JPH04334192A - Noise reduction device and noise reduction image pickup device - Google Patents

Abstract

PURPOSE:To improve the S/N considerably by solving the deterioration in a video signal due to occurrence of after-image by means of noise reduction and to reduce noise especially when an input video signal is a component signal. CONSTITUTION:A component video signal (comprising a luminance signal, an R-Y signal and a B-Y signal) from input terminals 1, 22, 28 are received and A/D converters 2, 23, 29 convert each into a digital signal. In this case, let a sampling frequency of the luminance signal be fs and let a sampling frequency of two color difference signals be nearly 1/4 of that of the luminance signal. Each component signal is inputted an independent recursive filter type noise reduction circuit. The input signal is given to a 2X4-degree Hadamard conversion processing circuit group 21 or 2X8-degree Hadamard conversion processing circuit groups 25B, 25R and the sampling frequency and the degree of the Hadamard conversion are varied with the input signal to obtain high improvement of S/N without deteriorating the input video signal.

Description

[Detailed description of the invention]

0001

[Industrial Application Field] The present invention utilizes fields, frames, or line correlations of video output from televisions, videos, video cameras, etc., thereby preventing deterioration of the original image such as afterimages even in moving images. The present invention relates to a noise reduction device that reduces noise, and particularly relates to the sampling frequency and signal processing band of each signal when the input signal is a component signal.

0002

2. Description of the Related Art A conventional noise reduction device is disclosed, for example, in Japanese Patent Application Laid-Open No. 158574/1983.

[0003] A conventional noise reduction device will be explained below. FIG. 8 is a block diagram showing the configuration of a conventional noise reduction device.

In FIG. 8, 1 is an input terminal for a composite video signal, 2 is an A/D converter that converts the composite input video signal from an analog signal to a digital signal, and 3 is an input terminal for two input signals. 4 is a subtraction circuit for subtracting the other input signal from the input video signal, and 4 is a frame memory for delaying the noise-reduced signal from the input video signal by one frame. Since the phase of the color signal of the NTSC color video signal is inverted every frame, 5 is a color signal phase shift circuit to compensate for this, and only the color signal of the video signal delayed in the frame memory 4 is phase shifted. Invert. 6 is a subtraction circuit that obtains a difference signal (frame difference signal) by subtracting the input video signal and the video signal delayed by one frame; 7 is a Hadamard transform that extracts vertical components, horizontal components, and diagonal components from the frame difference signal. It is a circuit. Since the output of the Hadamard transform circuit 7 has different distributions of noise components and motion components,
Reference numeral 8 denotes a nonlinear processing circuit that utilizes this fact to extract noise components. 9 is a Hadamard inverse transform circuit, and since the noise signal extracted from the nonlinear processing circuit 8 is obtained by Hadamard transform, it is returned to the original time axis by Hadamard inverse transform. 10 is a D/A converter that converts a digital signal into an analog signal.

In the conventional noise reduction device configured as described above, when a composite 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 is passed through the subtracter 3. A non-correlated component, which will be described later, is subtracted by , resulting in a video signal component that ideally does not contain noise components, which is stored in the frame memory 4 and delayed for one frame. These are each pixel point of m frames shown in FIG. Since the phase of the color signal in this video signal delayed by one frame is reversed from that of the original 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. , a subtracter 6 obtains a difference signal (frame difference signal) between two video signals having the same chroma phase. 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 will be a signal that is a combination of a signal component with no frame correlation and a noise component.

A method for obtaining only the noise component from this frame difference signal will be described below. This frame difference signal is converted into a low frequency component by a Hadamard transform circuit 7.
It can be divided into a vertical component, a horizontal component, etc. in this case,
One pattern of input picture elements is shown in Figure 9, and the input is as follows.

[0007]

[Math 1]

[0008] If the transformation output of the 2×4 Hadamard transform is F24, then (Equation 2) and (Equation 3) are obtained.

[0009]

[Math 2]

0010

[Math 3]

According to the above equations (1) to (3), the Hadamard transform output F24 is obtained from the 2×4-order input picture element X24. H2 and H4 are as shown in (Equation 4) and (Equation 5).

0012

[Math 4]

[0013]

[Math 5]

The output from the Hadamard transform circuit 7 is a converted output of eight components. On the other hand, since noise has no correlation, it is almost equally distributed among the eight component frequencies of the output of the Hadamard transform circuit 7. Since the noise level at the output of the Hadamard transform circuit 7 corresponds to the noise level of the input signal as is well known, only small-level noise components can be extracted from each of these components through the nonlinear processing circuit 8. Since each component extracted by this nonlinear processing circuit 8 is obtained by Hadamard transform, it is returned to the original time axis by passing through Hadamard inverse transform circuit 9, and a parallel digital noise signal is obtained here. Become. The signal obtained here is one in which only the noise component is extracted from the frame difference signal that has no frame correlation, and is supplied to the subtraction circuit 3, where a noise-free digital video signal is obtained by subtracting the noise component from the input video signal. It will be done. Finally, the digital video signal is converted into the original analog signal by the D/A converter 10 and output.

[0015] A noise reduction device using such a method is as follows:
In principle, it is possible to reduce noise not only for still images but also for moving images when inputting a composite video signal without significantly deteriorating the input video.

[0016]

The noise reduction device described in the conventional example can, in principle, reduce noise without producing an afterimage in still images as well as in moving images by inputting a video signal.

However, in the conventional configuration described above, when the input signal is a component signal, an optimal configuration for signals having different bands, such as a luminance signal and a color difference signal, has not been proposed.

The present invention solves the above conventional problems and provides an optimal circuit configuration when the input signal is a component signal, and also provides a noise reduction device that is most advantageous in terms of circuit scale. The purpose is to

[0019]

[Means for Solving the Problem] In order to achieve this object, the noise reduction device of the present invention quantizes a plurality of component video signals at independent frequencies, and then decomposes the noise into independent and different characteristic components. A noise reduction device using a removing means, each of the noise removing means comprising a first subtracting means for obtaining a difference component between each input video signal and an output signal of the n (n>0) field delay means; a feature extraction means for decomposing the output of the subtraction circuit No. 1 into a plurality of feature components; a nonlinear processing means for performing nonlinear processing on at least one of the feature components; and a difference between the output of the nonlinear processing means and the input video signal. The present invention is comprised of a second subtraction means for obtaining a signal, and the n-field delay means for delaying the output signal from the second subtraction means.

[0020]

[Operation] With the above-described configuration, the present invention makes the quantization frequencies of each of the component input video signals independent, and makes the feature extraction components independent and different, thereby effectively extracting information from each signal component of the component input video signal. noise can be removed.

[0021]

Embodiments Hereinafter, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 shows a block diagram of a noise reduction device according to a first embodiment of the present invention. In FIG. 1, the components 1 to 6 are the same as those shown in FIG.
They work the same way. 21 is the circuit configuration shown in FIG. 2, 7 is a 2×4-order Hadamard transform circuit, and 80 is a group of nonlinear processing circuits. 9 is a 2×4-order Hadamard transform circuit, and 2
1 is a circuit configuration including a series of processes, and 7 and 9 are circuits similar to the Hadamard transform circuit 7 and Hadamard inverse transform circuit 9 shown in FIG. 8, and have the same function. 10 is D shown in FIG.
It is similar to the /A conversion circuit 10 and functions similarly. 2
2 is an input terminal for color difference signal B-Y, 28 is color difference signal R-Y
input terminals 23 and 29 are A/D converters 24 that convert input signals B-Y from analog signals to digital signals;
, 30 is a frame memory that stores one frame of each of the B-Y and R-Y signals, 25 is the circuit configuration shown in FIG. 3, 40 is a 2×8-order Hadamard transform circuit, and 41 is a group of nonlinear processing circuits. , 42 is a 2×8-order inverse Hadamard transform circuit,
This is a circuit configuration that includes a series of processes. 26, 31 is D/A
In the converter, 27 is an output terminal for the color difference signal B-Y, and 32 is an output terminal for the color difference signal RY.

The operation of the noise reduction device of this embodiment configured as described above will be explained below. When a luminance signal is input as a component signal from an input terminal 1, it is sampled by an A/D converter 2 at 4 fsc (however, the frequency is not limited to this) and converted from an analog signal to a digital signal. This signal passes through a subtracter 3 to subtract non-correlated components, which will be described later, and ideally becomes a luminance signal that does not contain noise components. This luminance signal is stored in the frame memory 4 and delayed for one frame, and a subtracter 6 obtains a frame difference signal between the two luminance signals. This frame difference signal is a signal that is a combination of a signal component (motion component) with no frame correlation and a noise component. This frame difference signal enters the 2×4-order Hadamard transform processing circuit group 21, and the 2×4-order Hadamard transform circuit 7 converts the frame difference signal into eight components such as a low frequency component, a vertical frequency component, and a horizontal frequency component. Divided into dimensional frequency components. At this time, the spectrum appearing in the horizontal direction is
When the sampling frequency is 4fsc, an energy distribution appears centered on the spectrum shown in FIG. This is a spectrum on the band of the luminance signal, and the distribution interval of the spectrum is approximately constant. The frame difference signal input to the 2×4-order Hadamard transform circuit 7 is subjected to 2×4-order Hadamard transform and replaced with a signal component on the frequency axis. Among the outputs of the 2×4 Hadamard transform circuit 7, the video signal has correlation on the frequency axis, so energy is distributed on a biased spectrum. On the other hand, since noise has no correlation on the frequency axis, it is almost evenly distributed within the frequency distribution of the eight components of the output of the Hadamard transform circuit 7. As is well known, the noise level at the output of the Hadamard transform circuit 7 corresponds to the noise level of the input signal, so by passing it through the nonlinear processing circuit group 80, the spectrum of each output of the Hadamard transform has a different nonlinear By performing the processing, only low-level noise components can be extracted from the eight components. Each component extracted by this nonlinear processing circuit group 80 is 2×
Since it is obtained by the fourth-order Hadamard transformation, 2
The signal is returned to the original time axis by passing through the ×4-order Hadamard inverse transform circuit 9. The signal obtained here is one in which only the noise component is extracted from the frame difference signal that has no frame correlation, and is supplied to the subtraction circuit 3 as described above, and is converted into a noise-free digital signal by subtracting the noise component from the input luminance signal. A luminance signal will be obtained. And D/A
The digital video signal is converted into the original analog signal by the converter 10 and outputted from the output terminal 11.

Next, processing of color difference signals, which are component signals, will be described. B-Y and RY are input from input terminals 22 and 28, respectively. Since the band of the chrominance signal is approximately 1/4 of the band of the luminance signal, it is sufficient for the sampling frequency to be 1/4 of the sampling frequency of the luminance signal. Therefore, the input color difference signal is sent to the A/D converter 23.
, 29, fs is a sampling frequency of 1/4 of the luminance signal.
The analog signal is sampled at a sampling frequency of c (3/4 fsc when the sampling frequency of the luminance signal is 3 fsc), and the analog signal is converted into a digital signal. B-
The two color difference signals inputted as the Y input signal and the R-Y input signal pass through subtracters 3B and 3R to subtract non-correlated components, which will be described later, and ideally become color difference signal components that do not contain noise components. Become. The respective color difference signals are stored in the frame memories 24 and 30 and delayed for one frame, and frame difference signals of the color difference signals are obtained by the subtracters 6B and 6R. This frame difference signal is a signal that is a combination of a signal component with no frame correlation and a noise component. This frame difference signal is processed by the 2×8-order Hadamard transform processing circuit group 25.
B, 25R, and is converted into signal components on the frequency axis by a 2×8-order Hadamard transform.

[0025] Here, the 2×8-order Hadamard transform will be explained in detail. input

[0026]

[Math 6]

Then, the output of the 2×8 Hadamard transform is

[0028]

[Math 7]

[0029]

[Math. 8]

[0030] (Equation 8) provides the output of the 2×8-order Hadamard transform. H2 and H4 are (Math. 9), (
It is as shown in Equation 10).

[0031]

[Math. 9]

[0032]

[Math. 10]

The color difference signal obtained by dividing the frame difference signal into frequency components such as a low frequency component, a vertical frequency component, and a horizontal frequency component by the 2×8-order Hadamard transform circuit 40 is as follows.
fsc (3/4 fsc is also considered), and in this case, the signal processing speed in the circuit is 1/4 of that of the luminance signal, and due to the circuit scale, the Hadamard transform order of the color signal is higher than that of the luminance signal. It becomes possible to take
A two-dimensional Hadamard transformation of order 2×8 is performed. As the transform order increases and the number of output spectra of the Adal transform increases as a result, the degree of S/N improvement increases. This 2×
If the input pixel method of the 8th order Hadamard transform is shown in Figure 5,
The output of the 2×8-order Hadamard transform circuit 40 has an energy distribution centered around the spectral components shown in FIG. This spectrum is located on the band of the color signal, and can decompose the input color difference signal into approximately equally spaced frequencies. The video signal output from the 2×8-order Hadamard transform circuit 40 is a converted output of 16 components having 16 types of frequency characteristics, and is an output signal having an energy distribution biased toward a certain frequency on the frequency axis. On the other hand, since noise has no correlation on the frequency axis, it is almost evenly distributed within the frequency distribution of the 16 components of the output of the 2×8 Hadamard transform circuit 40. As is well known, the noise level at the output of the 2×8-order Hadamard transform circuit 40 corresponds to the noise level of the input signal, so through the nonlinear processing circuit group 41,
Only low-level noise components can be extracted from each of the 16 components. Since each component extracted by the nonlinear processing circuit group 41 is obtained by a 2×8-order Hadamard transform, the 2×8-order Hadamard inverse transform circuit 42
By passing through, you are returned to the original time axis. The signal obtained here is one in which only the noise component is extracted from the frame difference signal having no frame correlation, and as mentioned above, it is supplied to the subtraction circuits 3B and 3R, and the noise is removed by subtracting the noise component from the input color difference signal. In this case, an RY, BY digital color difference signal is obtained. Then, the digital color difference signals are converted into the original analog signals by A/D converters 26 and 31 and outputted from output terminals 27 and 32.

Finally, component video signals (luminance signal, R-Y signal, B-
Y signal) output is obtained.

Although this embodiment uses a frame memory, it is also possible to implement this embodiment using a field memory.

Next, a second embodiment of the present invention will be described with reference to the drawings. FIG. 7 shows a block diagram of a noise reduction imaging device using the noise reduction device according to the first embodiment of the present invention. In FIG. 7, 51 is a CCD of an imaging device, 52 is a camera signal processing circuit, 53 is a luminance signal noise reduction circuit group, 54 is an RY signal noise reduction circuit group, 55 is a BY signal noise reduction circuit group, 56 is a luminance signal output terminal, 57 is an R-Y signal output terminal, and 58 is a B
-Y signal output terminal.

The operation of the noise reduction imaging device of this embodiment configured as described above will be explained below. The input optical signal is converted from an optical signal to an electrical signal by the CCD 51 of the solid-state imaging device at the clock frequency fS of the imaging device. A luminance signal, an RY signal, and a BY signal are obtained from the input signal by the camera signal processing circuit 52 as component signals. These respective component signals are input to respective noise reduction circuit groups 53, 54, and 55, and noise is reduced. These three noise reduction circuit groups are similar to each of the noise reduction circuits described in the first embodiment, and perform similar processing. However, in this case, the sampling frequency of the luminance signal is the same as the clock frequency fs of the imaging device, and the sampling frequency of the color difference signal is approximately 1/4 of the sampling frequency of the luminance signal, which is 1/4 fs. Component signals with reduced noise (luminance signal, R-Y signal, B
-Y signal) is output from output terminals 56, 57, and 58.

Although this embodiment uses a frame memory, it is also possible to implement this embodiment using a field memory.

[0039]

As described above, the present invention samples each component video signal at an independent sampling frequency and provides an independent noise reduction circuit for each component video signal, thereby reducing the bandwidth of the video signal and the brightness of the component signals having different properties. Even in signals and color difference signals, noise can be efficiently reduced without deteriorating the input signal.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of a noise reduction device in a first embodiment of the present invention.

[FIG. 2] A block diagram showing the internal configuration of the 2×4-order Hadamard transform circuit group 21 in FIG. 1.

[FIG. 3] 2×8-order Hadamard transform circuit group 25B in FIG. 1,
Block diagram showing the internal configuration of 25R

[FIG. 4] Spectrum diagram showing the output of the 2×4 Hadamard transform circuit 7 in FIG. 2

[FIG. 5] 2×8-order Hadamard transform processing circuit group 25 in FIG. 1
Input pattern diagram showing input to B, 25R

[Figure 6] Figure 3
A spectrum diagram showing the output of the 2×8-order Hadamard transform circuit 40 of

FIG. 7 is a block diagram of a noise reduction imaging device using a noise reduction device according to a second embodiment of the present invention.

[Figure 8] Block diagram showing the configuration of a conventional noise reduction device

[Figure 9] Figure 8
Input pattern diagram showing the input to the 2×4 Hadamard transform circuit 7

[Explanation of symbols]

1 Input terminals 2, 23, 29 A/D converter 3 Subtractor 4, 24, 30 Memory 5 Color signal phase shift circuit 6, 6B, 6R Subtractor 7 Hadamard transform circuit 8 Nonlinear circuit 9 Hadamard inverse transform circuit 10, 26 , 31 D/A converter 11 Output terminal 21 2×4-order Hadamard transform processing circuit group 22 B
-Y signal input terminals 25B, 25R 2×8-order Hadamard transform processing circuit group 27, 57 B-Y signal output terminal 28 RY signal input terminals 32, 58 RY signal output terminal 40 2×8-order Hadamard transform circuit 41 Nonlinear circuit 42 2×8th order Hadamard inverse transform circuit 51 CCD
(Solid-state imaging device) 52 Camera signal processing circuit 53 Luminance signal noise reduction circuit group 54 RY signal noise reduction circuit group 55 B-Y luminance signal noise reduction circuit group 56 Luminance signal output terminal

Claims (14)

[Claims] 1. Means for dividing a plurality of component video signals into at least two sets, quantization means for quantizing each set at an independent frequency, and independent noise removal means for each of the component video signals. The noise reduction device includes a first subtraction device for obtaining a difference component between each input video signal and an output signal of the n (n>0) field delay device, and the first subtraction circuit. a feature extracting means for decomposing the output of the above into a plurality of feature components; a non-linear processing means for performing non-linear processing on at least one of the feature components; and a step for obtaining a difference signal between the output of the non-linear processing means and the input video signal. 2 subtraction means, and the n-field delay means for delaying the output signal from the second subtraction means, the noise having an output of the n-field delay means or an output of the second subtraction means as an output signal. Reduction device. 2. The noise reduction device according to claim 1, wherein the plurality of component input video signals are a luminance signal and two color signals. 3. The noise according to claim 1, wherein the plurality of component input video signals are a luminance signal and two color difference signals, and the quantization frequency of the luminance signal is higher than the quantization frequency of the two color difference signals. Reduction device. 4. A claim characterized in that the plurality of component input video signals are a luminance signal and two color difference signals, and the quantization frequency of the two color difference signals is 1/4 of the quantization frequency of the luminance signal. 1. The noise reduction device according to 1. 5. Claim 1, wherein the plurality of component input video signals are a luminance signal and two color difference signals, and the number of feature components extracted by the feature extraction means is greater for the color difference signals than for the luminance signals. Noise reduction device as described. 6. The plurality of component input video signals are a luminance signal and two color difference signals, and the number of feature components extracted by the feature extraction means is such that the number of luminance signals is 1/2 of the number of color difference signals. The noise reduction device according to claim 1, characterized in that: 7. The plurality of component input video signals are a luminance signal and two color difference signals, the quantization frequency of the luminance signal is 4 fsc (fsc = 3.58 MHz), and the quantization frequency of the two color difference signals is fsc. The noise reduction device according to claim 1, characterized in that: 8. means for replacing an input optical signal with an electrical signal at a clock frequency fs; means for extracting the replaced electrical signal from the optical signal into at least two sets of a plurality of digital component video signals; means for dividing the video signal into at least two sets; and quantization means for quantizing at least one set of the component video signals at a quantization frequency of fs, and quantizing the other sets at independent frequencies. , a noise reduction device which passes the component video signal through each independent noise removal means, wherein the noise reduction means calculates the difference component between each input video signal and the output signal of the n (n>0) field delay means. a first subtraction means for obtaining a first subtraction circuit; a feature extraction means for decomposing the output of the first subtraction circuit into a plurality of feature components; a nonlinear processing means for performing nonlinear processing on at least one of the feature components; and the n-field delay means for delaying the output signal from the second subtraction means, the output of the n-field delay means or A noise reduction imaging device that uses the output of the second subtraction means as an output signal. 9. The noise reduction imaging device according to claim 8, wherein the plurality of component input video signals are a luminance signal and two color signals. 10. The noise according to claim 8, wherein the plurality of component input video signals are a luminance signal and two color difference signals, and the quantization frequency of the luminance signal is higher than the quantization frequency of the two color difference signals. Reduction imaging device. 11. A claim characterized in that the plurality of component input video signals are a luminance signal and two color difference signals, and the quantization frequency of the two color difference signals is 1/4 of the quantization frequency of the luminance signal. 8. The noise reduction imaging device according to 8. 12. Claim 8, wherein the plurality of component input video signals are a luminance signal and two color difference signals, and the number of feature components extracted by the feature extraction means is greater for the color difference signals than for the luminance signals. The noise reduction imaging device described. 13. The plurality of component input video signals are a luminance signal and two color difference signals, and the number of feature components extracted by the feature extraction means is such that the number of luminance signals is 1/2 of the number of color difference signals. The noise reduction imaging device according to claim 8. 14. The plurality of component input video signals are a luminance signal and two color difference signals, the quantization frequency of the luminance signal is 4 fsc (fsc = 3.58 MHz), and the quantization frequency of the two color difference signals is fsc. The noise reduction imaging device according to claim 8, characterized in that: