JP2814188B2 - Noise compression device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a noise compression device,
In particular, the present invention relates to a technique for compressing the amplitude of a noise component included in a video signal of each of the three primary colors to remove noise.

[0002]

2. Description of the Related Art FIG. 11 shows a conventional noise reduction method for real-time noise reduction of a video signal.
A noise compression device using a noise coring method such as that described above is known. This method removes small-amplitude component noise in high-frequency components of a video signal.

In FIG. 11, from an input terminal 1 to an output terminal 6, a low-pass filter (hereinafter, referred to as "LPF") 2 for extracting a low-frequency signal component having a frequency equal to or lower than a predetermined frequency of a video signal.
A subtracter 3 for subtracting the low-frequency signal component extracted by the LPF 2 from the original input signal; a limiter circuit 4 for applying a limiter to a signal of the subtracter 3 at a predetermined limiter level; An adder 5 that adds the output signal and the output signal of the LPF 2 is sequentially connected.

[0004] This operation is described as follows.
First, for example, from a solid-state imaging device or an imaging tube to the input terminal 1,
A video signal including noise as shown in FIG. The LPF 2 extracts a low-frequency signal component including only low-frequency components as shown in FIG. 12B from this signal. The subtractor 3 subtracts the low-frequency signal component extracted by the LPF 2 from the original input signal, and obtains a high-frequency signal component in the video signal as shown in FIG. The small amplitude component in this high frequency signal component is a noise component.

In the limiter circuit 4, the limiter level is arbitrarily set, noise components having an amplitude smaller than the limiter level are removed, and a high-frequency signal component from which a noise component is removed as shown in FIG. Can be Finally, the low-frequency signal component extracted by the LPF 2 and the high-frequency signal component from which noise has been removed by the limiter circuit 4 are mixed by the adder 5, and the image from which noise has been removed as shown in FIG. A signal is output from the output terminal 6.

[0006]

However, in such a conventional noise coring type noise compression apparatus, discrimination is performed at a predetermined level, and a signal below the predetermined level is unconditionally cut. If the image is hidden inside, the outline of the image may be blurred.

That is, when the amplitude level of the noise is relatively large, the output signal of LPF2 is the same as in FIG.
Such signal becomes, the higher-band signal component obtained by the subtracter 3
Indicates that the signal of the contour part is a signal hidden in the noise component.
This makes it impossible to distinguish between signal and noise.

[0008] If, at the limiter circuit 4, combined with the limiter level to the noise level, since the edge component from being removed together with the noise becomes a flat signal, or
In addition, the signal output from the adder 5 is a signal as if the low-frequency signal component from which the contour component of the LPF 2 was cut is output as it is, and becomes a video signal with a blurred contour portion.

SUMMARY OF THE INVENTION The present invention has been made in view of such a conventional problem, and has been developed in a video signal processing apparatus such as a color television camera capable of obtaining image signals of three primary colors from three image sensors. By making use of the fact that the noise in the primary color video signal has no correlation with each other, the noise component and the signal component in the high frequency signal component included in the video signal of each of the three primary colors are reliably distinguished, and the contour component is An object of the present invention is to provide a noise compression device capable of removing only a noise component without removing it.

[0010]

Means for Solving the Problems] Therefore, in the present invention, 3 Hara
For each video signal separated in color, and a higher-band signal component and the low frequency signal component is divided at a predetermined frequency by the high-frequency signal component and low
Filter means for extracting each of the signal components ;
Each high-frequency signal component of the primary color video signals are mixed at a predetermined ratio, the luminance high band signal generating means for generating a luminance high-band signal component, said
A noise discriminating signal generating means for generating a noise discriminating signal for discriminating between a noise component and a signal component based on the luminance high band signal component, and a multiplication for multiplying the noise discriminating signal for each of the three primary color high band signal components and means configured high-frequency signal components of three primary colors noise distinction signal is multiplied by the multiplication means, the adding means for adding the low-frequency signal component, including every three primary colors
A noise compression device, wherein the noise discriminating signal generating means includes:
Absolute value conversion means for converting the luminance high-frequency signal component into an absolute value,
Amplifies the absolute value of the luminance high frequency signal component with a predetermined amplification factor
Amplifying means, and the amplified high luminance signal component at a predetermined value.
And a clip means for clipping .

Each of the video signals separated into the three primary colors
Are amplified at different ratios, the luminance
The high-frequency signal generating means is a reciprocal of a ratio of the video signals,
A structure for mixing the high frequency signal components of the three primary color video signals;
May be .

[0012]

According to the above arrangement, each of the three filter means is used.
A high-frequency signal component and a low-frequency signal component divided at a predetermined frequency for each primary color video signal are extracted. The extracted high frequency signal components of the three primary color video signals are mixed at a predetermined ratio by the luminance high frequency signal generating means.

Each noise component included in a three-primary-color video signal generated by being separated into three primary colors is generated and output independently by, for example, a solid-state image pickup device or an image pickup tube. Since they are processed by circuits, they are in principle not correlated with each other. Therefore, if the high-frequency signal components of the three primary color video signals are mixed at a predetermined ratio, only the noise component is canceled out, and a high-luminance signal component with reduced noise components is created.
You. Accordingly, the signal-to-noise ratio of the high luminance signal component is improved.

The high-luminance signal component with reduced noise component is:
Absolute value conversion by the absolute value conversion means
In order to achieve the matching of
After being amplified, it is clipped to a predetermined value by clipping means.
It is. As a result , a noise discrimination signal that clearly discriminates a noise component from a signal component is obtained. At this time, the predetermined amplification degree
Is the noise component contained in the absolute value
Set the minutes to be less than 100%. And amplification
So that the maximum value of the activated high-frequency signal component becomes 100%
Then, the high luminance signal component is clipped. Then the generated
The noise discrimination signal obtained is 100% if it is a signal component.
If it is a noise component, it will be less than 100%. Then, the noise distinction signal, the multiplying means multiplies each high-frequency signal components of the three primary colors, a high-frequency signal components of three primary colors noise distinction signal is multiplied, the low every three primary colors by adding means If added to the gamut signal component, it is possible to obtain a three-primary-color video signal in which only the noise component is compressed without changing the signal component .

Each of the video signals separated into the three primary colors is different.
If the video signal is amplified at a certain ratio, the ratio of each video signal
Mixes each high-frequency signal component of the three primary color video signals with the reciprocal of
This maximizes the signal-to-noise ratio and improves the noise compression efficiency.

[0016]

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below with reference to FIGS. Note that the same elements as those in FIG. 11 are denoted by RGB after the same numbers, and description thereof is omitted. In FIG. 1 showing this embodiment, for example, R (RED), G (GR
EEN) and B (BLUE), the video signals obtained by the solid-state image pickup device and the image pickup tube are supplied to terminals 1R, 1G, and 1 respectively.
B is input.

The video signal processing circuit for the R signal has an input terminal 1
From R to the output terminal 6R, the LPF 2R, the subtracter 3R, the multiplier 4R, and the adder 5R are sequentially connected, and the same applies to the video signal processing circuits for the G and B signals. It has a circuit configuration. The high-luminance signal generation circuit 7 as high-luminance signal generation means mixes the high-frequency signal components of the R, G, and B signals extracted by the subtracters 3R, 3G, 3B at a predetermined ratio, respectively. This is a circuit for obtaining a high luminance signal component.

The absolute value amplifier 8 is an absolute value conversion means and amplifying means, absolute values of the luminance high-band signal component obtained by the luminance high band signal generation circuit 7, is a circuit for amplifying with a predetermined amplification degree . A clip circuit 10 serving as a clip unit is a circuit that clips the signal amplified by the absolute value amplifier 8 at a predetermined amplitude level . This clip level is a level corresponding to the coring level in the noise coring method. By doing so, a noise discrimination signal that discriminates between a noise component and a signal component included in the three primary color video signals is generated. This noise discrimination signal is supplied to multipliers 4R and 4R, respectively.
G, 4B. The terminal 9 is a terminal for inputting a clip level. Here, in the portion corresponding to the clip level, the multipliers 4R, 4G, and 4B multiply by a fixed amount (for example, 1). When the clip level is lower than the clip level, the value is determined by comparison with the clip level (for example, 0 or more and 1 or less). Is multiplied. The absolute value amplifier 8 and the clip circuit 10 correspond to a noise discriminating signal generating means.

Next, the operation will be described. Input terminal 1R, 1G
And 1B are input with RGB video signals as shown in FIG. The LPFs 2R, 2G, and 2B extract low-frequency signal components (hereinafter, referred to as low-frequency signals) RL, GL, and BL from the input signal as shown in FIG.
In G and 3B, the low frequency signals RL, GL, and BL are subtracted from the original RGB input signal, respectively, to obtain the signals shown in FIGS.
As shown in (B) and (C), high frequency signal components (hereinafter, referred to as high frequency signals) RH, GH, and BH in the three primary color video signals.
Get.

The luminance high-frequency signal generation circuit 7 detects the presence or absence of a correlation between the respective signals so that the high-frequency signal R
H, GH, and BH are mixed at a predetermined mixing ratio (substantially the same ratio), and a high-luminance signal component (hereinafter, referred to as FIG. 5)
YH is created. High frequency signal R
The reason why H, GH and BH are mixed is to enlarge the difference between the uncorrelated noise component contained in the three primary color video signals RH, GH and BH in the television camera and the correlated, for example, contour component. The amplitude of the noisy and uncorrelated noise components is compressed and the correlated contour components are added.

For example, the signal-to-noise ratio (S / N ratio) of a luminance signal produced by mixing three primary color (RGB) video signals will be described in the case of the NTSC system. expressed. Y = 0.3R + 0.59G + 0.11
B. . . . . . . . . . . (1) Here, when the video signals RGB are mixed, the noise component becomes
Since there is no correlation between the three primary color (RGB) video signals, they are represented as a sum of power, and the contour signal components are represented as a sum of amplitudes since they have a correlation with each other. Therefore, when the signal-to-noise ratio (S / N ratio) is represented by the mixture ratio of equation (1), the following equation (2) is obtained.

20 log ((0.3 ² +0.59 ² +0.11 ² ) ^1/2 /(0.3+0.59+0.11))≒−3.4 dB. . . (2) As described above, in the expression (2), the luminance signal Y has three primary colors (R
GB) Signal-to-noise ratio (S / N ratio) is lower than video signal
This shows that the signal is improved by 3.4 dB and the S / N ratio is improved.

In order to improve the S / N ratio, the mixing ratio may be set as in the following equation (3). Y = (R + G + B) /
3. . . . . . . . . . . . . . . . . . (3) By setting as described above, the S / N ratio becomes as shown in Expression (4). ^{20log ((1 2 +1 2 +1} 2) 1/2 /(1+1+1))≒-4.8dB. . . (4) Therefore, the noise component of the signal RGB is improved by about −4.8 dB in the luminance signal Y.

Although the noise level of the video signal of each of the three primary colors may be different before being input to the present apparatus,
This case will be described later. As shown in Expressions (2) and (4), when the luminance signal Y is created by mixing the three primary color (RGB) video signals, the S / N ratio is surely improved although the improvement ratio varies depending on the mixing ratio. , The amplitude of the noise component becomes smaller.

When the high-range signal of the luminance signal is mixed by the NTSC system, since the band exceeds the band of the NTSC system, no inconvenience occurs even if the mixing ratio of the NTSC system shown in the equation (1) is not used. . In this case, the mixing ratio of the luminance signal YH may be arbitrary, but in order to improve the S / N ratio most,
As shown in the following equation (5), the same ratio as in equation (3) may be used.

Y = (RH + GH + BH) /
3. . . . . . . . . . . . . . . (5) For example, as shown in FIGS. 4A, 4B, and 4C, assuming that the amplitude of noise in the high band signals RH, GH, and BH signals is 5%, When mixed at a ratio, the amplitude of the noise component in the high luminance signal YH is improved by about -4.8 dB, so that the amplitude of the noise component is given by the following equation (6).

5 × 10 ^{(−4.8 / 10)} = 1.6 (%). . . . . . . (6) The amplitude of the noise portion is about 1.6% from 5%, and a high luminance signal YH as shown in FIG. 5 is obtained. Next, the absolute value amplifier 8 generates an absolute value of the high luminance signal YH. The absolute value of the high luminance signal YH is determined by using the amplitude of the signal component as an optimum input to the multipliers 4R, 4G, and 4B. Amplification is performed at a predetermined amplification rate so as to have a predetermined level.

For example, as shown in FIG.
Is 5%, the amplitude of the signal component of the high-frequency luminance signal YH is 100% if the predetermined amplification factor is 20 times.
, And the noise amplitude becomes a signal level of about 32%. This signal is clipped by the clipping circuit 10 at a clip level of 100% input from the terminal 9, and a noise discrimination signal Z for discriminating between noise components and signal components as shown in FIG. 6 is generated.

The amplification factor of the absolute value amplifier 8 is determined by the relationship between the input levels of the multipliers 4R, 4G, and 4B and the input level required by the clipping circuit 10. Usually, the multipliers 4R, 4G, and 4B Is determined by setting the clip level to the maximum allowable level from the relationship of the dynamic range in the above, and setting the clip level to what percentage of the specified value of the video signal.

In the multipliers 4R, 4G, and 4B, the noise discrimination signal Z is converted into high-frequency signals RH, GH, and BH of each of the three primary colors.
Are multiplied, respectively, FIGS. 7 (A), (B) , as shown in (C), compared with the high frequency signal RH, GH and BH, it correlation
There signals are greatly compressed, the signal having a correlation is slightly compressed signal rh, gh, is bh obtained. These signals rh, gh, and bh are added to the three primary color video low-frequency signals RL, GL, and BL, and output from the output terminals 6R, 6G, and 6B as shown in FIG.
As shown in ( ₁₎ , signals R ₀ , G ₀ , and B ₀ whose contour signals are not compressed in spite of the compression of the noise component are output.

Specifically, using equations, the following equations (7) to
It becomes like (9). R ₀ = RH × Z + RL
. . . . . . . . . . . . . . . (7) G ₀ = GH × Z + GL
. . . . . . . . . . . . . . . (8) B ₀ = BH × Z + BL
. . . . . . . . . . . . . . . (9) Here, the features of the present invention will be described with reference to the conventional noise coring method and the input / output characteristics of the high frequency component of the present invention.

First, the noise level in the input signal is 5 at the maximum.
Assume that there is a% amplitude. In the case of the noise coring method (FIG. 11), the limiter level is set to 5%.
FIG. 9 shows the input and output of the high frequency component in the signal. Although the input and output high frequency components have positive and negative signs, FIG. 9 shows only the positive case. If the input is below the limiter level as in FIG. 9, the output will be zero. In this case, the high-frequency component is uniformly removed regardless of whether it is noise or a contour.

The case of the present invention will be described. For the sake of simplicity, it is assumed that the same signal is input as a RGB signal (a black-and-white subject is photographed), and that the noise level in the input signal has an amplitude of up to 5% each. . Also, the luminance high band signal generation circuit 7 mixes the signals as shown in Expression (3), and sets the amplification factor of the absolute value amplifier 8 to 20 times. In the case of the present invention, a high-frequency component in a signal can be considered separately from noise and contour.

First, the case where the high-frequency component represents an outline will be described. When the high-frequency component in the input signal represents an outline, the high-luminance signal generation circuit 7 has a correlation between RH, GH, and BH.
The level of H does not decrease. Accordingly, the noise discrimination signal Z is obtained by multiplying the contour signal by 20 times, and the high frequency components in the output signal are as shown in Table 1.

[0035]

[Table 1]

In FIG. 10 showing the input / output characteristics of the present embodiment, the characteristic A is a graph of the input / output characteristics of the high frequency component shown in Table 1. Next, a case where the high frequency component represents noise will be described. When expressing the high frequency component in the input video signal,
In the high luminance signal generation circuit 7, RH, GH, BH
, There is no correlation, the level of the luminance high-frequency signal YH becomes small, and the noise discrimination signal Z becomes 20 times that,
Table 2 shows the improvement rate of the high frequency component and the noise in the output signal.

[0037]

[Table 2]

The characteristic B in FIG. 10 is a graph of the input / output characteristic of the high frequency component in Table 2. Thus, according to the present invention, when the high frequency component in the input signal is a contour signal,
Although the contour signal is slightly deteriorated, it does not disappear at all like the noise coring method. If the high-frequency component in the input signal is noise, the compression is larger than the degradation amount of the contour signal, and the smaller the level of the high-frequency component, the greater the improvement. That is, the effect on the contour signal of the high frequency component in small, parts found to be clearly noise minute of the high frequency component can be increased the improvement, or noise of the high range component Since the improvement rate is reduced for a part whose contour is not clear, it is possible to perform more appropriate processing than the noise coring method.

According to this configuration, by utilizing the fact that there is no correlation between the noise components included in the high-frequency signal components of the three primary-color video signals, the high-frequency signal components of the three primary-color video signals are mixed, and the absolute value is calculated. Amplify and clip at a predetermined amplitude level to generate a noise discrimination signal, multiply each high-frequency signal component and add it to the low-frequency signal component to compress uncorrelated noise Therefore, the noise in the three primary color video signals can be reliably distinguished from the original video signal, and the noise can be compressed according to the presence or absence of the correlation.

Since the clip level (corresponding to the coring level of the coring method) has relatively linear input / output characteristics for signal components having no correlation, the clip level in the coring method is low. You don't have to worry about determining the level as you do when setting the ring level. Therefore, there is an advantage that level adjustment using a variable resistor is unnecessary in an actual circuit design.

In particular, even when the video signal is hidden by noise, the noise and the video signal are reliably distinguished from each other,
It is possible to obtain an image in which the outline is not deteriorated, the outline is not blurred, and the noise is small. In the present embodiment, LP is used to divide the high frequency signal and the low frequency signal of the video signal.
F and the subtractor were used.
Even if a (high-pass filter) is used, a high-frequency signal can be easily extracted.

Next, a case where the noise levels of the video signals of the three primary colors input to the present apparatus are different will be described. In this embodiment, the case has been described where the noise components of the video signals of the three primary colors input to the present device are at the same level. However, in order to adjust the white balance, the video signals of the three primary colors are output before the present device. In some cases, they have been amplified at different amplification rates. In this case, even if the video signal of each of the three primary colors has the same noise level before amplification for white balance adjustment, the noise level will be different when amplified and input to the present apparatus.

For example, assuming that the amplification factors are a: b: c, the RGB signals input to the apparatus are as follows. R = rl × a + (rh + Nr) ×
a. . . . . . . . . . . (10) G = gl × b + (gh + Ng) ×
b. . . . . . . . . . . (10) B = bl × c + (bh + Nb) ×
c. . . . . . . . . . . (10) rl, gl, bl: low-frequency signals in each signal before amplification rh, gh, bh: high-frequency signals in each signal before amplification (outline) Nr, Ng, Nb: in each signal before amplification RL = rl × a RH = (rh + Nr) × a GL = gl × b GH = (gh + Ng) × b BL = bl × c BH = (bh + Nb) × c Assuming that the mixing ratio when mixing the high luminance signal YH is d: e: f, YH = ((rh + Nr) × a × d + (gh + Ng) × b × e + (bh + Nb) × c × f) / (D + e + f).

Here, the noise YH in the high luminance signal YH
N is a value represented by the following equation. YHN = (Nr × a × d + Ng × b × e + Nb × c × f) / (d + e + f) Here, to make the S / N ratio the best, the final noise level should be the same ratio. A × d: b × e: c × f = 1: 1: 1. That is, d = 1 / a e = 1 / b f = 1 / c.

Therefore, when the noise level of the video signal of each of the three primary colors is different before this apparatus, the amplification ratios of the video signals of each of the three primary colors have substantially the same ratio.
The amplification factor may be adjusted in the high luminance signal generation circuit 7. By doing so, the high frequency signals RH, GH, and BH are mixed at substantially the same ratio, and the efficiency of noise compression can be improved.

[0046]

As described above, according to the present invention, the high frequency signal components of the separated three primary color video signals are mixed, amplified by taking an absolute value, and clipped at a predetermined value to form a noise discrimination signal. Then, the noise component can be compressed by multiplying each high-frequency signal component by this noise discrimination signal and adding the resulting signal to the low-frequency signal component, thereby making it possible to compress the noise component in the three primary color video signal and the original video signal. The distinction can be made reliably.

In particular, when the video signal is hidden by noise components, the outline of the video signal is blurred by the conventional method. However, according to the present invention, the signal-to-noise ratio is improved when a high luminance signal is generated. Since the generated noise discrimination signal is generated, the discrimination between the noise component and the video signal is reliably performed, and a video with no blurred outline can be obtained. Also,
Each video signal separated into three primary colors before this device has a different ratio
When amplified by the ratio, the reciprocal of the ratio of each video signal
To mix each high-frequency signal component of the three primary color video signals.
Thus, the efficiency of noise compression can be improved.

[Brief description of the drawings]

FIG. 1 is a block circuit diagram showing one embodiment of the present invention.

FIG. 2 is a waveform diagram of a three primary color video signal input to the circuit of FIG. 1;

FIG. 3 is a waveform diagram of an output signal of each LPF of FIG. 1;

FIG. 4 is a waveform diagram of an output signal of each subtractor in FIG. 1;

FIG. 5 is a waveform diagram of an output signal of the high luminance signal generating circuit of FIG. 1;

FIG. 6 is a waveform chart of the noise discrimination signal of FIG. 1;

FIG. 7 is a waveform diagram of an output signal of each multiplier of FIG. 1;

FIG. 8 is a waveform diagram of an output signal of the device of FIG. 1;

FIG. 9 is an explanatory diagram of conventional input / output characteristics.

FIG. 10 is an explanatory diagram of input / output characteristics of the present embodiment.

FIG. 11 shows a conventional block circuit.

12 is a waveform chart of a signal in each circuit of FIG . 11 ;

[Explanation of symbols]

 7 Brightness high frequency signal generation circuit 8 Absolute value amplifier 9 Clip circuit

Claims (2)

(57) [Claims] To 1. A each video signal separated into the three primary colors, and the higher-band signal component and the low frequency signal filtering means and components by dividing a predetermined frequency the high-frequency signal component and a low signal component respectively extracted A high luminance signal generating means for mixing the extracted high frequency signal components of the three primary color video signals at a predetermined ratio to generate a high luminance signal component, and a noise component based on the high luminance signal component; Signal generating means for generating a noise discriminating signal for discriminating the signal and the signal component; multiplying means for multiplying the noise discriminating signal for each high-frequency signal component of each of the three primary colors; An adding means for adding the multiplied high-frequency signal components of the three primary colors to the low-frequency signal components for each of the three primary colors , wherein the noise discriminating signal generating means comprises: Absolute value means for converting the luminance high-frequency signal component into an absolute value; Amplifies the absolute value of the luminance high-frequency signal component with a predetermined amplification factor.
Amplification means for clipping the amplified high luminance signal component at a predetermined value.
Noise compression apparatus characterized by being configured to include a lip means. 2. Each of the video signals separated into the three primary colors is different.
In the case where each of the video signals is amplified at a ratio of
The reciprocal is used to mix each high frequency signal component of the three primary color video signals.
The noise compression device according to claim 1, wherein the noise compression device has a configuration.