JPH04185176A - Ghost eliminating device - Google Patents

Abstract

PURPOSE:To improve the stability at ghost elimination by selecting a video input signal to be inputted to the ghost elimination means and trying the processing again from the start when a ghost change is caused. CONSTITUTION:A GCR signal or a vertical synchronizing signal is subject to synchronization addition by a CPU 5 or subject to field sequence processing, then the result is transferred to a RAM 1 and the input to a ghost elimination circuit 3 is switched to input an output of the RAM 1 to the ghost elimination circuit 3, in which ghost is eliminated, and a ghost change detection circuit 6 is provided, then the input signal of the ghost elimination device is inputted to the ghost elimination circuit 3 when no ghost change exists. Moreover, after synchronization addition and field sequence processing are once implemented, the resulting signal is read repetitively from the RAM 1 to apply ghost elimination. Thus, the processing time is reduced and the stability of ghost elimination is improved.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a ghost removal device that performs ghost removal using a GCR signal or a vertical synchronization signal as a reference signal.

Conventional Technology The first generation of EDTV broadcasting, which aims to improve picture quality while maintaining compatibility with current television systems, is about to begin, and ghost removal is attracting a lot of attention. Items required in this technology include an image quality evaluation value after ghost removal and removal time, and how quickly the amount of remaining ghosts can be reduced.

As an example of a conventional ghost removal device, the Technical Report of the Television Society (RE80-6, pp. 9-14, 1972)
There is a ghost canceller reported in February 2013).

This is to perform ghost pressure detection using the differential signal of the leading edge of the vertical synchronization signal of the television signal as a reference waveform. The ghost is removed by performing a correlation calculation on the time axis using the detected ghost signal with the input signal of the ghost canceller and successively correcting the tap coefficients of the transversal filter. Further, as a ghost removal device using a GCR signal, there is a ghost canceller reported in the Technical Report of the Television Society (ROFT89-6, pp, s1-36). This uses a transversal filter as the ghost canceller in the ghost canceller, but the input and output of the transversal filter is transferred to the CPU via memory and all subsequent ghost removal calculations, including synchronous addition and field sequence processing, are carried out. I do.

An example of a conventional ghost removal device will be described below with reference to the drawings.

FIG. 3 is a schematic block diagram showing the configuration of a conventional ghost removal device. In FIG. 3, 5 is a CPU, 12 is a transversal filter, 20 is an A/D converter, and 21 is a D
/A converter, 22 is a waveform memory. The operation of the ghost removal device configured as above will be explained.

The input video signal is A/D converted by an A/D converter 20 and input to a transversal filter 12 and a waveform memory 22, respectively. The input/output of the transversal filter 12 is connected to the CPU 5 via the waveform memory 22.
be taken in. In the first generation EDTV broadcasting currently being carried out, the reference signals are the WRB signal and O pedestal signal shown in Figures 4(a) and 4(b). A sequence of pedestal signal - WRB signal -> 0 pedestal signal -> WRB signal, which goes around in 8 fields, is superimposed and transmitted in the same horizontal period. The signal shown in FIG. 4(C) can be obtained by performing the calculation shown in equation 1 below on the signals of these 8 fields. However, Fn (n=1 to 8
) represents the nth field signal. Hereinafter, the processing between fields according to the transmission sequence as shown in the first equation will be referred to as field sequence processing.

F=1/41 (Fl-F5)+ (F6-F2)+(
F3-F7)+ (F8-F4)l ・・・・・・(
1) Actually, Fig. 4(d) is obtained by differentiating the signal in Fig. 4(C).
The following ghost removal calculation is performed using the signal shown in as a direct reference signal for ghost detection.

Generally speaking, MSE (Mean 5 square Err
or) Law or Z F (Zero Forcing)
These methods perform calculations on the time axis according to a certain algorithm, sequentially modify tap coefficients, and finally obtain quantitative tap coefficients. The output signal of the transversal filter is +Ykl, the reference signal is (Rkl,
) If the difference signal between the output signal of the transversal filter and the reference signal is (Ekl), and the total number of taps is M+N+1, then the n-th tap coefficient Ct of the transversal filter is
i I(mouth' is modified based on the following equation 2 in the MSE method and equation 3 in the ZF method. However, α and β are coefficients for determining the amount of correction.

1cil in-I) -(Ci) (n)
-a"ΣY k -i-E k -(2)K-IchiyonIc i l ""-1c i l (n'-β・E
i --(3) The CPU 5 performs the synchronous addition shown in the first equation,
After performing the field sequence processing, the second or third equation is calculated to repeat the correction of the tap coefficients.

A series of these processes is performed by software, and a flowchart of this process is shown in FIG. Here, in ghost detection, processing is continued until it is determined that the amount of residual ghost has become sufficiently small.

Generally, when the S/N of the input video signal is low, the residual ghost level will drop below the noise signal level and the residual ghost level will no longer be a ghost. It becomes impossible to distinguish between noise and noise, and as a result, this level becomes the limit for ghost removal. Therefore, in order to further reduce the amount of residual ghosts, it is necessary to sufficiently perform synchronous addition to improve the S/N ratio for the above-mentioned reasons. However, the processing time increases in proportion to the number of synchronous additions, resulting in a longer ghost removal time. Furthermore, because the video input signal is re-fetched to the CPU each time the tap coefficients are corrected, the possibility of the system being affected by factors that cause instability, such as divergence, increases (on the time axis). When trying to find the optimal tap coefficients by successive corrections, the tap coefficients oscillate due to the influence of noise and other disturbances, even though a certain procedure is followed, and depending on the combination of tap coefficients, the worst transversal filter 12 diverges.

In response to this, the inventors previously proposed in Japanese Patent Application No. 1-322123 that a reference signal that has been subjected to synchronous addition and field sequence processing is held separately from the input/output waveform memory of the ghost removal means, and used for ghost removal calculations. We have shown that it is possible to significantly improve the removal time and stability problems by adding a separate memory for this purpose. However, with this configuration, there is no means to know the ghost change in the input video signal input to the ghost removal device while the memory is being read and ghost removal is being performed. That is,
When an attempt is made to switch the input video signal to the ghost removal means using the switching means after ghost removal, it is not possible to determine whether the switching is allowed.

If the ghost has changed from the initial state of processing, switching may result in the output from the ghost removal means becoming a more distorted signal than before ghost removal.

Problems to be Solved by the Invention As mentioned above, in order to reduce the residual amount of ghosts as much as possible and improve the ghost removal performance with the configuration of the conventional ghost removal device, it is necessary to perform sufficient synchronous addition to improve the S/N. On the other hand, increasing the number of synchronous additions will result in a longer ghost removal time.Also, since corrections are made sequentially on the time axis, not only will the ghost removal time increase; It is also more easily affected by external disturbances, increasing the possibility of divergence.

In view of the above-mentioned problems, the present invention reduces the amount of ghosts remaining after ghost removal as much as 8 times, reduces the removal time significantly compared to the conventional method of capturing a signal every time a tap coefficient is corrected, and also improves stability during ghost removal. To provide a ghost removal device that can improve performance.

Means for Solving the Problem To achieve this purpose, the GCR signal or vertical synchronization signal is synchronously added and subjected to field sequence processing by the CPU, and then transferred to the RAM, and the input to the ghost removal means is switched to change the output of the RAM. The system is configured to perform ghost removal by inputting the signal to the ghost removal means and to detect a ghost change so that if there is no ghost change, the input signal of the ghost removal device is switched to be input to the ghost removal means. .

In the present invention, once synchronous addition and field sequence processing are performed using the above-described configuration, this signal is repeatedly read out in the RAM to perform ghost removal calculations. This greatly reduces processing time without having to import the signal every time the tap coefficient is successively corrected, and improves the S/N ratio by ensuring a sufficient number of - times of synchronous addition, making it unaffected by fluctuations caused by disturbances in the input signal. do it like this. Furthermore, by detecting ghost changes, it is possible to avoid errors in determining whether to switch the input video signal to the ghost removal means after ghost removal.

EXAMPLE Hereinafter, a ghost removal device according to an example of the present invention will be described with reference to the drawings. FIG. 1 is a schematic block diagram of the circuit configuration of a ghost removal device in a first embodiment of the present invention. In FIG. 1, l is a RAM, 2 is a switching circuit,
3 is a ghost removal circuit, 4 is an input/output waveform memory, 5 is a CPU, and 6 is a ghost change detection circuit. The operation of the ghost removal device configured as described above will be described below.

Initially, the switching circuit 2 selects the a side, and the ghost removal circuit 3 receives an input video signal. The input/output of the ghost removal circuit 30 is taken into the CPU 5 via the waveform memory 4. The CPU 5 repeatedly takes in input video signals and performs sufficient synchronous addition to improve the S/N ratio. Thereafter, the field sequence processing shown in equation 1 is performed to calculate the signal shown in FIG. 4(d). This signal is transferred from the CPU 5 to the RAMI for writing. After this, the switching circuit 2 selects the b side and repeatedly reads out the RAMI. This RAMI signal is taken into the CPU 5 in the same way as when the input video signal was first taken in. The CPU 5 performs a difference with a reference signal stored internally in advance, performs calculations shown in the second and third equations, and controls the ghost removal circuit 3. A flowchart of the processing at this time is shown in FIG.

Here, assuming that N fields are required for synchronous addition, M times are required for correction, and L fields are required for ghost detection operator tap coefficient correction, the conventional method and the configuration of the present invention are compared when all processes are performed. , the processing time ratio a is as shown in the fourth equation below.

α=N+M*L/ (N+L)*M ・・・・・・
・(4:・Here, - synchronous addition is 128 as a value for boat fishing.
Field, number of corrections 20 times, ghost detection ÷ calculation +
If we consider the case where the tap coefficient correction is set to 1 field, α becomes 1/17, which is a significant reduction.

On the other hand, during the period when the switching circuit is switched to the b side and the ghost removal calculation is being performed, the signal taken into the CPU is not the input video signal, so the ghost state of the input video signal, in other words, the ghost is changing. I don't know. If the ghost change detection circuit detects a ghost change in the input video signal when switching the input video signal to the ghost removal means after the tap coefficient has been corrected M times and the ghost removal has been sufficiently performed, if there is no change, the input video signal will remain as it is. There is no problem with switching, and if the ghost changes, you can just reset the CPU and restart the process from that point. Hereinafter, specific embodiments of the present invention will be described with reference to the drawings. FIG. 2 is a block diagram showing a different embodiment of the present invention. In Fig. 2, 10 is a GCR signal extraction circuit, and 11 is a FIFO 512.
is a transversal filter, 13 is a selector, 14 is a counter, 15.16 is a switch, 17 is a memory, 18
1 is a correlation circuit, and 19 is a determination circuit.

The GCR signal extraction circuit 10 generates a gate pulse signal during the period when the GCR signal is superimposed.

The input video signal is input to the transversal filter 12 with the switch 15 selecting the a side. As for the input and output of the transversal filter 12, a counter output is supplied to a memory address by a gate pulse period selector 13, and is written into the memory 4 by DMA processing. After this, the address from the CPU is supplied to the memory 4, and the CPU
5 and performs synchronous addition and field sequence processing shown in the first equation. All of these are processed by software inside the CPU. After performing these series of processes a necessary number of times, select b with the changeover circuit switch 16, and select FIFO II.
Transfer to. After the transfer is completed, the switch 1.6 is switched to the a side, the PIFO II is read out, and the signal written by the CPU 5 is repeatedly output. On the other hand, the switch 15 is also switched to the a side at the same time, and the output of the FIFO II is input to the transversal filter 12. After transferring to PIFO II in this way, by using the FIFO II signal, it is possible to process the signal that has already been subjected to synchronous addition and field sequence processing each time, reducing processing time and operating stably as if using a fixed signal source. It becomes possible to do so.

In parallel with the ghost removal processing described above, the correlation circuit 18 performs a correlation calculation between the GCR signal of the current field and the GCR signal of the previous field outputted from the memory 17, and compares the correlation output with a threshold level set in advance by the determination circuit 19. Make a comparison. This determination result is output to the CPU 5 to determine whether the ghost has changed. As a result, if it is determined that there is no ghost change, the switch 15 is switched to the a side.

Effect of the invention As described above, according to the present invention, -degree synchronous addition is performed.

After field sequence processing, this signal can be used in combination with the input video signal to significantly shorten processing time and improve stability during ghost removal. Furthermore, by detecting ghost changes during this processing period, it becomes possible to switch the input video signal to the ghost removal means after ghost removal without any problem.

[Brief explanation of the drawing]

FIG. 1 is a schematic block diagram of a ghost removal device according to an embodiment of the present invention, FIG. 2 is a block diagram of a specific embodiment of the ghost removal device of the present invention, and FIG. 3 is a block diagram of a conventional ghost removal device. Block diagram, Figure 4 (a) is a signal waveform diagram of the WRB signal, Figure 4 (b) is a signal waveform diagram of the O pedestal signal, Figure 4 (C) is a signal waveform diagram after field sequence processing, Figure 4 ( d) is a signal waveform diagram obtained by differentiating (C),
FIG. 5 is a flowchart of conventional ghost removal processing.
FIG. 6 is a flowchart of ghost removal processing according to the present invention. 1...RAM, 2...Switching circuit, 3.
... Ghost removal circuit, 4.17 ... Memory, 5 ... CPU, 6 ... Ghost change detection circuit, 10 ... GCR signal Extraction circuit, 11... FIFO 112... Transversal filter, 13... Selector, 1
4... Counter, 15.16... Switch, 18... Correlation circuit, 19... Judgment circuit. Name of agent: Patent attorney Haruaki Koebi and two others ( ヱ 'todoω / Figure 4 Figure 5 Figure 6

Claims (1)

[Claims] A ghost removal device that uses a GCR signal or a vertical synchronization signal as a reference signal to remove ghosts, the device comprising ghost removal means for removing ghosts, an input video signal, and a RAM.
switching means for switching between the output signal of the input video signal and inputting the output signal to the ghost removal means; ghost change detection means for detecting a change in the ghost state of the input video signal; Synchronous addition of the reference signal to the output waveform memory and the input/output waveform memory to enable ghost removal calculation;
the RA;
After sending a signal to M from the CPU that has been subjected to synchronous addition and inter-field processing according to the sending sequence, the switching means inputs this signal to the ghost removal means to perform ghost removal and eliminate the ghost change. switching to input the video input signal to the ghost removal means only when the detection means determines that there is no ghost change;
A ghost removal device characterized in that if it is determined that there is a ghost change, the process is restarted from the beginning.