JP3429615B2 - Image signal noise reduction circuit - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image signal noise reduction circuit, and more particularly to MUSE (Mult).
apple sub Nyquist Sampling
The present invention relates to a noise reduction circuit that performs digital signal processing of an image in an Encode) high-precision television signal decoder, an EDTV receiver, and the like.

[0002]

2. Description of the Related Art A moving image and a still image are distinguished from each other and subjected to different signal processing, and then the moving image signal and the still image signal are mixed based on motion information of the image, such as a MUSE decoder or an EDTV receiver. In the motion adaptive image processing system, the still image part uses a cyclic noise removal method using a frame memory. However, in the moving image portion, the cyclic noise removal method using the frame memory cannot be used because it causes a problem such as a double image.

FIG. 6 shows the configuration of a conventional moving image signal noise reduction circuit. In FIG. 6, reference numeral 78 is an input terminal for inputting an image signal, 79 is a low-pass filter (hereinafter referred to as LP
Abbreviated as F. ), 80 is a subtraction circuit, 81 is a non-linear processing circuit, 82 is an addition circuit, and 83 is an output terminal.

The operation of the conventional noise reduction circuit for moving picture signals is shown below. The signal input from the input terminal 78 is LP
It is distributed to the F79, the subtraction circuit 80, and the addition circuit 82.
The LPF 79 removes high frequency components from the input signal and outputs only low frequency components. The subtraction circuit 80 subtracts the signal input to the input terminal 78 from the output signal of the LPF 79, that is, the low frequency part of the input signal. As a result, a high frequency signal component (negative) is extracted from the output of the subtraction circuit 80. After this high-frequency signal component (negative) is processed by the non-linear processing circuit 81,
The addition circuit 82 adds the original input signal and outputs the noise-reduced signal from the output terminal 83.

FIG. 7 shows an example of the input / output characteristics of the non-linear processing circuit 81 used in such a conventional noise reduction circuit. As is clear from FIG. 7, the output is set to “0” as an error in the extremely low input level range (portion A). Next, in the range where the input level is high to some extent (portion B), a level proportional to the input level is output as noise. Further, in the range where the input level is high (portion C), the input signal is a mixture of noise and contour components of the image, and the proportion of contour components increases as the input level increases.
The higher the input level, the lower the level that is output. That is, by the nonlinear processing circuit 81, B
Only signals in the range of C and C are determined to be noise, and the addition circuit 82 removes them from the original input signal.

In this case, the setting of the respective ranges A, B, C having different input / output characteristics is empirically determined, and
Since the noise component and the contour component are to be distinguished only by the level of the input signal, it is difficult to remove the noise without degrading the image.

Regarding conventional noise reduction techniques,
For example, a "noise reduction circuit" disclosed in Japanese Patent Laid-Open No. 6-113315 is known. This noise reduction circuit is configured to detect the contour of the image, remove noise in the flat portion of the image by a low-pass filter, and output the input signal as it is at the contour portion of the image. However, in the case of this method, the noise in the contour portion of the image is not removed.

[0008]

In the case of such a conventional noise reduction circuit, since noise is reduced by removing high frequency components, high frequency components of the video signal are also removed. As a result, there is a problem that the image quality of the moving image deteriorates (blur). That is, there is a trade-off relationship that the image quality of a moving image deteriorates instead of obtaining the noise reduction effect. In addition, the noise removal effect is less than that of the frame cyclic method used for the still image noise removal method, and S / N (signal to noise) in the moving image portion and the still image portion in the image after the motion adaptation processing is performed. The difference in the ratio is noticeable, which causes deterioration of image quality.

The present invention has been made to solve the above-mentioned problems, and is free from blurring of moving images, and is equal to or more than the frame recursive noise elimination system for still images. It is an object of the present invention to provide a noise reduction circuit for a moving image that obtains a noise removal effect.

[0010]

According to the output of a contour detection circuit, a noise reduction circuit according to a first aspect of the present invention changes the respective ratios of the output of a low pass filter and the output of a median filter. If the ratio of the contour signal to the noise level signal is large, the ratio of the output of the median filter is increased, and if the ratio of the contour signal to the noise level signal is small, the ratio of the output of the low-pass filter is increased. It is characterized in that the output is increased. With this configuration, in the flat part of the image,
By increasing the ratio of the low-pass filter output signal and increasing the ratio of the median filter output signal in the contour part, noise is removed in the contour part of the image while maintaining the contour of the image, and A noise reduction effect can be obtained. As a result, it is possible to obtain a noise reduction effect equivalent to or higher than that of the frame recursive noise reduction method used for still images in moving images. It is possible to eliminate the difference in the S / N ratio between the moving picture and the moving picture.

[0011] Noise reduction circuit according to claim 2, cascaded in two stages noise reduction circuit according to claim 1, the
Cascade connection of the output of the first-stage noise reduction circuit connected in cascade
It is characterized in that it is used as the input of the noise reduction circuit of the second stage that is continued . With this configuration, the noise removal effect can be increased.

According to a third aspect of the present invention, there is provided a noise reduction circuit further comprising a second arithmetic circuit connected to the output of the noise reduction circuit according to the second aspect, and the second arithmetic circuit responds to the output of the noise level signal. Te, connected in cascade to the output of the noise reduction circuit of the first stage connected in cascade second stage of the noise reduction circuit
The output and adds while varying the respective proportions, when the noise level signal is large, increases the proportion of output of the noise reduction circuit of the second stage, when the noise level signal is small 1 It is characterized in that the ratio of the output of the noise reduction circuit in the stage is increased and then output. With this configuration, it is possible to obtain a noise reduction effect that matches the noise level, and as a result, it is possible to suppress deterioration in image quality due to the unnecessary noise removal effect.

According to a fourth aspect of the present invention, there is provided a noise reduction circuit further comprising a noise reduction circuit according to the third aspect, further comprising a third arithmetic circuit and a switching circuit connected to each other.
Obtain the absolute value of the difference between the input and output of the noise reduction circuit in the second stage, compare the absolute value with the noise level signal, and make a significant difference for both .
If there is a difference, depending on the switching circuit so as to connect the input and final output of the noise reduction circuit of the first stage,
Further, when there is no significant difference between the two , it is characterized in that the switching circuit switches the output of the second arithmetic circuit to the final output. With this configuration, it is possible to reduce blurring of an image due to a mistake in the contour detection circuit.

[0014]

BEST MODE FOR CARRYING OUT THE INVENTION Next, an embodiment of a noise reduction circuit according to the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram for explaining the configuration of a first embodiment of a noise reduction circuit according to the present invention. In FIG. 1, 1 is an input terminal for inputting an image signal,
Reference numeral 2 is a contour detection circuit, 3 is a low-pass filter, 4 is a median filter, 5 is a subtraction circuit, 6 is a multiplication circuit, 7 is an addition circuit, 8 is a noise level input terminal, and 9 is an output terminal.

The signal input from the input terminal 1 is supplied with a contour detection circuit 2, a low pass filter 3 and a median filter 4.
Entered in and.

FIG. 2 shows the configuration of the contour detection circuit 2. Figure 2
10 is an input terminal, 11, 12, 13, and 14 are pixel delay circuits, 15 and 16 are addition and averaging circuits, 17 is a subtraction circuit, 18 is an absolute value circuit, 19 is a noise level signal input terminal, 20 is a coefficient circuit, Reference numeral 21 is a divider circuit, and 22 is an output terminal. With respect to the signal input from the input terminal 10, the difference between pixels is obtained by the averaging circuits 15 and 16 and the subtraction circuit 17, and the absolute value thereof is obtained as the contour signal of the pixel. The contour signal controls the degree of noise removal and the degree of image quality deterioration (image blur). That is, by dividing this signal by a value proportional to the noise level, when the noise level is large, that is, when the S / N ratio is low, noise removal is prioritized, and in the opposite case, image preservation is prioritized. To obtain the contour signal.

Returning to FIG. 1, based on the signal from the contour detecting circuit 2 obtained in this way, the subtracting circuit 5, the multiplying circuit 6 and the adding circuit 7 have
The ratio of the median filter output is increased, and the parts other than the contour of the image are mixed so that the ratio of the low pass filter output is increased. The operation of this mixing will be described below. When the low pass filter output signal is "L", the median filter output signal is "M", and the contour detection circuit output signal is "E", the signal obtained from the adding circuit 7 is (ML) * E + L. Becomes Here, when this equation is modified, it becomes M * E + L * (1-E) ... (1). Here, when 0 ≦ E ≦ 1, when the contour signal is large, that is, when E approaches 1, the equation (1) becomes M, that is, the ratio of the median filter output signal increases, and when E approaches 0. Is L, that is, the ratio of the low-pass filter output signal increases.

Here, regarding the median filter,
While there is a feature that noise can be removed without blurring the contour of the image, the noise to be removed is exclusively impulse noise and is not so effective against random noise. On the other hand, the low-pass filter has a great effect of removing random noise,
There is a property that the outline of an image is blurred. This 2
By combining the merits of the filters having different properties, it is possible to suppress the deterioration of image quality and increase the noise removal effect. That is, by increasing the ratio of the median filter output in the contour portion of the image, noise is removed while preventing blurring of the contour of the image although the noise removal effect is small. A large noise removal effect is obtained by increasing the ratio of the filter output.

FIG. 3 is a block diagram for explaining the configuration of the second embodiment of the noise reduction circuit according to the present invention. In FIG. 3, 23 is an input terminal for inputting an image signal, 24 and 31 are contour detection circuits, 25 and 32 are low pass filters, 26 and 33 are median filters, 27 and 3
Reference numeral 4 is a subtraction circuit, 28 and 35 are multiplication circuits, 29 and 36 are addition circuits, 30 is a noise level input terminal, and 37 is an output terminal.

The operations from the input terminal 23 to the adder circuit 29 and from the adder circuit 29 to the output terminal 37 are the first.
The description is omitted because it is the same as the embodiment.

As described above, by connecting the noise removing circuits having the structure of the first embodiment in two stages in series, a large noise removing effect can be obtained. Moreover, noise can be reduced over a wide frequency band by changing the characteristics of the low-pass filters 25 and 32 and the characteristics of the median filters 26 and 33, respectively.

FIG. 4 is a block diagram for explaining the configuration of the third embodiment of the noise reduction circuit according to the present invention. In FIG. 4, 38 is an input terminal for inputting an image signal, 39 and 46 are contour detection circuits, 40 and 47 are low-pass filters, 41 and 48 are median filters, and 42 and 4
9, 52 are subtraction circuits, 43, 50, 53 are multiplication circuits, 4
Reference numerals 4, 51 and 54 denote addition circuits, 45 denotes a noise level input terminal, and 55 denotes an output terminal.

Since the operation from the input terminal 38 to the adder circuit 51 is the same as that of the second embodiment, its explanation is omitted.
The subtraction circuit 52, the multiplication circuit 53, and the addition circuit 54 mix the output signal from the first-stage noise removal filter with the output signal from the second-stage noise removal filter in accordance with the noise level. . That is, when the noise level is high, the ratio of the output signal of the second-stage noise removal filter is increased to enhance the noise removal effect, and when the noise level is low, the ratio of the output signal of the first-stage noise removal filter is increased. Reduce the noise removal effect. As a result, it is possible to obtain a noise removing effect suitable for the noise level, and it is possible to suppress image quality deterioration to a minimum.

FIG. 5 is a block diagram for explaining the configuration of the fourth embodiment of the noise reduction circuit according to the present invention. In FIG. 5, 56 is an input terminal for inputting an image signal, 57 and 64 are contour detection circuits, 58 and 65 are low-pass filters, 59 and 66 are median filters, and 60 and 6
7, 70 and 73 are subtraction circuits, 61, 68 and 71 are multiplication circuits, 62, 69 and 72 are addition circuits, 63 is a noise level input terminal, 74 is an absolute value circuit, 75 is a comparison circuit, and 76 is a signal switching circuit. , 77 are output terminals.

The third from the input terminal 56 to the adder circuit 72 is
The description is omitted because it is the same as the embodiment. The subtractor circuit 73 takes the difference between the input signal of the first-stage noise removal filter and the output signal from the adder circuit 62, and this is converted into an absolute value by an absolute value circuit 74 (hereinafter, this signal is referred to as an error signal). .). Originally, it is desirable that the contour detection circuit 57 perfectly detects the contour, but in reality, it is difficult to perfectly detect the contour for every image. Therefore, the error signal is used to detect a mistake in the contour detection circuit. Since a mistake in the contour detection circuit is also affected by the noise level, the comparator circuit 75 produces a signal in a form that also takes noise level information into consideration. Since the error signal is the difference between the input and output signals of the first stage noise removal filter, the obtained signal becomes a noise signal, and its value should be a value proportional to the noise amount. Therefore, the error signal and the noise level are compared by the comparison circuit 75, and if there is a significant difference between the noise level and the error signal, it is determined that the difference is caused because the contour of the image is blunted due to a contour detection error. , "B" is selected by the signal switching circuit 76, and noise removal is not performed. In the comparison circuit 75, whether or not there is a significant difference between the noise level and the error signal is specifically determined as follows. The noise level signal is multiplied by an arbitrary constant obtained experimentally, and this is compared with the error signal. If the error signal is larger, there is a significant difference.

By this operation, even if an error occurs in contour detection, deterioration of image quality can be suppressed to a minimum.

[0028]

According to the noise reduction circuit of the first aspect, the proportion of the low-pass filter output signal is increased in the flat portion of the image, and the proportion of the median filter output signal is increased in the contour portion. As a result, noise can be removed in the contour portion of the image while maintaining the contour of the image, and a comprehensive noise reduction effect for the entire screen can be obtained without blurring the screen. As a result, it is possible to obtain a noise reduction effect equivalent to or higher than that of the frame recursive noise reduction method used for still images in moving images. It is possible to eliminate the difference between the S / N ratio of the moving picture.

According to the noise reduction circuit of the second aspect, the output of the noise reduction circuit of the first stage connected in cascade is vertically
The noise removal effect can be enhanced by using the input of the continuously connected second stage noise reduction circuit .

According to the noise reduction circuit of the third aspect, when the noise level signal is large, the ratio of the output of the noise reduction circuit of the second stage is increased, and when the noise level signal is small, it is increased by one stage. By increasing the output ratio of the noise reduction circuit of the eye and outputting it, it is possible to obtain the noise reduction effect that matches the noise level, and as a result, the noise removal effect that is greater than necessary Deterioration can be suppressed.

According to the noise reduction circuit of the fourth aspect, the absolute value of the difference between the input and the output of the first- stage noise reduction circuit is obtained, and the absolute value is compared with the noise level signal,
If there are significant differences between them are depending on the switching circuit
The input and the final output of the first- stage noise reduction circuit should be connected, and if there is no significant difference between the two, the switching circuit should connect the output of the second arithmetic circuit and the final output. By switching to, it is possible to reduce blurring of the image due to a mistake in the contour detection circuit.

[Brief description of drawings]

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

FIG. 2 is a block diagram showing a configuration of a contour detection circuit used in a noise reduction circuit according to the present invention.

FIG. 3 is a block diagram showing a configuration of a second embodiment of a noise reduction circuit according to the present invention.

FIG. 4 is a block diagram showing a configuration of a third embodiment of a noise reduction circuit according to the present invention.

FIG. 5 is a block diagram showing a configuration of a fourth embodiment of a noise reduction circuit according to the present invention.

FIG. 6 is a block diagram showing a configuration of a conventional noise reduction circuit.

FIG. 7 is a graph showing an input / output characteristic example of a non-linear circuit used in a conventional noise reduction circuit.

[Explanation of symbols]

1, 10, 23, 38, 56, 78 Image signal input terminal 2, 24, 31, 39, 46, 57, 64 Contour detection circuit 3, 25, 32, 40, 47, 58, 65, 79 Low pass filter 4,26,33,41,48,59,66 Median filter 5,17,27,34,42,49,52,60,6
7, 70, 73, 80 Subtraction circuit 6, 28, 33, 43, 50, 53, 61, 68, 71
Multiplier circuits 7, 29, 36, 44, 51, 54, 62, 69, 7
2,82 Adder circuit 8,19,30,45,63 Noise level input terminal 9,37,55,77,83 Output terminal 11,12,13,14 Pixel delay circuit 15,16 Addition averaging circuit 18,74 Absolute value Circuit 20 Coefficient circuit 21 Division circuit 22 Contour signal output terminal 75 Comparison circuit 76 Signal switching circuit 81 Non-linear processing circuit

─────────────────────────────────────────────────── ─── Continuation of front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) H04N 5/14-5/217

Claims (4)

(57) [Claims] 1. A noise reduction circuit for an image signal, comprising: a low-pass filter having an image signal as an input; a median filter having an image signal as an input; and an image signal and a noise level signal as inputs. A contour detection circuit that detects a contour portion as a contour signal and outputs a ratio of the contour signal to the noise level signal, and an output of the low-pass filter and an output of the median filter according to the output of the contour detection circuit. Is added while changing the respective ratios, and when the ratio of the contour signal to the noise level signal is large, the ratio of the output of the median filter is increased to increase the ratio of the contour signal to the noise level signal. And a first arithmetic circuit configured to increase the ratio of the output of the low-pass filter and to output the output. Noise reduction circuit of the signal. 2. A cascaded in two stages noise reduction circuit as claimed in claim 1, noise reduction at the first stage of the cascaded
Noise reduction circuit of the image signal with <br/> input of the second stage of the noise reduction circuit connected in cascade output of the circuit. 3. Connect the further second arithmetic circuit to the output of the noise reduction circuit according to claim 2, by the second arithmetic circuit, according to an output of said noise level signal, the vertical
The output of the first-stage noise reduction circuit connected in series and the cascade connection
The output of the connected noise reduction circuit of the second stage is added while changing the respective ratios, and when the noise level signal is large, the ratio of the output of the noise reduction circuit of the second stage is increased. When the noise level signal is small, the noise reduction circuit for the image signal is configured to increase the ratio of the output of the noise reduction circuit in the first stage and output the noise. 4. Connect the further third arithmetic circuit and a switching circuit to the noise reduction circuit according to claim 3, the arithmetic circuit of the third, input and output of the noise reduction circuit before Symbol 1-stage Then, the absolute value of the difference is calculated, the absolute value is compared with the noise level signal, and when there is a significant difference between the two,
So as to connect the input and final output of the noise reduction circuit before Symbol first stage by said switching circuit, also significant to both
If there is no difference between the second
An image signal noise reduction circuit that switches so as to connect the output of the arithmetic circuit and the final output.