JP2009033603A - Noise reduction apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a noise reduction apparatus for reducing mosquito noise that exerts a little adverse influence upon texture in a video signal. <P>SOLUTION: A noise reduction apparatus uses a high frequency extraction section 1 for extracting a high frequency component of a video signal inputted; a strong edge region detection section 2 for detecting a strong edge region in which a magnitude of the high frequency component is equal to or greater than a first threshold and for defining a region of a predetermined range around the strong edge region as a noise reduction candidate region (bolded region); a weak edge detection section 3 for detecting a weak edge region in which the magnitude of the high frequency component is less than the first threshold and is equal to or greater than a second threshold smaller than the first threshold; a noise reduction region detection section 4 for detecting the noise reduction region, excepting for the weak edge region, from the noise reduction candidate regions; and a noise reduction section 5 for applying noise reduction processing to the noise reduction region. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a noise reduction apparatus for reducing mosquito noise that occurs during decoding of encoded image data.

As is well known, in compression encoding systems such as JPEG and MPEG, the higher the compression rate, the more likely mosquito noise is generated around the edge of the decoded image.
This mosquito noise is conspicuously generated in a region including many high-frequency components of a spatial frequency, represented by the boundary between the normal image and the graphics screen when a graphics image such as a telop is superimposed on the normal image. Further, the mosquito noise tends to be easily noticeable in a flat area of the image.

  On the other hand, mosquito noise generated in the texture portion is relatively inconspicuous. If the texture contains many strong edges, the texture itself is a source of mosquito noise, and it is adversely affected by itself, but it is not easy to stand out. Further, the texture does not include such a strong edge, but if there is a strong edge nearby, the texture is affected by mosquito noise from the strong edge.

Various methods for reducing the mosquito noise as described above have been proposed.
For example, in the conventional technique described in Patent Document 1, a technique for performing filtering by enlarging an effective edge is proposed. Further, in the prior art described in Patent Document 2, when edge pixels and a flat part coexist in the same block, a method for correcting the level difference of the flat part to be small is proposed. Further, Patent Document 3 proposes a method of using a smoothed edge detection signal as an edge strength in a noise removal filter.
JP-A-5-227431 JP-A-11-112978 Japanese Patent Laid-Open No. 2004-226651

However, any of these conventional techniques is basically a process of simply dividing an image into an edge and a flat part and reducing noise on the flat part around the edge.
Therefore, in the above prior art, when the area where the texture that is originally intended to be displayed faithfully (texture area) is determined as a flat portion around the edge, noise reduction processing is performed on the texture area, There was a problem that the original texture could not be reproduced faithfully.

The present invention has been made to solve the above problems.
The purpose is to provide a noise reduction device that prevents the noise displayed on the area near the strong edge from being subjected to noise reduction processing and can reproduce the original texture as faithfully as possible. is there.

In order to solve the above problems, the present invention provides the following apparatuses.
(1) high-frequency extraction means for extracting high-frequency components of the input video signal; (1);
On the screen when the video signal is displayed, a first area in which the magnitude of the high frequency component is equal to or greater than a first threshold is detected, and an area in a predetermined range centered on the first area is detected. (1) the first detection means set as the second area;
Second detection means for detecting a third region in which the magnitude of the high frequency component is less than a first threshold and is equal to or greater than a second threshold smaller than the first threshold; (3)
(4) (5), noise reduction means for performing noise reduction processing on the area obtained by removing the third area from the second area;
A noise reduction device comprising:

  According to the present invention, it is possible to prevent the texture displayed in the region near the strong edge from being subjected to noise reduction processing, and to reproduce the original texture as faithfully as possible.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram showing the configuration of the noise reduction apparatus of this embodiment.
FIG. 4 shows an example in which the luminance signal applied to this embodiment is displayed on the screen. In FIG. 4, locations indicated by reference numerals 10 and 11 are a frame line 10 and a frame line 11 inserted as telops, which are strong edge regions. Further, the portion indicated by reference numeral 20 is the texture 20 that is the original display target.

5 and 6 are diagrams showing an image when the present embodiment is applied to the luminance signal displayed on the screen of FIG. 4 described above.
Hereinafter, this embodiment will be described in detail with reference to FIGS. 1 and 4 to 6.

  The luminance signal (input luminance signal) input to the noise reduction apparatus of this embodiment will be described below as a signal decoded with a resolution of 8 bits (255 gradations) for each pixel. However, the form of the input luminance signal is not limited to this.

  In FIG. 1, the high-frequency extraction unit 1 sequentially sets each pixel of the input luminance signal as a noise reduction processing target pixel, and, for example, −1 for each 5 × 5 pixel block centering on the noise reduction processing target pixel. Filter processing by a Laplacian filter having tap constants of / 16, -1/4, 5/8, -1/4, -1/16 is performed in the horizontal direction and the vertical direction. Thereby, a high frequency signal which is a high frequency component of the input luminance signal can be extracted. Note that the filter used in the high-frequency extraction unit 1 is not limited to the Laplacian filter described above, and may have any configuration as long as a high-frequency signal can be extracted.

  The strong edge region detection unit 2 converts the supplied high frequency signal to an absolute value, and the level of the high frequency signal converted to the absolute value is equal to or higher than a first threshold Th1 that is set in advance to a predetermined value. The region is detected as a strong edge region (first region). The specific value of the first Th1 is, for example, “28”.

  Then, it is determined that the detected strong edge region is a region that is highly likely to be a source of mosquito noise. Furthermore, a region in a predetermined range centered on the strong edge region is set as a region (noise reduction candidate region) that seems to require mosquito noise reduction processing. The predetermined range is, for example, a 7 × 7 pixel block centered on each pixel detected as a strong edge region. When such processing is performed, a plurality of blocks (noise reduction candidate regions) usually exist along strong edge regions detected as lines, so that the edges (lines) appear on the screen. The image becomes thicker. Therefore, in the present description, this processing is referred to as “thickening processing”. The noise reduction candidate region (second region) obtained by the thickening process is referred to as a “thickening region”.

  FIG. 5 is a diagram showing an image of the thickened area. The frame line 10 and the frame line 11 are strong edge regions, and the surrounding gray portions 30 and 31 are a thick lined region 30 and a thick lined region 31, respectively.

  Then, a thickened area detection signal indicating the thickened area determined by the above processing is supplied to the noise reduction area detecting unit 4. The thick line area detection signal is, for example, a binary signal in which the thick line area is “1” and the other areas are “0”. In addition to this, any signal may be used as long as it is a signal indicating a thickened region.

  The weak edge region detection unit 3 converts the supplied high frequency signal into an absolute value, and the level of the high frequency signal converted into the absolute value is a second value whose value is set in advance smaller than the first threshold Th1 described above. An area that is equal to or greater than the threshold Th2 and less than the first Th1 is detected as a weak edge area (third area). Note that a specific value of the second threshold Th2 is, for example, “14”.

  Although this weak edge region has an edge, the edge is not a strong edge as detected in the strong edge region, but is a relatively weak edge, so it is highly likely that the edge is inherent to the texture. Judge. Therefore, it is desirable not to perform the mosquito noise reduction process on this weak edge region. In FIG. 5, it is assumed that the texture 20 is detected as a weak edge region.

  Then, a weak edge signal detection signal indicating the weak edge region determined by the processing as described above is supplied to the noise reduction region detection unit 4. The weak edge area detection signal is, for example, a binary signal in which the weak edge area is “1” and the other areas are “0”. Any other signal may be used as long as it indicates a weak edge region.

  The noise reduction area detection unit 4 is configured to detect an input luminance signal based on the thick line area detection signal supplied from the strong edge area detection unit 2 and the weak edge area detection signal supplied from the weak edge area detection unit 3. Then, a noise reduction target area, which is an area on which mosweet noise reduction processing is performed, is detected.

Specifically, a region where the thick line region detection signal is 1 and the weak edge region detection signal is 0 is detected, and the detected region is set as a noise reduction target region.
Then, a noise reduction control signal indicating the noise reduction target region determined by the above processing is supplied to the noise reduction processing unit 5. The noise reduction control signal is, for example, a binary signal in which the noise reduction target area is 1 and the other areas are 0. In addition to this, any signal may be used as long as it indicates a noise reduction target region.

The noise reduction processing unit 5 performs noise reduction processing on the input luminance signal based on the noise reduction control signal supplied from the noise reduction region detection unit 4.
Specifically, the noise reduction process is performed on the area where the noise reduction control signal is 1, and the noise reduction process is not performed on the area where the noise reduction control signal is 0.

As a mosquito noise reduction process, for example, a method using the following intermediate value process is effective.
Raster scanning is performed within a 5 × 5 pixel block centered on the noise reduction target pixel. Then, in the case where the noise reduction target pixel is included in the noise reduction target region, an intermediate value of peripheral pixels included in the noise reduction target region is acquired, and the pixel value of the noise reduction processing target pixel is changed to this intermediate value. To do. The specific processing content of this method will be described in detail in the description of the operation procedure of the noise reduction processing described later.

FIG. 6 is a diagram showing an image of the result of performing the above processing in FIG.
In the same figure, portions indicated by reference numerals 40 to 43 are portions of the texture 20 existing in the thick lined region 31, and this portion is not subjected to noise reduction processing. This portion is defined as a noise reduction stop region 40-43. Therefore, the noise reduction target area is an area obtained by removing the noise reduction stop areas 40 to 43 from the thick line area 30 and the thick line area 31.

In the figure, the noise reduction stop areas 40 to 43 are displayed as relatively large areas so that they can be easily understood, but only the weak edge area that is the texture 20 is actually the original noise reduction stop area. .
<Operation procedure of noise reduction area detection>
Hereinafter, a specific operation procedure of noise reduction area detection processing realized by using the high-frequency detection unit 1, the strong edge region detection unit 2, the weak edge region detection unit 3, and the noise reduction region detection unit 4 of FIG. 1 and the flowchart of FIG.

The subject of the operation in this flowchart is a control means (not shown) for controlling each means in FIG. 1 unless otherwise specified.
In step S101, each pixel of the input luminance signal is sequentially set as a noise reduction processing target pixel, and a Laplacian filter process is performed for each block of 5 × 5 pixels centering on the noise reduction processing target pixel to generate a high frequency signal. Extract.

  Next, in step S102, whether or not a strong edge region is detected from the high frequency signal extracted in step S101 is determined based on whether or not the absolute value of the high frequency signal is greater than or equal to the first threshold Th1. If detected (Y), the process proceeds to the next step S103. If not detected (N), the process proceeds to step S106.

In step S103, a region in a predetermined range centering on the strong edge region detected in step 102 is set as a region (thickened region) where mosquito noise reduction processing is considered necessary.
In step 104, it is determined whether or not a weak edge region (texture region) is detected in the thick lined region, based on whether or not the absolute value of the high frequency signal is less than the second threshold Th2. If not detected (N), it is determined that this area is a noise reduction target area, and the process proceeds to step S105. If it is detected (Y), it is determined that the area is not a noise reduction target area, and the process proceeds to step S106.

In step S105, the polarity “1” (noise reduction target area) is set in the noise reduction control signal.
In step S106, the polarity “0” (not the noise reduction target region) is set in the noise reduction control signal.

The noise reduction region detection processing is realized by executing the above-described series of processing for each pixel in the luminance signal.
<Operation procedure of noise reduction processing>
Next, a specific operation procedure of the noise reduction processing realized by the noise reduction processing unit 5 of FIG. 1 will be described with reference to the flowcharts of FIG. 1 and FIG.

The subject of the operation in this flowchart is a control means (not shown) for controlling each means in FIG. 1 unless otherwise specified.
In the present embodiment, as described above, a noise reduction processing method using intermediate value processing is used.

  In step S201, a raster scan is performed within a 5 × 5 pixel block centered on the noise reduction target pixel of the input luminance signal, and associated with the noise reduction control signal input corresponding to each pixel in this block. .

  In step S202, whether or not the noise reduction target pixel is included in the noise reduction target region is determined based on the polarity of the noise reduction control signal associated in step S201. When the noise reduction target pixel is included in the noise reduction target region (Y) (polarity = 1), the process proceeds to the next step S203. If not included (N) (polarity = 0), the noise reduction processing is not performed, and the processing for the noise reduction processing target pixel is terminated.

  In step S203, with reference to the polarity of the noise reduction control signal associated in step S201 for each pixel in the block, a pixel having a polarity of 1 (included in the noise reduction target region) is extracted. Then, the number of extracted pixels is counted, and it is determined whether or not the count value is equal to or greater than a third threshold value Th3. Note that a specific value of the third threshold Th3 is, for example, “0”. When the count value is less than Th3 (N), the noise reduction process is not performed, and the process for the noise reduction process target pixel is terminated. If the count value is equal to or greater than the threshold Th3 (Y), the process proceeds to the next step S204.

In step S204, the values of the pixels included in the noise reduction target area extracted in step S203 are compared to obtain an intermediate value.
In step S205, a difference value between the intermediate value acquired in step S204 and the noise reduction target pixel value is calculated. And it is judged whether the absolute value of this difference value is more than 4th threshold value Th4. A specific value of the fourth Th4 is, for example, “15”. When the absolute value is equal to or greater than the fourth threshold Th4 (N), the noise reduction process is not performed and the process for the noise reduction process target pixel is terminated. If the absolute value is less than the fourth threshold Th4 (Y), the process proceeds to the next step S206.

In step S206, the value of the noise reduction process target pixel is changed to the intermediate value acquired in step S204, and the process ends.
The noise reduction process is realized by executing the above series of processes for each pixel in the luminance signal.

In this embodiment, the noise reduction processing method based on the intermediate value processing is used. However, the noise reduction processing method is not limited to this, and any effective method for reducing mosquito noise may be used.
<Application example>
FIG. 7 is a diagram illustrating a configuration example when the noise reduction device of the present embodiment is applied to a reproduction device such as a DVD player or a DVD recorder, or a reception device such as a digital broadcast receiver.

  The bit stream input unit 200 is based on a DCT (Discrete Cosine Transform) based on a representative of MPEG or the like, which is reproduced from a tape or a disk in the case of a playback device or received by a tuner unit in the case of a reception device. A bit stream of a video signal encoded by a video encoding method is input.

In the decoding unit 300, the bit stream input to the bit stream input unit 200 is decoded to become a baseband digital video signal.
The mosquito noise is reduced in the noise reduction unit 100 using the noise reduction apparatus of the present embodiment mainly for the luminance signal of the digital video signal. The noise reduction unit 100 may be bypassed depending on the signal content and viewing purpose. For example, when the bitstream encoding rate is sufficiently high and the possibility of occurrence of mosquito noise is low, it may be bypassed. On the other hand, signals with many telops may be actively used.

  When the output unit or the display unit 400 is an output unit mounted on a DVD player, a DVD recorder, or the like, the digital video signal that has been adaptively subjected to noise reduction processing is digitally output as it is, or D / A converted and output as an analog video signal. In the case of a display unit such as a display or projector installed in a digital broadcast receiver, the digital video signal adaptively subjected to noise reduction processing as described above is D / A converted into an analog video signal. indicate.

  The present invention includes a program for causing a computer to realize the functions of the apparatus described above. These programs may be read from a recording medium and loaded into a computer, or may be transmitted via a communication network and loaded into a computer.

It is a block diagram which shows the structure of the Example of the noise reduction apparatus of this invention. It is a flowchart which shows the operation | movement procedure of a present Example. It is a flowchart which shows the operation | movement procedure of a present Example. It is a figure which shows the example which displayed the luminance signal which gives a present Example on a screen. It is a figure which shows the image of the thick line area | region when the present Example is given to the luminance signal shown in FIG. It is a figure which shows an image when the present Example is given to the luminance signal object image shown in FIG. It is a block diagram which shows the application example of a present Example.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 High region extraction part 2 Strong edge area | region detection part (1st detection means)
3 Weak edge region detector (second detection means)
4 Noise reduction area detector (Noise reduction means 1)
5 Noise reduction processing part (Noise reduction means 2)
10, 11 Strong edge area 20 Texture (weak edge area)
30, 31 Thickened regions 40 to 43 Noise reduction stop region 100 Noise reduction unit 200 Bit stream input unit 300 Decoding unit 400 Output unit or display unit

Claims (1)

High-frequency extraction means for extracting high-frequency components of the input video signal;
On the screen when the video signal is displayed, a first area in which the magnitude of the high frequency component is equal to or greater than a first threshold is detected, and an area in a predetermined range centered on the first area is detected. First detection means for setting as a second region;
Second detection means for detecting a third region in which the magnitude of the high frequency component is less than a first threshold and is equal to or greater than a second threshold smaller than the first threshold;
Noise reduction means for applying a noise reduction process to a region excluding the third region from the second region;
A noise reduction device comprising: