JPH07212626A - Noise reduction device - Google Patents

Abstract

PURPOSE:To provide how to configure a high pass filter and how to reduce noise concretely when noise reduction processing suitable for a measured noise quantity is implemented by providing a measurement function of noise to a signal transmission path. CONSTITUTION:A signal entirely the same as or having correlation with a video signal of a line L1 is received by a line L2. A noise detection section 2-2 applies subtraction processing between the same signal waveforms arranged periodically in the video signal to obtain a difference signal and an additional noise component of a transmission system appearing at a signal period when a substantial value is to be zero is detected. A noise reduction section 2-1 divides the input video signal of the line L1 into plural frequency bands, adjusts noise limit sensitivity to each of them based on the detection quantity to reduce the noise and then synthesizes them again to obtain an output signal.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION This invention relates to a television (TV).
The present invention relates to a high-performance noise reduction device that can be applied to various video devices such as a receiver and a video tape recorder (VTR).

[0002]

2. Description of the Related Art FIG. 5 shows two configuration examples of a conventional noise reduction device. FIG. 5A shows the first configuration example. From the line L1, the frequency fm as shown in FIG.
Luminance signals having spectral components in the range up to z) are supplied. Block 1-1 is a high-pass filter, and is shown in FIG.
With such frequency characteristics, the signal component in the high frequency region of the input luminance signal is extracted and output. The next block 1-2 is a signal converter composed of a ROM or the like and has the input / output characteristics as shown in FIG. 7 (c) and functions to pass only a minute component such as noise. ing. The other input terminal L3 is an external setting terminal for adjusting up to what level the signal is extracted as noise, that is, the noise limiting sensitivity.

Blocks 1-3 are subtractors, and line L
The subtraction processing is performed between the input signal from 1 and the detected noise component from the block 1-2 to output. This output signal includes the influence of the non-linear processing in the block 1-2 and also includes the component outside the band exceeding the band fm of the video signal. Therefore, the low-pass filtering in the next block 1-4 is performed. The output signal of which noise is reduced by removing the component outside the band by the filter is output from the line L2. FIG. 5B is a second configuration example. Block 1-21 is a delay circuit (delay time 1H), 1-22 is a subtractor, 1-23 is an adder,
1-24 is a high-pass filter, 1-25 is a signal converter, 1-2
6 is a subtractor, 1-27 is a low-pass filter, and 1-28 is an adder.

The input signal from line L1 is a composite video signal. This input signal is separated into a luminance signal and a chrominance signal by a so-called comb filter composed of blocks 1-21 to 1-23. The function of the subtractor in block 1-22 is to subtract the signal delayed by one line in the delay circuit 1-21 from the input signal from the line L1 to extract the color signal. Input signal from line L1 and delay circuit 1
The function of the adder in block 1-23 is to add the signals delayed by one line at -21 to extract the luminance signal.
Blocks 1-24 to 1-27 are 1-1 to 1-1 in FIG.
Noise reduction circuits with the same functions as 1-4,
The high frequency component is separated from the luminance signal supplied from the block 1-23, and the minute component therein is removed as noise. The input line L3 of the block 1-25 is a noise limiting sensitivity adjustment input terminal like the input line L3 of the block 1-2 of FIG.

The color signal from the block 1-22 and the noise-removed luminance signal from the block 1-27 are supplied to the block 1-28, and the luminance signal and the color signal are added and combined and output. Therefore, the composite video signal in which the high frequency noise in the luminance signal component is reduced is output to the output line L2. The problem with these conventional examples is that they do not have a noise level determination function, so sensitivity setting for noise removal was either manual setting from line L3 or fixed setting in a safe place. is there. Therefore, noise reduction processing is performed even when there is little noise, the waveform is distorted due to non-linear processing, and sufficient noise reduction is not performed when there is a lot of noise. There were problems such as having to do it.

[0006]

The problem to be solved by the present invention is not the noise reduction processing by the fixed or manual sensitivity setting as in the conventional example, but the noise measurement function having the noise measurement function. The point is how to perform a suitable noise reduction process, how to construct a high-pass filter for noise reduction, how to reduce noise, and what kind of measures should be taken for that purpose.

[0007]

Therefore, the present invention solves the above-mentioned problems by constructing a noise reduction device by the following means.

(1) A signal period in which a difference signal is obtained by performing a subtraction process between the same signal waveforms periodically arranged in the first input signal which is the noise detection target signal, and the original value should be zero. A noise-sensitivity is set based on the detected noise amount, and a second noise reduction target signal that is correlated with the first input signal is detected.
The input signal of is divided into a plurality of regions by the plurality of filters, and the output of each filter is subjected to the noise reduction processing based on the noise limiting sensitivity, and then the noise-reduced first A noise reduction device characterized by being configured to re-synthesize two input signals.

(2) A signal period in which an original value should be zero by performing subtraction processing between the same signal waveforms periodically arranged in the first input signal which is the noise detection target signal to obtain a difference signal. A noise-sensitivity is set based on the detected noise amount, and a second noise reduction target signal that is correlated with the first input signal is detected.
A noise reducing device for reducing noise of the input signal, wherein an odd number of sampled value sequences of the second input signal are DCT-transformed,
After performing noise reduction processing based on the noise limiting sensitivity using the even-numbered transform terms of the DCT transform term, DC
A noise reduction device, characterized in that it is configured to perform T inverse transformation and re-synthesize the noise-reduced second input signal.

(3) A subtraction process is performed between the same signal waveforms periodically arranged in the first input signal which is the noise detection target signal to obtain a difference signal, and the original value is passed through the bandpass filter. An additional noise component of the transmission system that appears in a signal period that should be zero is detected, a noise limiting sensitivity is set based on the detected noise amount, and a noise reduction target signal that has a correlation with the first input signal is obtained. The two input signals are subjected to noise reduction after dividing the occupied frequency band of the signal into a plurality of regions by a plurality of filters and performing noise reduction processing based on the above noise limiting sensitivity on each filter output. The second
The input signal is recombined, and the bandpass filter is configured by using a coefficient value of a predetermined one of the plurality of filters for the second input signal as a representative value, A noise reduction device characterized in that it is configured to detect the amount of noise and to determine the noise limiting sensitivity.

(4) A subtraction process is performed between the same signal waveforms periodically arranged in the first input signal which is the noise detection target signal to obtain a difference signal, and the original value is passed through the bandpass filter. An additional noise component of the transmission system that appears in a signal period that should be zero is detected, a noise limiting sensitivity is set based on the detected noise amount, and a noise reduction target signal that has a correlation with the first input signal is obtained. 2 is a noise reduction device for reducing noise of an input signal, wherein an odd number of sampled value sequences of the second input signal is DCT-transformed, and the even-numbered transform term of the DCT transform term is used to After the noise reduction processing based on the noise limiting sensitivity, the DCT inverse transform is performed to re-synthesize the noise-reduced second input signal.
The band-pass filter is configured by using a predetermined one transform coefficient value in the even-numbered DCT transform terms excluding zero as a representative value, performs the noise amount detection, and obtains the noise limiting sensitivity. A noise reduction device characterized by being configured.

(5) A subtraction process is performed between the same signal waveforms periodically arranged in the first input signal which is the noise detection target signal to obtain a difference signal, and the difference signals are obtained by the first plurality of filters. The occupied frequency band of the differential signal is divided into a plurality of regions, the additional noise component of the transmission system is detected for each divided region, the noise limiting sensitivity for each divided region is set based on the detected noise amount, The second input signal, which has a correlation with the first input signal and serves as a noise reduction target signal, is divided into a plurality of regions by the second plurality of filters, and the occupied frequency band of the second signal is divided into a plurality of regions. Noise-reduction processing based on the noise limiting sensitivity obtained from the first plurality of filters is performed on the plurality of filter outputs of the first filter, and then the second noise-reduced second input signal is recombined. A noise reduction device characterized by being configured.

(6) A difference signal is obtained by performing a subtraction process between the same signal waveforms that are periodically arranged in the first input signal that is the noise detection target signal, and the original value is passed through the bandpass filter. An additional noise component of the transmission system that appears in a signal period that should be zero is detected, a noise limiting sensitivity is set based on the detected noise amount, and a noise reduction target signal that has a correlation with the first input signal is obtained. 2 is a noise reduction device for reducing noise of an input signal, wherein an odd number of sampled value sequences of the second input signal is DCT-transformed, and the even-numbered transform term of the DCT transform term is used to After the noise reduction processing based on the noise limiting sensitivity, the DCT inverse transform is performed to re-synthesize the noise-reduced second input signal.
The band-pass filter is composed of a plurality of band-pass filters using the respective transform coefficient values of the even-numbered DCT transform terms, and determines the noise limiting sensitivity by obtaining each noise amount for the even-numbered DCT transform terms. , A noise reduction device characterized by being configured to perform noise reduction.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 shows an embodiment of the noise reducing apparatus of the present invention, and FIG.
~ Fig. 4 is a detailed configuration example thereof, and Figs. 6 to 18 are operation explanatory views. FIG. 5 is a diagram of the conventional example already cited. In explaining the operation, a simplified simulated expression method is also used for convenience. Even when a digital circuit is taken as a specific circuit example, the signal waveform of the circuit may be shown as an analog waveform in order to make the explanation of the operation easier to understand. Further, for convenience of explanation, the delay of the signal due to the processing time of each circuit itself, and the delay circuit or the like normally used only for simply correcting the delay will be omitted except when necessary for explanation.

FIG. 1 is an overall view of a noise reduction device, FIG. 2 is a configuration example of a noise reduction section, FIG. 3 is a detailed configuration example of a noise reduction section, and FIG.
2A and 2B are a configuration example of a noise detection unit and a detailed configuration example thereof. Figure 1
In 2-1, 2-1 is a noise reduction unit and 2-2 is a noise detection unit. The noise detection unit detects the amount of noise and determines the noise limiting sensitivity, and the noise reducing unit removes the noise based on the noise limiting sensitivity. In FIG. 1, a noise reduction target signal (a signal for reducing noise) added to the line L1 and a noise detection target signal (a signal for detecting noise) added to the line L0.
The signals of line L1 and line L0 are exactly the same signal, or the signal of line L0 is, for example, a composite video signal, which is YC separated and then converted to the Y signal of line L1. Signals that are correlated with each other, such as being converted. Note that the noise treated as a target in this proposal is assumed to be noise added in the video signal transmission system.

A concrete example of the noise reduction section 2-1 is shown in FIG.
This is a circuit that performs DCT conversion using an odd number of sample values of the noise reduction target input signal, divides the signal frequency band, and performs noise reduction processing on the signal for each band. In FIG. 2, as an example thereof, an example using nine sample values as shown in FIG. 7F will be taken up. In FIG. 2, a block 2-11 is a delay circuit for obtaining nine sample values from an input signal, blocks 2-12 to 2-16 are circuits for obtaining even-numbered conversion terms by DCT conversion, and a block 2-17. ~ 2-
Reference numeral 1a is a signal converter that reduces noise by removing minute signal components, block 2-1b is a DCT inverse transform circuit, and block 2-1c is a low-pass filter that removes out-of-band frequency components. As an example of the input signal to the line L1 in FIG. 2, a luminance signal having frequency components up to the frequency fm = 4 MHz as shown in FIG. 7A will be treated. First block 2-1
1 is a delay circuit. FIG. 3A shows an example of the circuit, in which the signal input from the line L3a1 is delayed every time T (however, the delay time T has the same value as the sampling period Ts. 1)), and eight delay circuits (3-a1 to 3-a8) for outputting.

[0017]

[Equation 1]

The signal input from the line L3a1 in FIG. 3 and a total of nine signals delayed by T time are the line L3.
It is sent to the next stage via a2. The relationship between each sample value shown in FIG. 7F and the output of each block is as follows. Input of block 3-a1 ... x (8) (Input from line L3a1) Output of block 3-a1 ... x (7) Output of block 3-a2 ... x (6) Output of block 3-a3 ... x (5 ) Output of block 3-a4 ... x (4) Output of block 3-a5 ... x (3) Output of block 3-a6 ... x (2) Output of block 3-a7 ... x (1) Of block 3-a8 Output ... x (0) The next block 2-12 to 2-16 is DCT (Discrete
cosine transform) A converter. The DCT transform is expressed by the following equation.

[0019]

[Equation 2]

FIG. 8 shows the impulse response, that is, the frequency characteristic of each DCT conversion term when N = 9. The way to look at the characteristics in the figure is as follows. Characteristic of symbol ... Frequency characteristic of X (0) Characteristic of symbol ... Frequency characteristic of X (1) Characteristic of symbol ... Frequency characteristic of X (2) Characteristic of symbol ... Frequency characteristic of X (3) Characteristic of symbol ... X ( Frequency characteristics of 4) Symbol characteristics ... Frequency characteristics of X (5) Symbol characteristics ... Frequency characteristics of X (6) Symbol characteristics ... Frequency characteristics of X (7) Symbol characteristics ... X (8) frequency characteristics However, the DCT inverse transform equation is also shown in the following equation.

[0021]

[Equation 3]

Nine sample values x (0) to x of the input signal
DCT conversion is performed using (8), noise reduction processing is performed, and D
When the inverse CT is performed and the sample value to be reproduced is x (4), the DCT conversion term used for reproduction x (4) has a relationship with the cosine term (that is, the cosine term of the odd-numbered term becomes zero). Be)
From X (0), X (2), X (4), X (6), X
There are only five conversion terms in (8). Therefore, the number of calculation blocks for DCT conversion is 9
Effectively, only 5 blocks 2-12 to 2-16 are needed. This fact, that is, performing a noise reduction process with a smaller number of operations by using a larger number of sampled values is one of the ingenuities and features in applying the DCT transform to the present invention.
Is one. Each of the blocks 2-12 to 2-16 has exactly the same circuit configuration as shown in FIG. In FIG. 3B, the nine signals x (0) to x (8) applied via the input line L3b1 are respectively block 3-.
b1 to 3-b9, where DC for each signal
The T-transform coefficient value (= coefficient value of x (n) in Equation 2) is multiplied, weighted and output, and the next combiner 3-ba
Are added and synthesized in the above, and output from the line L3b2. The difference between blocks 2-12 to 2-16 is block 3-b.
It is a coefficient value of DCT conversion set to 1 to 3-b9.

In block 2-12, the DCT transform term X (0) when m = 0 in the equation 2 is obtained, and in block 2-12
13 is the DCT conversion term X (2) when m = 2, block 2-14 is the DCT conversion term X (4) when m = 4, and block 2-15 is the DCT conversion term X when m = 6. (6),
In block 2-16, the DCT conversion term X when m = 8
(8) is required. Blocks 2-12, 2-13, 2
DCT conversion term outputs of -14, 2-15, and 2-16 are X (0), X (2), X (4), X (6), and X (8), respectively.
And is an even-numbered conversion term. X (0) is the DC term,
The terms other than X (0) are AC terms, and the spectrum components obtained by dividing the frequency band into substantially equal parts are output as shown in FIG.
The signal converters of blocks 2-17 to 2-1a are connected to each AC term. Although each signal converter has the same circuit configuration, an example thereof is shown in FIG. The circuit configuration is almost the same as that of the conventional example of FIG.
2 and 1-3 are blocks 3-c1 and 3- of FIG.
It has the same function as c2. The high-pass filter shown in FIG. 5A is not shown in FIG. 3C, but a DCT converter (blocks 2-13 to 2-16) which is a bandpass filter is used instead.

The block 3-c1 is a signal converter composed of a ROM or the like, and has an input / output characteristic as shown in FIG. 7 (d), and functions to pass only a minute component such as noise. ing. The noise limiting sensitivity K (m) at this time (the upper limit value of the signal amplitude to be passed) is externally set from the other input terminal L3c3. The line L3c3 is connected to the line L3 from the noise detection unit 2-2 in FIG. In the description of the signal converter 1-2 of the conventional example, the conversion characteristic as shown in FIG. 7C was shown as an example. Here, although the conversion characteristic is similar, as shown in FIG. 7D. Use conversion characteristics.
Assuming that the input signal to the signal converter 1-2 is qi and the output signal is qo, the noise limiting sensitivity K from the line L3
It is expressed by the following equation using (m) as a parameter.

[0025]

[Equation 4]

Therefore, the input qi is the noise limiting sensitivity K (m).
When it is smaller than, it is output as it is as a noise component and q
When i is larger than K (m), it is interpreted as not noise, and it works so that the output becomes zero. As will be described later, the noise limit sensitivity K (m) is a value determined from the standard deviation σ (m) obtained from the amount of noise detected by the noise detector, and is completely correlated with the noise component. When the noise is small, the value is small, and when the noise is large, the maximum Kma.
Since the value becomes large in the range up to x, the output qo correlated with the noise amount can be obtained. The block 3-c2 is a subtracter, and the block 3-c1 is connected to the input of the line L3c1.
The detection noise component of is subtracted and output from the line L3c2. As a result, the minute signal component that is interpreted as noise is removed. This output is added to the DCT inverse converter 2-1b at the next stage.

Signal converters 2-17, 2-18, 2-1
9, 2-1a have the same circuit configuration, but the noise limiting sensitivity K (m) set via the line L3 can have different values, so that, as shown in FIG. ，
It is possible to perform different noise limiting processing for the characteristics of and. The reason why the noise limiting process is not performed on X (0) from the block 2-12 is that the DC term X (0) is generally mixed with a positive setup value as a feature of the image signal. Therefore, it is difficult to remove noise, and because the DC term requires a considerable degree of accuracy, it is better not to perform non-linear processing if possible, but this noise reduction processing for X (0) is prohibited. Since this is not the case, noise reduction can be performed by a circuit similar to the other even terms if necessary. The next block 2-1b in FIG. 2 is the DCT inverse transformer, and block 2-1
DCT conversion DC term X (0) from 2 and four DCT conversion AC terms X (2), X (4), X after noise reduction processing
From (6) and X (8), DCT inverse transformation is performed according to equation 3 to obtain a value x (4) (see the following equation (5)).

[0028]

[Equation 5]

Since the variable of the cosine term of this expression is an integral multiple (m) of π / 2, the cosine term becomes zero and has no value when m is an odd number. Therefore, m = 0, 2, 4, 6, 8 X
If the value of (m) is used, x (4) can be reproduced. Figure 3
(D) is a circuit example of the DCT inverse converter 2-1b.
The signal X from the block 2-12 is sent to the line L3d1.
(0), the signal X (2) from the block 2-17 on the line L3d2, the signal X (4) from the block 2-18 on the line L3d3, and the block 2-19 on the line L3d4.
From the signal X (6) from the line L3d5 to the block 2-
The signal X (8) from 1a is added, each signal is amplified by the multipliers 3-d1 to 3-d5 by the DCT inverse transform coefficient value, and added and combined by the combiner 3-d6, and the line L3d6 is added. Is output via. DCT obtained in this way
If the noise reduction processing is not performed, the inverse transform term x (4) means that the input signal is reproduced as it is, but when the noise reduction accompanied by the non-linear processing is performed, the signal including a large amount of out-of-band components is obtained. Becomes

The low-pass filter of the block 2-1c is a circuit for blocking this out-of-band component, and has a function of removing frequency components of frequency fm or higher as shown in FIG. 7 (a). FIG. 3E shows an example of the circuit configuration of this low pass filter.
The line L3e1 is an input line and the line L3e2 is an output line. The blocks 3-e1 to 3-e6 have a delay time T.
= Ts (sampling period) delay circuit. The input / output signals of the respective delay circuits are added to the next multipliers 3-e7 to 3-ed, weighted by the coefficient value as the low-pass filter, and added and combined by the next combiner 3-ee to output the line L3e2. Is output via. Thus, FIG. 3E shows a low-pass filter having a transversal filter configuration.

Next, the noise detecting section which is another feature of the present invention will be described. Block 2-2 in FIG. 1 is a noise detection unit. FIG. 4A shows the first circuit configuration example. As the second input signal (input 2) which is the noise detection target signal, the line L0 is supplied with a composite video signal having a contained frequency component up to fm (about 4 MHz) as shown in FIG. 7A. In this composite video signal, there is a signal portion having the same waveform in a predetermined cycle.

One example thereof is the signal portion of three lines before and after the vertical synchronizing signal in the vertical blanking period. FIG. 6A shows the first field (even field), and FIG. 6B shows the second field.
FIG. 5 is a waveform diagram of a vertical synchronization signal period of a field (odd field) and equalizing pulse periods of 3H (= 3 lines) before and after the same, but these waveforms are completely the same. Therefore, if the difference is obtained between the signals separated by one field (262.5H), the section Tn2 of t = ta2 to tb2 becomes a no signal period as shown in FIG. 6C or 6D. If there is noise in this no-signal period, the noise is not canceled by subtraction because of its decorrelation, but rather the peak value is doubled and output. Therefore, noise detection can be performed in the section Tn2. Actually, the time interval required to measure the noise amount is
Shorter periods are sufficient. For example, as shown in Figure 6,
T = tc2 to td2 during the equalizing pulse period before the vertical synchronizing signal
1H period, that is, the range of the section Td2 can be used for noise detection. When the sampling frequency is set to fs (= 4fsc), about 900 sampled values are obtained. Therefore, it is necessary to measure the noise amount to obtain the standard deviation value σ (m) and the noise limiting sensitivity K (m). This is a sufficient number of data.

Noise Detection Circuit In FIG. 4A, the composite video signal x2 which is the noise detection target signal from the line L0.
(T) is the subtractor of block 4-a1 and block 4-a2
Is added to the delay circuit. This delay circuit delays an input signal by one field (= 262.5H) and outputs the delayed signal. The line L0 from the next subtracter 4-a1
From the input x2 (t) and the signal x one field before
A signal d (t) (see the following equation (Equation 6)) which is the difference from 2 (t−T) is output.

[0034]

[Equation 6]

When the input signal x2 (t) is shown in FIG. 6 (a), x2 (t-Td) in FIG. 6 (b) is subtracted from FIG.
An output like d (t) in (c) is obtained. When FIG. 6B shows the input signal x2 (t), x2 of FIG.
(T-Td) is subtracted to obtain an output like d (t) in FIG. 6 (d). The output signal of this subtractor is shown in FIG.
The other is either (d). When there is no noise in this way, the Tn2 period in the range of t = ta2 to tb2 is a no signal period. Even if there is a ghost or the like, there is a correlation between the fields as in the case of the signal, which is canceled out and becomes zero.
However, when there is noise added in the transmission system, there is no correlation between the fields, so that it is output as a significant value. This noise detectable range Tn2
Thus, sufficient noise measurement accuracy can be obtained. The output d (t) of the subtractor 4-a1 is added to the next block 4-a3. This block is a delay circuit, and FIG. 3A shows an example of its circuit configuration. It is the same circuit configuration example as that used in the noise reduction unit. The line L3a1 is an input line, delay circuits with delay time T = Ts (Equation 1) are cascaded, and a total of nine signals at each time T are output via the output line L3a2.

Input of block 3-a1 ... d (8) (Input from line L3a1) Output of block 3-a1 ... d (7) Output of block 3-a2 ... d (6) Output of block 3-a3. d (5) Output of block 3-a4 ... d (4) Output of block 3-a5 ... d (3) Output of block 3-a6 ... d (2) Output of block 3-a7 ... d (1) Block 3 -A8 output ... d (0)

The next four blocks 4-a4 to 4- in FIG.
Reference numeral a7 denotes the same DCT converter as 2-13 to 2-16 in FIG. 2 (see FIG. 3B), and the even-numbered DCT conversion terms D (2), D (4), and D (6). , D (8) by the following equation. DCT transform term X in the noise reduction unit
The same noise detection filter as that used when (2), X (4), X (6) and X (8) is obtained is configured.

[0038]

[Equation 7]

The outputs of the blocks 4-a4 to 4-a7 are the next standard deviation circuits 4-a8 to 4-a directly connected to them.
connected to b. These standard deviation circuits have the same configuration as shown in FIG. 4 (c). The block 4-c1 is a squaring circuit for squaring an input signal and outputting the squared signal, 4-c2
Is an adder, 4-c3 is a delay circuit that delays the input signal by one clock, and 4-c4 is a function that normalizes the input signal by the number of sample values used for addition, and further its square root. Is a data conversion circuit having a function of obtaining The delay circuit 4-c3 including a latch circuit and the like starts its operation at the same time when the noise measurement period Td1 (FIG. 6) starts, forms a loop with the adder 4-c2, and is input from the block 4-c1. The squared values of the signals are integrated and, after the end of the period Td2, output from the block 4-c4 via the line L4c2 as the standard deviation value in the period Td2. With these circuits, as shown in the following equation, the square root of the average value of the sum of squares of Di (m) in the Td2 section of FIG. 6, that is, the standard deviation value σ (m) is obtained by the following equation.

[0040]

[Equation 8]

The next blocks 4-ac to 4-af connected to the blocks 4-a8 to 4-ab are memory circuits, which function to store and output the obtained standard deviation value σ (m). is doing. As described in the description of FIG.
Since the waveforms of (c) and (d) are obtained for each field, the standard deviation value σ (m) is also updated for each field. The next blocks 4-ag to 4-aj connected to the blocks 4-ac to 4-af are data conversion circuits, and an example of the conversion characteristics is shown in FIG. The standard deviation value σ (m) is input on the horizontal axis, and the noise limiting sensitivity K is on the vertical axis.
(M). This conversion characteristic is expressed by the following equation.

[0042]

[Equation 9]

As a result, a value (output 2) obtained by multiplying the noise amount σ (m) by p (m) is sent to the next-stage noise reduction section 2-1 via the line L3 as the noise limiting sensitivity K (m). I can do things. By using p (m), an individual coefficient value can be set for each DCT conversion term, so that the optimum noise limiting sensitivity K (m) can be set for each DCT conversion term.

FIG. 4B shows a second configuration example of the noise detecting section. In FIG. 4A, the DCT conversion, the standard deviation circuit, the data conversion circuit, and the storage circuit, which have four sequences, are selected as one representative sequence of noise characteristics to obtain the noise detection sensitivity K (m). . In the example of FIG. 4B, m = in FIG.
4 series circuits will be used. This is because the characteristic in FIG. 8 is in the high frequency component region in the video signal band, and is in the region where the image quality is particularly affected. The blocks 4-b1 and 4-b2 in FIG.
The circuit is exactly the same as the blocks 4-a1 and 4-a2 in (a), and has a subtractor and one field (262.5H).
Delay circuit. Therefore, the input x2 from line L0
The difference signal d (t) between (t) and x2 (t-Td) is obtained (see Formula 6).

The next block 4-b3 is 4-in FIG. 4 (a).
The delay circuit is the same as a3, and FIG. 3A is an example of the circuit. Based on the signal d (t) from the block 4-b1, a total of nine signals (sample values) separated by time T are simultaneously output. The next block 4-b4 is a DCT conversion circuit, which is a DCT conversion circuit for obtaining D (4) in FIG. 4A and the same bandpass filter as the block 4-a5. This bandpass filter has a frequency characteristic as shown in FIG. 8 and functions to extract the noise component D (t) in a region corresponding to the high frequency region (2 to 4.5 MHz) of the video signal band. The next block 4-b5 is the same standard deviation circuit as the block 4-a9, and has the circuit configuration of FIG. Here, the square root of the mean value of the sum of squares of the period Td2 of the input signal D (t), that is, the standard deviation value σ is obtained by the following equation.

[0046]

[Equation 10]

The next block 4-b6 is a memory circuit,
It has a function of storing and outputting the obtained standard deviation value σ. As described above, the standard deviation value σ is updated for each cross field. The next block 4-b7 is a data conversion circuit, and FIG. 7 (e) is an example of its conversion characteristic (see the equation 9).
The noise limiting sensitivity K (m) thus obtained is supplied as the output 2 to the noise reduction unit 2-1 via the line L3, that is, the blocks 2-17 to 2-1a, and the noise reduction processing according to the noise amount is performed. Can be done. Unlike FIG. 4, the noise measurement uses the measurement values in the representative region when the frequency band is divided, and based on the standard deviation value σ obtained, block 2-1
The noise limiting sensitivity K (m) for 7 to 2-1a can be individually set by the coefficient p (m) as in the following equation.

[0048]

[Equation 11]

Next, the result of the simulation performed to see the effect of the present invention will be shown. The target circuit is the noise reduction unit 2-1 of FIG.
-2 in FIG. 4A, the input signal (input 1) that is the noise reduction target signal from the line L1 is the luminance signal, and the line L
An input signal (input 2) which is a noise detection target signal from 0 is a composite video signal (vertical synchronization signal part and equalized pulse signal part)
And FIG. 9A is an example of the waveform of the luminance signal input to the line L1 when there is no noise. Amplitude 1 and width 10μ
7 is a pulse waveform obtained by subjecting a rectangular wave of s to a high frequency cutoff process similar to that of FIG. FIG. 9B shows the waveform of FIG. 9A with its amplitude displayed in logarithm. FIG. 9C shows d
It is a spectrum distribution figure of B display. It can be seen that the waveform is a band-limited waveform that rapidly attenuates from a frequency of 4 MHz to 4.5 MHz. The following operation waveform charts, FIGS.
Also has the same display form.

FIG. 10 shows an example in which noise is added to FIG. It is a figure which can understand well the noise attachment of a small signal part when FIG.10 (b) is compared with FIG.9 (b). This FIG. 10 is shown in FIG.
When the output signal from the line L2 is obtained by processing with the noise reducing device of FIG. FIG. 13 (a) to FIG.
It can be seen that fine high-frequency components are reduced as compared with (a). Further, comparing FIG. 13 (b) with FIG. 10 (b), the amplitude of noise is reduced and the density thereof is also coarse, and the effect of reducing the noise component can be seen. FIG. 12 shows a signal converter 2-17,
It is a synthetic | combination output waveform diagram of the noise component used as the reduction object extracted by 2-18, 2-19, and 2-1a. This waveform diagram is not a waveform diagram directly obtained from the configuration diagram of FIG. 2, but blocks 2-17 and 2-1 are included for the sake of clarity.
It is a waveform diagram obtained by synthesizing the outputs of 8, 2-19, and 2-1a. Signal converters 2-17, 2-18, 2-19,
Since the process of 2-1a is a non-linear process, FIG.
As can be seen from the above, the existence of an out-of-band component (frequency component of 4.5 MHz or higher) is recognized.

FIG. 11 is a waveform diagram of the differential signal d (t) in the noise detector 2-2, which is the subtracter 4-a1 of FIG. 4 (a).
Is an output waveform of. In FIG. 11A, the waveform is enlarged and displayed in the amplitude direction in order to facilitate understanding of the state of the noise component (compared to FIG. 10A and the like). Comparing FIG. 11 (b) and FIG. 10 (b), it can be seen that the noise components are at the same level (actually 3 dB up). After this noise component detection, the standard deviation value σ (m) and the noise limiting sensitivity K
Although (m) is obtained, the conversion as shown in the following equation is performed using the above-mentioned equation 9. The result of the noise reduction process based on the noise limiting sensitivity K (m) is shown in FIG.

[0052]

[Equation 12]

FIG. 14 and FIG. 17 are waveforms corresponding to FIG. 10 and FIG. 13 and are input / output waveform diagrams of the noise reducing section when the noise amplitude is halved. Compared to FIG. 10, 13 noise components are clearly reduced. The collapse of the ringing waveform at the pulse edge is also extremely small. Next, let us look at the operation of the noise reduction device when a clean signal with no noise is input. When the waveform of FIG. 9 is input to the line L1, there is no noise, so the value of the noise limiting sensitivity K (m) from the line L3 becomes zero, and the noise reduction processing is not performed. Therefore, the final output signal obtained from the line L2 is as shown in FIG. 18, which has exactly the same waveform as that of the input signal shown in FIG.

As described above, since the noise components are detected independently and the noise reduction processing is performed by the noise limiting sensitivity K (m) based on the noise amount, when the noise is small, the noise reduction effect becomes weak and the nonlinear Mixing distortion into the signal by processing,
It can be kept to the minimum necessary. This was not possible with the prior art. As described above, in the present invention, the noise detection unit 2-2 of FIG. 1 determines the noise limiting sensitivity based on the noise measurement result in the same signal waveform portion in the video signal, and the noise reducing unit 2-based on this noise limiting sensitivity. In step 1, noise reduction processing is performed for each sample point. When the vertical synchronizing signal part and the equalizing pulse parts before and after it in FIG. 6 are used for noise measurement, the noise change can be detected every 1 field (1/60 seconds). You can respond quickly when making changes.

Although the standard deviation value σ (m) is used in obtaining the noise limiting sensitivity K (m), equivalent performance can be obtained by using the average value of the sum of absolute values instead. . D
Although the number of sample values used for CT conversion is set to 9 in the embodiment, the number is not necessarily limited to 9 and may be 7 or 5. However, it is considered that an odd number is better because the characteristics are well balanced and there are many parts that can be omitted in terms of circuitry. Lastly, it should be decided based on the balance between cost and performance. Similarly, the reason why the number of sample points used for DCT conversion is set to an odd number is to suppress the circuit scale and to consider the degree of influence of nonlinear processing for noise reduction on a signal.
The T-transform, noise measurement, noise reduction process, and DCT inverse transform can be performed without any damage. With the same gist, various orthogonal transforms such as Hadamard transform and discrete sign transform can be easily applied. The point is that the occupied frequency region of the signal is divided into a plurality of regions by the signal conversion unit including a plurality of filters such as various orthogonal transformation units, the noise measurement is performed in the no-signal period, the noise limitation sensitivity is determined, and the noise limitation sensitivity is determined. This means that the noise-reduced input signal can be recombined after performing the noise reduction processing based on the sensitivity.

[0056]

As described above, the noise reduction device of the present invention has the following effects. (B) The subtraction process is performed between the same signal waveforms that are periodically arranged in the video signal to obtain a difference signal, and the additional noise component of the transmission system that appears in the signal period when the original value should be zero is detected. Therefore, objective and accurate noise measurement can be performed without being affected by the signal content. (B) Since the noise limiting sensitivity is set and the noise is removed by using the objective noise measurement result, it is possible to perform the optimum noise reduction processing according to the noise amount. The distinction between noise and signal is significantly improved as compared with the prior art. (C) The signal frequency band is divided by DCT conversion, noise measurement is performed for each frequency band, noise limit sensitivity is set, and noise is removed. Therefore, it is possible to perform appropriate and fine noise reduction processing. . The distinction between noise and signal is significantly improved as compared with the prior art. (D) Even if various orthogonal transforms other than the DCT transform are used, D
It is possible to perform high-precision noise reduction processing similar to CT conversion. (E) When the amount of noise is small, the effect of noise reduction processing also weakens, so the influence of nonlinear distortion on the video signal also decreases.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of an entire noise reduction device of the present invention.

FIG. 2 is a block configuration diagram showing a noise reduction unit of the noise reduction device of the present invention.

FIG. 3 is a block configuration diagram showing a detailed configuration example of a noise reduction unit of the present invention.

FIG. 4 is a block configuration diagram showing a configuration example and a detailed configuration example of a noise detection unit of the noise reduction device of the present invention.

FIG. 5 is a block configuration diagram showing two conventional examples of a noise reduction device.

FIG. 6 is a waveform diagram for explaining the operation of the entire noise reduction device of the present invention.

FIG. 7 is a waveform diagram for explaining the operation of the entire noise reduction device of the present invention.

FIG. 8 is a waveform diagram for explaining the operation of the entire noise reduction device of the present invention.

FIG. 9 is a waveform diagram for explaining the operation of the entire noise reduction device of the present invention.

FIG. 10 is a waveform diagram for explaining the operation of the entire noise reduction device of the present invention.

FIG. 11 is a waveform diagram for explaining the operation of the entire noise reduction device of the present invention.

FIG. 12 is a waveform diagram for explaining the operation of the entire noise reduction device of the present invention.

FIG. 13 is a waveform diagram for explaining the operation of the entire noise reduction device of the present invention.

FIG. 14 is a waveform diagram for explaining the operation of the entire noise reduction device of the present invention.

FIG. 15 is a waveform diagram for explaining the operation of the entire noise reduction device of the present invention.

FIG. 16 is a waveform diagram for explaining the operation of the entire noise reduction device of the present invention.

FIG. 17 is a waveform diagram for explaining the operation of the entire noise reduction device of the present invention.

FIG. 18 is a waveform diagram for explaining the operation of the entire noise reduction device of the present invention.

[Explanation of sign]

 L0, L1, L2, L3 Line 2-1 Noise reduction section 2-2 Noise detection section

[Procedure amendment]

[Submission date] August 22, 1994

[Procedure Amendment 1]

[Document name to be amended] Statement

[Correction target item name] 0039

[Correction method] Change

[Correction content]

The outputs of the blocks 4-a4 to 4-a7 are the next standard deviation circuits 4-a8 to 4-a directly connected to them.
connected to b. These standard deviation circuits have the same configuration as shown in FIG. 4 (c). The block 4-c1 is a squaring circuit for squaring an input signal and outputting the squared signal, 4-c2
Is an adder, 4-c3 is a delay circuit that delays the input signal by one clock, and 4-c4 is a function that normalizes the input signal by the number of sample values used for addition, and further calculates the square root thereof. A data conversion circuit having a function and configured by a ROM or the like. The delay circuit 4-c3 including a latch circuit and the like starts its operation at the same time when the noise measurement period Td2 (FIG. 6) starts, forms a loop with the adder 4-c2, and is input from the block 4-c1. The square values of the signals are integrated, and after the end of the period Td2, block 4-c4
Is output as the standard deviation value in the period Td2 via the line L4c2. With these circuits, as shown in the following equation, 2 (di) of Di (m) in the Td2 section of FIG.
The square root of the average value of the sum of multiplications, that is, the standard deviation value σ (m) is obtained by the following equation.

[Procedure Amendment 2]

[Document name to be amended] Statement

[Correction target item name] 0053

[Correction method] Change

[Correction content]

FIG. 14 and FIG. 17 are waveforms corresponding to FIG. 10 and FIG. 13 and are input / output waveform diagrams of the noise reducing section when the noise amplitude is halved. 17 compared to FIG.
The noise component of is clearly reduced. The collapse of the ringing waveform at the pulse edge is also extremely small. next,
Let's look at the operation of the noise reduction device when a clean signal with no noise is input. When the waveform of FIG. 9 is input to the line L1, there is no noise, so the value of the noise limiting sensitivity K (m) from the line L3 becomes zero, and the noise reduction processing is not performed. Therefore, the final output signal obtained from the line L2 is as shown in FIG. 18, which has exactly the same waveform as that of the input signal shown in FIG.

[Procedure 3]

[Document name to be corrected] Drawing

[Name of item to be corrected] Fig. 4

[Correction method] Change

[Correction content]

[Figure 4]

Claims (6)

[Claims] 1. A difference signal is obtained by performing a subtraction process between identical signal waveforms that are periodically arranged in a first input signal that is a noise detection target signal, and a difference signal is obtained during a signal period when the original value should be zero. An additional noise component of the transmission system that appears is detected, a noise limiting sensitivity is set based on the detected noise amount, and a plurality of second input signals that are correlated with the first input signal and are noise reduction target signals are set. A filter divides the occupied frequency band of the signal into a plurality of regions, performs noise reduction processing based on the noise limiting sensitivity on each filter output, and then re-generates the noise-reduced second input signal. A noise reduction device characterized by being configured to be combined. 2. A difference signal is obtained by performing a subtraction process between the same signal waveforms that are periodically arranged in the first input signal that is the noise detection target signal, and the difference signal is obtained during the signal period when the original value should be zero. An additional noise component of the appearing transmission system is detected, a noise limiting sensitivity is set based on the detected noise amount, and noise reduction of a second input signal which is a noise reduction target signal in correlation with the first input signal A noise reduction device for performing a DCT conversion on an odd number of sampled value sequences of the second input signal, and using the even-numbered conversion terms of the DCT conversion terms to reduce noise based on the noise limiting sensitivity. A noise reduction device characterized in that it is configured to perform DCT inverse transformation after processing and re-synthesize the noise-reduced second input signal. 3. A difference signal is obtained by performing a subtraction process between the same signal waveforms periodically arranged in the first input signal which is the noise detection target signal, and the original value is zero through a bandpass filter. A second noise that becomes a noise reduction target signal that has a correlation with the first input signal is detected by detecting an additional noise component of a transmission system that appears in a signal period that should be The input signal of
The occupied frequency band of the signal is divided into a plurality of regions, noise reduction processing based on the noise limiting sensitivity is performed on each filter output, and then the noise-reduced second input signal is recombined. And the bandpass filter comprises the second bandpass filter.
Of the plurality of filters with respect to the input signal of 1) is used as a representative value, the noise amount is detected, and the noise limiting sensitivity is determined. Noise reduction device. 4. A difference signal is obtained by performing a subtraction process between identical signal waveforms which are periodically arranged in a first input signal which is a noise detection target signal, and an original value is zero through a bandpass filter. A second noise that becomes a noise reduction target signal that has a correlation with the first input signal is detected by detecting an additional noise component of a transmission system that appears in a signal period that should be Noise reduction device for reducing the noise of the input signal, the odd number sample value sequence of the second input signal is DCT-transformed, and the noise of the noise is reduced by using an even-numbered transform term of the DCT transform term. After performing noise reduction processing based on the limiting sensitivity, the DCT inverse transform is performed to re-synthesize the noise-reduced second input signal, and the band-pass filter is an even-numbered DCT other than the zero. One predetermined conversion factor in the conversion terms Configured using a value as a representative value, performs the noise amount detection, noise reduction device, characterized in that configured so as to determine the noise limit sensitivity. 5. A difference signal is obtained by performing subtraction processing between identical signal waveforms periodically arranged in a first input signal which is a noise detection target signal, and the difference signal is obtained by a first plurality of filters. The occupied frequency band of the signal is divided into a plurality of regions, the additional noise component of the transmission system is detected for each divided region, and the noise limiting sensitivity for each divided region is set based on the detected noise amount. A second input signal, which is a noise reduction target signal and has a correlation with the first input signal, is divided into a plurality of regions by the second plurality of filters, the occupied frequency band of the second signal;
After the noise reduction processing based on the noise limiting sensitivity obtained from the first plurality of filters is performed on the outputs of the second plurality of filters, the second noise-reduced second input signal is regenerated. A noise reduction device characterized by being configured to be combined. 6. A difference signal is obtained by performing a subtraction process between the same signal waveforms periodically arranged in the first input signal which is the noise detection target signal, and the original value is zero through a bandpass filter. A second noise that becomes a noise reduction target signal that has a correlation with the first input signal is detected by detecting an additional noise component of a transmission system that appears in a signal period that should be Noise reduction device for reducing the noise of the input signal, wherein the odd number sampled value sequence of the second input signal is DCT-transformed, and the noise of the noise is reduced by using an even-numbered transform term of the DCT transform term. The noise reduction process based on the limiting sensitivity is performed, and then the inverse DCT transform is performed to resynthesize the second noise-reduced second input signal. The bandpass filter is configured to include the even-numbered DCT transform term. Using the respective conversion coefficient values of A noise reduction device comprising a plurality of bandpass filters, configured to perform noise reduction by obtaining noise amounts for the even-numbered DCT conversion terms to determine noise limiting sensitivity.