JPH02250565A - Ghost eliminating device - Google Patents

Abstract

PURPOSE:To shorten the ghost erazing time and to improve the ghost erasing level by erasing collectively the normal ghost signals. CONSTITUTION:The sum total of the ghost cancel reference signals GCR' supplied to a subtractor 13 from a reference wave GCR' generating part 14 is set at a prescribed level S. Then the subtraction is carried out between the video signal having its approximate ghost signal eliminated by an approximate ghost eliminating filter and undergone the equalization of waveform and the signal GCR'. This difference signal (e) is multiplied by 1/S and used as a filter coefficient of a normal ghost eliminating filter. In this case, the output signal of the filter is turned into a ghost cancel signal having a level equal to and a code opposite to a normal ghost signal. The normal ghost signals are collectively erased. As a result, the ghost erasing time is shortened and at the same time the ghost eliminating level is improved.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention relates to a ghost removal device that uses a transversal filter to remove a ghost signal from a video signal.

[Prior Art] For example, when radio waves emitted from a transmitting antenna hit a building or other structure or a mountain, reflect, and reach a receiving antenna, a ghost signal is included in the video signal and becomes a ghost in the image. This will cause the image quality to deteriorate.

Currently, in response to the trend toward higher image quality, it is being considered that broadcasting stations multiplex ghost cancellation reference signals and transmit them so that ghost signals can be easily removed on the receiving side.

Furthermore, as described above, a ghost removal device has been proposed that uses a transversal filter to remove a ghost signal contained in a video signal. FIG. 2 shows an example.

In the figure, l is an input terminal to which a video signal Sv output from, for example, a video detection circuit (not shown) is supplied, and the video signal Sv supplied to this input terminal 1 is supplied to an adder 2.
supplied to During the vertical blanking period of this video signal SV,
Ghost cancellation reference signal GCR1 For example, 5inX
/X pulse is inserted.

Further, the video signal Sv supplied to the input terminal l is supplied to the transversal filter 3, and is adjusted by the transversal filter 3 so that the time and level match those of the nearby ghost signal included in the video signal SV, and the polarity is Inverted to form a ghost cancellation signal.

Here, a close ghost is a ghost that has a short time.
In general, this refers to a ghost with a delay time of about ±1 μsec.

The ghost cancellation signal formed by this transversal filter 3 is supplied to an adder 2, where the ghost cancellation signal is subtracted from the video signal S■, and from this adder 2, a video signal Sv' from which the nearby ghost signal has been removed is output. Output.

Further, the video signal Sv' outputted from the adder 2 is supplied to the adder 4, and the output signal of this addition B4 is supplied to the transversal filter 6, which normally removes ghost signals.
The transversal filter 5 adjusts the signal so that it matches the normal ghost signal contained in the video signal S2 in time and level, and inverts the polarity to form a ghost cancellation signal.

The ghost cancellation signal formed by the transversal filter 5 is supplied to the adder 4, and the video signal S
The ghost cancellation signal is subtracted. Then, at an output terminal 6 derived from the output side of the adder 4, a video signal SV# from which the normal ghost signal is removed in addition to the proximity ghost signal is obtained.

Further, the output of the adder 4 is supplied to a filter coefficient generating section 7. The filter coefficient generating section 7 calculates filter coefficients (tap coefficients) of the transversal filters 3 and 5, and the filter coefficients are supplied to the transversal filters 3 and 5 and set therein.

The filter coefficient calculation method is MSE (MeanSq
uared Error) method, Z F (Zero
There are two methods, such as the error minimization method, which repeats adjusting the filter coefficients so that the error between the output signal of the ghost removal filter and the ghost cancellation reference signal is reduced, as represented by the Forcing method, and the division method, as represented by the division method. There is a method of bulk elimination.

When the ZF method is used as a method for calculating filter coefficients, the filter coefficient generating section 7 is configured as shown in FIG.

In the same figure, the output signal of the adder 4 is supplied to the input terminal 11, and this output signal is supplied to the ghost cancellation reference signal extraction section 12, where the detected portion of the ghost signal corresponding to the ghost cancellation reference signal OCR is extracted. This portion is supplied to the subtracter 13. Note that the ghost cancellation reference signal GCR inserted in the vertical retrace period of the video signal Sv is not a sin X/X pulse but an s
in X/× bar, the output signal P of the extractor 12 is differentiated and then supplied.

Further, the subtracter 13 is supplied with a ghost cancellation reference signal GCR' (sinX/X pulse) from the reference waveform generator 14. In this case, the reference waveform generator 14:
Ghost cancellation reference signal GCR' using ROM etc.
is generated.

The subtracter 13 outputs an error signal e obtained by subtracting the revost cancellation reference signal GCR' from the output signal P of the extractor 12. This error signal e is supplied to a multiplier 15 and multiplied by a constant α.

This constant α is set to, for example, 0°01 in order to stabilize the system.
α<0.5.

The output signal αe of the multiplier 15 is supplied to an adder 16, and the output signal of the adder 16 is supplied to the a-side fixed terminal of the changeover switch 17.

The output signal of this changeover switch 17 is supplied to a filter coefficient memory 18 consisting of a shift register. b
I! ll fixed terminal. This changeover switch 1
The am fixed terminal 9 is in an electrically floating state, and its output signal is supplied to the output terminal 20.

In this case, the changeover switches 17, 19 are connected to all when modifying the data in the filter coefficient memory 18. In this case, a new filter coefficient obtained by adding the signal αe from the multiplier 15 to the old filter coefficient Co read from the filter coefficient memory 18 is written in the filter coefficient memo 18.

On the other hand, the changeover switch 17.19 is connected to the b side when outputting the filter coefficient to the transversal filter 3.5 (see FIG. 2). In this case, the filter coefficients read from the filter coefficient memory I8 are outputted to the transversal filter 3.5 via the changeover switch 19 and the output terminal 20. Note that the filter coefficient memory 18
When it is constructed from a RAM or the like instead of a shift register, the changeover switches 17 and 19 can be omitted.

1M, BE that the invention aims to solve In the error minimization method like the ZF method described above, the filter coefficients need to be modified many times, so the ghost cancellation time from the start of ghost signal removal to the completion of removal is shortened. I couldn't shorten it too much. In addition, in the error minimization method, the quantized difference signal e is multiplied by a small constant α, so even if the difference signal e is not 0, αe=o may occur. There was a large amount of unerased ghost signals.

On the other hand, the conventional batch elimination method required high-speed processing of a large amount of data, such as Fourier transform, and therefore required large-scale eight-domain hardware.

Therefore, it is an object of the present invention to shorten the ghost erasing time, reduce the amount of unerased ghost signals, and reduce the scale of the hardware.

Means for Solving ci*a] The present invention is configured using a proximal ghost removal filter that removes proximal ghost signals from an input video signal and a transversal filter, and removes a normal ghost signal from the input video signal. a normal ghost removal filter, a reference waveform generating means for generating a ghost cancellation reference signal whose sum is a predetermined number, and subtraction of the video signal from which the proximity ghost signal has been removed by the proximity ghost removal filter and the ghost cancellation reference signal. and filter coefficient forming means for forming filter coefficients of the transversal filter of the normal ghost removal filter by dividing the output signal of the subtraction means by the predetermined number l.

[Function] In the above configuration, the sum of the ghost cancellation reference signals generated by the reference waveform generating means is set to be a predetermined number S, and the proximity ghost signal is removed by the proximity ghost removal filter to equalize the waveform. The resulting video signal and the ghost cancellation group Il1 are subtracted, and the difference signal e is
It is multiplied by bl 1 /S and used as a filter coefficient of a normal ghost removal filter.

In this case, the output signal of the ghost removal filter is usually
A ghost cancellation signal having the same magnitude and opposite sign as the normal ghost signal is generated, and the normal ghost signal is canceled all at once.

[Embodiment] An embodiment of the present invention will be described below with reference to FIG. In this figure 1, the 31st! The same reference numerals are given to parts corresponding to I, and detailed explanation thereof will be omitted.

In the figure, a ghost cancellation reference signal GCR' (s
(in x/x pulses) is set to be a predetermined number S.

Further, the filter coefficient memory 18 is dedicated to filter coefficients set for the transversal filter 3. That is, the filter coefficients read from the filter coefficient memory 18 are output to the transversal filter a via the changeover switch 19 and the output terminal 20.

Further, the error signal e from the subtracter 13 is supplied to a multiplier 21 and multiplied by a constant 1/S. The output signal e/S of this multiplier 21 is supplied to a filter coefficient memory 22 composed of a shift register.

This filter coefficient memory 22 is dedicated to filter coefficients set for the transversal filter 5. That is, the filter coefficients read from the filter coefficient memory 22 are output to the transversal filter 5 via the output terminal 23.

In this case, after the adder 3 can obtain the video signal Sv' from which the nearby ghost signal has been sufficiently removed, the output signal of the multiplier 21 is written as a filter coefficient into the filter coefficient memory 22, and this is used as the transversal signal. The signal is controlled to be output to the filter 5.

The reason for controlling in this way is as follows.

- For boat fishing, the reference signal GCR' from the reference waveform generator 14 is a 5i signal with a flat frequency characteristic within the video signal band.
The nx/x pulse or the 5inx/x pulse with a slight roll-off is used, but as mentioned above, by sufficiently removing the proximity ghost signal, the video signal S
The reference signal OCR inserted into the reference signal OCR is waveform-equalized to the reference signal OCR', and has the same level and frequency characteristics as the reference signal OCR'. The transversal filter 5 delays this waveform-equalized reference signal OCR and multiplies it by a coefficient to create a ghost cancellation signal for canceling the normal ghost signal. It is necessary that the reference signal GCR is waveform-equalized. In other words, the reference signal OCR
If the level of the ghost signal does not reach the specified value, it is not possible to create a ghost cancellation signal for collective cancellation.

In addition, in the case where the input terminal 11 is supplied with either the output signal of addition @ 2 or the output signal of addition @ 4, the ghost signal is usually canceled all at once, so the filter coefficient of the transversal filter 5 is This is because the output signal of adder 2 and the output signal of adder 4 are equal during calculation.

This example is constructed as described above, and the rest is the same as the example shown in FIG.

In this way, in this example, the adder 3 sufficiently removes the nearby ghost signal, and the reference signal OCR becomes the reference signal OC
After the waveform of R' is sufficiently equalized, the output signal e/S of the multiplier 21 is stored as a filter coefficient in the filter coefficient memory 22.
This is output to the transversal filter 5. As a result, the transversal filter 5 generates a ghost cancellation signal for canceling the normal ghost signals all at once. Therefore, at the output terminal 6, a video signal S'' from which the normal ghost signal has been removed in addition to the proximity ghost signal is obtained.

Next, the reason why the normal ghost signals can be canceled all at once in this example will be explained.

In this example, the transversal filter 5 generates a ghost cancellation signal for canceling the normal ghost signal by performing necessary delay and coefficient multiplication on the reference signal OCR whose waveform has been equalized to the reference signal OCR'.

In the batch cancellation method, assuming that the frequency characteristics of the reference signal GCR' are flat within the video signal band, if the level represented by the DC component of the created ghost cancellation signal matches that of the ghost signal, then the ghost Can cancel the signal.

By the way, the DC gain of the transversal filter is
For example, in a transversal filter where the sum of the filter coefficients is 1, a certain direct filli
! If d is input, the output level will be d. This means that if the sum of the data of the reference signal OCR' is set to 1 and the data of the reference signal GCR' is directly used as a filter coefficient, the input DC component can be output with a gain of 1. Further, assuming that the frequency characteristic of the reference signal OCR' is flat, the input signal is output as 1 in the band portion where the characteristic is flat.

Assuming that the reference signal GCR is a 5inx/x pulse,
Since the video signal Sv is band-limited, the reference signal GC
The ghost signal component of R should be expressed by adjusting and combining the level and time of the reference signal OCR. Now, when the reference signal OCR has been waveform-equalized to the reference signal OCR' with a total sum of 1, there exists a single ghost signal with a phase of 0 degrees, where the sum of data of the error signal e in the normal ghost signal section is represented by h. Assume that h takes a certain value from O to l, and it can be seen that this is the ratio of the magnitude of the ghost signal to the reference signal. In other words, if the sum of the reference signals OCR' is set to 1, the normal ghost signal can be eliminated all at once by using the error signal e as it is as a filter coefficient for normal ghost removal.

However, it is not practical to set the sum of the reference signals OCR' to 1 due to limitations in the quantization range.

For example, consider the case where 8 bits of filter coefficients are assigned from 1128/128 to 127/128. In this case, l corresponds to quantization level 128. Since the video signal Sv is waveform-equalized and has a magnitude that depends on the reference signal OCR', the sum of the reference signals GCH is set to 1.
Setting the number to 28 limits the quantization range, and the quantization range cannot be set to the optimum state in many cases.

Therefore, as in this example, if the total sum of the reference signal OCR' is S and the error signal e is multiplied by 17S by multiplication 1!21 and then used as the filter coefficient of the transversal filter 5, the quantization range can be optimized. While setting, reference signal G
As in the case where the sum of CR' is set to 1, normal ghost signals can be erased all at once.

As described above, according to this example, since ghost signals are normally erased all at once, it is possible to shorten the ghost signal erasing time, reduce the amount of ghost signals left unerased, and improve the ghost erasing level.

Further, since the error signal e multiplied by 1/S directly serves as the filter coefficient for normal ghost removal, the configuration is simple and the scale of the hardware can be reduced.

[Effect of Gomei] As explained above, according to the present invention, ghost signals are usually erased all at once, so the ghost erasing time can be shortened, and the amount of ghost signals left unerased can be reduced, improving the ghost erasing level. be able to. Further, since the error signal divided by a predetermined number directly serves as the filter coefficient of the transversal filter of the normal ghost removal filter, the configuration is simple and the scale of the hardware can be reduced.

18.22 and φ filter coefficient memory

[Brief explanation of drawings]

FIG. 1 is a block diagram of a filter coefficient generator according to an embodiment of the present invention, FIG. 2 is a block diagram of an example of a ghost removal device, and FIG. 3 is a block diagram of an example of a filter coefficient generator. 1.11- 2.4.16- 3, 5 ・ 6, 20. 23 φ 7 φ 12 ・ 15.21 17゜・Input terminal・Adder・Transversal filter・Output terminal・Filter coefficient generation section・Ghost cancellation reference signal extraction section・Subtractor・Reference waveform generation section・Multiplier・Selector switch

Claims (1)

[Claims] (1) a near ghost removal filter that removes a near ghost signal from the input video signal; a normal ghost removal filter that is configured using a transversal filter and removes the normal ghost signal from the input video signal; a reference waveform generation means for generating a ghost cancellation reference signal of a number, a subtraction means for subtracting the ghost cancellation reference signal and a video signal from which the proximity ghost signal has been removed by the proximity ghost removal filter; 1. A ghost removal device comprising: a filter coefficient forming means for forming a filter coefficient of a transversal filter of the normal ghost removal filter by dividing the output signal by the predetermined number.