JPH044677A - noise reduction device - Google Patents

Abstract

PURPOSE:To realize a noise reduction device with less deterioration in the afterimage characteristic and large S/N improvement effect by selecting an optimum characteristic of nonlinear processing type implemented to each signal component whose characteristic is extracted. CONSTITUTION:An output level proportional to an input level with a proportion constant i(i>0) is outputted from the title device up to a prescribed input level, almost a prescribed value A or an output proportional to an input signal with a proportion constant c(i>¦c¦>0) is outputted with respect to an input level being a reference level (a) or below, and almost a prescribed value B or an output proportional to an input signal with a proportion constant d(i>¦d¦>0) is outputted with respect to an input level being a reference level (b) or over. Thus, fog of edges due to after-image is eliminated and no noise flicker is caused at the edge.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention uses field correlation or frame correlation of video signals from television cameras, videos, video cameras, etc. to degrade the afterimage characteristics even in moving images. It relates to a device that reduces noise without any noise, and more specifically, it provides the optimal value for the nonlinear processing method used in this system to achieve the best noise reduction for the noise that appears at the edges of the input image when it moves. The present invention relates to a noise reduction device capable of reducing noise.

BACKGROUND OF THE INVENTION Conventional noise reduction devices include, for example, Japanese Patent Application Laid-Open No. 1986-
The one shown in Japanese Patent No. 158574 is known.

FIG. 5 is a block diagram showing a conventional noise reduction device. In FIG. 5, 1 is an input terminal for a video signal, 2 is an A/D converter that converts the input video signal from an analog signal to a digital signal, and 3 is a noise from the input video signal of one input signal to the other input signal. A subtraction circuit 4 that subtracts a signal is a frame memory that delays the output signal of the subtraction circuit 3, which has reduced noise from the input video signal, by one frame. N.T.
The phase of the color signal of the SC color video signal is inverted every frame, but 5 is a color signal phase shift circuit to compensate for this, and the color signal of the video signal delayed in the frame memory 4 is phase shifted. Invert. 6 is a subtraction circuit that subtracts the input video signal and the video signal delayed by one frame to obtain a difference frame difference signal; 7 is a subtraction circuit that converts the frame difference signal output from the subtraction circuit 6 from a serial digital signal to a parallel digital signal. 8 is a Hadamard transformer that performs Hadamard transform, which is orthogonal transform, on the parallel digital signal output from the serial-parallel converter circuit 7, and extracts vertical components, horizontal components, and diagonal components from the frame difference signal. It is a conversion circuit. The output of the Hadamard transform circuit 8 has a different distribution of the noise component and the motion component, and 9 is a nonlinear processing circuit that extracts the noise component by utilizing this fact. 10 is a Hadamard inverse transform circuit, and since the noise signal extracted from the nonlinear processing circuit 9 is obtained by Hadamard transform, it is returned to the original time axis by Hadamard inverse transform. 11 is a parallel-to-serial conversion circuit that returns the parallel digital signal to the original serial digital signal, and 12 is a D/At converter that converts the digital signal output from the subtraction circuit 3 into an analog signal.

In the conventional noise reduction device configured in this way,
Let's explain its operation. When a video signal is input from the input terminal 1, it is converted into a digital signal by the A/D converter 2, and this digital signal passes through the subtraction circuit 3 to subtract non-correlated components, which will be described later, and ideally contains noise components. If there is no video signal component, it will be stored in the frame memory 4,
Delayed for one frame. Since the video signal delayed by one frame has a color signal phase inverted from that of the one frame signal, the phase is compensated by the color signal phase shift circuit 5, and only the phase of the color signal is inverted. After, subtraction circuit 6
A frame difference signal of the difference between two video signals having the same chroma phase can be obtained. Originally, when the input video signal is a still image, this frame difference signal becomes a noise component itself, and noise can be extracted without the need for the circuit described below.

However, if the input video signal is a moving image, this frame difference signal becomes a signal that is a combination of signal components with no frame correlation and noise components.

Next, a method for obtaining only the noise component from this frame difference signal will be described. This frame difference signal is converted from a serial digital signal to a parallel digital signal by a serial-to-parallel conversion circuit 7, and is divided into components that express characteristics well as signals such as a low frequency component, a vertical component, a horizontal component, etc. by a Hadamard transform circuit 8. It will be done. Now, it is assumed that the Hadamard transformation is a two-dimensional Hadamard transformation of order 4×2. In this case, one pattern of input picture elements becomes as shown in FIG. 6, and the input becomes xto Xll X12 XIN.

The conversion output of the 2×4 Hadamard transform is assumed to be F24.

F 24 = Ht·X4°H4 According to the above equation, the Hadamard transform output Fza is obtained from the 4×2-order input picture element h4. Hz and H4 are as follows.

+1-1 The output from the Hadamard transform circuit 8 is a converted output of eight components. On the other hand, since noise has no correlation, it is almost evenly distributed among the eight component wave numbers of the output of the Hadamard transform circuit 8. As is well known, the noise level at the output of the Hadamard switching circuit 8 corresponds to the noise level of the input signal, so it is possible to extract only small-level noise components from each of these components through the nonlinear processing circuit 9. Since each component extracted by this non-linear processing circuit 9 is obtained by Hadamard transform, it is returned to the original time axis by passing through the Hadamard inverse transform circuit 10, and is converted into a parallel digital noise signal. Then, the parallel-to-serial conversion circuit 11 converts the parallel digital noise signal into a serial digital signal similar to the input form.The signal obtained here is a signal obtained by extracting noise components from a frame difference signal that has no frame correlation. As mentioned above, the input video signal is supplied to the subtraction circuit 3, and a noise-free digital video signal is obtained by subtracting the noise component from the input video signal.Finally, the A/D converter 12 converts the digital video signal into the original signal. is converted into an analog signal and output.

In principle, a noise reduction device using such a method can achieve B.
9! It is possible to perform noise reduction not only for still images in image signal input but also for moving images without significantly deteriorating the input video.

Problems to be Solved by the Invention The noise reduction device described in the conventional example can, in principle, reduce noise without producing an afterimage in video signal input, not only in still images but also in moving images.

However, depending on the nonlinear processing method, the visual S/N improvement effect and the afterimage characteristics of the edge portion vary greatly, and so far, the optimal characteristics of such nonlinear characteristics have not been proposed.

The present invention improves this point, and by providing the optimal characteristics of the nonlinear processing method of the noise reduction device described in the conventional example, the deterioration of the residual characteristic is small and the noise reduction effect is large, and the S/N improvement effect is large. The object of the present invention is to provide a reduction device.

Means for Solving the Problems In order to resolve the above problems, the noise reduction device of the present invention takes the difference component between the input video signal and the output signal of the n (n>0, where n is an integer) field delay means. a first subtraction means; a feature extraction means for extracting a feature component of the output signal of the first subtraction means; a nonlinear processing means for extracting a noise component from the output of the feature extraction means; and an output of the nonlinear processing means. second subtraction means for obtaining a difference signal from the input video signal, the n-field delay means delays the output of the second subtraction means by n fields, and the non-linear processing means is arranged so that the input signal is at a reference level. When a (a>O) or less and reference level b (b<O) or more, an output signal proportional to the input signal with a proportionality constant i (where i>0) is output, and the input signal or reference level 8 is output. It is configured to output a substantially constant value A when the input signal is above, and to output a substantially constant value B when the input signal is below a reference level b.

Further, the noise reduction device of the present invention is configured to decompose and extract the output of the VF feature extraction means or the first subtraction means into a plurality of feature components, and also includes: A nonlinear processing group having a plurality of nonlinear processing circuits for extracting noise components from each output is provided, and this nonlinear processing group uses an input signal of each nonlinear processing means or a reference level an
(an>0) and above the reference level bn (bn<0), the proportional constant in (however, in
>0), outputs a substantially constant value An when the input signal is above the reference level an, outputs a substantially constant value Bn when the input signal is below the reference level bn,
and reference levels an, bn of each of the nonlinear processing means.
, a proportionality constant in, and substantially constant values An and Bn are configured such that at least one of them is different for each of the plurality of nonlinear processing means.

Operation With the above configuration, the nonlinear processing method performs nonlinear processing on the output of the feature extraction circuit, or the input signal is at the reference level a (a>0
) and above the reference level b (b > 0), an output signal proportional to the proportionality constant i is output. However, when the reference level is 8 or more, an approximately constant value An is output, or below the reference level b. Since this is performed by a non-processing circuit having optimal non-linear characteristics that outputs a substantially constant Bn, it is possible to perform noise reduction with little afterimage deterioration and with a large S/N improvement effect.

Example An embodiment of the present invention will be described below based on the drawings.

FIG. 1 shows a block diagram of a noise reduction device in a first embodiment of the present invention. In FIG. 1, parts given the same reference numerals as in FIG. 5 are the same as in FIG. 5, and have the same functions. Reference numeral 21 denotes a nonlinear processing circuit that performs nonlinear processing on the output from the Hadamard transform in the form of optimal characteristics. has been used.

The operation of the noise reduction device of the first embodiment configured as described above will be described below. When a video signal is input from the hexagonal terminal 1, the A/D converter 2 outputs 4 fsc.
(a case of 3 fsc is also considered) is sampled, and the analog signal is converted into a digital signal. The subtraction circuit 3 subtracts the video signal of the previous frame from the input video signal, thereby subtracting the uncorrelated component, and ideally a video signal containing no noise component is obtained. This video signal component is stored in the frame memory 4 and delayed for one frame. Since the color signal of the NTSC color video signal has a phase inversion for each frame, this is compensated for by the color signal phase shift circuit 5, and only the color signal of the delayed video signal is phase-inverted by the frame memory 4.

A subtraction circuit 6 obtains a frame difference signal representing the difference between the two video signals. This frame difference signal is a signal that is a combination of a signal component without frame correlation (that is, a motion component) and a noise component. This frame difference signal is converted from a serial digital signal to a parallel digital signal by a serial-parallel conversion circuit 7, and a Hadamard transform circuit 8, which is a signal percentage image extraction means, converts this frame difference signal into low frequency components and vertical frequency components. , into signal components representing characteristics such as horizontal frequency components. Since each of these components is a noise component and a motion component, only small-level noise components can be extracted from each of these components by passing it through the nonlinear processing circuit (2). Since each component extracted by the nonlinear processing circuit 21 is obtained by Hadamard transform, it is returned to the original time axis by passing through the Hadamard inverse transform circuit 10, and a parallel digital noise signal is obtained.

Next, the parallel-serial conversion circuit 11 converts the parallel digital noise signal into a serial digital signal similar to the input form. The signal obtained here is approximately the one obtained by extracting only the noise component from the frame difference signal.

Then, as described above, this signal is supplied to the subtraction circuit 3 and subtracted from the input signal, thereby obtaining a noise-free digital signal. Finally, the A/D converter 12 converts the digital video signal into the original analog signal and outputs it.

This embodiment is also possible using field memory.

Furthermore, it is clear that the IP is not limited to NTSC composite signals and can also be used as baseband signals.

Note that in this case, the color signal phase shift circuit is unnecessary.

The nonlinear processing circuit 21 of the noise reduction device having the above configuration will be described in detail. In general, there are various characteristics of this nonlinear processing circuit, as shown in Figure 2 (al to 1dlK).The characteristics of nonlinear processing that can be considered in principle are A positive output is performed, and a negative output is performed in response to a negative input.In principle, numerical S/N improvement can be expected for these nonlinear processes.However, the present inventor According to experiments, it was found that there is a relationship between visual evaluation of nonlinear characteristics and afterimage characteristics.

The results will be explained below.

The process shown in FIG. 2(a) outputs an output level proportional to the input component by a proportionality constant k(k)O). - According to experiments using this nonlinear processing circuit, there was a problem that an afterimage phenomenon occurred in edge portions where the level difference was small, and the edges became blurred and unclear.

The nonlinear processing shown in FIG. 2 fbl is a substantially sine curve, and outputs a 0 level above a certain level or below a certain level. According to experiments using this nonlinear processing circuit, there were problems in that edges with small level differences tend to be blurred, and noise flickering was observed near the edges.

The nonlinear processing shown in Figure 2 (cl) outputs an output level that is proportional to the input level by a proportionality constant n (n > 0) at a certain input level, and at a certain level above or below the input level. A proportional constant (m<O
) outputs a proportional output level, and outputs a 0 level above a certain level or below a certain level, respectively. According to experiments using this nonlinear processing circuit, there was a problem that edges with a small level difference tend to be blurred, and noise flickers were observed near the edges, almost the same as in FIG. 2 (bl).

The nonlinear processing shown in Figure 2 (di) outputs an output level that is proportional to the input level with a proportionality constant i (i>0) up to a certain input level, and above the reference level a12, it outputs an approximately constant value A. Or c (i>lcl
> 0), outputs a value proportional to the reference level ba, and below the reference level ba, it outputs an approximately constant value B or d relative to the input signal (i>ldl>
0) outputs a proportional value.

According to experiments using this nonlinear processing circuit, the edges are blurred due to the afterimage phenomenon that was observed in the processing circuits described above.
It was also confirmed that there was no noise flickering at the edges. As described above, although it is considered to be effective in principle, the inventor's experiments have confirmed that the nonlinear processing method shown in FIG. 2(d) provides the visually best results.

Therefore, the present invention proposes the characteristics of (dl) as the optimal characteristics for a nonlinear processing circuit.

3 and 4 show a circuit diagram of a noise reduction device according to a second embodiment of the present invention and a detailed diagram of its main parts.

In FIGS. 3 and 4, parts given the same reference numerals as in FIG. 5 are the same as in FIG. 5, and have the same functions.

In FIG. 3, 31 is a Hadamard transform circuit group 31 provided between the series/parallel switching circuit 7 and the parallel/serial conversion circuit 11.
As shown in FIG. 4, a Hadamard transform circuit 8, a nonlinear processing circuit group 32 to which the output of the number is input from the Hadamard transform circuit 8, and a nonlinear processing circuit group 32
Hadamard inverse transform circuit 1 which receives each output of
0, and the nonlinear processing circuit 32 performs different nonlinear processing on each of the plurality of outputs from the Hadamard transform circuit 7.

Each nonlinear processing circuit of this nonlinear processing circuit group 32 is
As explained in the first embodiment, the processing method shown in FIG. 2 fdl is the most effective. Each of these nonlinear processing circuits has a proportionality constant 1n (in>0
) outputs an output value proportional to the reference level anJ21.
In the above, cn(i
n>ICn1>0), outputs a proportional value and sets the reference level b. In the following, d is approximately constant value Bn or d for the input signal.
This is a nonlinear processing circuit that outputs a value proportional to n (in>1dnl>O).

The operation of the noise reduction device of this embodiment configured as described above will be described below. When a video signal is input from the input terminal 1, the analog signal is converted into a digital signal by the A/D converter 2. The subtraction circuit 3 subtracts the video signal of one frame before from the input video signal, thereby subtracting non-correlated components, and ideally B! which does not contain noise components. l! It becomes an image signal. This video signal component is stored in the frame memory 4 and delayed for one frame.

Since the color signal of the NTSC color video signal has a phase inversion for each frame, this is compensated by a color signal phase shift circuit 5, and only the color signal of the video signal delayed by the frame memory 4 is inverted in phase. A subtraction circuit 6 obtains a frame difference signal representing the difference between the two video signals. This frame difference signal is a signal component without frame correlation (that is, a motion component)
This signal is a combination of the noise component and the noise component. This frame difference signal is converted from a serial digital signal to a parallel digital signal by a serial-parallel conversion circuit 7, and a Hadamard conversion circuit 8 converts this frame difference signal into multiple components such as a low frequency component, a vertical frequency component, and a horizontal frequency component. Divide into ingredients. Since each of these components is a noise component and a motion component,
By passing each component through the nonlinear processing circuit group 32,
Only small-level noise components can be extracted from each of these components.

Here, the characteristics of the Hadamard transform output will be explained in detail. In general, a video signal has a percentage of horizontal movement and little vertical movement. For vertical motion components, it is better to extract most of the noise components to the nonlinear circuit output even if a small amount of motion components are mixed in. therefore,
It is desirable to set the reference levels bn and cn high;
Although the afterimage deterioration in the vertical direction is slightly increased, there is no problem with the pressure image, and a high S/N improvement can be achieved. On the other hand, it is better for a nonlinear processing circuit for horizontal motion components to have less motion components mixed in, even if the ability to extract noise components is reduced. Therefore, it is desirable to set the reference levels bn and cn low. Therefore, the S/N improvement ability may be reduced or the afterimage phenomenon may be suppressed.

In this way, each component decomposed into Hadamard-transformed feature components is subjected to nonlinear processing with different characteristics depending on each feature component, and noise components are extracted. Since each component extracted from each nonlinear processing circuit is obtained by Hadamard transform, it is returned to the original time axis by passing through Hadamard inverse transform circuit 10, and a parallel digital noise signal is obtained. . Next, the parallel-serial conversion circuit 11 converts the parallel digital noise signal into a serial digital signal similar to the input form. The signal obtained here is approximately the one obtained by extracting only the noise component from the frame difference signal. Then, as described above, this signal is supplied to the subtraction circuit 3 and subtracted from the input signal, thereby obtaining a noise-free digital signal. Finally, the A/D converter 12 converts the digital video signal into the original analog signal and outputs it.

This embodiment is also possible using field memory.

In addition, it is necessary to limit the NTSC composite signal to
, it is clear that it can also be constructed with baseband signals. Note that in this case, the color signal phase shift circuit is unnecessary.

Effects of the Invention As described above, according to the present invention, by providing the optimal characteristics of the nonlinear processing type performed on each signal component extracted as a feature, the deterioration of the afterimage characteristics is small and the S/N improvement effect is achieved. It is possible to obtain a noise reduction device with a large amount of noise, and its practical effects are great.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of the noise removal device according to the first embodiment of the present invention, FIG. 4 is a block diagram showing details of the Hadamard transform circuit group in the noise removing device, FIG. 5 is a block diagram of a conventional noise reducing device, and FIG. 6 is a block diagram showing the details of the Hadamard transform circuit group in the noise removing device.
II cumulative unit figure. 1... Signal input terminal, 2... A/D converter, 3, 6
... Subtraction circuit, 4... Frame memory, 5... Color signal phase shift circuit, 7... Series-parallel conversion circuit, 8
...Hadamard conversion circuit, 10...Hadamard inverse conversion circuit, 11...-Parallel-serial conversion circuit, 12...
D/A converter, 13...-output terminal, 21... nonlinear processing circuit, 31... Hadamard conversion processing circuit group, 32
...Nonlinear processing circuit group.

Claims (1)

[Claims] 1. A first subtraction means for taking the difference component between the input video signal and the output signal of the n (n>0, where n is an integer) field delay means, and the output of the first subtraction means. a feature extraction means for extracting a characteristic component of a signal; a nonlinear processing means for extracting a noise component from the output of the feature extraction means; and a second subtraction means for obtaining a difference signal between the output of the nonlinear processing means and the input video signal. The n-field delay means delays the output of the second subtraction means by n fields, and the non-linear processing means has an input signal at a reference level a (a>0) or lower and a reference level b (b<0). 0) or more, outputs an output signal that is proportional to the input signal by a proportionality constant i (however, i > 0), and when the input signal is equal to or higher than the reference level a, outputs a substantially constant value A, and the input signal A noise reduction device configured to output a substantially constant value B when is below a reference level b. 2. A first subtraction means that takes the difference component between the input video signal and the output signal of the n (n>0, where n is an integer) field delay means, and a plurality of (N ≠1)
); a nonlinear processing group having a plurality of nonlinear processing means for extracting noise components from each of the plurality of feature component outputs of the feature extraction means; and outputs and inputs from the nonlinear processing group. a second subtraction means for obtaining a difference signal from a video signal, the n field delay means delays the output of the second subtraction means by n fields, and the nonlinear processing group When the input signal is below the reference level a_n (a_n>0),
When the input signal is equal to or higher than the reference level b_n (b_n<0), an output signal proportional to the input signal by a proportionality constant i_n (however, i_n>0) is output, and the input signal is at the reference level a_n.
When above, a substantially constant value A_n is output, and when the input signal is below the reference level b_n, a substantially constant value B_n is output, and the reference levels a_n, b_ of each of the nonlinear processing means are output.
A noise reduction device configured such that at least one of n, proportionality constant i_n, and substantially constant values A_n and B_n is different for each of a plurality of nonlinear processing means.