JPH0432373A - Detection circuit for waveform distortion - Google Patents

Abstract

PURPOSE:To avoid cost increase and complicated control by approximating one of a GCR signal and a waveform distortion component subject to correlation calculation with a 2's power and using the approximated value to apply bit shift or bit rotate to the other value thereby realizing correlation multiplication. CONSTITUTION:The correlation arithmetic operation is executed by a software of a microcomputer 20 and a value of a pulse Pp of a GCR signal subject to correlation arithmetic operation is approximated by 2's power and the value of the pulse Pq is subject to bit shift by using the approximated value to realize multiplication in the correlation arithmetic operation. Then the correlation arithmetic operation of pulses Pp, Pq by a routine is executed by the microcomputer 20, and ghost information RSLT is extracted. Thus, the waveform distortion detection circuit is realized, in which the cost is not increased because of no need of an exclusive correlation arithmetic operation circuit and the control is not complicated.

Description

[Detailed description of the invention] [Industrial application field] The present invention relates to a circuit for detecting waveform distortion of a video signal.

[Summary of the invention]

The present invention aims to speed up the calculation by approximating data when performing correlation calculation using software, for example, in a ghost removal circuit.

[Conventional technology]

In a television receiver, ghost wave components can be removed from a received video signal as follows.

That is, on the transmitting side, a reference signal for ghost cancellation (hereinafter referred to as a GCR signal) is added to the video signal.

Then, on the receiving side, the received video signal is gated to extract the GCR signal and its ghost wave component,
This extracted GCR signal and its ghost wave component,
A ghost wave component is extracted by comparing the GCR signal formed on the receiving side, and the passage characteristics of a transversal filter, for example, are controlled so that the extracted ghost wave component is eliminated.

As the GCR signal at this time, a signal 5GCR as shown in FIG. 4 is considered.

That is, in the same figure, HD is the horizontal synchronizing pulse, BR
5T indicates a burst signal, and as shown in FIG. be done. Also, the rise characteristic is 5i
This is the ringing characteristic of nX/X.

Furthermore, the second GCR signal PDS has a pedestal waveform (0 level), as shown in FIG.

As shown in FIG. 5A, the 8-field period of the video signal is used as a repeating period, and the
18th line or 2nd line of 8th field period
The GCR signal GCR is inserted into the 81st line, and the remaining 2nd line
, the signal PDS is inserted into the 18th line or the 281st line of the 4th, 5th, and 7th field periods,
A video signal with this GCR signal 5GCR inserted is transmitted.

Then, the first to eighth GCR signals 5GCR are converted into a signal S
, ~S, then if the receiving side performs the calculation shown in FIG. 2B, the calculation result will be the signal GCR.

Also, if there is a ghost, this calculation result will include the signal GC
A ghost wave component Sg of R is also included.

Therefore, ghost removal can be performed from the signal GCR (and Sg) resulting from this calculation.

In this case, the burst signal BR5T, color signal, and horizontal synchronizing pulse HD, which are separated by 8 field periods, are in phase with each other, so when calculating the signals S1 to S,
Burst signal BRST, color signal and horizontal sync pulse HD
are canceled out respectively.

Therefore, the signal GCR (and ghost wave component Sg) resulting from the calculation does not include the burst signal BRST, the color signal, and the horizontal synchronization pulse HD, so it is difficult to remove the so-called front ghost and rear ghost, waveform equalization, etc. It can handle up to 45 microseconds. Moreover, false detection does not occur even for long ghosts up to about 80 μsec.

FIG. 6 shows an example of a ghost removal circuit using this GCR signal 5GCR.

That is, (1) shows the video detection circuit of the television receiver, and the above-mentioned GCR signal 5GCR is output from this detection circuit (1).
A color composite video signal SY to which SY is added is extracted, and this signal SY is supplied to an A/D converter (2) to convert one sample into, for example, an 8-bit digital video signal SY. It is supplied to the D/A converter (4) through a 640 stage (640 tap) transversal filter (3) and converted into the original analog video signal sY, and this signal SY is sent to the terminal (
5).

At this time, in the detection circuit (10), the GCR
A ghost wave component is detected from the signal 5GCR, and the detection output controls the passage characteristics of the filter (3) to remove the ghost wave component.

That is, the calculation shown in FIG. 5B can be rewritten as shown in FIG.
This indicates that it is sufficient to integrate the R signals 5GCR in order.

Therefore, the digital video signal SY from the A/D converter (2) is supplied to the gate circuit (11) and the GCR signal 5GCR (including the previous and subsequent detection periods) is taken out, and this signal 5GCR is sent to the buffer memory (12). GCR signal 5GCR for each field period is held for each field period.

Then, the GCR signal 5GCR of this memory (12) is
It is supplied to an arithmetic circuit (21). This calculation circuit (21)
Although the following circuits (22) to (28) are actually constructed by the microcomputer (20) and software, they are expressed by hardware here.

That is, in the arithmetic circuit (21), the memory (12)
The GCR signal 5GCR held in is sequentially added or subtracted every field period according to the formula shown in FIG. 5C, and as shown in FIG. A signal GCR (including its ghost wave component Sg) is extracted, and this signal GCR is passed through a differentiating circuit (2
2) to form a differential pulse PP as shown in FIG. 2B, and this pulse PP is supplied to a correlation calculation circuit (23).

In this case, since the signal GCR has a bar waveform, it is set as a pulse pp to provide a pulse response.

Further, the video signal SY from the filter (3) is supplied to the gate circuit (13), a GCR signal 5GCR (including the previous and subsequent detection periods) is taken out, and this signal 5GCR is supplied to the buffer memory (14). The GCR signal 5GCR of the field period is held for each field period.

Then, the GCR signal 5GCR of this memory (14) is
The signal GCR (including its ghost wave component Sg) is supplied to the arithmetic circuit (24) and is the arithmetic result of the 8-field period, as shown in FIG. is taken out, and this signal GCR is supplied to the subtraction circuit (25), and at the same time, a reference waveform signal GCR (FIG. 4A) is taken out from the reference GCR signal forming circuit (26), and this signal GCR is supplied to the subtraction circuit (25). 25). Therefore, as shown in FIG. 8B, the subtraction circuit (25) extracts the ghost wave component Sg of the received signal GCR (this ghost wave component Sg is also an error component that could not be removed).

Then, this ghost wave component Sg is sent to the differentiating circuit (27)
The pulse pq is supplied to a differential pulse pq as shown in C in the figure, and this pulse pq is supplied to a correlation calculation circuit (23).

This correlation calculation circuit (23) performs a product-sum calculation of pulses pp and pq to obtain ghost information, and this calculation circuit (23) performs a routine (30) as shown in FIG. Processing takes place.

However, as shown in FIG. 7B, the number of samples of pulse pp is 64 samples. Also, i: tap number of filter (3). i=0 to 639m: sampling point for pulse Pq during the ghost removal period, for example -2 μ seconds to +45 seconds n: sampling point of pulse Pp. n=o to Xn: Quantized value of pulse Pp Ym: quantized value of pulse Pq R5LT: Calculation result.

In addition, the arithmetic circuit (23) merely illustrates the processing of the microcomputer (20) isometrically as hardware;
In reality, the routine (30) is processed by the microcomputer (20).
Executed by

Then, when the processing of this routine (3o) is executed,
In the step (32), the calculation result R5LT is output from the calculation circuit (23) and supplied to the conversion circuit (28), and the calculation result RSLT is applied to the first filter (3).
It is converted into a tap coefficient (tap gain) Ti for the th tap.

This coefficient Ti is then supplied to the filter (3), and the pass characteristics of the filter (3) are controlled so that the ghost wave component Sg of the GCR signal 5GCR extracted from the filter (3) is canceled out and removed. .

Therefore, at this time, the ghost wave component of the video signal sy is also canceled out and removed in the filter (3), and the video signal SY containing no ghost wave component is taken out at the terminal (5).

Literature: 1989 National Conference Journal of the Television Society "Ghost Cancellation Standard Signal System" [Problem to be Solved by the Invention] However, as described above, when the correlation calculation in the calculation circuit (23) is executed by the routine (30), As is clear from the figure, the multiplication in step (31) must be performed 40,960 times (=640 taps×64 samples), which is the product of the number of taps of filter (3) and the number of samples of pulse PP.

However, since multiplication is one of the slowest processes for the microcomputer (20), executing multiplication more than 40,000 times by software would require a lot of time and would be impractical.

In this respect, if this correlation calculation circuit (23) is configured by hardware, high-speed calculation can be performed. However, this requires a dedicated correlation calculation circuit, which increases the cost and complicates its control.

This invention attempts to eliminate these problems.

[Failure to solve the problem]

Therefore, in this invention, in correspondence with the embodiments described later, the correlation calculation is executed by a microcomputer (20) and software (40), and the value of the pulse pp of the GCR signal to be subjected to the correlation calculation is divided by 2. This is approximated by a power, and the pulse PqO value is bit-shifted using this approximate value to realize multiplication in the correlation calculation.

[Effect]

Although the multiplication is performed by software, the time required for the correlation operation is reduced because the multiplication is performed by bit shifting.

〔Example〕

FIG. 1 shows a routine (40) executed by the microcomputer (20), and this routine (40) performs a correlation calculation between pulses PP and Pq to extract ghost information RSLT.

In other words, the processing of the microcomputer (20) is the routine (40).
Start from step (41), then step (42)
In step (43), n=o is set.
In , the Nth quantized value Xn of the pulse PP is approximated to a value xn that is a power of two.

In this case, since the pulse PP is quantized to 8 bits in this example, the approximation in step (43) is performed, for example, with the relationship shown in FIG. Also,
Such an approximation can be achieved by using a data table as shown in FIG.

Subsequently, in step (44), the value n is incremented by 1, and then in step (45), n≧
It is checked whether the number is 64, and if n is 64, the process returns from step (45) to step (43).

Thus, by steps (42) to (45), all quantization values X of pulse pp. −X63 is approximated to a power of 2 (!: X o−X 63.

Then, when all the approximations have been performed, the process proceeds from step (45) to step (51), and step (51)
) to (53), the values i, m, n, R3
LT initialization is performed.

Subsequently, in step (54), the mth pulse Pq
The th quantized value Ym is bit-shifted to the left by the value indicated by the exponent of the value xn to obtain the value Z. For example, xn=32
If so, since this is 2 to the 5th power, the value Ym is bit-shifted to the left by 5 bits.

That is, the value Ym is bit-shifted by log,xn bits to obtain the value Z. However, "'0pa" is set in the lower bit of the value Z which becomes vacant due to the bit shift.Furthermore, when χn=o, 2=0 is set.

In this case, this focus shift is equivalent to multiplying the value Ym by the value xn, and therefore the value Z is the product of the value Ym and the approximation xn of the value Xn.

Next, in step (55), the value Z resulting from the bit shift is added to the previous value R5LT, and subsequently, in step (56), the values n and m are each incremented by 1. And then step (57
), it is checked whether n≧64,
When n<64, the process returns from step (57) to step (54).

In this way, by steps (52) to (57), the values Ym (m=i) and (!X
The sum of products with n (n = 0 = 63) is the value R5LT
It is required as.

Then, when the product sum value R5I,T for n=o~63 is obtained, the process starts from step (57) to step (6).
Proceeding to step d), the product-sum value R5LT, that is, the calculation result R5LT for tap number i is supplied to the conversion circuit (28).

Then, in the conversion circuit (28), the supplied calculation result R3LT is converted into the first tap coefficient Ti,
This coefficient Ti is supplied to the filter (3), and the filter (
The filter (3) is used so that the ghost wave component Sg of the GCR signal 5GCR extracted from the filter (3) is canceled out and removed.
) is controlled.

In addition, in reality, the value RS in this conversion circuit (28)
Conversion from LT to value Ti is also performed by the microcomputer (2o), and the value Ti is sent from the microcomputer (2o) to the filter (3).
supplied to

Subsequently, in step (62), the value i is incremented by 1, and then in step (63), i≧
640, and if i<640, the process moves from step (63) to step (52).
Return to

Thus, for every tap number i, step (
52) to (61), that is, the calculation result R5LT
is determined and the filter (3) ζ tap coefficient Ti is supplied.

Then, when the coefficient Ti is supplied for all tap numbers i, the process starts from step (63) to step (64).
) and ends this routine (40).

Thus, according to the present invention, the correlation calculation for removing the ghost L-wave component is performed. In this case, especially according to the present invention, the microcomputer (20) and the software routine (40) perform the correlation calculation. Since I am running
There is no need for a dedicated correlation calculation circuit that increases costs or complicates its control.

Also, when multiplying the value Ym and the value Xn in a correlation calculation, the multiplication is realized by approximating the value Xn by a value xn that is a power of 2, and bit-shifting the value Ym corresponding to the value xn. The bit shift is performed by the microcontroller (
Since it is one of the fastest processes for transversal filter (3), it can significantly reduce the time required for multiplication, and even if the number of multiplications is 40,000 or more, it is still practical. Characteristics can be converged.

FIG. 3 shows another connection example of the transversal filter.

That is, the video signal S from the A/D converter (2)
Y is supplied to the first subtraction circuit (71), and
The signal is supplied to the subtraction circuit (71) through the first transversal filter (72).

The output signal of the subtraction circuit (71) is further supplied to a second subtraction circuit (73), and the output signal of this subtraction circuit (73) is passed through the second transversal filter (74) to the subtraction circuit. (73), and the output signal of this subtraction circuit (73) is sent to the D/A converter (4).
supplied to

Further, the detection signal from the detection circuit (10) is transmitted through the filter (
72) and (74) as a control signal for their pass characteristics.

Therefore, since the subtraction circuit (71) and filter (72) are configured in a feedforward type loop,
From the subtraction circuit (71), a video signal sy from which adjacent ghost wave components including front ghosts have been removed is taken out.

Further, since the subtraction circuit (73) and the filter (74) are configured in a feedback type loop, the subtraction circuit (73) and the filter (74) are configured in a feedback type loop.
73), the video signal SY from which the long ghost wave component has been removed is extracted.

Therefore, the video signal SY from which the close ghost wave component and the long ghost wave component have been removed can be taken out at the terminal (5).

Note that in the above, the value The focus of the value Xn may be shifted in accordance with the approximate value.

Also, bit rotation may be used instead of bit shift.

〔Effect of the invention〕

According to this invention, since the correlation calculation is executed by the microcomputer (20) and the software routine (40), a dedicated correlation calculation circuit is required, which increases the cost and makes the control complicated. Never.

Also, when multiplying the value Ym and the value Xn in the correlation calculation, the value Xn is approximated by a value xn that is a power of 2, and (iYm
The multiplication is realized by bit shifting corresponding to the value Even if the number of multiplications is as large as 40,000 times or more, the characteristics of the transversal filter (3) can be converged within a sufficient amount of time for practical use.

FIGS. 1 and 3 are flowcharts and system diagrams of an example of this invention,
FIGS. 2 and 4 to 9 are diagrams for explaining the same.

(1) is a video detection circuit, (2) is an A/D converter, (
3) , (72), and (74) are transversal filters, (4) is a D/A converter, (10) is a detection circuit,
(11), (13) are gate circuits, (12), (14)
is a buffer memory, and (20) is a microcomputer.

Claims (1)

[Claims] A received video signal is supplied to a transversal filter, a GCR signal is extracted from the output signal of the transversal filter, a waveform distortion component is extracted from the extracted GCR signal, and a waveform distortion component is extracted from the extracted waveform distortion component. In the waveform distortion component removal circuit, the video signal from which the waveform distortion component has been removed is extracted from the transversal filter by controlling the pass characteristic of the transversal filter based on the video signal from which the waveform distortion component remains. A GCR signal is extracted from the GCR signal, a correlation calculation is performed between the extracted GCR signal and the extracted waveform distortion component, and the result of this correlation calculation is supplied to the transversal filter as a control signal for the above-mentioned pass characteristic. A waveform in which one value of the GCR signal and the waveform distortion component to be correlated is approximated by a power of 2, and the other value is bit-shifted or bit-rotated using this approximated value to realize the multiplication in the above correlation calculation. Distortion detection circuit.