JP3286120B2 - Noise removal 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 processing apparatus and the like for removing noise (noise) from an input image by using a median filter (order filter), which is a kind of nonlinear digital filter, to improve image quality. Related to a noise removal circuit.

[0002]

2. Description of the Related Art Heretofore, as a technique related to this type of median filter, there has been a technique described in the following literature, for example. Literature: NHK Giken R & D, [7] (1989-11),
Kobayashi et al. "Nonlinear Digital Filter for Image-LSI"
P. 27-36 As described in the above-mentioned literature, an image passing through a transmission line like a television signal (hereinafter, referred to as a TV signal) includes, for example, a salient state seen in a BS reception image at the time of rainfall. Impulse noise may be mixed for various reasons, such as noise or burst noise in a VTR (video) reproduced image. In particular, since a TV signal requires real-time processing, it is difficult to remove the above-mentioned noise without deteriorating the resolution of the original image only by a linear operation using a digital filter.
Therefore, in the above document, a median filter (sequential filter), which is a kind of nonlinear digital filter, is used as an optimal method for removing the impulse noise.
We propose a noise elimination technology using. In general, a digital filter is a linear digital filter that delays an input by a plurality of stages, multiplies the input by a certain coefficient, and outputs the sum (linear operation). In particular, T
In V signal processing, there are many processes in which the input / output relationship is not represented only by a linear relationship, such as a square, a logarithm, and a threshold process.
An arithmetic unit including these nonlinear elements is called a nonlinear digital filter. The median filter is hardware that realizes a function similar to a non-linear filter for expanding and contracting an image in a non-linear processing algorithm. FIG. 2 shows a schematic configuration diagram of this.

A median filter 10 determines a sequence relationship between neighboring pixels from a kernel forming circuit 11 for extracting neighboring pixels based on a kernel selection signal a from an input image signal (ie, a sequence relationship between a plurality of input data). A parallel sorter 12 for determining the relationship) and a selector 13 for selecting and outputting a pixel of a specific order from the determined order relationship based on the order selection signal b. In the median filter 10, the kernel forming circuit 11 extracts a pixel existing in the vicinity of the pixel of interest in the sampled image (this is referred to as “kernel”). Order with respect to signal level. Then, the selector 13 selects and outputs a pixel signal corresponding to a certain specific rank (rank). These operations are non-linear operations.
Therefore, the median filter 10 is a kind of nonlinear digital filter. When removing the impulse noise in the image signal using such a median filter 10, for example, filtering of 3 pixels × 3 lines is performed in a field of the image signal, or 3 pixels × 3 lines between fields. A noise elimination circuit such as applying line filtering is conceivable.

[0004]

However, in the above-described noise elimination circuit, 3 pixels × 3 lines (9 samples)
Due to the filter processing for outputting the intermediate point of the above, there is a problem that a line of one line or a point within four pixels is determined to be noise and is removed. Therefore, it has been difficult to obtain a noise elimination circuit that can suppress image quality degradation as much as possible and effectively remove only noise. SUMMARY OF THE INVENTION It is an object of the present invention to solve the problems of the prior art and to provide a noise elimination circuit capable of suppressing image quality degradation as much as possible and effectively removing only noise.

[0005]

Means for Solving the Problems The first and third inventions are:
In order to solve the above problem, in a noise removal circuit using a median filter, a current signal to be subjected to a filtering process of an image signal and a previous signal before the current signal are input,
Motion detecting means for detecting whether the image of the current signal is a still image or a moving image; vertical edge detecting means for detecting the presence or absence of a vertical edge of the image using the previous signal; and horizontal scanning of the image using the previous signal. A horizontal edge detecting means for detecting the presence or absence of an edge; and inputting the image signal, and selecting a predetermined pixel signal in the vicinity of the pixel of interest from pixel signals of adjacent vertical x-lines (x is a natural number). And a vertical median filter for inputting the image signal, and selecting and outputting a predetermined pixel signal in the vicinity of the target pixel from pixel signals of adjacent horizontal y pixels (where y is a natural number). A median filter; delay means for delaying the image signal by a filter processing delay time in the vertical median filter and the horizontal median filter; and a selection means. The selection means, based on the detection results of the motion detection means, the vertical edge detection means, and the horizontal edge detection means, outputs an output signal of the vertical median filter when there is no vertical edge in a still image, and outputs a vertical edge in a still image. And outputs one of the output signals of the horizontal median filter when there is no horizontal edge and one of the output signals of the delay means when there is a vertical edge and a horizontal edge in a still image and in the case of a moving image. It is.

In the second and third inventions, in a noise elimination circuit using a median filter, a memory for delaying an image signal and outputting a previous signal before a current signal to be filtered by the image signal, The current signal to be filtered and the previous signal output from the memory are input, and the image of the current signal is a still image or a moving image. The image processing apparatus includes a vertical edge detecting unit that detects the presence or absence of a vertical edge of the image, a horizontal edge detecting unit that detects the presence or absence of a horizontal edge of the image using the preceding signal, and a determining unit. The determination means outputs a first determination signal when there is no vertical edge in the still image based on the detection results of the motion detection means, the vertical edge detection means, and the horizontal edge detection means, and the vertical edge is detected in the still image. Second when there is no horizontal edge
And outputs a third determination signal when there is a vertical edge and a horizontal edge in a still image and in a moving image. Further, in the second and third inventions, the first, second or third determination signal is delayed by a predetermined time, respectively, and the first, second or third selection signal corresponding to the first, second or third determination signal is output. Delay means, and the image signal is input,
A vertical median filter for selecting and outputting a predetermined pixel signal in the vicinity of a pixel of interest from pixel signals of adjacent vertical x-lines (where x is a natural number); A horizontal median filter that selects and outputs a predetermined pixel signal near the pixel of interest from a pixel signal of y pixels (where y is a natural number), and a filter processing delay time in the vertical median filter and the horizontal median filter. Second delay means for delaying the image signal and selection means are provided. The selection unit is configured to select the first, second, or third selection signal output from the first delay unit, based on the first selection signal, based on the output signal of the vertical median filter, and based on the second selection signal. One of the output signals of the second delay means is selected and output based on the output signal of the horizontal median filter or the third selection signal.

[0007]

According to the first and third aspects of the present invention, since the noise elimination circuit is configured as described above, when the current signal and the previous signal in the image signal are input to the motion detecting means, the motion detecting means Thus, whether the image of the current signal is a still image or a moving image is detected. Further, when the previous signal is input to the vertical edge detecting means and the horizontal edge detecting means, the presence or absence of a vertical edge of the image is detected by the vertical edge detecting means, and the presence or absence of the horizontal edge of the image is further detected by the horizontal edge detecting means. Is detected. Further, when the image signal is input to the vertical median filter, the horizontal median filter, and the delay unit, a predetermined pixel signal near the target pixel is selected by the vertical median filter and the horizontal median filter, and further the pixel signal is output by the delay unit. Is delayed for a predetermined time. Then, the selection means selects the output signal of the vertical median filter when there is no vertical edge in the still image based on the detection results of the motion detection means, the vertical edge detection means and the horizontal edge detection means, and the vertical edge is detected in the still image. When there is no horizontal edge, the output signal of the horizontal median filter is selected. When there is a vertical edge and a horizontal edge in a still image, and in the case of a moving image, the output signal of the delay means is selected. Output the output signal.

According to the second and third aspects, when an image signal is input to a memory, the memory delays the input image signal to output a previous signal before the current signal, This is given to the vertical edge detecting means and the horizontal edge detecting means. The motion detection means detects whether the image of the current signal is a still image or a moving image from the current signal of the image signal and the output signal of the memory, and sends the detection result to the determination means. The vertical edge detecting means detects the presence or absence of a vertical edge of the image from the previous signal, and sends the detection result to the determining means. Further, the horizontal edge detection means detects the presence or absence of a horizontal edge of the image from the previous signal, and sends the detection result to the determination means. The determining means outputs a first determination signal when there is no vertical edge in the still image based on the detection results of the motion detecting means, the vertical edge detecting means, and the horizontal edge detecting means. When there is no edge, a second determination signal is output, and when there is a vertical edge and a horizontal edge in a still image and in a moving image, a third determination signal is output. The first of these,
The second or third determination signal is delayed for a predetermined time by the first delay unit, and the corresponding first, second or third selection signal is output and sent to the selection unit. On the other hand, the image signal is input to the vertical median filter, the horizontal median filter, and the second delay unit. Then, the vertical median filter and the horizontal median filter select a predetermined pixel signal near the target pixel and send it to the selection unit. Further, the second delay unit delays the input image signal by a predetermined time and sends it to the selection unit. The selection means selects an output signal of the vertical median filter based on the first selection signal among the first, second, or third selection signals output from the first delay means, and outputs the selected signal to the second selection signal. The output signal of the horizontal median filter is selected based on the selected signal, the output signal of the second delay unit is selected based on the third selection signal, and one of them is output.

[0009]

FIG. 1 is a block diagram showing a configuration of a noise removing circuit according to an embodiment of the present invention. This noise removal circuit has
A delay means (for example, a frame memory) 21 for outputting a previous frame S21 delayed by one frame by storing an input image signal (ie, a current frame) Si for a frame is provided. A circuit 22, a vertical edge detection circuit 23, and a horizontal edge detection circuit 24 are connected. One frame is composed of, for example, two fields. The motion detection circuit 22 receives the current frame of the input image signal Si and the previous frame S21 output from the frame memory 21, and detects whether the image of the current frame is a still image or a moving image, and It has a function as a motion detecting means for outputting the information S22. As a detection method of the motion detection circuit 22, for example, a difference between frames is obtained and compared with a certain threshold value.
When the comparison result is equal to or smaller than the threshold value, it is determined that the image is a still image. Also, in order to reduce the influence of noise, the motion detection circuit 22 may be configured by including an isolated point removal circuit or an area expansion filter in the motion detection result. The vertical edge detection circuit 23 has a function as a vertical edge detection unit that detects the presence or absence of a vertical edge of an image and outputs edge information S23 by using the previous frame S21 output from the frame memory 21. Horizontal edge detection circuit 2
4 is the previous frame S output from the frame memory 21
21 has a function as horizontal edge detection means for detecting the presence or absence of a horizontal edge of an image and outputting edge information S24. As a method of detecting a horizontal edge, for example, there is a method of taking a difference value between adjacent pixels, comparing the difference value with a threshold value, and determining a horizontal edge when the comparison result is equal to or larger than the threshold value.

The motion detecting circuit 22, the vertical edge detecting circuit 2
The determination circuit 25 is connected to the output side of the horizontal edge detection circuit 24, and a delay circuit 26 is connected to the output side of the determination circuit 25. The judgment circuit 25
Based on the motion / static information S22 from the motion detection circuit 22 and the edge information S23 and S24 from the vertical edge detection circuit 23 and the horizontal edge detection circuit 24, the information contents are determined, and the vertical edge of the still image is determined. When there is no vertical edge, the first determination signal S25a is output. When there is a vertical edge in the still image and there is no horizontal edge, the second determination signal S25b is output.
It has a function as a judging means for outputting a third judgment signal S25c when there is a vertical edge and a horizontal edge in a still image and in a moving image. The delay circuit 26 connected to the output side of the determination circuit 25 is provided with a first, second or third determination signal S25a, S25a in order to take timing with a circuit described later.
25b and S25c are respectively delayed by a predetermined time, and the corresponding first, second or third selection signals S26a and S2
6b and S26c as a first delay unit. The noise elimination circuit is provided with a vertical median filter 27, a horizontal median filter 28, and a delay circuit 29 for inputting the input image signal Si, and a selection circuit 30 is connected to an output side thereof. The vertical median filter 27 receives the input image signal Si
, The adjacent vertical x lines (eg, x =
This is a filter that selects a predetermined pixel signal near the target pixel from the pixel signal of 3) (sample points are 3 pixels) and outputs the selected signal. The horizontal median filter 28 receives the input pixel signal S
i, the adjacent horizontal y pixels (for example, y =
This is a filter that selects and outputs a predetermined pixel signal near the pixel of interest from the pixel signal of 9). These median filters 27 and 28 have, for example, a conventional circuit configuration as shown in FIG. Further, the delay circuit 29 has a function as a second delay unit for delaying the input image signal Si by a predetermined time in order to set the input image signal Si for the delay of the processing of the median filters 27 and 28. Have.

The selection circuit 30 connected to the output side of the median filters 27 and 28 and the delay circuit 29
Based on the first, second and third selection signals S26a, S26b and S26c output from the delay circuit 26, one of the output signals of the median filters 27 and 28 and the delay circuit 29 is adaptively switched and output. It has a function as selection means for outputting the image signal So. When the selection signal given from the delay circuit 26 is the first selection signal S26a, the output signal of the vertical median filter 27 is selected, and the second selection signal S26b Is supplied, the output signal of the horizontal median filter 28 is selected, and the third selection signal S26
When c is given, the output signal of the delay circuit 29 is selected, and one output image signal So is output. Next, the operation of the noise elimination circuit configured as described above will be described with reference to FIGS.

FIG. 3 is a diagram for explaining the operation of the three-line vertical median filter 27 shown in FIG. 1. The filter is formed by using three pixels of the (n + 1) -th line and the (n-1) -th line with respect to the target pixel of the n-th black circle. The action of applying is shown. FIG.
FIG. 5 is an explanatory diagram of the operation of the 9-pixel horizontal median filter 28 in FIG. 1, showing an operation of applying a filter to data of an n-th pixel of a focused black circle using data of n−4 to n + 4 pixels. Have been. When the input image signal Si is input, it is input to the frame memory 21, the motion detection circuit 2
2, sent to the vertical median filter 27, horizontal median filter 28, and delay circuit 29. The frame memory 21 outputs the previous frame S21 by delaying the input image signal Si by one frame, and sends the previous frame S21 to the motion detection circuit 22, the vertical edge detection circuit 23, and the horizontal edge detection circuit 24. Using the current frame of the input image signal Si and the previous frame S21 from the frame memory 21, the motion detection circuit 22 detects whether the image of the current frame is a still image or a moving image, and obtains the motion / static information. S22 is sent to the determination circuit 25. Vertical edge detection circuit 23
Detects the presence or absence of a vertical edge of an image using the previous frame S21 from the frame memory 21, and detects the edge information S
23 is sent to the determination circuit 25. Further, the horizontal edge detection circuit 24 reads the previous frame S21 from the frame memory 21.
Is used to detect the presence or absence of a horizontal edge of the image, and the edge information S24 is sent to the determination circuit 25.

The determination circuit 25 includes motion / static information S22 from the motion detection circuit 22, and edge information S23 and S24 from the vertical edge detection circuit 23 and the horizontal edge detection circuit 24.
Based on the above, when it is determined that there is no vertical edge in the still image, a first determination signal S25a for selecting the vertical median filter 27 is output, and when it is determined that there is a vertical edge in the still image and there is no horizontal edge, A second determination signal S25b is output to select the horizontal median filter 28, and when a still image is determined to have edges both in the vertical and horizontal directions, and when it is determined to be a moving image, the third determination signal S25b is selected to select the delay circuit 29. Is output. These first, second and third determination signals S25a, S25b, S
25c is delayed for a predetermined time by the delay circuit 26, and the corresponding first, second, and third selection signals S26a, S26
b and S26c are sent to the selection circuit 30. On the other hand, in the vertical median filter 27, as shown in FIG. 3, adjacent three lines of pixel signals (sample points are three pixels) are used.
Sampling is performed on the input image signal Si, and the output signal is sent to the selection circuit 30. In the horizontal median filter 28, as shown in FIG. 4, the input image signal Si is filtered using the pixel signals of nine adjacent pixels, and the output signal is sent to the selection circuit 30. Further, in the delay circuit 29, the input image signal Si is time-matched by the filter processing delay of the median filters 27 and 28, and the output signal is sent to the selection circuit 30. Selection circuit 3
0, for example, the first selection signal S2
6a, the output signal of the vertical median filter 27 is selected and output as an output image signal So. When the second selection signal S26b is provided, the output signal of the horizontal median filter 28 is selected and output. Is output as the output image signal So, and when the third selection signal S26c is given, the output signal of the delay circuit 29 is selected and output as the output image signal So.

The feature of the present embodiment is that two types of filters, a vertical median filter 27 and a horizontal median filter 28, are adaptively applied to edge detection circuits 23 and 24 using information of the previous frame S21 only for a still image. Switch. Characteristics of impulse noise include:
"There is no correlation between lines", "There is no correlation between fields and frames", and "There is some correlation between pixels (since the magnitude of noise is always one or more pixels)"
A point is conceivable. Therefore, in order to effectively remove noise, the input image signal Si may be filtered by the vertical median filter 27 (three pixel samples). However, in the vertical direction (for example,
If there is a line having a width of one line) and the filtering is performed by the vertical median filter 27, the edge portion may be removed as noise. Therefore, when there is an edge in the vertical direction, the horizontal median filter 28 of 9 pixels in the horizontal direction is switched. Nine pixels have the effect of removing noise over four pixels. On the other hand, if it is determined that there is an edge in the horizontal direction, the input image signal Si is output through the delay circuit 29 in a through state without filtering the median filters 27 and 28 in order to prevent image quality deterioration. .

Such edge determination is also performed by the filter 27,
If the current frame to be multiplied by 28 is used, the edge is erroneously determined even at the time of impulsive noise, and as a result, the median filters 27 and 28 do not work well. In order to eliminate this problem, as described above, one of the characteristics of the impulse noise is that there is no correlation between fields and between frames. Occur randomly, and using the fact that they are not at the same location in the previous frame, edge detection circuits 23 and 24 perform edge detection on pixels at the same position in the previous frame S21. The median filters 27 and 28 applied to the frame are switched. At this time, in the case of a moving image, the signal of the current frame and the signal of the previous frame S21 are different. Therefore, the motion detection circuit 22 is used. Edge detection is performed, and when a moving image is determined, filter-through is performed. As described above, in the image determined to be a still image, the previous frame S2
The edge detection circuits 23 and 24 perform edge detection on one signal, and the selection circuit 30 switches between two types of median filters 27 and 28 according to the detection result. As a result, it is possible to eliminate image quality deterioration due to malfunction of the median filters 27 and 28 and to effectively remove impulse noise.

As described above, the effects of this embodiment can be summarized as follows. In the present embodiment, the motion detection circuit 22 using the current frame and the previous frame S21 one frame before is used.
And a vertical edge detection circuit 23 using the previous frame S21
And output information S22 and S with the horizontal edge detection circuit 24.
The selection circuit 30 adaptively switches the vertical median filter 27, the horizontal median filter 28, and the delay circuit 29 by 23 and S24. As a result, edges and the like are not erroneously detected as noise, and
The edge detection circuits 23 and 24 are also connected to the previous frame S at the time of a still image.
Since edge detection is performed using the signal 21, the impulse noise can be effectively removed without erroneously detecting the impulse noise as an edge. Note that the present invention is not limited to the above embodiment, and various modifications are possible. For example, there are the following modifications.

(A) The vertical median filter 27 is 3
Although the line / horizontal median filter 28 uses a filter of 9 pixels, a filter having another number of lines or another number of pixels may be used. (B) Edge detection is performed on the previous frame S21,
According to the detection result, two types of median filters 2
7 and 28 are switched, but the previous frame S2
As 1, a frame other than the current frame, such as the frame before the previous frame or the like, may be used. Alternatively, focusing on the fields constituting the frame and performing edge detection or the like using a field other than the current field, such as a field before the current field or a field before the current field, or the field before the previous field, the same as in the above embodiment. Almost the same effects can be obtained.

[0018]

As described in detail above, the first to third embodiments
According to the invention, the motion detecting means, the vertical edge detecting means,
The vertical median filter, the horizontal median filter, and the filter through (the filter bypass) through the delay means are adaptively switched based on the detection result of the horizontal edge detection means, so that the edges and the like are regarded as noise. There is no false detection. Furthermore, since the edge is detected by the edge detecting means using the previous signal at the time of the still image detected by the motion detecting means, the impulse noise is not erroneously detected as an edge, and the impulse noise is effectively detected. It can be removed.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a noise removing circuit according to an embodiment of the present invention.

FIG. 2 is a schematic configuration diagram of a conventional median filter.

FIG. 3 is an operation explanatory diagram of the vertical median filter of FIG. 1;

FIG. 4 is a diagram illustrating the operation of the horizontal median filter of FIG. 1;

[Explanation of symbols]

 21 Frame Memory 22 Motion Detection Circuit 23 Vertical Edge Detection Circuit 24 Horizontal Edge Detection Circuit 25 Judgment Circuit 26, 29 Delay Circuit 27 Vertical Median Filter 28 Horizontal Median Filter 30 Selection Circuit

────────────────────────────────────────────────── ─── Continued from the front page (72) Yoshiharu Hoshino, Inventor Yoshiharu Hoshino 2-2-1 Jinnan, Shibuya-ku, Tokyo Japan Broadcasting Corporation Broadcasting Center (56) References JP-A-6-260889 (JP, A) JP-A Heihei 4-105467 (JP, A) (58) Field surveyed (Int. Cl. ⁷ , DB name) H04N 5/21

Claims (3)

(57) [Claims] 1. A motion detecting means for inputting a current signal to be subjected to a filtering process of an image signal and a previous signal before the current signal, and detecting whether an image of the current signal is a still image or a moving image; A vertical edge detecting means for detecting the presence or absence of a vertical edge of an image using a signal; a horizontal edge detecting means for detecting the presence or absence of a horizontal edge of the image using the previous signal; and A vertical median filter for selecting and outputting a predetermined pixel signal in the vicinity of the pixel of interest from pixel signals of x lines in the direction (where x is a natural number); A horizontal median filter that selects and outputs a predetermined pixel signal in the vicinity of the pixel of interest from pixel signals (where y is a natural number); and the vertical median filter and the horizontal median filter. Delay means for delaying the image signal by a filter processing delay time, and the vertical median filter when there is no vertical edge in a still image based on the detection results of the motion detection means, the vertical edge detection means, and the horizontal edge detection means. The output signal of the horizontal median filter when there is a vertical edge and no horizontal edge in a still image, or the output signal of the delay means when there is a vertical edge and a horizontal edge in a still image and in the case of a moving image. And a selecting means for selecting and outputting any one of them. 2. A memory for delaying an image signal and outputting a previous signal before a current signal to be filtered for the image signal, a current signal to be filtered for the image signal, and output from the memory. A motion detection unit that inputs a previous signal and detects whether the image of the current signal is a still image or a moving image; a vertical edge detection unit that detects presence or absence of a vertical edge of the image using the previous signal; A horizontal edge detection unit for detecting the presence or absence of a horizontal edge of an image using a signal; and, based on detection results of the motion detection unit, the vertical edge detection unit, and the horizontal edge detection unit, when there is no vertical edge in a still image, A first determination signal is output, and a second determination signal is output when there is a vertical edge in a still image and there is no horizontal edge. Determining means for outputting a third determination signal; and first means for delaying each of the first, second or third determination signals by a predetermined time and outputting a corresponding first, second or third selection signal. A vertical median filter that receives the image signal, selects a predetermined pixel signal near the pixel of interest from an adjacent vertical x-line (where x is a natural number) pixel signal, and outputs the selected signal. A horizontal median filter that receives the image signal, selects a predetermined pixel signal in the vicinity of the pixel of interest from pixel signals of adjacent y pixels (where y is a natural number), and outputs the horizontal median filter; Second delay means for delaying the image signal by a filter processing delay time in the filter and the horizontal median filter; first, second or third output from the first delay means
Out of the selection signals, the output signal of the vertical median filter based on the first selection signal, the output signal of the horizontal median filter based on the second selection signal, or the third.
And selecting means for selecting and outputting any one of the output signals of the second delay means based on the selection signal. 3. The noise removal circuit according to claim 1, wherein the current signal is a current field or a current frame, and the previous signal is one or more previous fields or a current frame of a current field. The noise elimination circuit that is the previous frame immediately before.