JP3430087B2 - Ghost removal device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a ghost removing device installed in a receiving device such as a television receiver.

[0002]

2. Description of the Related Art In addition to signal components directly received by an antenna, video signals received by a television receiver include
It also includes signal components received with some delay after passing through various reflection paths by moving objects such as trees, buildings and vehicles in the vicinity. Therefore, apart from varying degrees, multiple images generally appear in the receiving screen. A signal component that causes a multiple image to appear in the reception screen is called a ghost, and a phenomenon in which the multiple image becomes so large as to be annoying and the image quality deteriorates is called a ghost disorder.

Such a ghost removing device includes a transversal filter for generating a pseudo ghost, a tap gain control circuit for detecting a ghost occurrence state and controlling a tap gain of the transversal filter, and a signal synthesizing circuit. Composed. This transversal filter generates a pseudo ghost by simulating the mechanism of ghost generation based on multiple reflections such as signal delay, attenuation, and mutual addition by cascaded delay circuit groups, coefficient circuit groups, and addition circuits.

This pseudo ghost is usually generated with the opposite polarity, is added to the original reception television signal in the signal combining circuit, and is canceled by the ghost component contained therein.
Generally, a ghost changes from moment to moment due to various factors such as a change in the wavelength of a received radio wave due to a change of a receiving channel and a change in the traffic situation of a moving body such as a vehicle, a ship, or an aircraft that passes or travels in the vicinity. Therefore, the tap gain of the transversal filter is adaptively controlled every moment so as to always have an optimum value.

Due to this adaptive control, a reference waveform for detecting the ghost occurrence condition on the transmitting side is inserted into the television signal. Further, the tap gain control circuit on the receiving side generates the optimum tap gain by detecting the ghost occurrence state from the analysis result of the distortion of the reference waveform extracted from the received television signal, and supplies the optimum tap gain to the transversal filter.

The above-mentioned ghosts are roughly classified into proximity ghosts and non-proximity ghosts according to the degree of proximity to the original signal. That is, as shown in FIG. 3, when the original signal when there is no ghost has a waveform A like the dotted line, the proximity ghost causes the waveform distortion illustrated by the solid line portion α, and the non-proximity ghost is The waveform distortion exemplified by the solid line portion β is generated. Proximity ghost includes various factors on the transmission characteristics that cause waveform distortion, while
Non-proximity ghosts are based on the ghost-specific phenomenon of the form of detour propagation paths.

Therefore, the non-proximity ghost is usually called a ghost. The non-proximity ghost appears apart from the original signal on the time axis, and due to multiple reflection, the parent ghost,
The child ghost and the grandchild ghost gradually decay and tend to appear repeatedly at predetermined time intervals. Therefore, in removing this non-proximity ghost, it is desirable to cause the pseudo ghost generated from the received television signal and the original received television signal to perform cyclic combining.

On the other hand, since the proximity ghost overlaps the original signal on the time axis, a portion preceding the original signal also appears in the generated proximity pseudo ghost. Therefore, cyclic synthesis cannot be applied to the generated pseudo ghost and the original signal. Further, regarding the proximity ghost, the waveform distortion caused by the ghost and the waveform distortion caused by the transmission characteristic are inseparably generated,
Proximity ghost removal is also considered to be a type of waveform equalization.

Therefore, in order to remove the proximity ghost and the non-proximity ghost, first, the proximity ghost is removed and then the non-proximity ghost is removed by connecting the exclusive removal devices in cascade instead of using one removal device. 2
It is done in stages. That is, it is configured by a cascade connection of a non-proximity ghost elimination section with a non-recursive connected transversal filter and a non-proximity ghost elimination section with a recursive connected transversal filter.

FIG. 4 shows the second patent of the applicant's earlier application.
It is a block block diagram of the ghost removal apparatus disclosed by No. 62,015.

This ghost removing device is composed of a proximity ghost removing section 31, a proximity ghost detecting section 32, a normal ghost removing section 33 and a normal ghost detecting section 34, and a ghost removing target receiving television supplied from an input terminal IN. Output terminal OUT after applying ghost elimination processing to the signal
It is configured to output from.

As shown in FIG. 5, the proximity ghost removing section 31 uses a transversal / pseudo-ghost ghost generation transversal
It has a non-recursive configuration including a filter 31a, a delay circuit 31b and an adder circuit 31c. Further, as shown in FIG. 6, the normal ghost removing section 33 has a cyclic type configuration including a pseudo normal ghost generating transversal filter 33a and an adding circuit 33b.

As shown in FIG. 7, the pseudo ghost generating transversal filters 31a and 33a convert a digital signal obtained by A / D converting a received television signal into one signal.
Delay devices 42a and 42 for delaying each sampling period
42n, and multipliers 43a, 43b, 4 for multiplying the signal extracted from each delay device by a tap coefficient.
.. 43n, and an adder 44 that combines the outputs of the multipliers.

In this way, the entire ghost detection area is divided into the near ghost detection area and the normal ghost detection area,
The configuration for detecting and removing each in stages is based on the following principle.

That is, the ghost detection reference waveform (GCR waveform) extracted from the received television signal and the time series signal sequence before and after it are defined as g (t), and the frequency spectrum is defined as G (jω). Further, if the transfer characteristic of the ideal ghost elimination device is H (jω) and the frequency spectrum of the ghost detection reference waveform before being distorted by the ghost is R (jω), then G (jω) × H (Jω) = R (jω) (1)

From the equation (1), the transfer characteristic H (jω) of the ideal ghost eliminating device is H (jω) = R (jω) / G (jω) (2).

G (jω) can be obtained by Fourier transforming the impulse response. For example, it can be easily obtained by using a sin x / x pulse waveform as the reference waveform. Also, sin x / x bar signals and vertical sync signals can be handled as step responses, so they can be converted to impulse responses by differentiating them, and G
(Jω) can be obtained.

If a signal in the near ghost region is extracted from the time series signal h (t) obtained by inverse Fourier transforming H (jω) in the equation (2), this is the impulse response h1 (t of the near ghost removing section. ). Further, from the time series signal g (t) of the GCR signal, the time series signal g1 (t) in the proximity region
By cutting out, the tap coefficient group of the pseudo-near ghost generation transversal filter can be calculated based on the low-order Fourier transform and the inverse Fourier transform.

The frequency spectrum Y (j
ω) is the impulse response h1 of the proximity ghost elimination unit
Transfer characteristic H1 (j obtained by Fourier transform of (t)
From (ω), it is calculated as Y (jω) = G (jω) · H1 (jω) (13).

Therefore, the transfer characteristic H2 (jω) of the normal ghost removing unit for equalizing Y (jω) to R (jω).
H2 (jω) = R (jω) / Y (jω) = 1 / [1-X (jω)] (14) where X (jω) = 1-G (jω) · H1 (jω) / R (jω) (15)

Therefore, the time-series signal group obtained by inverse Fourier transform of X (jω) in the equation (15) is tapped by the transversal filter for generating the pseudo normal ghost in the normal ghost removing section which is constructed as a cyclic type. It may be set as a coefficient group. In practice, in view of the finite band of R (jω), the band of X (jω) is limited by R (jω), and X (jω) ≈R (jω) -G (jω) · H1 ( The time series signal group obtained by performing the inverse Fourier transform of jω) (16) may be set as the tap coefficient group of the pseudo normal ghost generating transversal filter.

Therefore, the processing contents of the proximity ghost detection unit 32 and the normal ghost detection unit 34 of FIG. 4 are as shown in FIG. The processing contents of the proximity ghost detector and the normal ghost detector will be described with reference to the flowchart shown in FIG. 8 and the window characteristic diagram shown in FIG.

First, in the proximity ghost detector 32,
The time-series signal g (t) is extracted over the near region and the normal region of the ghost detection reference waveform (GCR waveform) included in the received television signal converted into the digital signal (step 51). Next, the time-series signal g1 in the proximity region is displayed using the window having the characteristics shown in FIG.
(T) is cut out (step 52), and the Fourier coefficient group G1 (jω) is generated by Fourier transform (step 53).

Since the number of sampling points of the time-series signal is small, a small value is sufficient for the order of the Fourier transform, and the order is typically about 128. Next, division is performed with respect to the Fourier coefficient group R (jω) of the undistorted reference waveform held in advance to generate the coefficient group H1 (jω) (step 54).

The order of this coefficient group H1 (jω) is 12
The inverse Fourier transform of 8 is performed to generate the time series signal h1 (t) (step 55). This time series signal h1
(T) is cut out as a time-series signal c (t) by using the characteristic window shown in FIG. 9C (step 5).
6) The tap coefficient group is supplied to the transversal filter 31a for generating the pseudo proximity ghost in the proximity ghost removing section.

Further, in the above step 56, FIG.
The time-series signal k (t) cut out by the window having the characteristic shown in (D) has a sampling point of zero amplitude over the normal region added in order to match the order with the processing data of the normal ghost detection unit 34. Has a degree of 102
Fourier transform of 4 is performed (step 57). This Fourier coefficient group K (jω) is normally supplied to the ghost detection unit 34. This supply path is formed between the output terminal Eo and the input terminal Ei of FIG.

On the other hand, in the normal ghost detecting section 34, the time series signal g (t) of the GCR signal is extracted by the processing step which can be shared with the processing step 51 of the proximity ghost detecting section 32, as shown in FIG. The proximity area and the normal area are cut out using the characteristic window shown in (B) (step 61). Since this time-series signal g (t) includes a large number of sampling points, it is typically Fourier-transformed with an order of 1024 (step 62) and passed to the processing step 63 as a Fourier coefficient group G (jω).

In processing step 63, processing step 62
Fourier coefficient group G (jω) that has passed through and Fourier coefficient group K (j that has passed through processing step 57 of the proximity ghost detection unit 32.
ω) is calculated, and this product is subtracted from the distortion-free GCR Fourier coefficient group R (jω) held in advance to generate X (jω) in the above-mentioned equation (16). .

The Fourier coefficient group X (jω) is subjected to the inverse Fourier transform in step 65 after the high-level side is suppressed by the non-linear processing in step 64, and is used for generating a pseudo normal ghost forming the normal ghost removing section 33. It is supplied to the transversal filter 33a.

[0030]

As described above, the pseudo ghost generating transversal filter 31a is used.
And 33a, as shown in FIG. 7, delay devices 42a, 42b, 42c ... 4 which delay the digital signal obtained by A / D converting the received television signal by one sampling cycle.
2n and multipliers 43a, 43b, 43c, ... 43n for multiplying the signal extracted from each delay device by a tap coefficient
And an adder 44 that combines the outputs of the multipliers.

The multipliers 43a, 43b, 43c ...
Since 43n is composed of a digital circuit, the number of processing bits here greatly affects the hardware scale. The number of processing bits can be determined by the performance required of the ghost elimination device. That is, as the number of bits increases, more accurate ghost removal or ghost removal with a larger amplitude becomes possible.

Further, even if the number of bits is the same, by changing the gain of the A / D-converted television signal input / output to / from the multiplier, it is possible to focus on either accuracy or amplitude. Become. Typically, the number of bits of the tap coefficient supplied to the multiplier is 8 bits with a sign, and the gain of the A / D-converted television signal input / output to / from the multiplier is determined by the pseudo proximity ghost transversal filter 31a. 0 dB, pseudo normal ghost transversal
The filter 33a sets -6 dB.

When the multiplier is configured with the above-mentioned number of bits and gain, the maximum ghost level capable of removing ghost is represented by the D / U ratio represented by the ratio of the magnitudes of the desired wave and the disturbing wave.
The proximity ghost is 0 dB, and the normal ghost is 6 dB.
Also, the minimum ghost level D /
The U ratio is 42 dB for the proximity ghost and 48 for the normal ghost.
It becomes dB.

However, since the proximity ghost has a small time difference between the original signal and the ghost signal, the relationship between the size of the ghost and the actual image quality deterioration is not clear. That is,
As shown in FIG. 10, when the original signal has a waveform A like a dotted line and the waveform distortion illustrated by the solid line portions α and β occurs due to the proximity ghost, the actual image quality is higher than that of the ghost D / U ratio. It feels little deterioration.

Since the proximity ghost has a small delay time difference between the original signal and the ghost signal, the effect of suppressing the ghost signal due to the directional characteristics of the receiving antenna in radio wave propagation cannot be expected, and an excessive proximity ghost may occur. For example, in FIG.
When a ghost as shown in (4) is to be removed, at the time when the signal shown by a in the figure is output, it is shown by a.
It is necessary to increase the amplitude to the amplitude indicated by b.

As shown in FIG. 5, the proximity ghost removing section 31 is configured to add the output of the pseudo proximity ghost generating transversal filter 31a and the input signal delayed by the delay circuit 31b by the adder 31c. Therefore, the gain required for the multiplier in the pseudo-close proximity ghost generation transversal filter 31a when there is a proximity ghost as shown in FIG. 10 is as follows for the multiplier having a delay time equal to the delay time of the delay circuit 31b : b / a-1.

[0037] Thus, for example, when the amplitude b is in more than twice the amplitude a is as shown in FIG. 10, the gain required for the multiplier 0
It will exceed dB. In reality, a gain exceeding 0 dB cannot be realized, so that not only the ghost is not removed, but there is a serious problem that the input signal is deteriorated in some cases.

If an attempt is made to cope with a ghost having a large amplitude by increasing the gain of the multiplier, the resolution will be poor and the residual ghost will increase. Therefore, if the maximum amplitude of the ghost signal is increased and the resolution is set to sufficiently reduce the residual ghost, the number of bits of the multiplier must be increased, which has a drawback of increasing the hardware scale.

[0039]

In order to solve the above-mentioned problems, according to the ghost elimination device of the present invention, a tap coefficient group is provided in the transversal filter for pseudo-near ghost generation of the non-recursive proximity ghost elimination section. The proximity ghost detection unit to be set performs a Fourier transform on the digital signal extracted from the proximity ghost area of the received television signal, divides the Fourier coefficient group of the ghost detection reference waveform held in advance by this Fourier coefficient group, and then inverses the result. A time-series signal group is generated by performing a Fourier transform, and the amplitude excess of the signal whose amplitude exceeds the amplitude that can be supplied to the pseudo proximity ghost generation transversal filter is calculated. After the signals are distributed to nearby signals in the sequence, the tap coefficient group is added to the transversal filter for generating the pseudo proximity ghost. And characterized in that it comprises means supplying.

With the above configuration, the tap coefficient is limited to the amplitude that can be supplied to the transversal filter for generating the pseudo-near ghost, but the excess amount of the amplitude caused by this limitation is tapped adjacently in time series. By assigning to the coefficients, the total sum of the coefficient values of the tap coefficient group does not change. Therefore, according to the present invention, it is possible to prevent the DC component error from occurring in the proximity ghost removing section, which means that at least an image amplitude error occurs in the video signal output from the proximity ghost removing section. It shows that there is no.

Further, the time interval of the taps which are adjacent in time series in the pseudo proximity ghost generating transversal filter is equal to one sampling period of the digital signal. The reciprocal of the sampling period is a frequency band in which the digital signal processing is possible, and is constituted by a time in which there is no problem in performing the digital processing of the video signal. That is, the effect of moving a part of the tap coefficient values to adjacent taps is limited to relatively high frequency components in the frequency band of the video signal.

As described above, since the proximity ghost has a small time difference between the original signal and the ghost signal, the relationship between the size of the ghost and the actual image quality deterioration is not clear. Although a slight waveform distortion does not cause a problem, an amplitude error is easy to detect, which is a serious problem. In the present invention, for a signal that exceeds the range that can be supplied to the pseudo proximity ghost generation transversal filter, the amount that exceeds the range is distributed to adjacent tap coefficients in time series, thereby eliminating the proximity ghost. The amplitude error of the video signal output from the unit is suppressed to achieve stable operation.

[0043]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a flow chart showing the procedure of processing according to an embodiment of the present invention. The ghost removing apparatus of the present invention has a basic configuration in which the adjacent ghost removing section 31 and the normal ghost removing section 33 are connected in cascade and are supplied to the input terminal IN, as in the conventional apparatus shown in FIG. The original television for ghost removal is configured to be supplied to the proximity ghost detection unit 32 and the normal ghost detection unit 34.

Further, the proximity ghost removing section 31 includes a transversal filter 31a for generating pseudo proximity ghosts, a delay circuit 31b, as in the conventional apparatus shown in FIG.
The adder circuit 31c and the adder circuit 31c are connected in a non-cyclic manner. Further, the normal ghost removing unit 33 is also shown in FIG.
Similar to the conventional device shown in FIG. 3, the pseudo normal ghost generation transversal filter 33a and the adder circuit 33b are cyclically connected to each other.

The transversal filters 31a and 33a both have a delay group of one sampling period, a multiplier group for multiplying the output of each delay unit by a tap coefficient, as in the conventional device shown in FIG. It is composed of an adder that combines the outputs of these multiplier groups.

The processing contents of the proximity ghost detection unit and the normal ghost detection unit according to the present invention will be described below with reference to the flowchart shown in FIG. 1 and the window characteristic diagram of FIG.

First, in the proximity ghost detector 32,
A time-series signal g (t) extending over a near region and a normal region of the ghost detection reference waveform (GCR waveform) included in the received television signal converted into a digital signal is extracted (step 11). Next, the time-series signal g1 in the proximity region is displayed using the window having the characteristics shown in FIG.
(T) is cut out (step 12), and the Fourier coefficient group G1 (jω) is generated by Fourier transform (step 13).

The order of this Fourier transform may be a small value because the number of sampling points of the time series signal is small, and is typically about 128. Next, division is performed with the Fourier coefficient group R (jω) of the undistorted reference waveform held in advance to generate the coefficient group H1 (jω) (step 14). The order 128 for this coefficient group H1 (jω)
The inverse Fourier transform is performed to generate the time series signal h1 (t) (step 15).

This time series signal h1 (t) is shown in FIG. 9 (C).
A dispersion process that is cut out as a time-series signal c (t) using the window having the characteristics shown in (16) and is within the maximum amplitude that can be supplied to the pseudo-close proximity ghost generation transversal filter in the close proximity ghost elimination section. Is performed (step 18), and the tap coefficient group is supplied to the transversal filter for generating pseudo proximity ghost.

Further, in step 16 described above, FIG.
The time-series signal k (t) cut out by the characteristic window shown in (D) has a sampling point of zero amplitude over the normal region added to match the order with the processing data of the normal ghost detection unit, and typically, Is subjected to Fourier transform of order 1024 (step 17). This Fourier coefficient group K (jω) is normally supplied to the ghost detection unit 34. This supply path is formed between the output terminal Eo and the input terminal Ei of FIG.

The distributed processing in step 18 will be described with reference to FIG. FIG. 2 (A) shows a part of the tap coefficient group c (t) that has been cut out in step 16 before the distributed processing, and the amplitude of each tap coefficient is the length of the arrow in FIG. Is equivalent to

The broken line represented by M in FIG. 2 indicates the position corresponding to the maximum amplitude of the tap coefficient that can be supplied to the pseudo-near ghost generation transversal filter. Here, the tap coefficient c (3) exceeds the maximum amplitude M that can be supplied, and its magnitude is α. In the distributed processing according to the present invention, the amplitude of c (3) is limited to the maximum amplitude M that can be supplied, and the amplitude α is divided into adjacent taps c (2) and c (4) in half. Sort. As a result of the distributed processing, the tap coefficient group shown in FIG. 2B is obtained.

On the other hand, the normal ghost detecting section 34 has the same structure as the conventional apparatus shown in FIG. That is, the time-series signal g (t) of the GCR signal is extracted by the processing step that can be shared with the processing step 11 of the proximity ghost detection unit 32, and the characteristic window shown in FIG. 9 (B) is used. The adjacent area and the normal area are cut out (step 21).

Since this time-series signal g (t) includes a large number of sampling points, it is typically Fourier-transformed with an order of 1024 (step 22), and the Fourier coefficient group G (j
ω) is passed to processing step 23. In processing step 23, the Fourier coefficient group G that has been subjected to processing step 22
(Jω) and processing step 17 of the proximity ghost detection unit 32
The product with the Fourier coefficient group K (jω) which has passed through is calculated, and this product is subtracted from the Fourier coefficient group R (jω) of the undistorted GCR stored in advance (16).
The expression X (jω) is generated.

This Fourier coefficient group X (jω) is subjected to the inverse Fourier transform in step 25 after the high-level side is suppressed by the non-linear processing in step 24, and is used for pseudo normal ghost generation which constitutes the normal ghost removing section 33. It is supplied to the transversal filter 33a.

When the time-series signal c (t) output from the above step 16 greatly exceeds the maximum amplitude that can be supplied to the pseudo proximity ghost generating transversal filter 31a, or when adjacent in time series. If both of the signals are above the maximum amplitude M that can be supplied, it may not be possible to limit all of the time series signals c (t) within the range of the maximum amplitude M that can be supplied by the distributed processing.

In such a case, it is effective to repeat the above dispersion processing. Further, when the number of times of repeating the distributed processing does not end within the predetermined number, or when the amplitude of the time series signal c (t) has a very large value, the transversal filter 31a for pseudo proximity ghost generation is generated. It is possible to prevent transient malfunction due to noise or the like and to improve the stability of the operation by not supplying the tap coefficient group as the tap coefficient group.

[0058]

The ghost elimination device of the present invention outputs the tap coefficient group from the proximity ghost detection section within the maximum amplitude range of tap coefficients that can be supplied to the transversal filter for pseudo proximity ghost generation in the proximity ghost removal section. Therefore, a stable operation can be realized even with an excessive proximity ghost.

[Brief description of drawings]

FIG. 1 is a flowchart showing an embodiment of a processing procedure of the present invention.

FIG. 2 is a conceptual diagram for explaining the processing operation of FIG.

FIG. 3 is a waveform diagram for explaining the concept of a proximity ghost and a normal ghost.

FIG. 4 is a block diagram showing a configuration of a ghost removing device to which the present invention is applied.

5 is a block diagram showing a configuration of a proximity ghost removing section 31 of FIG.

6 is a block diagram showing a configuration of a normal ghost removing section 33 of FIG.

FIG. 7 is a block diagram illustrating the configuration of the transversal filter for generating pseudo ghosts shown in FIGS. 5 and 6;

FIG. 8 is a flowchart showing a procedure of conventional processing.

9 is a conceptual diagram showing characteristics of various windows used for cutting out time-series signals in the processes of FIGS. 1 and 8. FIG.

FIG. 10 is a waveform diagram for explaining the concept of an excessive proximity ghost.

[Explanation of symbols]

IN Input terminal of television signal for ghost removal OUT Output terminal of television signal subjected to ghost removal 31 Proximity ghost removal section 31a Pseudo-proximity ghost generation transversal filter 31b Delay circuit 31c Adder circuit 32 Proximity ghost detection section 33 Normal Ghost Remover 33a Pseudo Normal Ghost Generation Transversal Filter 33b Adder 34 Normal Ghost Detector 42a-42n Delay Circuit 43a-43n Multiplier 44 Adder

Continuation of the front page (56) Reference JP-A-4-77180 (JP, A) JP-A-58-177081 (JP, A) JP-A-4-46468 (JP, A) JP-A-8-317254 (JP , A) JP 56-104586 (JP, A) JP 4-336881 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) H04N 5/14-5/217

Claims (4)

(57) [Claims] 1. A transversal filter for generating a pseudo proximity ghost by detecting a proximity ghost that appears in the vicinity of a ghost detection reference waveform included in a received television signal.
A proximity ghost detection unit for generating a tap coefficient group to be set to a normal ghost detection unit for generating a tap coefficient group to be set to the detected pseudo-normal transversal filter for generating ghost normal ghost appearing after the close ghosts , The transversal for generating the pseudo proximity ghost
A non-cyclic type near-field ghost removing unit having a filter and a combining means for combining the generated pseudo-near ghost with a television signal to be ghost-removed, a transversal filter for generating the pseudo-normal ghost, and this generated In a ghost elimination device having a recursive normal ghost elimination unit having a combining means for synthesizing a pseudo normal ghost with a television signal of a ghost elimination target outputted from the proximity ghost elimination unit, the proximity ghost detection unit is , Fourier transforming the digital signal extracted from the near ghost region of the received television signal, dividing the Fourier coefficient group of the ghost detection reference waveform held in advance by this Fourier coefficient group, and then performing the inverse Fourier transform. Means for generating a series signal group and the generated time series signal To the group, the touch
If the signal amplitude of the flop coefficient exceeds the maximum amplitude that can be supplied to the pseudo close ghost generation transversal filter is present, tap exceeds the maximum amplitude with limit the signal to the maximum amplitude It is provided with a tap coefficient distribution processing means for performing tap coefficient distribution processing by distributing excess coefficients to nearby signals on a time series and then supplying them to the pseudo proximity ghost generation transversal filter as a tap coefficient group. A ghost removing device characterized by the above. 2. The tap coefficient dispersion processing means repeatedly performs the tap coefficient dispersion processing when all the signals of the time series signal group cannot be restricted within the maximum amplitude range by one tap coefficient dispersion processing. The ghost removing device according to claim 1, further comprising means. 3. The tap coefficient distribution processing means, when the number of repetitions of the tap coefficient distribution processing is equal to or greater than a predetermined number, sets the tap coefficient group to the pseudo proximity ghost generation transversal filter. The ghost removing device according to claim 2, wherein the ghost removing device is not supplied. 4. The tap coefficient dispersion processing means, when the amplitude of the tap coefficient of the time-series signal group exceeds a predetermined range, applies the tap to the pseudo proximity ghost generating transversal filter. The ghost removing device according to claim 1, wherein the coefficient group is not supplied.