JP3321828B2 - Noise reducer - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to, for example, MUSE (Mul
BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a video signal processing device applied to a tiple sub-nyquist-sampling encoding ("tiple Sub-Nyquist-Sampling Encoding") system, and more particularly to a noise reducer that improves S / N by utilizing autocorrelation between frames of a video signal.

[0002]

2. Description of the Related Art A television image signal is a signal in which image information is repeated at a frame period, and the autocorrelation between the frames is extremely strong. On the other hand, since the noise (noise) component included in the video signal has almost no autocorrelation as described above, when the video signal is temporally averaged for each frame period, the energy of the video signal component hardly changes. Only the energy of the noise component decreases.

[0003] The noise reducer is configured to improve the S / N of a video signal based on the above principle.
Basically, it is a recursive filter having a delay unit for one frame of a television video signal, and is configured as a circuit for averaging the input video signal temporally for each frame period. In addition,
Although the above method is effective for still images, it has drawbacks such as an afterimage effect in moving images,
The noise reducer generally employs a method of detecting a moving image portion from an input video signal and changing the S / N improvement amount.

FIG. 7 is a diagram showing an example of a circuit configuration of a noise reducer adopting a method of changing the S / N improvement amount. In FIG. 7, 1 is an adder, 2 is a one-frame delay memory, 3 is a subtractor, 4 is a multiplier, and 5 is a comparator. The noise reducer having such a configuration obtains the difference (A−B) between the input video signal A and the one-frame delay signal B delayed by one frame in the one-frame delay memory 2 by the subtractor 3, and calculates the difference between the frames. The signal (AB) is multiplied by the multiplier 4 with a filter coefficient K for determining the S / N improvement amount, and a difference signal {(AB) K} multiplied by the filter coefficient K is input to the input signal. A is added and output to the decoding processing system, and one frame delay memory 2
Write to.

[0005] The above-mentioned inter-frame difference signal (AB) includes:
A noise component having no inter-frame autocorrelation and a changed difference between adjacent frames of the video signal component are obtained. The noise component is generally obtained in a still image portion, and its amplitude is small. On the other hand, the difference that has changed between adjacent frames is obtained in the moving image portion, and its amplitude is large. The comparator 5 compares the magnitude of the amplitude of the inter-frame difference signal (A-B) with a predetermined threshold value V _TH , and when the threshold value V _TH is larger, the multiplier 4 , The time averaging of the video signal is sufficiently performed,
It operates to increase the S / N improvement. On the other hand, when the amplitude of the inter-frame difference signal ( _AB ) is larger than the threshold value V _TH , it is determined that the video signal has changed, and
Of the filter coefficient K is made small, for example, to zero, and the amount of noise reduction is reduced so that the change of the input video signal can be obtained as it is at the output. In this way,
The afterimage effect in the moving image part is reduced.

[0006]

However, in the above-described noise reducer, the inter-frame difference signal (AB) is used.
Since the still image and the moving image are determined according to the magnitude of the amplitude of the predetermined threshold _VTH ,
There is a disadvantage that it is extremely difficult to separate small-amplitude motion from noise, and it is difficult to increase the amount of noise reduction.

Further, when the screen is frozen, noise generated at random in time stops, and there is a problem that this noise becomes conspicuous.

The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a noise reducer which can improve the S / N without increasing the afterimage even if the amount of noise reduction is increased. .

[0009]

SUMMARY OF THE INVENTION In order to achieve the above object, the present invention provides a method for improving the S / N of a video signal.
The motion of the input video signal,
Outputs a still image signal when the image is still or semi-still
Motion detection means, and the larger the value, the greater the S / N improvement
Two coefficients, a first coefficient and a value from the first coefficient
Switching of the output of the second coefficient with small
A coefficient switching circuit that responds to the
When a still image signal is received by the motion detecting means,
Outputting a first coefficient, the freeze signal and the still image signal
The second factor if none of the items has been received
The control signal for instructing to output
Control means for outputting a predetermined frame from the input video signal.
The difference signal obtained by subtracting the frame delay signal delayed by
Subtraction means for obtaining the difference signal from the subtraction means.
The value of the first coefficient or the second coefficient by the coefficient switching circuit
Averaging means for performing temporal averaging according to
Adding means for adding the difference signal and the input video signal by the converting means
After delaying the input video signal by a predetermined frame,
A frame delay signal is generated and output to the subtraction means.
And a frame delay memory.

[0010]

According to the present invention, the apparatus further comprises a frame memory, wherein the control means, when outputting a control signal instructing to output the first coefficient, outputs a write instruction to the frame memory at a predetermined timing. And the result of decoding the video signal output from the adding means is written.

In the present invention, the input video signal is offset by shifting the sampling point between frames, and the frame delay memory generates a two-frame delay signal obtained by delaying the input video signal by two frames. And outputs the result to the subtraction means. In the present invention, the input video signal includes motion information, and the motion detection means outputs a first coefficient when the motion information of the input video signal indicates a still image or a quasi-still image, and outputs the first coefficient. Indicates a moving image, a control signal for instructing to output the second coefficient is output to the coefficient switching circuit. Preferably, the motion detecting means detects a change in the video signal of the entire screen, and outputs the control signal to the coefficient switching circuit according to the detection result. Preferably, the motion detection means detects a change in a video signal on a screen by comparing sample points set in advance over the entire frame between adjacent frames.

[0013]

According to the present invention, the filter coefficient is switched only when the input video signal is substantially stationary.

According to the present invention, only when the motion information of the transmitted signal indicates, for example, completely stationary or quasi-stationary,
The filter coefficients are switched.

According to the present invention, the motion of the entire screen is detected from the input video signal, and as a result of this detection, the filter coefficient is switched only when the still image is detected.

According to the present invention, sample points corresponding to the same position in adjacent frames are compared with each other. If the difference is smaller than a predetermined value, it is determined that the sample is stationary, and only at that time, the filter coefficient is switched. .

[0017]

Embodiment 1 FIG. 1 is a block diagram showing a first embodiment of a noise reducer according to the present invention employed in a receiving apparatus of the MUSE system. The same components as those in FIG. 7 showing a conventional example are the same. It is represented by a sign. That is, 1 is an adder,
2a is a 2-frame delay memory, 3 is a subtractor, 4 is averaging
Multiplier as means , 11 is analog / digital (A
/ D) converter, 12 is a de-emphasis circuit, 13 is a reverse transmission gamma (γ ^-1 ) correction circuit, 14 is a control signal separation circuit, 15 is an OR circuit, and 16 is a coefficient switching circuit. The control signal separation circuit 14
And an OR circuit 15 constitute a motion detecting means.

In the MUSE system, the motion of the entire screen is detected by a transmitting encoder (not shown), and this motion information is transmitted as a control signal. The encoder performs so-called gamma correction and emphasis on a high-frequency portion of a transmission signal, converts a digital signal into an analog signal, and transmits the analog signal to a receiving side. Figure 2 shows the MUS
3 shows a configuration example of a transmission signal for one frame in the E system. As shown in FIG. 2, the transmission signal has the above-described transmission control signal in units of one field, and the control by the transmission control signal is effective for the next field in which the transmission control signal exists. I have.

FIG. 3 shows the contents of the transmission control signal. Transmission control signal, as shown in FIG. 3, control content for each bit of the 31 bits of the bit b ₁ ~b ₃₂ and is defined, expressed by 3-bit bit b _16, b _17, b ₁₈ " The above-described motion information is assigned to the values “0 to 7”. Specifically, “0” indicates “normal (movement)”, “1” indicates “complete still image”, “2” indicates “quasi-still image”, and “3” indicates “scene”. change",
In the case of "4 to 7", it is configured to "indicate the degree of movement". Further, for example, AM / FM mode signal bit b ₂₀ is that there emphasis when "0 (FM)", "1
(AM) "indicates no emphasis.

Next, the function of each section will be described. The adder 1 adds the output signal of the multiplier 4 to the output signal A of the γ ^-1 correction circuit 13 and outputs the added signal to a decoding processing system (not shown).
Write to.

The two-frame delay memory 2a delays the video signal by two frames in order to temporally average the video signal for each frame period. The two-frame delay is different from the one-frame delay in FIG. 7 for the following reason. That is, M
On the receiving side of the USE system, inter-field / inter-frame interpolation (corresponding to a still area) and field interpolation (corresponding to a motion area) are selectively used according to the motion in an image.
It is necessary to detect a motion area. The motion region is obtained by detecting a high-frequency component (corresponding to a strong motion) when viewed in time frequency. In brief, as shown in FIG. 7, a difference signal between data of one frame before and data of the current frame may be obtained. However, in the case of the MUSE method,
Since offset sampling is performed between frames and the sampling point is shifted between frames to give an offset, even if one frame difference is simply taken, there is no partner to take the difference. This means that accurate movement cannot be detected. Therefore, in the MUSE method, since the same point is sampled two frames before, the two frames are delayed by two frames.
It is configured so that a difference from the previous frame can be obtained.

The subtracter 3 obtains the difference (AC) between the output signal A of the γ ^-1 correction circuit 13 and the two-frame delay signal C delayed by two frames in the two-frame delay memory 2a, and outputs the difference to the multiplier 4. Output.

The multiplier 4 multiplies the inter-frame difference signal (AC) obtained by the subtractor 3 by a filter coefficient K for determining an S / N improvement amount, and multiplies the filter coefficient K by the filter coefficient K. The signal {(AC) K} is output to the adder 1. The filter coefficient K of the multiplier 4 is switched by the output of the coefficient switching circuit 16, and when the transmitted video signal is an all still image or a quasi-still image, the filter coefficient K1 is set to K1 = 0.9. In the case of a moving image, the filter coefficient K2 =
Set to 0.8. That is, only when the transmitted video signal is an all still image or a quasi-still image, the coefficient value is increased to increase the amount of noise reduction, and for other moving images, the coefficient value is decreased to decrease the amount of noise reduction. ing.

The A / D converter 11 converts the transmission video signal IA transmitted from the transmission side as an analog signal into a digital signal and outputs the digital signal to the de-emphasis circuit 12 and the control signal separation circuit 14.

The de-emphasis circuit 12 de-emphasizes the video signal emphasized on the transmission side,
^-1 is output to the correction circuit 13.

The γ ^-1 correction circuit 13 reversely corrects the de-emphasized video signal because the video signal is still subjected to the γ correction on the transmission side, and the adder 1 and the Output to the subtractor 3.

The control signal separating circuit 14 separates the control signal as shown in FIG. 3 from the transmission signal converted into the digital signal,
By analyzing the ₁₆ ~b _18, when only the bit b ₁₈ is high level "1" or bit b ₁₇ only a high level "1"
, "Complete still image" or "quasi-still image" respectively
And outputs a high-level complete still signal S1 or quasi-stationary signal S2 to the OR circuit 15.

The coefficient switching circuit 16 has an output terminal t ₁ connected to the control terminal of the multiplier 4 and a terminal t ₂ connected to the filter coefficient K.
1 side, the terminal t ₃ are connected to the K2-side filter coefficient. When the coefficient switching circuit 16 receives the output signal (control signal) of the OR circuit 15 at a high level, the output terminal t
_{When 1} is connected to the terminal t ₂ and the output signal of the OR circuit 15 is received at a low level, the output terminal t ₁ is connected to the terminal t ₃ .

Next, the operation of the above configuration will be described. The motion of the entire screen is detected by a transmitting encoder (not shown), and the motion information and various control information are recorded in each corresponding bit of the transmission control signal.
The video signal to be transmitted is subjected to processes such as gamma correction and emphasis for the high-frequency portion of the transmission signal, converted from a digital signal to an analog signal, and transmitted to the receiving side.

The video signal IA received on the receiving side is first converted from an analog signal into a digital signal by an A / D converter 11 and output to a de-emphasis circuit 12 and a control signal separation circuit 14. In the de-emphasis circuit 12, the video signal emphasized on the transmission side is de-emphasized, and the γ- ¹ correction circuit 13
Is output to In the γ ⁻¹ correction circuit 13, since the de-emphasized video signal is still subjected to γ correction, inverse γ correction is performed on the video signal, and the corrected signal is added to the adder 1 as the video signal A. And output to the subtractor 3.

In the subtracter 3, the difference (AC) between the video signal A and the two-frame delay signal C delayed by two frames in the two-frame delay memory 2 a is obtained and output to the multiplier 4. In the multiplier 4, this inter-frame difference signal (A-C)
Filter coefficient K for determining the S / N improvement amount for
And a difference signal {(AC) K} multiplied by the filter coefficient K is output to the adder 1. Thereby, in the adder 1, the input signal A is added to the difference signal {(AC) K}, and the added signal [A + {(AC)] is added.
K}] is output to the decode processing system and written to the two-frame delay memory 2a.

The above-described inter-frame difference signal (AC) includes:
A noise component having no inter-frame autocorrelation and a changed difference between adjacent frames of the video signal component are obtained. Specifically, a noise component is generally obtained in a still image portion, and a difference changed between frames is obtained in a moving image portion. Therefore, when the transmission video signal IA is completely stationary or semi-stationary, the value of the filter coefficient K of the multiplier 4 is set to a large value K1 = 0.9 in order to increase the amount of noise reduction. On the other hand, in the case of a moving image, the value of the filter coefficient K of the multiplier 4 is set to a small value K2 = 0.8 in order to reduce the noise reduction amount. The switching of the filter coefficient K is performed according to the analysis result of the control signal separation circuit 14.

That is, the control signal separation circuit 14
In, the transmission control signal is separated from the input digital video signal, and the bit b ₁₆
Analysis of ~b ₁₈ is performed. Here, when only the bit b ₁₈ only a high level "1" or bit b ₁₇ of the control signal is determined to be high level "1", transmitted video signals are "completely still image" or "quasi As a still image, a high-level complete still signal S1 or a quasi-still signal S2 is output to the OR circuit 15. As a result, the output signal of the OR circuit 15 is at a high level and the coefficient switching circuit 16
Is output to On the other hand, as a result of the above analysis, if it is determined that the transmitted video signal is not a “complete still image” or a “quasi-still image” but a moving image, the complete still signal S1 and the quasi-static signal S2 become low level. Is output to the OR circuit 15. In this case, the output signal of the OR circuit 15 is output to the coefficient switching circuit 16 at a low level.

In the coefficient switching circuit 16, when the output signal of the OR circuit 15 is received at a high level, the output terminal t ₁ is switched to the terminal t.
Connected to ₂ . Thus, the filter coefficient K of the multiplier 4 is set to K1 = 0.9. On the other hand, when the output signal of the OR circuit 15 is received at a low level, the output terminal t ₁ is connected to the terminal t ₃ . Thereby, the filter coefficient K of the multiplier 4 is set to K2 = 0.8.

As described above, according to this embodiment,
In the broadcast system in which the transmission side detects motion of the entire screen and transmits the motion information as a transmission control signal to the reception side, the control signal separation circuit 14 on the reception side separates and analyzes the control signal from the transmission video signal and transmits When the motion information is a moving image, the coefficient value K is reduced to reduce the amount of noise reduction to reduce the afterimage, and only when the transmission motion information is a complete still image and a quasi-still image in which the motion of the entire screen is small, the multiplier 4 Since the noise reduction amount is increased by increasing the coefficient value K, the S / N can be improved without the side effect of generating an afterimage.

Table 1 below shows the S / N improvement of the noise reducer and the time constant T of the afterimage obtained by the above-described configuration, obtained for each value of the filter coefficient K based on the following equations (1) and (2). Is shown. S / N improvement = 10 log (1 + K) / (1-k) (1) T = −1 / (lnK) × 1/15 (2) In the expression (2), the present embodiment is 2 due to the frame delay arrangement (1/15) was doubled, in the case of one-frame delay configuration is (1/30) times.

[0037]

[Table 1]

As can be seen from Table 1, when the value of the filter coefficient K is set to the large value 0.9, the S / N improvement is improved by about 3 dB as compared with the case of the small value 0.8. At this time, the time constant T is increased by about 0.3 seconds,
Although the afterimage becomes longer, the filter coefficient K becomes 0.9.
In the case of (1), as described above, since it is only at the time of a complete still image or a quasi-still image, there is almost no occurrence of an afterimage, and there is no problem.

[0039]

Embodiment 2 FIG. 4 is a block diagram showing a second embodiment of the noise reducer according to the present invention. This embodiment is different from the above-described first embodiment in that the value of the filter coefficient K of the multiplier 4 is increased when the filter 4 is completely stationary or quasi-stationary to increase the noise reduction amount. On the other hand, in the present embodiment, when completely stationary or semi-stationary,
Further, the present invention is configured to increase the value of the filter coefficient K of the multiplier 4 to increase the noise reduction amount only when the user performs freeze or hard copy.

The reason for this configuration is that when performing freeze or hard copy, noise is particularly noticeable.
/ N improvement is desirable, but if the difference between the filter coefficient K1 at the time of complete stillness or quasi-staticity and the filter coefficient K2 at the time of a normal moving image is large, a change in image quality of the screen becomes conspicuous, Since the value of the filter coefficient K1 cannot be increased so much, the operation of increasing the value of the filter coefficient K of the multiplier 4 to increase the amount of noise reduction is described below.
This is because the change in image quality does not become unnatural even if the S / N improvement is increased only when freeze or hard copy is performed.

In FIG. 4, 17 is a freeze switch,
Freeze timing generating circuit 18 as a control means,
Reference numeral 19 denotes a frame memory, and reference numeral 20 denotes a graphic printer. In such a configuration, FIG.
As shown in FIG. 7, immediately after the freeze switch 17 is pressed, the control signal separating circuit 14 analyzes the motion information, and as a result, determines that the transmitted video signal is a complete still image or a quasi-still image. When the signal S1 or the quasi-still image signal S2 is output to the OR circuit 15 at a high level and the output signal of the OR circuit 15 at a high level is input to the freeze timing generation circuit 18, the freeze timing generation circuit 18 outputs the coefficient switching circuit 16 output terminals t ₁ and with a control signal CTL for supporting the terminal t ₂ as to connect the output, the write signal W for writing instruction output of the decoder (not shown) to the frame memory 19 with respect to
R is output. Thus, the video signal from which much noise has been removed is decoded, and its output is written to the frame memory 19. As a result, a good freeze output of S / N is obtained.
A good hard copy of / N can be obtained.

According to the present embodiment, even if the S / N improvement is increased, it is possible to obtain a freeze output or a hard copy in which a change in image quality is natural.

[0043]

Third Embodiment FIG. 6 is a block diagram showing a third embodiment of the noise reducer according to the present invention. This embodiment is not the case where the transmission side detects the motion of the entire screen and transmits this motion information as a control signal as in the MUSE method employed in the first and second embodiments. An example in which the present invention is applied when transmission is not performed is shown.

In FIG. 6, reference numeral 21 denotes a frame switching circuit;
Reference numeral 22 denotes a first sample point memory, 23 denotes a second sample point memory, and 24 denotes an image determination circuit.
In such a configuration, in the frame switching circuit 21,
Under the control of a control system (not shown), the output destination of the output video signal A of the γ ^-1 correction circuit 13 is switched between the first sample point memory 22 and the second sample point memory 23 for each frame. First and second sample point memories 22, 23
, For example, 20 × 2 of one screen of the output video signal A of the γ ^-1 correction circuit 13 input through the frame switching circuit 21
Zero sample points are recorded. The sample point memories 22 and 23 record information on sample points selected over the entire screen of an adjacent frame. The image determination circuit 24 compares corresponding sample point information over the entire screen between adjacent frames stored in the first and second memories 22 and 23, and determines the maximum value of the difference between the same sample points. When the difference is smaller than the difference generated by the preset noise, the current input video signal is determined to be a still image, and only then is the still signal STL output to the coefficient switching circuit 16. As a result, the output terminal t ₁ becomes the terminal t
₂ , the value of the filter coefficient K of the multiplier 4 is set to a large value, and the noise reduction amount is increased.

In this embodiment, as in the first embodiment, there is an advantage that the S / N can be improved without the side effect of the occurrence of an afterimage. In the present embodiment, the case where the sample points stored in the first and second sample point memories 22 and 23 are 20 × 20 sample points over the entire screen has been described as an example, but the present invention is not limited to this. It goes without saying that it is not a thing.

[0046]

As described above, according to the noise reducer of the present invention, the S / N can be improved without the side effect of generating an afterimage. Further, even if the S / N improvement is increased, a freeze output or a hard copy in which a change in image quality is natural can be obtained.

[Brief description of the drawings]

FIG. 1 is a configuration diagram showing a first embodiment of a noise reducer according to the present invention.

FIG. 2 is a diagram illustrating a configuration example of a transmission signal of one frame unit in the MUSE system.

FIG. 3 is a diagram for explaining the contents of a transmission control signal.

FIG. 4 is a configuration diagram showing a second embodiment of the noise reducer according to the present invention.

FIG. 5 is an explanatory diagram of the operation of FIG. 4;

FIG. 6 is a configuration diagram showing a third embodiment of the noise reducer according to the present invention.

FIG. 7 is a diagram illustrating a circuit configuration example of a conventional noise reducer.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Adder 2 ... 1 frame delay memory 2a ... 2 frame delay memory 3 ... Subtractor 4 ... Multiplier 11 ... Analog / Digital (A / D) converter 12 ... Deemphasis circuit 13 ... Reverse transmission gamma ((gamma) ^-1) ) Correction circuit 14 Control signal separation circuit 15 OR circuit 16 Coefficient switching circuit 17 Freeze switch 18 Freeze timing generation circuit 19 Frame memory 20 Graphic printer 21 Frame switching circuit 22 First sample point memory 23 ... Second sample point memory 24 ... Image determination circuit

Continuation of the front page (56) References JP-A-5-145800 (JP, A) JP-A-4-74080 (JP, A) JP-A-62-137977 (JP, A) JP-A-4-6961 (JP) JP-A-49-5517 (JP, A) JP-A-1-233825 (JP, A) JP-A-62-69779 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB (Name) H04N 5/21

Claims (6)

(57) [Claims] 1. A noise reducer for improving the S / N of a video signal, wherein the motion detector detects a motion of an input video signal and outputs a still image signal when the image signal is a still image or a quasi-still image. And coefficient switching for performing output switching of a first coefficient, which is two coefficients having a larger S / N improvement degree as the value is larger, and a second coefficient having a smaller value than the first coefficient, in accordance with an instruction of a control signal. A first coefficient is output when a freeze signal is received and a still image signal is received by the motion detecting means, and when any one of the freeze signal and the still image signal is not received, Control means for outputting a control signal for instructing to output a second coefficient to the coefficient switching circuit; subtraction means for obtaining a difference signal obtained by subtracting a frame delay signal delayed by a predetermined frame from the input video signal; Averaging means for temporally averaging the difference signal by the subtraction means according to the value of the first coefficient or the second coefficient by the coefficient switching circuit; A noise reducer comprising: an adding unit that adds a signal; and a frame delay memory that generates the frame delay signal obtained by delaying an input video signal by a predetermined frame and outputs the frame delay signal to the subtraction unit. 2. The apparatus further comprises a frame memory, wherein the control means, when outputting a control signal instructing to output the first coefficient, issues a write instruction to the frame memory at a predetermined timing. 2. The noise reducer according to claim 1, wherein the decoding result of the video signal output from said adding means is written. 3. The input video signal is provided with an offset by shifting a sampling point between frames, and the frame delay memory generates a two-frame delay signal by delaying the input video signal by two frames. The noise reducer according to claim 1, wherein the noise reducer outputs the signal to the subtraction means. 4. The input video signal includes motion information, and the motion detection means outputs a first coefficient when the motion information of the input video signal indicates a still image or a quasi-still image, and outputs the first coefficient. 4. The noise reducer according to claim 1, wherein when the signal indicates a moving image, a control signal for instructing to output the second coefficient is output to the coefficient switching circuit. 5. The noise reducer according to claim 1, wherein said motion detecting means detects a change in a video signal of an entire screen and outputs said control signal to said coefficient switching circuit according to a result of the detection. 6. The noise reducer according to claim 5, wherein said motion detection means detects a change in a video signal on a screen by comparing sample points set in advance over the entire frame between adjacent frames.