(
t), the (past) value of x △・n before the current time X(t-△n) is multiplied by an appropriate weighting coefficient W(△n), which is W(△n)・x( t-Δn), and the above x
(t) multiplied by an appropriate weighting coefficient W(0) W(
0)・x(t), and the (future) value x(t+△n) of x after △・n from the current point in time.
By subtracting W(△n) x(t-△n) by n=-n1, the current represents an equation for forming a video signal y(t) from which distortion components such as ghosts contained in the input video signal x(t) of the digital filter section 9 are removed.

By combining these temporally varying values at an appropriate ratio,
The possibility of removing distortion components by performing integral operations with appropriate weighting has long been known as the theory of acyclic digital filters.

However, how can we determine the optimal weighting coefficient W
There is no established theory as to whether (Δn) can be set. The present invention relates to a method for setting weighting coefficients that has no established theory, and provides a method for setting weighting coefficients for removing not only delayed ghosts but also preceding ghosts contained in a video signal. In a television video signal, as shown in FIG. 3A and an enlarged view of the same in FIG. It is known that it is easiest to observe ghost information.

In FIG. 3B, the t-TO portion corresponds to the slope part of the vertical synchronization signal, and the t-TO portion is 29.39 μs and 2T. It has a flat waveform portion of 32 μs. Therefore,
If there are irregularities on this flat part, it is a distortion component, so to speak. On the other hand, in FIG. 3B, if there is an unevenness on the right side, it is a delayed ghost, and if there is an unevenness on the left side, it is a leading ghost. It can be assumed that this was formed due to the influence of the slope in the area. Therefore, observation time periods of appropriate width are set up on the left and right sides with TO as the center (t=T1 to T2 in Figure 3B), and this is used as the reference signal section, and the unevenness of the flat part in this reference signal section is eliminated. Smoothing can be achieved by setting weighting coefficients in the directions. As a method for observing the presence or absence of unevenness in this flat area, first, the difference value of the reference signal part is taken, and the presence or absence of unevenness is investigated by performing a so-called differential operation.

However, according to this operation, the sloped portion at the TO will also be converted into data as a type of unevenness, so it is necessary to remove the influence of this sloped portion. Therefore, in the present invention, a standard signal is determined in advance as an ideal waveform of the reference signal part, and the difference value of the slope part of this standard signal is subtracted from the difference value of the reference signal part from the reference signal processing section 26. By doing so, the influence of the slope part in the reference signal part is equivalently canceled, and only the influence of unevenness due to the original ghost wave remains, and the weighting coefficient is set using the difference value obtained in this way. , to obtain a video signal from which ghosts have been removed. Furthermore, in the present invention, when setting the above-mentioned weighting coefficients, the weighting coefficients are not set so as to immediately eliminate the uneven portions in the flat portion of the reference signal portion, but are set while sequentially correcting the weighting coefficients so as to gradually eliminate them. It is something. If the weighting coefficient is set so as to immediately eliminate the irregularities in the flat part of the reference signal section, the ghost component in the video signal extracted from the low-pass filter 30 will diverge (if the amplitude is at an extremely large level). This is because it has been experimentally confirmed that
Next, the above principle will be explained based on a simple example of a leading ghost, but the following explanation also applies to a delayed ghost, a mixture of leading and delayed ghosts, a complex form of ghost, or other distortions. is applicable.

FIG. 4A shows a waveform of a reference signal portion having a time width from time T1 to time T2.

and amplitude A. It has an inclined part. FIG. 3B shows a waveform diagram near the reference signal (t-T, ), which is preceded by τ1 from the position TO of the slope portion of the reference signal portion and has a ghost superimposed thereon having an amplitude of b1 and the same phase as the desired wave.

However, T at this time. The amplitude at the slope part of is A, (=
AO-b1). This signal waveform and the fourth
The signal waveforms shown in Figures D, F, and G, respectively, appear at the output of the low-pass filter 30. In FIG. 4B, the sampling frequency F.

By sampling (=-) △ and taking the difference value between adjacent values in the difference circuit 27, the difference circuit 27
As shown in FIG. 4C, time T.

An impulse train having an amplitude a1 at time t1 and an amplitude b1 at time t1 is obtained. Time T. The impulse at time t1 is information on the slope portion of the reference signal portion, and the impulse at time t1 represents information on the ghost. On the other hand, if a standard signal similar to that shown in FIG.

An impulse (amplitude A.) representing the slope at is obtained. This amplitude A from the impulse train shown in FIG. 4C. An impulse of amplitude b1 representing the original ghost information by subtracting the impulse of (difference in height of the slope a1
- AO) is obtained, and based on this, the weighting coefficient controller 21 calculates the amplitude A of the standard signal.
A weighting coefficient normalized by O (ratio to amplitude A) is set to a value expressed by the following equation. W(0)
-(a1-AO)///AO(2)W(1τ1)-B
, /AO(3) Therefore, the output signal y(t) of the digital filter section 9 is expressed by the following equation.

y(t)-x(t)-W(0)・x(t)-W(-τ1
)・x(t+τ1) (4) When a video signal having a reference signal portion shown in FIG. 4B passes through the digital filter section 9 expressed by this equation, the reference signal of the output video signal is as shown in FIG. 4D. As shown, the ghost that was superimposed at time t1 is sufficiently compressed to have an amplitude of B2, and a new amplitude of τ2(
= 2τ1) T.

A ghost whose phase is inverted with an amplitude C2 sufficiently smaller than that of the initial ghost b1 appears at a position T2 which is far from the position . This is also called a grandchild ghost. The waveform shown in FIG. 4D is obtained through the following calculation process. First, we substitute equations (2) and (3) for equation (4).

Next, in the waveform shown in FIG. 4B, T1 to T, -τ1(
=T2), x(t)=x(t+τ1μA
Since this becomes O, substituting this into equation (5) yields.

Also, in the time range from t=T2 to tl, x(t)=AO,
Since x(t+τ,)=al, if we substitute this into (5 wipes), we get x(t)=A
l, x(t+τ1)2o, and furthermore, at t=TO-T2, x(t)2x(t+τ1)2o.

Therefore, time T. , tl, and t2 are A2, b2, and c2, then Equation (9), Equation (8),
(7 wipes, from formula (6). However, AO = A2 + B
It can be seen that the waveform shown in FIG. 4D can be obtained from equations 2-C2 (6) to 10).

As can be seen from Equation 1, the unevenness of the waveform shown in Figure 4D is a function of the amplitude B of the ghost in Figure 4B, which is -1! - (a value sufficiently smaller than l).

In this way, the distortion caused by the ghost shown in FIG. 4B is reduced while being diffused. FIG. 4E is an output signal waveform diagram of the difference circuit 21 obtained by calculating the difference between the reference signal portions of FIG. 4D.

As a result, the weighting coefficients are newly modified additively as follows. Therefore, the new relational expression between X and y is as follows.

Thereafter, by the same operation as above, the waveform distortion caused by the ghost is replaced with the value of the weighting coefficient, and is gradually destroyed as shown in Fig. 4 F, G, H, and finally the reference signal part becomes the standard The signal waveform will approach the assumed signal waveform shown in A in the same figure. In addition, A3, b in Figure 4 F
3, c3, d3 are respectively bl←d)2, and A3+B
3- is equal to C3+D3.

Also, A4, b4, c4, d4, and e4 shown in G in the same figure are respectively. Therefore, A5, b5, c5, d5, e5, f shown in figure H are each 1
becomes.

Also A4+B4-C4+D4-E4 and A5+B5-C5
+D and -E5+F are each equal to AO.

In this way, an acyclic digital filter is used, the reference signal part is observed using the output signal of the digital filter part 9, and the value of the weighting coefficient is set using the difference value of this reference signal part. When setting the weighting coefficient, the value of W(0) is subtracted by the difference value of the slope part of the standard signal, and is normalized by the amplitude of the standard signal, and the value of W(0) that was set before that is subtracted by the difference value of the slope part of the standard signal. In addition to additively adding to and modifying the value of
For coefficient values other than W(0), ghosts are finally removed by normalizing the difference value of the reference signal part with the amplitude of the standard signal and adding it additively to the previously set value. The video signal obtained is extracted from the low-pass filter 30.

Note that the input terminal 8 shown in FIG. 2 normally receives a video signal obtained by filtering a television signal using the vestigial sideband transmission method, but in this case, a video signal detected by a synchronous detector is input. The most desirable case is when there is an incoming signal.

This is because when the ghost wave and the desired wave are synchronously detected, the video signal of the desired wave and the ghost are additively superimposed and output, so the ghost always appears in the reference signal part in an accurate form. This is because the ghost removal effect can be maximized. However, input terminal 8
When an envelope-detected video signal comes in, the envelope detection itself is a nonlinear operation, making it difficult to easily estimate the waveform, and the video signal and ghost are additively superimposed. Since the absolute value component of the detected signal is detected and output, ghost information does not necessarily appear accurately in the reference signal section, and the device of the present invention has a sufficient ghost reduction effect compared to the case of synchronous detection. Although this cannot be expected, if the degree of modulation is low, a practically sufficient ghost reduction effect can be obtained even in the case of envelope detection. In addition, Figure 4 A-H
In the description of the apparatus of the present invention with the waveforms shown in FIGS.

Although the slope part in is represented by one difference value, in reality, this slope part is represented by multiple sample values, so the difference value as information on the slope part is also represented by multiple impulse trains. It will be. FIG. 5A shows the reference signal section shown in FIG. 4A in a more realistic form, and its differential impulse train is α as shown in FIG. 5B. ,α
There are three impulse trains: 1 and α2. Therefore, the correction of the slope part in W(0) performed in the above explanation also applies to adjacent weighting coefficient values. For example, if the reference signal portion including a ghost has a waveform as shown in FIG. 5C, by performing a difference operation, β as shown in FIG. , β1
, β2, γ0, γ1, γ2 are obtained, and the coefficient values at this time are as follows. w(to)−(β2−
α2)/AOW(0)−(βo−α.

)/AOW(-△)1(β1-α1)/AO W(-τ,+△)-γ2/AO W(-τl)2γo/AO W(-τ1-△)=γ1/AO However, AO+αo+α1+α2 , a1+βo+β1+爲, b1+γ.

+γ1+γ2 In this case, the differential impulse train of the slope part is 3
However, depending on the waveform, it increases further.

By repeating these operations, the waveform of the reference signal part can be brought close to the waveform of the standard signal assumed in advance, but if the reference signal part does not contain distortion such as ghosts, If there is some difference between the signal and the standard signal, the waveform of the reference signal section will gradually approach the waveform of the standard signal, which not only removes ghosts but also improves frequency characteristics. can be done.

Note that in the above explanation of the series of digital filter sections, the amount of attenuation when the video signal passes through the analog shift register 11 shown in FIG. Therefore, it is necessary to modify the value of the weighting coefficient by that amount, and the new weighting coefficient is obtained by multiplying the weighting coefficient w obtained by the above method by a constant that corrects the amount of attenuation, and equivalently, The levels of the parallel output signals of the analog shift register 11 may be made the same.

Furthermore, an amplifier 14 is provided to correct the attenuation amount for the output signal from the reference terminal 11a of the analog shift register 11, and if necessary, the output signal y(t) is amplified by an appropriate amount by the amplifier 16. , a more faithful reproduction of the deghosted video signal is possible.
Incidentally, as described above, it is the calculation section 19 that processes the differential zeta of the reference signal portion obtained by the reference signal processing section 26 and newly sets the weighting coefficient. At present, it is not possible to expect the processing speed of the calculation unit 19 to be high enough to process the weighting coefficients as described above in real time. Therefore, it is difficult to modify the weighting coefficient every cycle (for example, one second as the reciprocal of the field frequency) in which the reference signal portion appears. However, since most ghosts are close to statics that change on a time-by-time basis, the above-mentioned processing speed is usually not a problem. Of course, ghosts caused by flying objects such as airplanes are a bit of a problem as they change from units of Ms to units of seconds, but this is also temporary and is not something to be taken seriously, and it will depend on the future development of computers. With the realization of an arithmetic unit with even faster processing speed, such concerns are expected to disappear. In addition, in the above explanation of setting the weighting coefficient, the method of subtracting the difference value of the standard signal from the difference value of the reference signal part is adopted, but it is also possible to subtract the standard signal from the reference signal part and then differentiate. Conceivable. In the former method, a meaningful difference value of the standard signal is obtained only in the part including the slope part, and the difference value in other parts is zero. Therefore, in this method, when subtracting the difference value of the standard signal from the difference value of the reference signal, it is only necessary to process the portion of the standard signal that includes the sloped portion, so that high-speed processing is possible. On the other hand, the latter method cannot be said to be very good in terms of processing speed compared to the former method, but there are cases where the calculation unit 19 can be made faster, and the overall processing time for weighting coefficient processing is reduced. The latter method is also possible if the length is sufficiently long. In this case, the second
Although the difference circuit 27 shown in the figure becomes unnecessary, a difference calculation function is required in the calculation section 19 instead. Although the device of the present invention has the primary purpose of removing ghosts from television, it can also be used with various communication lines (for example, facsimile, videophone, image information communication, data communication with artificial satellites, microwave lines, etc.). This method can be similarly applied to the removal of reverberant distortion in , and exhibits the desired effect.

That is, a test pulse (a pulse whose shape is well known in advance and corresponds to the reference signal part) periodically inserted into the data signal is used as the reference signal part, and echo distortion is removed by the above-mentioned means. By adopting a method of reproducing the original signal, the quality of zeta can be significantly improved. Further, in the embodiment shown in FIG. 2, an analog shift register using a charge transfer device such as a CCD is used for the digital filter section 9 and the weighting coefficient control section 21, but the analog shift register described above is not limited to this. Distributed constant delay line, acoustic surface wave delay element,
Of course, a delay circuit configured to output delayed signals in parallel by connecting a plurality of delay cables or LC type delay lines may also be used.

As described above, the automatic distortion removal device according to the present invention selects in advance, as a reference terminal, the terminal that outputs a signal with a value between the maximum and minimum delay time among the plurality of parallel output terminals of the delay circuit. The output signal is used as the output signal of one of the delay circuits, and the differential value is obtained by calculating the difference between two adjacent sample values among each sampled value. A predetermined level change portion (slope portion) of the reference signal portion is generated from a predetermined reference signal portion and a standard signal previously set as a standard waveform of the reference signal portion.
Obtain a third difference value of the difference between the difference value of the reference signal portion and the standard signal, in which the difference value of
By normalizing the difference value based on the amplitude value of the standard signal and adding it additively to the previously set value of the corresponding weighting coefficient, the waveform of the reference signal part is finally changed to the standard signal. The value of the weighting coefficient is successively corrected and set so as to approximate the waveform of , and the distortion contained in the input signal of the acyclic digital filter is removed, so that it has the following features.

Not only one-delayed ghosts but also preceding ghosts can be removed.

2. Not only a single ghost but also complex waveform ghosts, distortion caused by circuits in the television receiver other than ghosts, and echo distortion in various communication lines can be removed.

3. The weighting coefficients are corrected to follow distortions such as ghosts that vary over time, and the distortions can be automatically removed.

4. Regardless of the presence or absence of distortion, the waveform of the reference signal portion is brought close to the preset standard signal waveform, so that distortion can be removed and frequency characteristics can be improved at the same time.

5. If the input signal of the acyclic digital filter is a video signal detected from a television signal transmitted using the vestigial sideband transmission method, the maximum ghost removal effect can be achieved when the detection output of the synchronous filter is used. , a practically sufficient ghost removal effect can be obtained even with the detection output of an envelope detector.

Since it will be possible to eliminate ghosts, which have been considered the most difficult technology in 6 television, it will be a major driving force in solving the ghost problem in urban and mountainous areas.
Therefore, it has great effects in a technical and industrial sense.

7. means for correcting the weighting coefficient value by multiplying it by a constant so as to correct the amount of attenuation that occurs before the input signal passes through the delay circuit and is output in parallel; The structure includes an amplifier that amplifies the output signal of the subtraction circuit by a value that corrects it, and an amplifier that amplifies the output signal of the subtraction circuit to an appropriate level. It can be output from a digital filter.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an example of a conventional ghost removal device, FIG. 2 is a block diagram of an embodiment of the device of the present invention, FIGS. 3A, B, 4A-H, and 5 A to D are signal waveform diagrams for explaining the operation of FIG. 2, respectively. 8... Video signal input terminal, 9... Digital filter section, 10... Timing signal generation section, 11, 24.
...Analog shift register, 19...Arithmetic unit, 21
. . . Weighting coefficient control unit, 26 . . . Reference signal processing unit.

Claims (1)

[Claims] 1. A delay circuit having a plurality of parallel output terminals that delay an input signal by a unit time and output the delayed signal, and a delay circuit that multiplies each parallel output signal of the delay circuit by a weighting coefficient set from the outside. A weighting coefficient circuit to output, an adding circuit for adding the parallel output signals of the weighting coefficient circuits, and a signal from which distortion has been removed by subtracting the output signal of the adding circuit and the output signal of one of the delay circuits. In a distortion removal device having an acyclic digital filter consisting of an output subtraction circuit, one of the plurality of parallel output terminals of the delay circuit that outputs a signal with a value between the maximum and minimum delay time is used as a reference. A difference means is selected in advance as a terminal, the output signal of the reference terminal is used as an output signal of one of the delay circuits, and the difference between two adjacent sample values of each sampled value is calculated as a difference value. The difference between a predetermined level change portion of the reference signal portion from a predetermined reference signal portion in the output signal of the acyclic digital filter and a standard signal preset as a standard waveform of the reference signal portion. Obtaining a third difference value of the difference between the difference value of the reference signal portion whose value has been canceled and the difference value of the standard signal, and normalizing the third difference value based on the amplitude value of the standard signal. By adding additively to the previously set value of the corresponding weighting coefficient, the value of the weighting coefficient is successively corrected so that the waveform of the reference signal portion finally approaches the waveform of the standard signal. 1. An automatic distortion removal device, characterized in that the automatic distortion removal device is configured to remove distortion contained in an input signal of the acyclic digital filter and output it from the subtraction circuit. 2. A delay circuit having a plurality of parallel output terminals that outputs a delayed input signal by a unit time, and a weighting coefficient circuit that multiplies each parallel output signal of the delay circuit by a weighting coefficient set from the outside and outputs the result. , an adder circuit that adds the parallel output signals of the weighting coefficient circuits, and a subtractor circuit that subtracts the output signal of the adder circuit and the output signal of one of the delay circuits to output a signal from which distortion has been removed. In the distortion removal device having an acyclic digital filter, one of the plurality of parallel output terminals of the delay circuit that outputs a signal with a value between the maximum and minimum delay time is selected in advance as a reference terminal, The output signal of the reference terminal is set as one output signal of the delay circuit, and the output signal from the reference terminal is adjusted by a value that corrects the amount of attenuation that occurs before the input signal passes through the delay circuit and is output in parallel. an amplifier that amplifies the
The acyclic digital filter is provided with an amplifier that amplifies the output signal of the subtraction circuit to an appropriate level, and a difference means that calculates the difference between two adjacent sample values among each sampled value to obtain a difference value. The reference signal in which the difference value of the predetermined level change portion of the reference signal portion is canceled from a predetermined reference signal portion in the output signal of the reference signal portion and a standard signal preset as a standard waveform of the reference signal portion. obtaining a third difference value of the difference between the difference value of the signal part and the difference value of the standard signal;
The third difference value is normalized based on the amplitude value of the standard signal and added to the previously set value of the corresponding weighting coefficient, and at the same time, the attenuation amount by the delay circuit is corrected. By multiplying the weighting coefficient by a constant, the value of the weighting coefficient is successively corrected and set so that the waveform of the reference signal portion finally approaches the waveform of the standard signal, and the value of the weighting coefficient is successively set. An automatic distortion removal device characterized in that it is configured to remove distortion contained in an input signal and output it from the subtraction circuit.