JPH06121193A - Noise removing circuit - Google Patents

Abstract

PURPOSE:To improve a picture quality by effectively suppressing a noise mixture to an animation part at a picture changing point from the animation part to a still picture part. CONSTITUTION:In a noise removing circuit 10, a noise removing filter 13 for the animation part used when an input picture is the animation part, and a noise removing filter 40 for the still picture part used when the input picture is the still picture part are appropriately selected and used according to the movement of the input picture. At that time, the output of one of the noise removing filter 13 for the animation part and the noise removing filter 40 for the still picture part is selected based on the movement detected output of a movement detecting circuit 30 of the input picture. The selected filter output is supplied as an input signal to the noise removing filter 40, and the mean input picture value of the number (n) of the continuous still frames in inputted still picture elements is outputted from the filter 40.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a noise canceling circuit suitable for use as a process immediately before encoding an input signal.

[0002]

2. Description of the Related Art In many cases, a device (codec device) for compressing / decompressing an input image signal is provided with a noise removing means for suppressing noise immediately before an encoder provided in the data compression system. Among noise removal circuits used as a means for removing noise, a noise removal circuit by three-dimensional processing is known.

FIG. 10 shows a noise removing circuit 1 showing an example thereof.
In the system diagram of 0, for example, an input signal (image signal) which is digitally converted is supplied to the terminal 11. This input signal includes a noise removal filter 12 for a still image portion used when the input image having pixels constituting the input signal is a still image portion, and a moving image portion used when the input image is a moving image portion. Noise removal filter 13 for use in noise reduction processing in accordance with the contents of the input image in each of the filters 12 and 13, and the output of each noise-removed filter is supplied to the switching means 20. .

The input image is further supplied to the motion detection circuit 30, the motion is detected for each pixel of the input image, and the switching means 20 is adaptively selected by the motion control output according to the motion of the input image. Therefore, the pixels in the still image portion are subjected to noise removal by a temporal filter that performs filtering processing in the temporal direction, and the pixels in motion are subjected to noise removal using a spatial filter.

As described above, when the filter is adaptively selected according to the movement of the input image to remove the noise, the noise can be removed without deteriorating the image quality.

The noise removal filter 12 for the still image portion is a simple average value filter and has a frame memory 15. The memory output of the previous frame is multiplied by k (0
<K <1) and the input image of the current frame supplied to the input terminal 11 multiplied by (1-k) in the coefficient unit 17 are subjected to addition processing by the adder 18, and the addition result is It becomes the filter output.

FIG. 11 is a diagram for explaining the operation of the noise removal filter 12 for the still image portion. If random noise indicated by black circles is mixed in specific pixels in each frame, k = 0.
When the noise level is 5, noise levels a, b, c,
It can be seen that d is sequentially reduced by a power of 2 and output.

An intermediate value filter such as a median filter, which is a type of spatial filter, is used as the moving image noise removal filter 13. The intermediate value filter uses the level of a pixel having an intermediate level between two pixels before and after including the pixel of interest as the level of the pixel of interest.
FIG. 2 is a diagram for explaining the operation of this filter.

When the noise n is mixed in the target pixel b (the original level of the pixel is also b) with respect to the input image change like the curve p, the intermediate value of a, n and c is in the target pixel. Selected as output pixel level. Therefore, c is selected. Since n <d in the next pixel,
At this time, the noise n is selected for the first time. Since this noise level is smaller than d, even if the noise n is selected, it is not so noticeable because it does not project extremely from the surrounding pixels.

Incidentally, the pixel level as shown in FIG. 12 is obtained at the normal output without a filter, and the noise n is selected at the low level, so that the noise level is prominent and the selected noise n becomes conspicuous. .

FIG. 13 illustrates the motion detection operation of the input image. In the figure, the surrounding 3 × 3 pixels including the target pixel e in the current frame (n frame) and the previous frame (n−1).
The difference (aa ', b- at the same pixel position in the frame)
b ′, ... i−i ′), and the sum S thereof is obtained.

The value of the sum S is compared with a predetermined threshold value. When the sum S is larger than the threshold value, the pixel of interest is judged to be a moving pixel (moving image part), and when it is smaller than the threshold value, it is regarded as a still pixel (still image part). To be judged.

[0013]

By the way, in the conventional noise removing circuit configured as described above, since the time filter (average value filter) is used as the noise removing filter 12 for the still image portion, it is averaged. Output will remain until the end.

For example, when noise (noise level is a) is mixed in the pixel 1 in the nth frame as shown in FIG. 11, the noise component does not become completely zero after several frames, and its influence remains for a long time. The afterimage phenomenon does not disappear immediately.

Since the influence of the noise mixed immediately before on the next frame is as large as 1/2, the image quality when moving from the moving image portion to the still image portion is not so effectively improved.

The noise removal filter 12 for the still image portion
The input signal to is using the input signal itself supplied to the input terminal. Therefore, particularly when switching from the noise removal filter for still image section 12 to the noise removal filter for moving image section 13, the input signal to this filter 13 uses the input signal from the terminal 11 in which noise is not suppressed at all. Therefore, it is the actual situation that the noise removing effect cannot be sufficiently obtained particularly when the image is switched from the moving image portion to the still image portion.

Therefore, the present invention solves such a conventional problem, and proposes a noise removing circuit which can effectively improve the image quality at the change point of the image from the moving image portion to the still image portion. Is.

[0018]

In order to solve the above-mentioned problems, in the invention according to claim 1, a noise elimination filter for a moving image portion used when the input image is a moving image portion, and the input image is static. In the noise removal circuit, in which the noise removal filter for the still image section used when the image section is used is adaptively selected and used according to the movement of the input image,
A motion detection circuit for an input image is provided, and either the moving image noise removal filter or the still image noise removal filter output is selected based on the motion detection output, and the still image noise removal is performed. The selected filter output is supplied to the filter as an input signal, and the noise removal filter for the still image portion outputs the average input image value of the continuous still frame number n in the input still pixels. It is characterized by.

According to a second aspect of the present invention, a noise removal filter for a moving image portion used when the input image is a moving image portion, and a still image portion used when the input image is a still image portion. A noise removal circuit, in which a noise removal filter is adaptively selected and used according to the motion of the input image, a motion detection circuit for the input image is provided, and the noise removal filter for the moving image section is based on the motion detection output. The output of the noise removal filter for the still image portion is adaptively mixed and output, and the output of the noise removal filter for the still image portion is selectively supplied to the noise removal filter for the moving image portion, and The number of continuous still frames in the still pixels input from the noise removal filter for the image part
It is characterized in that the averaged input image information determined by the above is output.

[0020]

The embodiment shown in FIG. 6 will be described. Based on the motion information from the motion detection circuit 30, the stationary counter 50
In, the weighting factors 1 / n and (n-1) / n are calculated from the number n of continuous still frames (see FIG. 2) at still pixels and the number n of continuous still frames. These weighting factors are used in the still image portion.

That is, the noise removal filter 4 for the still image portion
The frame memory 41 and the multiplier 42 provided at 0 calculate the final output signal of the previous frame that has been subjected to noise averaging processing in the time direction determined by the number n of continuous still frames in still pixels, and The input signal of the current frame is weighted by n in the multiplier 43,
The addition signal of the output signal of the current frame and the output signal of the previous frame is used as the output of the noise removal filter 40 for the still image portion.

In this way, at the transition point from the moving image portion to the still image portion, since n = 1, only the signal of the current frame is output from the adder, so that a filter output with no afterimage can be obtained, and similarly the still image can be obtained. Since n = 1 even at the transition point from the image portion to the moving image portion, only the signal of the current frame is output from the adder. Therefore, the image quality does not deteriorate in the moving part of the image.

In other cases, the output signal of the previous frame is obtained by weighting with n, so that an output signal in which noise is suppressed is obtained.

Further, since the input to the frame memory 41 is the output of the switching means 20, it is the final output signal of the previous frame that has undergone the spatio-temporal processing. Therefore, especially from the moving image portion to the still image portion. At the change point, noise suppressed by the moving image noise removal filter 41 is input to the frame memory 41.

As a result, the noise suppressing effect at the changing point is exhibited, and the deterioration of the image quality due to the noise at the changing point is improved.

[0026]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, a case where an example of a noise removing circuit according to the present invention is applied to a data compression system of a codec device will be described in detail with reference to the drawings.

FIGS. 1 to 5 are diagrams for explaining the noise removing circuit 10 according to the present invention. For convenience of explanation, the explanation will be started from these first.

In the noise removing circuit 10 shown in FIG.
The input signal input to the terminal 11 is supplied to the noise removal filter 13 for the moving image portion and the noise removal filter 40 for the still image portion. Either of the filter outputs is selected by the switching means 20 and is output to the output terminal 21.

As the noise removing filter 13 for the moving image part, a median filter which is a kind of spatial filter can be used as in the conventional case. A processing example of the median filter is shown in FIG. The example of FIG. 2 shows a case where noises n1 and n2 are continuously mixed in the input signal (waveform is p).

Reference numeral 30 denotes a motion detection circuit, a concrete example of which will be described later. The motion detection output obtained from the motion detection circuit is supplied to the switching means 20 and is also supplied to the stationary counter 50 so that continuous stationary frames for stationary pixels are obtained. The number n is detected.

The number n of continuous still frames will be described with reference to FIG. When M is a moving pixel and S is a still pixel as shown in the figure, when the target pixel is determined to be a moving pixel, n = 1 is always set, and the change point from the moving image portion to the still image portion and the still image portion are The change point from the image portion to the moving image portion is also set to n = 1, and in the still image portion, n is sequentially accumulated according to the number of continuous still frames. The figure shows the n value when the pixel of interest moves over three frames and the next three frames are stationary.

The still counter 50 calculates weighting factors 1 / n and (n-1) / n from the number n of continuous still frames, and these weighting factors are supplied to the still image section filter 40.

The still image section filter 40 has a frame memory 41 for averaging as shown in the figure, and a frame output (an output signal in which the noise of the previous frame is suppressed) is supplied to the first multiplier 42. It is multiplied by the weighting factor (n-1) / n.

The input signal of the current frame supplied to the terminal 11 is multiplied by the weighting coefficient 1 / n in the second multiplier 43, and the sum of these two multiplication outputs in the adder 44 is the final value of the filter 40. Used as a visual output. The addition output is also input to the frame memory 41. The input to the frame memory 41 is

[0035]

[Equation 1]

As shown in FIG. 2, since the number of frames averaged is different from the number of continuous still frames n, the noise removal effect increases as n increases, and the change point from the moving image portion to the still image portion increases. In such places, n = 1, so
Since the output of the first multiplier 42 becomes zero and the input signal itself is output from the filter 40, an afterimage does not occur.

Further, since the output of the adder 44 is cyclically input to the frame memory 41, when the input image changes from a moving image to a still image, the image information of the moving image portion is sequentially circulated. Since they are averaged, the influence of the moving image portion on the still image is reduced over time, and the noise removal effect is exerted over time.

FIG. 4 shows a concrete example of the motion detection circuit 30.
The motion detection shown in the figure shows a case where the threshold value for motion detection is adaptively changed at the edge portion and the flat portion of the image.

Therefore, the input signal is supplied to the edge detection circuit 31 as shown in the figure. In order to detect the edge, in this example, 3 including the target pixel set to the coefficient value as shown in FIG.
A Laplacian filter having a unit of detection block of × 3 is used. Then, the sum of the values obtained by multiplying the corresponding input pixels by the respective Laplacian coefficients (either -1 or 8) is used as the detection value (activity) L.

Describing with reference to the 3.times.3 unit block (nth frame) shown in FIG. 13, the detection value L is L = 1/1 / when i to i are the input levels in the respective pixels. 8 {8e- (a + b + c + d + f + g + h + i)} ... (4)

The detected value S is compared with a predetermined threshold value β, and when L ≧ β ... Edge part L <β ... Based on the comparison output CA which is the determination result, a threshold value α (two threshold values α
1 or α2 (α1> α2)) is selected.

In this example, as shown in FIG.
2 is provided, and two threshold values α1 and α2 are stored therein. For example, when it is determined that it is an edge portion, the large threshold value α1 is selected.

The movement of the input signal is determined by the threshold value α selected by the comparison output CA. That is, in the motion detecting unit 33, the following determination process is performed by using the two thresholds α1 and α
2 will be performed.

(1) When it is determined that L <β (flat portion) S ≧ α1 ... Moving image portion S <α1 ... Still portion (2) It is determined that L ≧ β (edge portion) When S ≧ α2 ... Video portion S <α2 ... Still portion As a result, when the input image is flat, that is, the presence of noise is noticeable inside the edge, in this case (1) Accordingly, the motion is detected using the larger threshold value α1.

According to the condition (1), the still image portion is more likely to be determined than the moving image portion. Therefore, in this case, the still portion filter 40 is selected. Since the noise for the stationary portion filter 40 is more suppressed than that for the moving image portion noise removal filter 13, the noise removal effect in the flat portion becomes greater.

In the edge portion, the one α2 having the smaller threshold value is used in accordance with (2), and therefore, in this case, it is easy to determine that the pixel is moving. Therefore, although the effect of removing noise at the edge portion is somewhat weakened, noise is not so conspicuous at the edge portion, and therefore the effect of setting a small threshold value is small.

If the still image filter 40 is used at the edge portion, the image is blurred because it is a temporal filter, which is not preferable. According to the above, the probability that the edge portion is determined to be a moving image is high, so that the still image portion noise removal filter 40 is hard to be selected, and the image is not blurred.

By adaptively changing the threshold value for motion detection according to the edge information in this way, the noise removal effect in the flat portion is enhanced, and the image quality deterioration in the edge portion is reduced. The values of α1, α2 and β described above can be appropriately selected according to the application. Instead of the ROM 32, the reference values of α1 and α2 may be selected.
The number of thresholds used is arbitrary.

FIG. 6 shows a noise removing circuit 10 according to the present invention.
A specific example of Since its basic configuration is the same as that of FIG. 1, its detailed description will be omitted. In the case of FIG. 1, the input to the frame memory 41 is the output itself of the adder 44, but in FIG. 6, the output signal obtained at the output terminal 21 is used.

The output signal obtained at the output terminal 21 is an output signal subjected to spatial noise processing by the moving image noise removal filter 13 or an output signal subjected to noise processing in the temporal direction by the still image filter 40. Therefore, each is an output signal of the previous frame in which noise is sufficiently suppressed. When the output signal obtained at the output terminal 21 is expressed by an equation,

[0051]

[Equation 2]

Therefore, the first and second multipliers 4
2, 43, an adder 44, and a frame memory 41 are used as a filter 40, and the output signal of which the noise has been sufficiently suppressed is used as an input, and the output signal of the previous frame weighted by the number of continuous still frames n The input signal of the frame is added with the input signal of the current frame weighted by the number n of continuous still frames, and the sum is output as the filter.

Therefore, when noise is mixed in the immediately preceding moving image portion at the time of switching from the moving image portion to the still image portion, the filter output in which the noise is suppressed by the filter 40 can be used as the output signal, and therefore, the moving image portion. Figure 1 shows the noise reduction effect at the transition point from the image to the still image part.
Greater than.

FIG. 7 shows another example of the present invention. In FIG. 1 or 6, the input signal itself of the current frame is used as the input signal of the noise removing filter 13 for the moving image portion. On the other hand, the noise removal circuit 10 according to the present invention shown in FIG. 7 uses the filter output of the still image section filter 40 as the input signal of the moving image section noise removal filter 13.

In this way, the noise-removing filter 13 for the moving image section receives the filter output that has been subjected to noise-removing processing in the time direction. The noise removal processing effect in the moving part can be enhanced more than in the case. That is, the same effect as that of FIG. 6 is obtained.

FIG. 8 shows a modified example of FIG. 7, in which a switching means 60 is provided in the preceding stage of the noise removing filter 13 for the moving image part, and the input signal of the current frame and the output signal of the previous frame are adaptive according to the movement. It can also be configured to switch to.

The input signal of the current frame is selected when it is the moving image portion, and the output signal of the previous frame is selected when it is the still image portion. In the still image portion, noise removal processing is performed based on the output signal of the previous frame in which noise is sufficiently suppressed, so effective noise removal is possible, and the output is further input to the frame memory 41 in a cyclic manner. Since it is configured by the formula, the noise suppression effect is great.

In each of the above-described embodiments, when either the noise-removed filter output in the time direction or the noise-removed filter output in the space is adaptively selected according to the motion detection output. Is.

In FIG. 9, one of the two filter outputs is not selected, but is used by adaptively mixing according to the motion. Therefore, in terms of hardware, the mixer 70 is provided instead of the switching means 20, and the mixing ratio thereof is controlled according to the movement amount.

The motion amount of the input image is mq, the filter output when the median filter is used as the moving image noise removal filter 13 is dm (k), and the output of the filter 40 (output of the adder 41) is ds (k). Then, the final output signal or the like obtained at the output terminal 21 can be expressed as follows.

[0061]

[Equation 3]

By adaptively controlling the mixture ratio of both filter outputs in accordance with the amount of movement in this way, it is possible to improve image quality deterioration at the time of switching between the moving image portion and the still image portion.

The mixing coefficients mq, (1-mq) may not be controlled linearly according to the amount of movement, but several kinds of mixing coefficients may be prepared and adaptively selected.

The above-described embodiments can be configured by arbitrarily combining with each other, and in that case, noise removal in the time direction and space may be more effectively performed.
Motion detection can also be performed using a single threshold as in the conventional case.

[0065]

As described above, in the noise removing circuit according to the present invention, the filter output in the time direction and the filter output in the space are adaptively selected according to the movement of the input signal.

According to this, it is possible to effectively suppress the noise mixed in both the moving image and the still image, so that it is possible to generate an output signal with good image quality.

In particular, the final output signal obtained at the output terminal is used as the input of the still image noise removal filter, or the output of the still image noise removal filter is selectively used as the input of the moving image noise removal filter. In this case, the noise can be sufficiently suppressed even when noise is mixed in the moving image portion at the changing point from the moving image portion to the still image portion, so the image quality at the image changing point can be improved. There is.

[Brief description of drawings]

FIG. 1 is a system diagram showing an example of a noise removal circuit used for explaining the present invention.

FIG. 2 is an explanatory diagram of filter processing when a median filter is used as a spatial filter.

FIG. 3 is an explanatory diagram of the number of continuous still frames in a specific pixel.

FIG. 4 is a system diagram of a motion detection circuit.

FIG. 5 is an explanatory diagram of a Laplacian filter for edge detection.

FIG. 6 is a system diagram of a noise removal circuit according to the present invention.

FIG. 7 is a system diagram of a noise removal circuit similar to that of FIG.

FIG. 8 is a system diagram of a noise removal circuit similar to FIG.

FIG. 9 is a system diagram of a noise removal circuit.

FIG. 10 is a system diagram of a conventional noise removal circuit.

FIG. 11 is an operation explanatory diagram of a filter for a still image portion.

FIG. 12 is an explanatory diagram of the operation of the moving image filter.

FIG. 13 is an explanatory diagram for motion detection.

[Explanation of symbols]

 10 Noise Removal Circuit 12, 40 Still Image Part Filter 13 Video Part Filter 20, 60 Switching Means 30 Noise Detection Circuit 31 Edge Detection Circuit 32 Threshold ROM 33 Motion Detection Unit 41 Frame Memory 42, 43 Multiplier 44 Adder 50 Motion counter 70 mixer

─────────────────────────────────────────────────── ───

[Procedure amendment]

[Submission date] October 1, 1993

[Procedure Amendment 1]

[Document name to be amended] Statement

[Correction target item name] 0041

[Correction method] Change

[Correction content]

The detected value L is compared with a predetermined threshold value β, and when L ≧ β, the edge portion L <β, it is determined that the portion is a flat portion. Based on the comparison output CA which is the determination result, a threshold value α (two threshold values α
1 or α2 (α1> α2)) is selected.

[Procedure Amendment 2]

[Document name to be amended] Statement

[Correction target item name] 0042

[Correction method] Change

[Correction content]

In this example, as shown in FIG.
2 is provided, and two threshold values α1 and α2 are stored therein. For example, when it is determined to be an edge portion, the small threshold value α2 is selected.

[Procedure 3]

[Document name to be amended] Statement

[Correction target item name] 0051

[Correction method] Change

[Correction content]

[0051]

[Equation 2]

[Procedure amendment 4]

[Document name to be amended] Statement

[Correction target item name] 0060

[Correction method] Change

[Correction content]

The motion amount of the input image is mq, the filter output when a median filter is used as the moving image noise removal filter 13 is dm (k), and the output of the filter 40 (output of the adder 44) is ds (k). Then, the final output signal or the like obtained at the output terminal 21 can be expressed as follows.

[Procedure Amendment 5]

[Document name to be amended] Statement

[Correction target item name] 0061

[Correction method] Change

[Correction content]

[0061]

[Equation 3]

[Procedure correction 6]

[Document name to be corrected] Drawing

[Name of item to be corrected] Figure 1

[Correction method] Change

[Correction content]

[Figure 1]

[Procedure Amendment 7]

[Document name to be corrected] Drawing

[Name of item to be corrected] Figure 2

[Correction method] Change

[Correction content]

[Fig. 2]

[Procedure Amendment 8]

[Document name to be corrected] Drawing

[Correction target item name] Figure 9

[Correction method] Change

[Correction content]

[Figure 9]

Claims (3)

[Claims] 1. A moving image noise removal filter used when the input image is a moving image portion, and a still image noise removal filter used when the input image is a still image portion A noise removal circuit that is adaptively selected and used according to the motion, is provided with a motion detection circuit for the input image, and based on the motion detection output, the noise removal filter for the moving image part and the noise removal filter for the still image part. Output of the selected image is supplied to the noise removal filter for still image section as an input signal. A noise removal circuit characterized in that an average input image value of the number of still frames n is output. 2. A noise removal filter for a moving image portion used when the input image is a moving image portion and a noise removal filter for a still image portion used when the input image is a still image portion A noise removal circuit that is adaptively selected and used according to the motion, is provided with a motion detection circuit for the input image, and based on the motion detection output, the noise removal filter for the moving image part and the noise removal filter for the still image part. The output of the noise removal filter for the moving image portion is selectively supplied to the noise removal filter for the moving image portion, and the output of the noise removal filter for the still image portion is selectively supplied from the noise removal filter for the still image portion. A noise removal circuit characterized in that averaged input image information determined by the number n of continuous still frames in input still pixels is output. 3. The noise removal filter for a still image part, wherein the first multiplication means for multiplying an input image by 1 / n, and the output of the average value memory and (n-1) / n are multiplied. 3. The noise removing circuit according to claim 1 or 2, wherein the noise removing circuit comprises two multiplying means and an adding means, and the output of the adding means is input to the average value memory.