JP2557274B2 - Image signal noise reduction device - Google Patents

Description

The present invention relates to an image signal noise reduction device. Specifically, an effective signal can be obtained by obtaining a more accurate judgment as to whether the input image signal belongs to a still area or a moving area, and performing noise processing with characteristics suitable for the result of the judgment. It is an object of the present invention to provide an image signal noise reduction device capable of reducing the distortion of the image signal.

[Prior Art] Conventionally, as a method for reducing noise included in an input image signal without impairing faithful reproducibility of the image, the input image signal belongs to a static region or a moving region. It has been proposed that the noise be processed with a characteristic according to each determination result after determining whether or not (motion determination).

As an example, not only the inter-frame difference value, which is the amount of change in the pixel that is the target of noise processing, but also the difference value of its neighboring pixels are incorporated into the evaluation value for motion determination, and the target image is determined. There has been a method of determining the noise removal characteristic from the result of the determination (conventional example 1, JP-A-1-143583). As the evaluation value for the motion determination here,
The absolute value of the inter-frame difference value is used, and the motion determination for each pixel is obtained from the absolute value of the inter-frame difference value of the target pixel and the absolute value of each inter-frame difference value of its neighboring pixels, and each determination is performed. The noise reduction characteristics are continuously changed according to the result.

As another example, determine the center point for noise reduction, compare all absolute values of the inter-field difference values at multiple points near the center point with a certain value, and then compare the results. From the center point, the noise reduction amount is increased if it is determined to be a static part,
In some cases, the noise reduction amount was reduced if it was determined to be a moving part (Prior art example 2, Japanese Patent Laid-Open No. 2-7773).

[Problems to be Solved by the Invention] According to the conventional example 1 and the conventional example 2 described above, in the case of performing the motion determination using the inter-frame difference value of only the pixel to be the noise processing target or each pixel forming the target block. A more accurate determination can be obtained, and the possibility of erroneously determining noise with a large amplitude in the static region as belonging to the moving region is reduced.

However, even if these conventional examples are used, when the noise component is large, there is a possibility that the motion determination of the input image signal is not always accurately performed. For example, in Conventional Example 1, the inter-frame difference value of the pixel to be processed is X _0, and the inter-frame difference value of n neighboring pixels is
x ₁ , x ₂ , x ₃ , ..., x _n , x _s1
= | Than a threshold which varies according to the size of the _{x s1 | x 0 | + |} x 1 | + | x 2 | + | x 3 | + ... + | x n | a predetermined threshold or | x ₀ When it is small, it is determined that the target pixel belongs to the still area. However, in the case of noise with many high frequency components,
Generally, each of the above-mentioned difference values x ₁ , x ₂ , x ₃ , ..., X _n includes positive and negative values. Therefore, if the absolute value of each such difference value is added as an evaluation value, the threshold value used for the motion determination must be set relatively large. As a result, it is determined that the motion signal having a small difference value belongs to the still region, and the signal is processed with a strong noise elimination characteristic. Therefore, the conventional example 1 has a problem to be solved in that the degree of distortion of the effective motion signal is increased, and the image is unnatural such as an afterimage.

Further, in Conventional Example 2, all of the absolute values of the inter-field difference values x ₁ , x ₂ , x ₃ , ..., X _n at a plurality of points near the center point where noise reduction is to be performed are all above a predetermined threshold value. When it is small, that is, when x _s2 = max (| x ₁ |, | x ₂ |, ..., | x _n |) × n is smaller than a predetermined threshold, it is determined that the center point belongs to the stationary region. ing. Therefore, it is necessary to set the threshold value to a relatively large value in order to deal with noise having many high frequency components. Therefore, the conventional example 2 also has an unsolved problem similar to that described in the conventional example 1.

[Means for Solving the Problem] The present invention has been made in view of the problem to be solved, and the target image and its vicinity are used as the evaluation value for the motion determination of the image to be subjected to the noise processing. The absolute value of the added value of the inter-frame difference values of pixels or the absolute value of the added value of the inter-frame difference values of only neighboring pixels other than the target pixel is obtained, and noise removal is performed using the threshold value corresponding to the obtained evaluation value. I decided to determine the characteristics.

In addition, the inter-frame difference value of the pixels that have already undergone the noise removal processing is included in the evaluation value calculation element.

[Operation] By using such a means, it becomes possible to set a smaller threshold value for motion determination regarding a pixel to be subjected to noise processing, and a difference value between frames is a very small motion signal. Since an accurate judgment can be obtained, the distortion of the effective signal is reduced and a more natural image can be obtained.

[Embodiment] A circuit configuration of an embodiment of the present invention will be described with reference to FIG. 1 and an arrangement example of pixels to be processed by the circuit shown in FIG.

Many circles in FIG. 3 indicate each pixel, and P0 indicated by a bold circle indicates a pixel to be noise-processed, and a pixel P0
P1, P2, P3, and P4 indicated by the bold circles around P are pixels used for noise processing in the pixel P0.

In FIG. 1, the subtracter 12 receives a signal 31 indicating the image of the current frame input from the input terminal 11,
Signal 41 from memory 21 showing the image of the previous frame
The difference value for each pixel is calculated. The signal 54 indicating the obtained difference value indicates the difference value of the pixel P4, which is
Memory 13 and evaluation value calculation circuit 17 for obtaining motion determination
Is input to The line memory 13 delays the input signal 54 indicating the difference value by a time shorter than that of one line by one pixel, and outputs the signal 54. The signal 53 indicating the difference value delayed by the line memory 13 indicates the difference value of the pixel P3, which is input to the pixel memory 14 and the evaluation value calculation circuit 17. The pixel memory 14 outputs the input signal 53 with a time delay of one pixel. The signal 50 indicating the difference value delayed by the pixel memory 14 indicates the difference value of the pixel P0 which is the current noise processing target, and this is further input to the pixel memory 15, the evaluation value calculation circuit 17 and the noise removal circuit 18. To be done. Pixel memory
Reference numeral 15 outputs the input signal 50 with a time delay of one pixel, similarly to the pixel memory 14. The signal 52 indicating the difference value delayed by the pixel memory 15 indicates the difference value of the pixel P2, which is input to the line memory 16 and the evaluation value calculation circuit 17. The line memory 16 delays the input signal 52 by a time shorter than that of one line by one pixel, similarly to the line memory 13, and outputs the signal. Line memory 16
The signal 51 indicating the difference value delayed by indicates the difference value of the pixel P1, which is input to the evaluation value calculation circuit 17.

In this way, the evaluation value calculation circuit 17 is supplied with the respective signals 50, 51, 52, 53, 54 indicating the inter-frame difference values for the pixels P0 to P4 obtained by the subtractor 12. Therefore, the evaluation value calculation circuit 17, from each input signal 50, 51, 52, 53, 54, the evaluation value for determining whether the pixel P0 targeted for noise processing belongs to a still area or a moving area. Gather together. The absolute value of the sum of the difference values indicated by the signals 50, 51, 52, 53, 54 is used as the evaluation value.

That is, the difference value indicated by the signal 51 corresponding to the pixel P1 is
x ₁ , the difference value x ₂ indicated by the signal 52 corresponding to the pixel P ₂ , the difference value indicated by the signal 50 corresponding to the pixel P ₀ is x ₀ , the difference value indicated by the signal 53 corresponding to the pixel P ₃ is x ₃ , the pixel P 4 is indicated. Assuming that the difference value indicated by the corresponding signal 54 is x ₄ , the absolute value x _s = | x ₀ + x ₁ + x ₂ + x ₃ + x ₄ | When x _s is obtained, it is compared with a known predetermined threshold value or a threshold value that changes according to the values of x _s and | x ₀ |, and the motion determination regarding the target pixel P0 is performed based on the comparison result. The signal 37 indicating the obtained determination result is the noise removal circuit 18
Sent to

Therefore, the noise elimination circuit 18 performs noise elimination processing on the input signal 50 indicating the difference value of the target pixel P0 with the characteristic according to the determination result. The signal 38 indicating the difference value of the pixel P0 thus subjected to the noise processing is input to the adder 20, and the line 41 obtained by delaying the signal 41 indicating the image of the immediately preceding frame from the frame memory 21 for a predetermined time. It is added to the signal 39 from the memory 19. The signal 40 indicating the added value is
Output from the output terminal 22 and also the frame memory
It is sent to 21 and stored. The delay time of the line memory 19 is one line. The delay time of the signal 50 that indicates the difference value of the pixel to be processed by the noise processing is calculated by the line memory.
A total of one line is provided by passing through 13 and the pixel memory 14. This is because the delay time by the evaluation value calculation circuit 17 and the noise removal circuit 18 is generally small enough to be ignored. Further, since the signal 40 from the adder 20 is delayed by one line, the delay time of the frame memory 21 is set to be shorter than one frame by one line.

As described above, in the present invention, the pixel P0 targeted for noise processing is
As the evaluation value for obtaining the motion determination regarding, the difference value x ₀ of the target pixel P0 and each difference value x _{1 to} x _{4 of the} neighboring pixels P1 to P4
The absolute value of the sum of and is used. That is, noise having many high frequency components generally includes positive and negative values of the difference values x _{0 to} x ₄ . Therefore, where _{| x 0 + x 1 + x} 2 + x 3 + x 4 | <| x 0 | + | x 1 | + | x 2 | + | x 3 | + | x 4 | That, or x _s <x _s1 | x ₁ + x ₂ + x ₃ + x ₄ | <max (| x ₁ |, | x ₂ |, | x ₃ |, | x ₄ |) × 4 That is, x _s0 = | x ₁ + x ₂ + x ₃ + x ₄ | Then, x _s0 <x _s2 holds.

Therefore, if the absolute value of the sum of the difference values x ₀ ~ x ₄ or x ₁ ~ x ₄ is the evaluation value for the motion determination, the sum of the absolute values of the difference values as in the conventional example, that is, Compared with the case where x _s1 (conventional example 1) or x _s2 (conventional example 2) is used as the evaluation value, the threshold value of the motion determination can be made smaller.

Since a value corresponding to the evaluation value will be set as this threshold, a smaller threshold than before can be set, and an effective motion signal with a small difference value is erroneously determined to belong to the stationary area. Therefore, it is possible to avoid processing with a strong noise removal characteristic. As a result, small effective motion signal distortions are reduced and more natural images can be obtained.

FIG. 2 shows a circuit configuration of another embodiment of the present invention, and the components corresponding to those in FIG. 1 will be described with the same symbols.

In FIG. 2, only the points different from the embodiment shown in FIG. 1 will be described. To the evaluation value calculation circuit 17, the difference values that have been subjected to the noise removal processing by the noise removal circuit 18 are input as signals 52B and 51B via the pixel memory 15 and the line memory 16, respectively. Therefore, each signal 50,5
The difference values indicated by 3,54 are x ₀ , x ₃ , and x ₄ , respectively, and each signal 51B,
52B represents the noise-processed difference value x ₁ ′,
Assuming x ₂ ′, the evaluation value calculation circuit 17 obtains | x ₀ + x ₁ ′ + x ₂ ′ + x ₃ + x ₄ | as the evaluation value for motion determination. In this case, if the noise has many high frequency components,
In general, since each difference value x ₀ + x ₁ ′ + x ₂ ′ + x ₃ + x ₄ includes a positive value and a negative _value , | x ₀ + x ₁ ′ is the same as described in the embodiment shown in FIG. ＋ x ₂ ′ ＋ x ₃ ＋ x ₄ | ＝ x _s ′, x _s ′ <x _s1 or | x ₁ ′ ＋ x ₂ ′ ＋ x ₃ ＋ x ₄ | ＝ x _s0 ′, x _s0 ′ <x _s2 Holds.

Therefore, according to the present embodiment as well, since a value corresponding to the evaluation value is set as the threshold value used for the motion determination regarding the pixel for which noise is to be removed, the threshold value is set to be smaller than that of the conventional examples 1 and 2. Is possible
The same effect as that of the embodiment shown in FIG. 1 can be obtained.

The line memories 13 and 16, the pixel memories 14 and 15, the evaluation value calculation circuit 17 and the noise elimination circuit 18 shown in FIGS.
The noise reduction circuit consisting of can be replaced and applied to the conventional examples 1 and 2 as it is.

Note that, in the above, the case where the pixel P0 that performs noise processing and the four neighboring pixels P1 to P4 are used as the elements of the evaluation value calculation for the motion determination is described as an example, but the line memory and the pixel to be arranged are arranged. By changing the number of memories, a larger number of neighboring pixels corresponding to that number can be used as an element for calculating the evaluation value, or even if the pixel to be processed is excluded and the evaluation value is calculated, The invention can be applied.

[Effects of the Invention] As is clear from the above description, according to the present invention, the threshold value used for the motion determination for the pixel that is the target of the noise processing can be set to a smaller value. It is less likely that a minute motion signal is erroneously determined to belong to a stationary area and is processed with a strong noise removal characteristic. Therefore, effective signal distortion is reduced and a more natural image is obtained, and the effect of the present invention is extremely large.

[Brief description of drawings]

FIG. 1 is a circuit configuration diagram of an embodiment of the present invention, FIG. 2 is a circuit configuration diagram of another embodiment of the present invention, and FIG. 3 is an evaluation value calculation circuit by the evaluation value calculation circuit shown in FIG. FIG. 6 is a pixel position diagram for explaining a positional relationship of pixels that are elements. 11 …… Input terminal, 12 …… Subtractor 13,16 …… Line memory 14,15 …… Pixel memory 17 …… Evaluation value calculation circuit, 18 …… Noise elimination circuit 19 …… Line memory, 20 …… Adder 21 …… Frame memory 22 …… Output terminals 31,37 to 41,50 to 54 …… Signal.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-1-143583 (JP, A)

Claims (2)

(57) [Claims] 1. An image signal of an input current frame and 1
Subtracting means (12) for obtaining a difference value for each pixel from the image signal of the immediately preceding frame, and a pixel (P
0) and the peripheral means (13 to 16) for obtaining the difference value for each pixel of the peripheral pixels, the difference value for each pixel from the subtraction means, and at least the peripheral pixels from the memory means. In order to determine the noise removal characteristic of the pixel (P0) to be subjected to the noise processing from the threshold value corresponding to the evaluation value, the absolute value of the added value of the difference values of is calculated as the evaluation value. An image signal including: an evaluation value calculating means (17) and a noise removing means (18) for removing noise included in the difference value for each pixel with the noise removal characteristics determined by the evaluation value calculating means Noise reduction device. 2. The input image signal of the current frame and 1
Subtracting means (12) for obtaining a difference value for each pixel from the image signal of the immediately preceding frame, and a part of the pixels to be subjected to the noise processing from the output of the subtracting means and the peripheral pixels First memory means (13, 14) for obtaining a difference value for each pixel, and for each pixel obtained by the subtracting means according to the noise removal characteristic determined corresponding to the evaluation value for determining the noise removal characteristic Noise removing means (18) for removing noise included in the difference value, and the noise of pixels other than the part of the pixels to be subjected to noise processing from the output of the subtracting means and the peripheral pixels. Second memory means (15, 16) for obtaining a difference value from which noise has been removed by the removing means; a difference value for at least the peripheral pixels from the subtracting means and the first memory means; Noise from two memory means The absolute value of the sum of the removed difference value,
Evaluation value calculation means for calculating the noise removal characteristic of the noise removal means, and an evaluation value calculation means for determining the noise removal characteristic of the pixel (P0) to be subjected to the noise processing from a threshold value corresponding to the evaluation value. (17) A noise reduction device for image signals, including and.