KR950012662B1 - Ghost modeling system for tv receiver - Google Patents

Abstract

The system conprises a low pass filter(21) which filters only low pass band of ghost eliminating signals, an interpolating unit(22) which increases frequencies, an amplitude modulating unit(23) which performs amplitude modulation, a side-band waveform filter(24) which cuts off some parts of lower side-band waveforms, a ghost generating unit(25) which generates ghost signals by delaying phase or decreasing amplitude, a noise generating unit(26) which mixes the ghost signals with noise, a frequency converting unit(27) which coverts carrier frequency to middle frequency, a Nyquist filter(28) which deletes unremoved interference in the output signals, a sampling unit(30) which takes a sample, a channel characteristic determining unit(31) which finds the characteristics of transfer channel, and a ghost removing filter(32) which removes ghost.

Description

TV Ghost's Modeling System

1 is a schematic diagram of a TBS signal transmission / reception system according to the present invention.

FIG. 2 is a diagram illustrating a system configuration for modeling a ghost generated in a tub signal transmission / reception system.

3 is a characteristic diagram of a filter for obtaining a residual sideband modulated signal.

4 is a frequency characteristic diagram of each filter in FIG. 2;

5 is a detailed view of a ghost generator and a noise generator in FIG.

FIG. 6 is a ghost elimination reference signal diagram of three columns of length 307 and elements of (1,0, -1).

7 is a flowchart of a ghost simulation in FIG.

8 is a ghost demodulation waveform diagram according to time delay, amplitude attenuation and phase delay.

* Explanation of symbols for main parts of the drawings

1,21: low pass filter 2,23: bilateral amplitude modulator

3,24: Residual side wave filter 22: interpolator

25: ghost generator 26: noise generator

27: intermediate frequency converter 28: Nyquist filter

29 demodulator 30 decay unit

31 channel characteristic discriminator 32 ghost removal filter

5l: time delay 52: attenuator

53: Hilbert Converter 54: 90。 Phase Delay

55,56: first and second multipliers 57,58: first and second adders

The present invention relates to the modeling of the TV (tv) ghost, in particular, to model the process of transmitting and receiving the video signal of the tub ghost modeling system to accurately generate the ghost in consideration of time delay, amplitude and phase without approximation It is about.

In the conventional ghost, if the value is negative according to the amount of time delayed based on the main signal, it becomes a front ghost, and if the value is positive, it becomes a post ghost. It is divided into many (more than 25㎲) long ghosts.

However, the ghost is known as the main cause of deterioration of image quality by appearing as a faint multiple image at the position moved to the left (front ghost or right (after ghost) direction from the original image on the tub screen, which is a high-rise building, mountain, airplane, etc. Generated by a multipath channel composed of reflectors, the multipath channel may also be formed by causes such as impedance mismatch in a cable broadcasting system.

In other words, the prior art considers only the delay time of the ghost, does not accurately consider the phase, and does not filter in the frequency domain, so it is very difficult to obtain an ideal filter characteristic (amplitude). There was a problem acting.

Therefore, in order to solve the conventional problem, the present invention is a next generation tub system such as EDTV (EXTENDED DEFlNITION TELE-VISION), HDTV (HIGH DEFINITION TELEVISION), and the like. The Hilbert transform allows accurate generation of ghosts of arbitrary phase as well as any time delay or amplitude, while the Fast Fourier Transform (FFT) allows filtering in the frequency domain. Except for the frequency resolution error due to the length of the tubing ghost modeling system that can be easily implemented by the frequency sampling method of the filter of the ideal form, it will be described in detail with reference to the accompanying drawings.

FIG. 1 is a schematic diagram of an NTSC-type tub signal transmission / reception system according to the present invention. As shown in FIG. 1, a low pass filter (1) for receiving an input video signal and removing interference and high-frequency components between broadcast channels; Of the two sideband amplitude modulators (2) for transmitting both sidebands of which amplitude modulation is performed on the signal filtered through the low pass filter (1), and the two sideband amplitude modulators (2) A residual sideband filter 3 which cuts a predetermined portion of the lower sideband portion, a transmission channel 4 for receiving a transmission signal processed through the residual sideband filter 3 so as to be receivable, and the transmission channel (4) a high frequency amplifier (5) for amplifying a high frequency transmitted through a predetermined level so as to be received, and a local oscillator (9) for receiving a signal amplified by the high frequency amplifier (5) and outputting a frequency suitable for a corresponding broadcasting station (9). ), A mixer 6 for mixing the station frequency of the local oscillator 9 and the frequency of the high frequency amplifier 5 and outputting the mixed frequency to the intermediate frequency amplifier 7 and the signal output through the intermediate frequency amplifier 7. And an image signal detector 8 for finding the original image.

In addition, the system configuration for modeling the tub ghost generated in the tub signal transmission and reception system of FIG. 1 includes a low pass filter 21 for removing interference and high frequency components between the voice channels, as shown in FIG. An interpolator 22 which receives the filtered signal through 21 and increases the sampling frequency by an interpolation ratio so that aliasing does not occur in the modulation process, and the interpolated signal is interpolated through the interpolator 22. A double-sided wave amplitude modulator 23 for raising amplitude at a carrier frequency fc and a residual side-wave filter for blocking a portion of the lower side wave portion from both sides of the amplitude-modulated side wave in the double-sided wave amplitude modulator 23 ( 24) a high-suit generator 25 for generating a ghost signal by delaying, amplitude attenuating, and phase-delaying the signal through the residual side wave filter 24, and the ghost generator 25. A noise generator 26 that generates noise by mixing the ghost signal, an intermediate frequency converter 27 for converting the carrier frequency fc through the noise generator 26 into an intermediate frequency f _1F , and the intermediate frequency converter. A Nyquist filter 28 for eliminating interference that has not been removed from the signal output through 27, and a demodulator 29 for demodulating the signal finally outputted through the Nyquist filter 28; A channel characteristic discriminator 31 which finds the characteristics of the transmission channel as a cross-correlation function of the silencer 30 taking one sample for each interpolation ratio, the ghost removal order signal and the output of the damper 30, and channel characteristic discrimination. A filter coefficient necessary for ghost removal is generated from the channel characteristics determined by the controller 31, and the ghost removal filter 32 is configured to remove ghost in real time by the filter coefficient.

Referring to the operation and effect of the present invention configured as described above in detail.

First, the relationship between the transmission and reception of the tub signal according to FIG. 1 will be described first.

When the video signal is input to the low pass filter 1, the interference and the high frequency components between the broadcast channels are removed, and then both side band amplitude modulators 2 transmit both the upper and lower side bands after amplitude modulation is performed on the filtered signal. Among the upper and lower band bands transmitted to the both side band amplitude modulators 2, the residual band filter 3 blocks a part of the lower band part, thereby performing an antenna (not shown) through the transmission channel 4 for the signal processed signal. Through the high frequency amplifier (5).

Then, the amplification output of the high frequency amplifier 5, which is amplified to be suitable for reception, is input from the mixer 6 and the local oscillator 9, respectively. In the local oscillator 9, the frequency is changed in accordance with a desired broadcasting station to the mixer 6. The mixer 6 mixes the frequencies of the high frequency amplifier 5 and the local oscillation section 9 and inputs them to the intermediate frequency amplifier 7.

The intermediate frequency amplifier 7 amplifies to an appropriate level to enter the intermediate frequency band and sends it to the video signal detector 8, where the detector 8 finds the original video signal.

Therefore, the first diagram operating as described above is represented as a mathematical model to enable computer simulation. When the ghost removal reference signal is input, ghost generation and transmission channel determination are performed, and then ghost is removed according to transmission channel determination. This will be described with reference to FIGS. 2 to 8 as follows.

When the ghost elimination reference signal is input from the low pass filter 21, the ghost elimination reference signal is subjected to low pass filtering to prevent interference between broadcast channels and to remove and transmit high frequency components. When the filtered ghost removal reference signal from which the high frequency component is removed is input to the interpolator 22, the sampling frequency is increased by an interpolation ratio to prevent aliasing that may occur during the modulation process. The interpolated signal is raised from both side amplitude amplitude modulator 23 to the carrier frequency fc. When outputting the upper and lower side waves with amplitude modulation, the residual side wave filter 24 shows the lower side wave among the upper and lower side waves. The part is blocked to some extent according to the filter characteristics and sent to the ghost generator 25.

As shown in FIG. 5, the ghost generator 25 delays the signal input through the residual side wave filter 24 through the time delay 51 for a predetermined time (τ) and again attenuates 52. If the amplitude (α) is adjusted and sent to the Hilbert transformer 53, the Hilbert transform is sent to the second multiplier 56. Is multiplied with the phase delayed by 90 ° through the 90 ° phase retarder 54 and sent to the first adder 57. At this time, the signal in which the time delay and the amplitude are set by the time delay unit 51 and the attenuator 52 is cos in the first multiplier 55. When multiplied by a phase and sent to the first adder 57, the first adder 57 receives a time delay?, An amplitude?, And a phase ( Is taken into account and the ghost signal g (n) is output.

The ghost signal g (n) is expressed as follows, where the ghost signal g (n) is represented as a continuous time of discrete time of the ghost signal g (n).

Therefore, when NG = 8, α ₁ = 0.1 τ ₁ = (30 / 4fsc) xi, and Ø ₁ = −45 ° xi (i = l, 2 ..., NG), the demodulated waveform without noise is eliminated. It is shown at 8 degrees, where a is Ø = -45 °, b is Ø = -90 °, and g is Ø = -315 ° or 45 °.

In the formula (l), the signal obtained by subtracting sigma from the second term on the right side is as follows.

Like the ^ here, the Hilbert transform is used.

By employing the Hilbert transform as described above, ghosts with time delay, amplitude and arbitrary phase can be accurately generated without approximation.

When the ghost generator 25 generates a ghost signal g (t) having a predetermined time delay, amplitude, and an arbitrary phase, the second adder 58 of the noise generator 26 mixes noise with the ghost signal to generate an intermediate frequency. Export to transducer 27.

The intermediate frequency converter 27 converts the input carrier frequency fc into the intermediate frequency f _1F . Equation (2) becomes as follows after the conversion process.

Equation (3) and the value is converted into a signal, and when the Nyquist is input to the filter 28, the frequency characteristics of the filter 28 is filtered and the demodulator 29 undergoes a demodulation process as described below.

In the above equation (4), when the ghost time delay τ ₁ satisfies Wcτ ₁ = 2πn (n = integer), the time delay τ ₁ has no effect on the phase delay Ø ₁ and finally demodulates it. I ghost signal

Given by

The demodulated signal through the demodulator 29 is sent to the channel characteristic discriminator 31, which takes one sample per interpolation ratio and lowers the sample freight from the damper 30, and outputs the ghost removal reference signal and the damper 30. The crosscor-relation function is obtained and divided by the average power of the ghost removal reference signal to find the characteristics of the transmission channel and send it to the ghost elimination filter 32 to determine the appropriate filter coefficient for ghost elimination according to the determined channel characteristics. Create and remove ghosts in real time.

The residual sideband modulation in the residual sideband filter 24 may also be performed as a process of residual sideband filtering after performing both sideband modulation. It is based on the frequency characteristic of the filter shown to (a) and (b) of FIG.

For convenience of explanation, when the discrete time is regarded as a continuous time, the ghost elimination reference signal x (n) may be represented by x (t), and the signal x _VSB (representing x (t) with residual sideband modulation (VSB) t)) is x _VSB (t) = x (t) Cos Vct- (t) It can be expressed as Sin Wct.

here (t) is a high-pass filtered signal with impulse response of x (t), and ^ denotes the Hilfert transform

The modeling of the ghost removal described above is as follows when the operation of the simulation is performed from step S1 to step S7 of FIG. 7.

As shown in FIG. 6, a ghost elimination reference signal x (n) having a length N consisting of only 1,0 and -1 elements having a ternary sequence of N = 307 is used. In the elimination reference signal x (n), the first column of the first frequent row is x (0) and the seventh column of the last row is x (N-1).

The autocorrelation function Rxx (n) of the ghost removal reference signal x (n) is given as follows.

As in step S1 shown in FIG.

Horizontal Scanning Frequency (fh) ← (15.75 / 1.001) KHz Color Subcarrier Frequency (fsc) ← 22.75fh

Sampling Frequency (fs) ← 4fsc

And then input the ghost elimination reference signal x (n) as shown in FIG. 6, the carrier-to-sleep ratio (CNR) from the ghost elimination reference signal x (n) input to the low pass filter 21. (dB), number of ghosts (NG), time delay (r) (㎲), oscillation (α) (dB) and phase (Ø) (deg) of each ghost, and then the ghost removal reference signal (x ( n)) is broadcasted by performing a Fast Fourier Transform (FFT) with 0 added to the back so that its length is NPT = 2048, and low pass filtering with the frequency characteristics shown in Fig. 4 (a). Inverse fast Fourier transform (FFT) is performed to remove the interference and high frequency components between channels and output the converted signal.

In step S5 again,

Interpolation Ratio (M) ← 4

Sampling Frequency (fs) ← Mfs

Sampling Cycle (Ts) ← 1 / fs

Data length (NPT) ← (M) NPT

Carrier Frequency (fc) ← Mfsc

Intermediate Frequency (f _lF ) ← M _fsc

Local Oscillator Frequency (f _LO ) ← (2M) fsc

The interpolator 22 receiving the ghost elimination reference signal passing through the low pass filter 21 inserts 0 as much as the interpolation ratio M for each sample of the ghost elimination reference signal so that the cutoff frequency is set to the sampling frequency ( fs) After performing lowpass filtering with the frequency characteristics shown in FIG. 4 (b), an inverse fast Fourier transform is performed to scale by M times to output an interpolated signal so that aliasing does not occur.

Here, when the interpolation ratio (M) is M = 2 when described mathematically as follows.

If the signal to be interpolated with length N is x ₁ (n), the signal with 0 inserted into x ₁ (n) is x ₂ (n), and the interpolated signal is x ₃ (n), and x ₁ (k ), x ₂ (k) and x ₃ (k) are discrete Fourier transforms of x ₁ (n), x ₂ (n) and x ₃ (n), respectively.

= 1/2 [x ₃ (n) + (-1) ⁿ x ₃ (n)]

x ₂ (k) = l / 2 [x ₃ (k) + x ₃ (kN)], 0 ≦ k ≦ 2N-1

x ₃ (k) = 2x ₂ (k) = x ₃ (kN)

In the above equation, x ₃ (kN) is a high frequency component of 2x ₂ (k),

x ₃ (k) = DPF [2x ₂ (k)], 0 ≦ k ≦ 2N-1

x ₃ (n) = DFT ⁻¹ [x ₃ (k)], where 0 ≦ k ≦ 2N−1, where LPF and DFT ⁻¹ represent each lowpass filtering and inverse discrete Fourier transform.

The interpolated signal through the interpolator 22 performs amplitude modulation in both side wave amplitude modulator 23, which is multiplied by (-1) of the interpolated signal and added a DC offset of an appropriate level, followed by 2Cos (2πfc τ _S n. ) Is multiplied for the interval n = 0,1 --NPT ^{-1, and} the amplitude modulated sidebands are transmitted to the residual sideband filter 24, which performs the NPT-point fast Fourier transform and transmits both sidebands. The lower frequency band is cut off, which means that the carrier frequency fc set in step S5 becomes fc = Mfsc = 4fsc, and thus the frequency of the residual sideband filter 24 starts to decrease at 4fsc-0.75MHz. Thus, at 4fsc-1.25MHz, the magnitude is completely zero.

In this way, the filtering is performed again and the NPT-point, inverse fast Fourier transform is performed, and the output is given to s (n).

In other words, the residual sideband modulation may be performed as a process of residual sideband filtering after performing both sideband modulation, but it is also possible to perform residual sideband filtering first and then perform both sideband modulation, which is shown in FIGS. 3 (a) and 3 (b). This is due to the frequency characteristics of the filter shown in FIG.

When the continuous time region is selected instead of the discrete time of the fun image, the ghost elimination reference signal may be represented by x (t) instead of x (n).

If WC = 2πfc and X (t) is the high-pass filtered signal with impulse response (V (t)), x _H (t), and if the Hilbert transform is ^, the residual sideband ( VSB) The modulated signal (x _VSB (t))

Can be displayed as As can be seen from FIGS. 3 (a) and 3 (b) of FIG. 3, V (W) is a high pass filter having a cutoff frequency of β and in FIG. 4 (c) of FIG. 4, β = 2π (0.5MHz). ) And Sgn (W) is a coded constant as

Where Re is the real part and F ^-1 is the inverse Fourier transform.

By the way, the carrier frequency fc should originally be the channel frequency of the broadcasting station. Here, the carrier frequency fc is defined as M times the color carrier frequency fsc for the purpose of reducing the amount of calculation.

What has been described above is from step 7 to step 8.

The signal transmitted through the low pass filter 21, the interpolator 22, the side plate amplitude modulator 23, and the residual side wave filter 24 is input to the ghost generator 25 and the noise generator 26. Referring to FIG. 5, the transmitted signal is delayed by a predetermined time (τ) through the time delay 51 of the ghost generator 25, and is again attenuated (α) by the attenuator 52. After the Hilbert transform is performed in the transducer 53, the signal through the 90 ° phase delay 54 is multiplied by the second multiplier 56 with respect to the CosØ phase.

At this time, the signal through the time delay 51 and the attenuation 52 is multiplied by the signal having the CosØ phase in the first multiplier (55). In this way, the signals multiplied by the first and second multipliers 55 and 56 are summed by the first adder 57 and output as the ghost signal g (n).

In this regard, units are first converted as in steps S9 and S10 of FIG. 7.

τ offset ← nint (2 × 10 ^-6 × fc) × M ∴ nint: Integer close to argument

τ (i) ← τ offset + nint [τ (i) × 10 ^-6 × fc] × M

∴ i = 1,2

If set to, the ghost signal g (n) can be indicated as follows.

As described above, when the ghost signal g (n) is generated by the ghost generator 25 and input to the noise generator 26, the additive white Gaussian noise Vpn is mixed as in step S11 of FIG.

When inputted to the intermediate frequency converter 27 of the mixed signal up to the additive white Gaussian lock (V (n)), the intermediate frequency conversion is performed by multiplying the additive white Gaussian noise (V (n)) by 2 COS (2τf _LO T _s n). After performing Fast Fourier Transform (FFT) and filtering at the frequency as shown in FIG. 4 (c) of FIG. 4 showing the frequency characteristics of the Nyquist filter 28, one signal (V '(n)) having an inverse FFT is obtained. Get

In addition, multiply the ghost signal g (n) by 2COS (2τf _LO T _S n) to perform the intermediate frequency conversion in the intermediate frequency converter 27, perform the fast Fourier transform, and send it to the Nyquist filter 28. After performing Nyquist filtering with the frequency characteristic shown in (d), a signal g '(n) obtained by performing an inverse FFT again is obtained.

The Nyquist filter 28 then sends an intermediate frequency change and the Nyquist filtered signal sg (n) to the demodulator 29, which signal sg (n) is represented as follows.

sg (n) = g '(n) + factor × V' (n)

Where factor is Is the standard deviation of V '(n), and gpv is the difference between peak and valley.

The demodulator 29 multiplies the signal sg (n) received from the Nyquist filter 28 by COS (2πf _LO T _S n), and then filters the whole pass with frequency characteristics as shown in FIG. 4 (e) in FIG. Demodulation is performed. This demodulated signal is denoted by sg '(n).

When the demodulated signal sg '(n) through the demodulator 29 is input to the attenuator 30, one sample is taken for each interpolation ratio and the output (y (n) ← sg' as in step S14 of FIG. (Mn)). The channel characteristic discriminator 31, which has received the output y (n) of the sound absorber 30, has an autocorrelation function R _xx (0) of the ghost cancellation reference signal, And the autocorrelation function R _XY (n) of the attenuator 30 Then, as in step S16, cross-correlation function (h (n)) ← R _XY (n) / R _XX (0)) is obtained to find the characteristics of the transmission channel.

When the channel characteristic determined by the channel characteristic discriminator 3l is sent to the ghost elimination filter 32, the ghost elimination filter coefficient is obtained from the cross-correlation function h (n) of the channel characteristic discriminator 31 to remove the ghost. do.

As described in detail above, the present invention accurately generates a ghost with arbitrary phase as well as any time delay and amplitude by employing a Hilbert transform, and accurately generates a Fast Fourier Transform (FFT). By filtering in the frequency domain, the ideal arbitrary filter can be easily implemented by frequency sampling method, and the time required for filtering is greatly reduced, and the zero phase filtering in the frequency domain by using fast Fourier transform (FFT) Since -phase filtering can be performed, the ghost can be simulated without time delay caused by various operations.

Claims (7)

The low pass filter 21 performs low pass filtering on the ghost elimination reference signal to remove the interference and high frequency components between the broadcasting channels, and the filtered signal is input through the low pass filter 21 to be applied during the modulation process. an interpolator 22 that increases the sampling frequency by an interpolation ratio so that no aliasing occurs, and a bilateral wave amplitude modulator that modulates an amplitude by raising an interpolated signal through the interpolator 22 to a carrier frequency fc. (23), a residual sideband filter (24) for blocking a portion of the lower sideband part of both sidebands amplitude-modulated by both sideband amplitude modulators (23), and a signal through the residual sideband filter (24). A ghost generator 25 generating a ghost signal by time delay, amplitude attenuation, and phase delay with respect to the ghost signal, a noise generator 26 generated by mixing noise with a ghost signal of the ghost generator 25, and The carrier frequency by a noise generator (26) (fc) of intermediate frequency (f _1F) to change to that eliminates the intermediate frequency converter 27, it is not removed the interference from signals output from the intermediate frequency converter (27) A Nyquist filter 28, a demodulator 29 for demodulating the signal finally outputted through the Nyquist filter 28, a damper 30 taking one sample per interpolation ratio, and a ghost The channel characteristic discriminator 31 which finds the characteristics of the transmission channel as a cross-correlation function between the rejection reference signal and the output of the attenuator 30 and the channel characteristic discriminated by the channel characteristic discriminator 31 are required for ghost rejection. A ghost ghost modeling system comprising a ghost elimination filter (32) that creates a filter coefficient and removes ghost as the coefficient. 2. The system of modeling a tub ghost according to claim 1, wherein the ghost cancellation reference signal adopts three rows of 307 lengths. The method of claim 1, wherein the ghost generator 25 is a time delay 51 for delaying a predetermined time (τ) with respect to the input signal, and the amplitude (α) for the signal delayed through the time delay 51 is adjusted. Attenuator 52, Hilbert transform 54 for performing a Hilbert transform on a signal whose time τ delay and amplitude? Are adjusted, and a region of 90 ° relative to the applied phase COSØ. 90 ° phase retarder 54, first multiplier 55 multiplying the signal through the attenuator 52 and phase COSØ, and 90 ° phase with the signal converted by the Hilbert transducer 53 A second multiplier 56 for multiplying the delayed phase through the delayer 54 and a first adder 57 for adding the signals outputted through the first and second multipliers 55 and 56; Tubing Ghost's Modeling System. 4. The system of claim 3, wherein the ghost signal is output as a sum of time delayed, amplitude attenuated, and phase delayed signals. The system of claim 1, wherein the noise generator (26) comprises a second adder (58) for applying a lock to the ghost signal generated by the ghost generator (25). 2. The low pass filter 21, the residual sideband filter 24, and the Naquist filter 28 perform a Fast Fourier Transform (FFT) to filter the respective frequency characteristics to reduce the calculation time. Turb Ghost's modeling system. The method of claim 1, wherein the channel characteristic discriminator (31) obtains the cross-correlation function of the ghost cancellation reference signal and the output of the silencer (30) and divides the average value of the ghost removal reference signal to determine the characteristics of the transmission channel. Characteristic system of Tub Ghost.