JPH08251450A - Ghost eliminating device - Google Patents

Abstract

PURPOSE: To eliminate quickly ghost by correcting a tap coefficient while taking a steep increase change increasing a steepness evaluation function representing the steepness of the distribution of tap coefficients into account so as to reduce the number of times of correction. CONSTITUTION: The device is provided with an error decrease change arithmetic section 2 to calculate an error decrease change (d) to attenuate an error evaluation function E to be a minimum value and a steep increase change arithmetic section 5 calculating a steep increase change (f) to increase a steepness evaluation function P and correcting a tap coefficient (c) based on the error decrease change (d) and the steepness increase change (f).

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 for removing a ghost component contained in a received signal of a television receiver or the like.

[0002]

2. Description of the Related Art Conventionally, a radio wave transmitted from a transmitting antenna of a broadcasting station directly arrives at a receiving antenna of a television receiver, or is reflected by a building such as a building or a mountain and is delayed with the receiving antenna. When these radio waves are received at the same time, images due to the radio waves are superimposed on the received signal with a shift, and so-called ghost failure occurs in which multiple images appear on the screen. Then, a television receiver may be provided with a ghost removing device in order to eliminate such a ghost obstacle.

The ghost removing device removes only the ghost component from the received signal by processing with a filter circuit having a characteristic opposite to the frequency characteristic of the transmission system which causes the ghost according to the same principle as the waveform equalization. To do. Although such a filter circuit can be realized by an analog circuit, it is not very practical. Therefore, hereinafter, a digital ghost removing apparatus will be described.

Since the filter circuit used in the ghost elimination device must be capable of adjusting the frequency characteristic in order to cope with various ghost disturbances, a transversal filter 11 as shown in FIG. 5 is generally used. This transversal filter 11 has N-1 delay devices 11a and N storage devices 11b, where N is the number of taps.
And N multipliers 11c and N-1 adders 11d. The delay device 11a is a circuit that sequentially delays the input signal by a fixed period (for example, one sampling period), and the storage device 11b is a circuit that stores _N tap coefficients c _{0 to} c _N−1 , respectively. Also,
The multiplier 11c is a circuit that multiplies the input signal or the signal delayed by each delayer 11a by the tap coefficients c _{0 to} c _{N -1} stored in each storage 11b, and the adder 1c
1d is a circuit for adding all the multiplication results of the multiplier 11c. Then, the addition result of these adders 11d becomes an output signal. Further, it is possible to obtain an arbitrary frequency characteristic by appropriately changing the tap coefficients c _{0 to} c _N-1 stored in each memory 11b.

The filter circuit uses the FIR [Finite Impulse] for the transversal filter 11 as shown in FIG.
Response] type configuration, or as shown in FIG.
It is used as an IR [Infinite Impulse Response] type configuration, or as shown in FIG. 8, a combination of these FIR type configuration and IIR type configuration is used. The FIR type configuration inputs an input signal to the transversal filter 11 and outputs the processing result as it is. Therefore, if the k-th (for example, the k-th sample) input signal is represented by x _k , the k-th output signal y _k of this filter circuit is expressed by Equation 1, and the impulse response becomes a finite and stable filter circuit.

[0006]

[Equation 1]

Further, since the IIR type configuration feeds back the processing result of the transversal filter 11 to the input signal, the impulse response is normally continuous infinitely, and a sharp frequency characteristic can be obtained with a small number of taps. Instead of being able to obtain, it becomes necessary to consider stability. Each of these configurations is appropriately selected according to the desired specifications of the filter circuit, but basically any configuration can obtain almost arbitrary frequency characteristics.

The ghost removing device removes a ghost by modifying the tap coefficient of the transversal filter 11 so that the filter circuit has a characteristic opposite to the frequency characteristic of the transmission system. The frequency characteristic of the transmission system is obtained by comparing the input signal with an internal reference signal having the original reference signal waveform generated inside the receiver, using the portion of the input signal whose waveform is known as the reference signal. To detect. As the reference signal, in addition to using an existing waveform signal such as an equivalent pulse superimposed on the vertical synchronizing signal period of the input signal or the leading edge of the vertical synchronizing signal, it is standardized for removing ghosts. Newly inserted GCR
In some cases, the [Ghost Canceling Reference] signal is used. Further, the method of determining the tap coefficient of the transversal filter 11 based on the comparison of the reference signals is roughly classified into the batch calculation method and the successive correction method. The collective operation method is a method in which a frequency characteristic required by a filter circuit is calculated from a Fourier transform of an input signal and an internal reference signal, and a tap coefficient of the transversal filter 11 is calculated based on a result of inverse Fourier transform of the frequency characteristic. Therefore, the tap coefficient can be determined by one calculation. However, in this batch operation method, the device becomes expensive when trying to shorten the processing time, and therefore, in general, the sequential correction method is often used.

The successive correction method is to correct the tap coefficient of the transversal filter 11 little by little each time the reference signal of the input signal appears, so that the filter circuit converges to the required frequency characteristic. That is, if the value of the i-th tap coefficient c _i after n times of correction is represented by c _i ⁽ⁿ⁾ and the displacement amount of one time of correction is represented by d _i , the value after _i + 1 times of correction c _i ^{(n +1)} is shown by the ^equation 2.

[0010]

[Equation 2]

At this time, the value of the displacement amount d _i is the tap coefficient c _i.
Must be able to converge to the required value.
Then, as a method for obtaining such a value of the displacement amount d _i , MSE [Mean Squa] whose convergence is mathematically proven
re Error] method is generally used.
The SE method will be described. First, the error signal e is defined by Equation 3 from the difference between the output signal y of the filter circuit and the internal reference signal r,

[0012]

(Equation 3)

The error evaluation function E is defined by Equation 4 as the square integral value of the error signal e.

[0014]

[Equation 4]

The tap coefficient c of the transversal filter 11 may be determined by the method of least squares so that the error evaluation function E has a minimum value.

Here, it has been proved by the method of least squares that the error evaluation function E of the equation 4 has a unique minimum point when the tap coefficient c _i is used as a parameter. Then, if the displacement amount d _i is determined according to the value obtained by partially differentiating the error evaluation function E by the tap coefficient c _i , the error evaluation function E reaches the minimum point while the correction by the equation 2 is repeated. .
Therefore, in the following, this displacement amount d will be used as the error reduction displacement amount d.
Called. When the error evaluation function E of the equation 4 is partially differentiated by the tap coefficient c _i , the equation 5 is obtained,

[0017]

(Equation 5)

Here, the partially differentiated error signal e _k is given by
, The internal reference signal r _k is irrelevant to the tap coefficient c _i , so that only the term of the output signal y _k is transformed into the remaining number 6.

[0019]

(Equation 6)

For simplicity, if the filter circuit is of FIR type, the output signal y _k of equation 6 is represented by equation 1 and the terms other than c _i · x _ki are irrelevant to the tap coefficient c _i . Therefore, it is transformed into Equation 7.

[0021]

(Equation 7)

That is, the partial differentiation of the error evaluation function E of the equation 4 by the tap coefficient c _i is represented by the correlation between the error signal e and the input signal x which is advanced by the period i.

[0023]

(Equation 8)

Therefore, if the error reduction displacement amount d _i is determined as in the equation 8, the tap coefficient c is changed by the equation 2 in the direction opposite to the partial differential of the error evaluation function E by an amount corresponding to the constant α. By repeating this, it is possible to obtain the tap coefficient c that minimizes the error evaluation function E of the equation 4 and converge the filter circuit to the required frequency characteristic. The value of the constant α in the equation 8 is empirically determined in consideration of the balance between the convergence speed and the stability in the equation 2. That is, if the value of the constant α is too small, the stability will increase, but it will take time to converge. On the other hand, if the value is too large, the value may converge quickly, but the stability becomes poor. There is a risk of oscillation. Therefore, the constant α
Is set to a value such that it can converge in the shortest possible time while ensuring sufficient stability.

FIG. 9 shows a specific configuration of a conventional ghost removing apparatus that determines the tap coefficient of a filter circuit based on the above-mentioned successive correction method. The filter circuit 1 is assumed to have the FIR type configuration shown in FIG. 6 for simplicity.

The input signal x received by the television receiver and AD-converted is input to the filter circuit 1 and expressed by
Is output as the output signal y. The internal reference signal generator 4 generates an internal reference signal r in synchronization with the input signal x each time a reference signal appears in the input signal x.
Then, the error signal calculation unit 3 calculates the error signal e from the difference between the output signal y and the internal reference signal r based on the equation 3 for each value from 0 to N−1 of the subscript k. The error reduction displacement amount calculation unit 2 temporarily stores the error signal e and the input signal x, and for each value from 0 to N−1 of the subscript i, the error signal e and the input signal x are calculated based on Equation 8. The error reduction displacement amount d is calculated by calculating the correlation with and the constant α times, and the error reduction displacement amount d and the initial value or the previous tap coefficient c ⁽ⁿ⁾ to the current tap coefficient c are calculated based on Equation 2. Calculate ^{(n + 1)} . The tap coefficient c ⁽ⁿ⁾ is read from the tap coefficient storage unit 6, and the newly calculated current tap coefficient c ^{(n + 1)} is sent to the tap coefficient storage unit 6. The tap coefficient storage unit 6 is a circuit that stores the tap coefficient c, and sequentially inputs the current tap coefficient c ^{(n + 1)} calculated by the error reduction displacement amount calculation unit 2 to update the storage content. Further, the tap coefficient storage unit 6 corrects the tap coefficient c of the filter circuit 1 based on the updated tap coefficient c. Then, every time when the reference signal appears in the input signal x, this process is repeated until the error evaluation function E of the equation 4 becomes small to some extent, so that the final tap coefficient c of the filter circuit 1 is determined, thereby eliminating the ghost. be able to.

A specific operation example of the ghost removing device will be schematically shown with reference to FIG. It is assumed that ghosts G ₁ and G ₂ appear in two places in the input signal x with respect to the reference signal R. The initial value of the tap coefficient c of the filter circuit 1 is assumed to be 1 only for the tap coefficient c ₀ and 0 for all others. Therefore, the uncorrected output signal y of the tap coefficient c is exactly the same as the input signal x. The internal reference signal r is a signal consisting of only the component of the reference signal R in the input signal x, and the error signal e is the internal reference signal r.
Since the difference between the output signal y and, a signal consisting of only the component of the ghost G _1, G _2. The error reduction displacement amount d
Ghosts G ₁ and G ₂ of the error signal e have a high correlation with the reference signal R of the input signal x.
_The absolute value becomes large near the positions of ₁ and G ₂ , and the tap coefficient c is updated accordingly. Further, when the tap coefficient c is once corrected in this way, the output signal y when the reference signal R appears next in the input signal x is the ghost G ₁ ,
The waveform has a slightly relaxed G ₂ . Then, when the correction processing of the tap coefficient c is repeated until the error evaluation function E of Equation 4 becomes sufficiently small, the tap coefficient c becomes a ghost G ₁ ,
Each becomes a value attenuated as the distance has a peak amplitude at the position of G _2, whereby the ghost G _1, G ₂ is removed from the output signal y.

In the above ghost removing apparatus, the filter circuit 1 has the FIR type configuration, but the filter circuit 1 can be also realized in the IIR type configuration or a combination thereof. In this case, for example, the tap coefficient c for the FIR type configuration corrected by the tap coefficient storage unit 6 may be converted into the tap coefficient c of another configuration having the same frequency characteristic, or
May be replaced with a mathematical formula according to the filter circuit 1 having another configuration.

[0029]

By the way, in the conventional ghost removing apparatus based on the above-described successive correction method, the distribution of the tap coefficient of the filter circuit 1 is increased as the number of corrections increases.
It becomes steep gradually as shown by the arrow. However,
Since the actual error reduction displacement amount d is a small value, the change of the tap coefficient is very gradual, and especially the error evaluation function E
After .apprxeq. Approaches the minimum value, the value of the error reduction displacement amount d becomes extremely small, and the change in the tap coefficient also extremely decreases.

Therefore, in the conventional ghost removing device,
There has been a problem that the number of tap coefficient corrections required until the error evaluation function E becomes sufficiently small becomes extremely large, and it takes a long time until the final tap coefficient is determined and the ghost is reliably removed. Moreover, when a signal with a small high frequency component such as an equivalent pulse is used as the reference signal for removing the ghost, the error reduction displacement amount d becomes a smaller value, so that the change of the tap coefficient becomes slower. It will take a longer time until the final tap coefficient is determined.

The present invention solves the above-mentioned conventional problems, and eliminates ghosts quickly by modifying the tap coefficient in consideration of the steeply increasing displacement amount that makes the tap coefficient distribution steep. It is an object of the present invention to provide a ghost removing device capable of

[0032]

A ghost removing device of the present invention is a ghost removing device for obtaining an output signal from which a ghost component is removed by processing a received signal with a filter circuit, and updating a tap coefficient of the filter circuit. In this case, the error reduction displacement amount calculating means for calculating the error reduction displacement amount of the tap coefficient capable of reducing the error between the output signal and the internal reference signal generated inside the receiver, and the tap coefficient of the filter circuit are A steep increase displacement amount calculating means for calculating a steep increase displacement amount of the tap coefficient capable of making the distribution of the tap coefficient steep when updated; and an error decreasing displacement amount calculated by the error decreasing displacement amount calculating means. And a tap coefficient updating means for updating the tap coefficient of the filter circuit based on the steep increasing displacement amount calculated by the steep increasing displacement amount calculating means. The object can be achieved.

[0033]

With the above structure, the tap coefficient updating means considers not only the error reduction displacement amount calculated by the error reduction displacement amount calculation means but also the steep increase displacement amount calculated by the steep increase displacement amount calculation means. Update the tap coefficient of. Therefore, compared with the conventional ghost elimination device that updates the tap coefficient based on only the error reduction displacement amount, the distribution of the tap coefficient becomes sharper rapidly, and the error evaluation function approaches a sufficiently small value with a small number of corrections. The time required for removing the ghost can be shortened.

[0034]

Embodiments of the present invention will be described below.

1 to 4 show an embodiment of the present invention. FIG. 1 is a block diagram showing the configuration of a ghost removing device, and FIG. 2 is a diagram showing changes in the steepness of tap coefficient distribution. FIG. 3 is a diagram showing how the error evaluation function is reduced in a ghost removing apparatus using a GCR signal as a reference signal,
FIG. 4 is a diagram showing how the error evaluation function is reduced in the ghost removing apparatus using the equivalent pulse as the reference signal. The constituent members having the same functions as those of the conventional example shown in FIG. 9 are designated by the same reference numerals.

The input signal x received by the television receiver and AD-converted is input to the ghost eliminating device,
As shown in FIG. 1, it is sent to the filter circuit 1 and the error reduction displacement amount calculator 2. The filter circuit 1 may have an FIR type configuration, an IIR type configuration, or a configuration in which these are combined, but here, the FIR type configuration is used for simplicity. The input signal x is filtered by the filter circuit 1 and is output as an output signal y represented by Formula 1. Then, this output signal y is sent to the subsequent circuits as a signal obtained by removing the ghost from the input signal x.

The output signal y is also input to the error signal calculator 3. The internal reference signal r generated by the internal reference signal generator 4 is also input to the error signal calculator 3. The internal reference signal generator 4 is a circuit that generates an internal reference signal r in synchronization with the input signal x each time a reference signal appears in the input signal x. The error signal calculator 3 is
The output signal y and the internal reference signal r are input, and the error signal e is calculated based on the equation 3 for each value of the subscript k from 0 to N-1. The error signal e calculated here is input to the error reduction displacement amount calculation unit 2. The error reduction displacement amount calculation unit 2 temporarily stores the input signal x and the error signal e, and for each value from 0 to N−1 of the subscript i, the input signal x and the error signal e are calculated based on Equation 8. The error reduction displacement amount d is calculated by calculating the correlation of
Is a circuit for calculating. The error reduction displacement amount d calculated here is input to the steep increase displacement amount calculation unit 5.

The steep increase displacement amount calculation unit 5 calculates the tap coefficient c.
The steep increase displacement amount f for making the distribution of steep is calculated, and the error decrease displacement amount d and the steep increase displacement amount f are calculated.
The tap coefficient c is corrected based on
Is a circuit for updating the tap coefficient c stored by The steeply increasing displacement amount f is obtained by multiplying the tap coefficient c _i ⁽ⁿ⁾ after n corrections by n + for each value from 0 to N−1 of the subscript i.
The value is set so that the steepness of the distribution of the tap coefficient c is increased by repeating the process of correcting the tap coefficient c _i ^{(n + 1)} after the correction once.

[0039]

[Equation 9]

Therefore, first, the steepness evaluation function P indicating the steepness of the distribution of the tap coefficient c is defined by Equation 10.

[0041]

[Equation 10]

At this time, the steepness value P _j for each value of the subscript j from 1 to 5 is expressed by equation 11.

[0043]

[Equation 11]

The term of the equation 12 in the steep value P _j is a kind of differential value indicating the slope of the change of the tap coefficient c in the interval j,

[0045]

(Equation 12)

The sum of squares of the differential value shown in the equation 11 becomes larger as the distribution of the tap coefficient c becomes steeper. Then, the steepness evaluation function P shown in Expression 10 is 1 to 5
The steepness of the distribution of the global tap coefficient c can be expressed by adding the steepness values P _j for the respective intervals j of.

[0047] When partially differentiated by the tap coefficients c _i of the steepness evaluation function P, since terms that do not contain the tap coefficients c _i of sharp values P _j are all 0, as shown in Formula 13.

[0048]

(Equation 13)

Therefore, the steeply increasing displacement amount f is represented by the subscript i.
When each value from 0 to N-1 is determined as in Equation 14, each tap coefficient c is calculated by Equation 9 as the steepness evaluation function P.
The value of the steepness evaluation function P can be increased to make the distribution of the tap coefficient c steeper even more.

[0050]

[Equation 14]

However, if the steepness is increased more than necessary, the error signal e becomes too large and the error evaluation function E is increased.
Since the convergence of the error signal becomes worse, it is preferable to multiply the constant β by a value that becomes smaller as the error signal e decreases. Therefore, in the present embodiment, the steeply increasing displacement amount f is calculated by Equation 15 using the error decreasing displacement amount d whose absolute value decreases as the error signal e decreases.

[0052]

(Equation 15)

Further, the steep increase displacement amount calculation unit 5 taps each value from 0 to N-1 of the subscript i by the formula 16 based on the error decrease displacement amount d and the steep increase displacement amount f. Correct the coefficient c.

[0054]

[Equation 16]

Therefore, the tap coefficient c is not only corrected by the error reduction displacement amount d so that the error evaluation function E becomes smaller, but also corrected by the steeply increased displacement amount f so that the steepness is increased. Then, the tap coefficient c modified here is sent to the tap coefficient storage unit 6. The tap coefficient storage unit 6 is a circuit for storing the tap coefficient c, and a new tap coefficient c from the steep increase displacement amount calculation unit 5 is stored.
Is sent, the stored contents are updated based on this and the tap coefficient c of the filter circuit 1 is corrected. Then, every time when the reference signal appears in the input signal x, this process is repeated until the error evaluation function E becomes small to some extent, so that the final tap coefficient c of the filter circuit 1 is determined, whereby the ghost can be removed. .

According to the ghost removing device having the above-described structure, the tap coefficient c of the filter circuit 1 is corrected not only to bring the error evaluation function E close to the minimum value but also to increase the steepness evaluation function P. As shown in FIG. 2, the steepness of the tap coefficient c rapidly increases as the number of corrections increases, as compared with the conventional ghost removing device that corrects the tap coefficient c based only on the error reduction displacement amount d, and the error is reduced. The number of corrections until the evaluation function E becomes sufficiently small can be reduced.

FIG. 3 shows how the error evaluation function E decreases when the tap coefficient c is repeatedly corrected by the ghost removing apparatus of the conventional example using the GCR signal as the reference signal. In this case, the value of the error evaluation function E decreases more rapidly in the present embodiment considering the steepness evaluation function P, and the number of corrections for decreasing the error evaluation function E to the same value is also performed in the present embodiment. It turns out that there are fewer. FIG. 4 shows how the error evaluation function E decreases when the equivalent pulse is used as the reference signal. In this case, the rate of decrease of the error evaluation function E at the initial stage of correction in the present embodiment is further increased, and it can be seen that the effect is particularly great when a signal with few high frequency components such as an equivalent pulse is used as the reference signal. .

As described above, according to the ghost removing apparatus of the present embodiment, not only the error reduction displacement amount d but also the steepness increasing displacement amount f that increases the steepness evaluation function P are taken into consideration when correcting the tap coefficient c. Therefore, the number of corrections of the tap coefficient c until the error evaluation function E becomes a sufficiently small value is reduced, and the ghost can be removed in a short time.

In this embodiment, the ghost elimination device is shown in which each functional block is divided as shown in FIG. 1, but it may be constituted by only one processing circuit having all these functions. Alternatively, it may be configured by a plurality of processing circuits divided into other functional blocks. In addition, all or some of these functions may be
Processing Unit] and DSP [Digital Signal Processo]
r] and a processing program may be combined.

Further, in the steepness evaluation function P shown in Expression 10 of the present embodiment, the interval j is set in the range of 1 to 5, but it is not limited to this and may be set in any range. Furthermore, although the steepness value P _j shown in Expression 11 is used as the sum of squares of the difference between the tap coefficients c, the sum is not limited to this, and is the sum of absolute values of the difference between the tap coefficients c as shown in Expression 17, for example. You can also do it.

[0061]

[Equation 17]

Further, in the present embodiment, the tap coefficient c is updated based on both the error decrease displacement amount d and the steep increase displacement amount f in each correction according to the equation 16, but the error according to the equation 2 is used, for example. It is also possible to repeat the operation in which the correction based on only the reduced displacement amount d is repeated a predetermined number of times and then the correction based on both the error reduced displacement amount d and the steeply increased displacement amount f is performed once. Further, in Equation 15, the absolute value of the error reduction displacement amount d is used in order to prevent the steepness of the distribution of the tap coefficient c from increasing more than necessary, but it becomes smaller as the error signal e decreases. Other values can be used as long as they are values. Further, in the present embodiment, the filter circuit 1 has the FIR type configuration, but the filter circuit 1 can be similarly realized in the case of the IIR type configuration or a configuration in which these are combined. In this case, for example, the tap coefficient c for the FIR type configuration calculated by the steep increase displacement amount calculation unit 5 is converted into the tap coefficient c of another configuration having the same frequency characteristic, or the equation 8 or the equation 1 is used.
5 may be replaced with a mathematical formula according to the filter circuit 1 having another configuration. Further, in the present embodiment, the ghost removing device in which the processing circuit is digitized has been described, but the present invention can be similarly applied to a ghost removing device using an analog circuit.

[0063]

As described above, according to the present invention, not only the error reduction displacement amount for making the error evaluation function close to the minimum value but also the steep increase displacement amount for making the distribution of the tap coefficient steep are taken into consideration. Since the tap coefficient of the filter circuit is updated in this way, the distribution of the tap coefficient becomes sharp rapidly, and the error evaluation function can be brought close to a sufficiently small value with a small number of corrections to shorten the time required to remove the ghost. it can. The effect of shortening the ghost removal time becomes particularly remarkable in the ghost removal device that uses a signal with a few high frequency components such as an equivalent pulse as the reference signal.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a ghost removing device according to an embodiment of the present invention.

FIG. 2 is a diagram showing an embodiment of the present invention and is a diagram showing a change in steepness of distribution of tap coefficients.

FIG. 3 is a diagram showing an embodiment of the present invention and is a diagram showing how an error evaluation function is reduced in a ghost elimination device using a GCR signal as a reference signal.

FIG. 4 is a diagram showing an embodiment of the present invention and is a diagram showing how an error evaluation function is reduced in a ghost eliminating device using an equivalent pulse as a reference signal.

FIG. 5 is a block diagram showing a configuration of a transversal filter.

FIG. 6 is a block diagram showing a filter circuit of FIR type configuration.

FIG. 7 is a block diagram showing a filter circuit having an IIR type configuration.

FIG. 8 is a block diagram showing a filter circuit combining a FIR type configuration and an IIR type configuration.

FIG. 9 is a block diagram showing a configuration of a ghost removing device, showing a conventional example.

FIG. 10 is a time chart showing an operation of the ghost removing device, showing a conventional example.

FIG. 11 is a diagram showing a conventional example and showing a change in steepness of distribution of tap coefficients.

[Explanation of Codes] 1 Filter Circuit 2 Error Decrease Displacement Calculator 5 Sharp Increase Displacement Calculator 6 Tap Coefficient Storage

Claims (1)

[Claims] 1. A ghost eliminating device for obtaining an output signal from which a ghost component has been eliminated by processing a received signal with a filter circuit, wherein the output signal and the inside of the receiver are generated when the tap coefficient of the filter circuit is updated. Error reduction displacement amount calculating means for calculating an error reduction displacement amount of the tap coefficient capable of reducing an error with the internal reference signal, and sharpening the distribution of the tap coefficient when the tap coefficient of the filter circuit is updated. And a steep increase displacement amount calculating means for calculating a steep increase displacement amount of the tap coefficient, an error decreasing displacement amount calculated by the error decreasing displacement amount calculating means, and a steep increasing amount calculated by the steep increasing displacement amount calculating means. A ghost elimination device comprising: a tap coefficient updating means for updating a tap coefficient of a filter circuit based on a displacement amount.