JP3950815B2 - Noise suppression device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a noise suppression device and a noise suppression method, and more particularly, to a noise suppression device and a noise suppression method capable of reducing noise generated with image coding.
[0002]
[Prior art]
In general, in the MPEG (Moving Picture Experts Group) method used for compression of digital image data, when image data having an insufficient transmission bit rate is input, the image is compressed in a form in which a high-frequency component is deleted. When this compressed data is expanded, the high-frequency component is deleted, so that noise called mosquito noise is generated at the edge portion of the image. Mosquito noise (also called ringing noise) is noise that flickers as if mosquitoes are flying at the edge of an image.
[0003]
4 to 8 are diagrams showing a conventional noise reduction device (see, for example, Patent Document 1) that can reduce this noise. The processing of each block will be described for the luminance data processing unit 100 with reference to FIGS. First, the input data Yin shown in FIG. 4 is input to the luminance part flag generation unit 101.
[0004]
FIG. 5 is a diagram illustrating a configuration of the luminance part flag generation unit 101, and input data Yin is sent to the edge detection unit 115. The input data Yin is also input to the high level detector 117. The high level detection unit 117 detects a pixel equal to or higher than a preset threshold value (for example, TH2 is 180) for the input data Yin and generates a flag 2. When pixels above the threshold are telops, the pixels before and after that are often at the black level.
[0005]
FIG. 6 is a diagram showing details of the edge detection unit 115. The edge detection unit 115 detects the absolute value of the difference between the input data Yin and data obtained by shifting Yin by one pixel by the delay unit, and the absolute value thereof. A signal EDGE is generated that is an edge for a pixel that exceeds a preset threshold (for example, TH1 is 40).
[0006]
The signal EDGE detected by the edge detection unit 115 is input to the edge left direction pixel filter range setting unit 116 and the edge right direction pixel filter range setting unit 118.
[0007]
FIG. 7 is a diagram for explaining an edge left pixel filter range setting unit, FIG. 7 (a) is a diagram for explaining the configuration of an edge left direction pixel filter range setting unit, and FIG. 7 (b) is a diagram for explaining its operation. It is a figure to do.
[0008]
The edge left pixel filter range setting unit 116 operates at the timing shown in FIG. 7B based on the input signal EDGE. As shown in the figure, based on the signal EDGE detected by the edge detection unit 115, N pixels (for example, 4 pixels) in the left direction from the signal EDGE are set as filter target pixels, and a signal flag 1 in a corresponding range is generated. To do. Since the flag 1 is generated with a delay from the input data Yin, it is necessary to relatively delay the target pixel data to be filtered. Therefore, the input data Yin is delayed by the delay unit 103 by a necessary amount. Further, a flag F-EDGE indicating a pixel position for changing the filter processing within the filter target range is generated.
[0009]
In FIG. 7, when the detected edge interval is measured by the edge interval measurement counter 1161 and less than N pixels (for example, less than 5 pixels), the flag 1 is not generated so as not to be a filter target pixel. . The edge interval (N pixels) can be set in advance as a filter execution pixel interval threshold value. By appropriately setting this value, it is possible to prevent a sensitive reaction when the pixel value suddenly changes due to external noise or the like. In addition, it is possible to prevent averaging (image blurring) between greatly different pixel values.
[0010]
FIG. 8 is a diagram illustrating the edge right direction pixel filter range setting unit. In the edge right pixel filter range setting unit 118, the flag 2 generated by the high level detection unit 117 based on the signal EDGE provided from the edge detection unit 115 and the signal flag 2 provided from the high level detection unit 117. N pixels (for example, 4 pixels) to the right of the signal EDGE detected by the edge detection unit 115 under the conditions in which the signal is generated are set as filter target pixels, and the corresponding signal flag 3 and the filter within the target range are filtered. B-EDGE indicating the pixel position for changing the process is generated.
[0011]
Note that the control signal EPN-CTL in FIG. 4 is a control signal indicating the quality of the image quality from another noise reduction device, and the luminance part enable signal generation unit 102 applies filter processing to each pixel based on this signal. Decide whether or not to do.
[0012]
As described above, in order to prevent the telop part (high luminance part) that is likely to generate mosquito noise from being filtered inside the telop, the flag 2 indicating the telop part is generated, and the N pixels before and after the telop are generated. Flags indicating the reference pixel position for generating the flag 1 and the flag 3 for performing the filtering process on the assumption that the mosquito noise is generated and further changing the filtering target pixel range within the filtering target range. F-EDGE and B-EDGE are generated.
[0013]
[Patent Document 1]
Japanese Patent Application No. 2002-198910 [0014]
[Problems to be solved by the invention]
In the conventional technique, a low pass filter is applied to four pixels around the edge. Further, as an exception process, when the edge interval is narrow (for example, less than 5 pixels), the filter is not applied. As a result, when the pixel value suddenly changes due to external noise or the like, it does not react excessively. In addition, it is possible to prevent blurring of the image due to averaging at pixels with greatly different pixel values.
[0015]
By the way, the image size of the image that is decoded and transmitted is reduced in some cases. For this reason, it is necessary to perform enlargement as the final processing of the MPEG decoder. When enlarging, the interval between the edges is enlarged as the image is enlarged. In this case, when the edge interval is 5 pixels or more, the low-pass filter is applied.
[0016]
As described above, when the enlargement ratio of the image is increased, the filtering process is also performed on a portion in which the pixel value is suddenly changed due to external noise or the like. In addition, averaging (low-pass filter processing) is performed between pixels having greatly different pixel values.
[0017]
9A and 9B are diagrams for explaining a specific example of processing by the low-pass filter at the time of image enlargement. FIG. 9A shows a case where the enlargement ratio is 1, and FIG. 9B shows a case where the enlargement ratio is 2. . When the enlargement ratio in FIG. 9A is 1, as shown in FIG. 9A, for example, when the interval between edges formed by a telop and a natural image is 5 pixels or more, for example, 5 before and after the telop. A low pass filter is applied to the pixel assuming that mosquito noise is generated (the filter effective signal is set to H level. Note that blurring occurs when a pixel is filtered on a natural image pixel at the boundary between a telop and a natural image. For this reason, the pixel is set not to be filtered by setting the filter effective signal to L level). Further, as shown in (b) of FIG. 9A, for example, when the interval between edges formed by a high-brightness natural image and a low-brightness natural image is less than 5 pixels, the high-brightness natural image For example, pixels before and after the image are not subjected to a low-pass filter because they are natural images with a sharp luminance change.
[0018]
When the enlargement ratio in FIG. 9B is 2, as shown in FIG. 9B, for example, when the interval between the edges formed by the telop and the natural image is 5 pixels or more, before and after the telop For example, a low-pass filter is applied to 5 pixels assuming that mosquito noise is generated.
[0019]
Further, as shown in (b) of FIG. 9B, for example, when an interval between edges formed by a high-brightness natural image and a low-brightness natural image becomes 5 pixels or more by an image enlargement operation. The low-pass filter is applied to, for example, five pixels before and after the high-luminance natural image. In this case, the steepness of the natural image portion (steep and precise natural image portion) with a sharp luminance change is lost. That is, the steepness in the enlarged image is lost by the filtering process.
[0020]
The present invention has been made in view of these problems, and provides a noise suppression device and a noise suppression method capable of performing an appropriate filter process according to an image enlargement ratio.
[0021]
[Means for Solving the Problems]
The present invention employs the following means in order to solve the above problems.
[0022]
Input image data, detect each edge of the image from the input image data, and if the pixels included between the detected edges are greater than or equal to a set value, the pixels included between the edges are filtered by a low-pass filter A flag generation unit that is designated as a filter processing target pixel to be applied, an enable signal generation unit that specifies a low-pass filter calculation pixel to be applied to each filter processing target pixel based on the specification of the flag generation unit, and the enable signal generation unit A low-pass filter that performs filtering on the input image based on the generated enable signal, and an enlargement ratio determining unit that determines the enlargement ratio of the image and changes the set value based on the enlargement ratio. The set value is set approximately in proportion to the enlargement ratio.
[0023]
As a result, when the low pass filter is applied to a plurality of pixels around the edge, the low pass filter can be prevented from being applied when the edge interval is narrow (for example, less than N pixels) as an exception process. The number N of pixels representing the edge interval is a value that is substantially proportional to the enlargement ratio of the image. Thereby, it is possible to prevent the steepness of the enlarged image from being lost by the filter processing.
[0024]
DETAILED DESCRIPTION OF THE INVENTION
10A and 10B are diagrams for explaining a specific example of processing by the low-pass filter at the time of image enlargement according to the present invention. FIG. 10A shows an enlargement factor of 1, FIG. 10B shows an enlargement factor of 2. This case is shown. When the enlargement ratio in FIG. 10A is 1, as shown in FIG. 10A, for example, when the interval between edges formed by a telop and a natural image is 5 pixels or more, for example, 5 before and after the telop. A low-pass filter is applied to the pixel, assuming that the mosquito noise is generated. Further, as shown in (b) of FIG. 10A, for example, when the interval between edges formed by a high-brightness natural image and a low-brightness natural image is less than 5 pixels, the high-brightness natural image The pixels before and after the image are not subjected to a low-pass filter because they are natural images with a sharp luminance change.
[0025]
Further, when the enlargement ratio in FIG. 10B is 2, as shown in FIG. 10A, for example, the interval between the edges formed by the telop and the natural image is 10 pixels (5 pixels × enlargement ratio) or more. For example, 10 pixels before and after the telop are subjected to a low-pass filter on the assumption that the mosquito noise is generated.
[0026]
Further, as shown in (b) of FIG. 10B, for example, an interval between edges formed by a high-brightness natural image and a low-brightness natural image is 10 pixels (5 pixels × enlargement) even by an image enlargement operation. If it is less than (rate), the low-pass filter is not applied to, for example, nine pixels before and after the high-luminance natural image. In this case, the steepness of the natural image portion (steep and precise natural image portion) with a sharp change in luminance is not lost by the filter operation.
[0027]
That is, when a low pass filter is applied to a plurality of pixels around the edge, the low pass filter is not applied if the edge interval is narrow (eg, less than N pixels) as an exception process. The number N of pixels representing the edge interval is a value proportional to the enlargement ratio of the image. Thereby, it is possible to prevent the steepness of the enlarged image from being lost by the filter processing.
[0028]
1 to 3 are diagrams for explaining a noise reduction apparatus according to an embodiment of the present invention. 1 to 3, the same parts as those shown in FIGS. 4 to 8 are denoted by the same reference numerals, and the description thereof is omitted. In FIG. 1, CTL is a control signal including a control signal EPN for generating the luminance part enable signal generation unit and a control signal HMODE for controlling a filter execution interval threshold used in the luminance part flag generation unit described later. is there.
[0029]
FIG. 2 is a diagram illustrating a configuration of the luminance part flag generation unit 101. The edge detection unit 115 detects an edge of the input data Yin, and an edge left direction pixel filter range setting unit 116 described later on the detected signal (EDGE). , And right edge pixel filter range setting unit 118. The input data Yin is also input to the high level detector 117. The high level detection unit 117 generates a flag 2 by detecting pixels that are equal to or higher than a preset threshold value for the input data Yin.
[0030]
FIG. 3 is a diagram illustrating an edge left direction pixel filter range setting unit and an edge right direction pixel filter range setting unit. FIG. 3A is a diagram illustrating a configuration of an edge left direction pixel filter range setting unit. 3B is a diagram illustrating the configuration of the edge right direction pixel filter range setting unit, and FIG. 3C is a diagram illustrating the threshold value setting units 1163 and 1183 thereof.
[0031]
As described with reference to FIGS. 7 and 8, N pixels (for example, four pixels) leftward from the signal EDGE are set as filter target pixels based on the signal EDGE detected by the edge detection unit 115, and corresponding thereto. A range signal flag 1 is generated. Since the flag 1 is generated with a delay from the input data Yin, it is necessary to relatively delay the target pixel data to be filtered. Therefore, the input data Yin is delayed by the delay unit 103 by a necessary amount. Further, a flag F-EDGE indicating a pixel position for changing the filter processing within the filter target range is generated.
[0032]
Further, when the detected edge interval is less than N pixels (for example, less than 5 pixels), the flag 1 is not generated so as not to be a filter target pixel. The threshold value of the edge interval (N pixels) can be arbitrarily set as the filter execution pixel interval threshold value from the outside. By setting this value in accordance with the enlargement ratio of the image, it is possible to prevent a sensitive reaction when the pixel value suddenly changes due to external noise or the like. In addition, it is possible to prevent averaging (image blurring) between greatly different pixel values.
[0033]
FIG. 3C illustrates the operation of the threshold setting units 11863 and 1183. As shown in the figure, the enlargement ratio of the image is calculated based on the control signal HMODE, and the filter execution pixel interval threshold value is set according to the enlargement ratio.
[0034]
Here, FIG. 11A is a diagram for explaining the details of the filter target determination unit (for the left direction), and FIG. 11B is a diagram for explaining the operation thereof. An edge interval count value is input from the edge measurement counter unit, and an EDGE signal is input from the edge detection unit to the filter target determination unit (for left direction) 1162. The filter execution pixel interval threshold value is a parameter for determining that the filter process is executed if the pixel interval between edges differs depending on the mode and the pixel interval between the edges is equal to or greater than the threshold value. The edge interval count value is taken into sig1 by the EDGE signal pulse. This is compared with the filter execution pixel interval value sig2, and if sig1 is not less than sig2, sig3 is set to 1. In other words, when EDGE comes, the edge interval count value is used to see how many pixels have passed since the previous EDGE. If this value is a certain value (for example, 4 pixels or more), the value immediately before the edge is obtained. A signal sig3 for permitting filter processing for four pixels is generated.
[0035]
When EDGE comes, sig5 starts counting up from 0, and when this becomes the same value as sig6, sig7 becomes 1. By entering the reset of the self-holding unit, sig4 becomes 0. When EDGE comes, sig4 becomes 1 by the self-holding function, but becomes 0 by the reset signal, so that 4 pixels after EDGE are enabled. This operates regardless of the number of pixels between edges, but sig3 is a signal that is enabled (1) depending on the number of pixels between edges. An AND of sig3 and sig4 is taken to generate a flag 1, and when the flag 1 is 1, an instruction is given to filter the corresponding pixel.
[0036]
Since the maximum value of the filter execution pixel interval value is 10, sig5 stops counting up at 11. The AND of sig7 and sig3 is F-EDGE and indicates a pixel position for changing the filter processing within the filter target range.
[0037]
FIG. 12A is a diagram for explaining the details of the filter target determination unit (for the left direction), and FIG. 12B is a diagram for explaining the operation thereof.
[0038]
An edge interval count value is input to the filter target determination unit (for left direction) 1182 from the edge measurement counter unit, and an EDGE signal is input from the edge detection unit. The filter execution pixel interval threshold value is a parameter for determining that the filter process is executed if the pixel interval between edges differs depending on the mode and the pixel interval between the edges is equal to or greater than the threshold value. Further, the flag 2 ′ is input from the high level detection unit 117.
[0039]
The edge interval count value is taken into sig1 by the EDGE signal pulse. This is compared with the filter execution pixel interval value sig2, and if sig1 is not less than sig2, sig3 is set to 1. In other words, when EDGE comes, the edge interval count value is used to see how many pixels have passed since the previous EDGE. If this value is a certain value (for example, 5 pixels or more), the value immediately before the edge is obtained. A signal sig3 for permitting filter processing for four pixels is generated.
[0040]
When EDGE comes, sig5 starts counting up from 0, and when this becomes the same value as sig6, sig7 becomes 1. By entering the reset of the self-holding unit, sig4 becomes 0. When EDGE comes, sig4 becomes 1 by the self-holding function, but becomes 0 by the reset signal, so that 4 pixels after EDGE are enabled. This operates regardless of the number of pixels between edges, but sig3 is a signal that is enabled (1) depending on the number of pixels between edges. An AND of sig3 and sig4 is taken to generate a flag 1, and when the flag 1 is 1, an instruction is given to filter the corresponding pixel.
[0041]
Since the maximum value of the filter execution pixel interval value is 10, sig5 stops counting up at 11. The AND of sig7 and sig3 is F-EDGE and indicates a pixel position for changing the filter processing within the filter target range.
[0042]
The flag 2 delays the flag 2 ′ detected from the high level detection unit 117 by 4 pixels, and when EDGE is continuously generated at a short interval (less than 5 pixels) on the right side of the high level detection, this signal is transmitted to the next edge. Extend it. This prevents erroneous detection of edges due to noise or the like.
[0043]
Flag 3 is obtained by masking flag 3 ′ obtained by delaying sig4 by 5 pixels with flag 2, and indicates a pixel to be filtered. When this is enabled (1), the pixels are filtered.
[0044]
D-EDGE is a mask of B-EDGE ′ obtained by delaying sig7 by two pixels with a flag 2, and indicates a pixel position for making the filter process variable within the filter target range.
[0045]
As described above, it is possible to prevent a filter from being applied to an exceptional image such as external noise having a narrow edge interval even when the exceptional image is enlarged. Further, the value of the edge interval to be filtered can be changed according to the enlargement ratio. As a result, it is possible to prevent the image from being blurred by an extra low-pass filter process.
[0046]
【The invention's effect】
As described above, according to the present invention, it is possible to provide a noise suppression device and a noise suppression method capable of performing an appropriate filter process according to an image enlargement ratio.
[Brief description of the drawings]
FIG. 1 is a diagram illustrating a noise reduction device according to an embodiment of the present invention.
FIG. 2 is a diagram illustrating a configuration of a luminance part flag generation unit 101;
FIG. 3 is a diagram illustrating an edge left direction pixel filter range setting unit and an edge right direction pixel filter range setting unit.
FIG. 4 is a diagram illustrating a conventional noise reduction device.
5 is a diagram showing a configuration of a luminance part flag generation unit 101. FIG.
6 is a diagram showing details of an edge detection unit 115. FIG.
FIG. 7 is a diagram illustrating an edge left direction pixel filter range setting unit.
FIG. 8 is a diagram illustrating an edge left direction pixel filter range setting unit;
FIG. 9 is a diagram illustrating a specific example of processing by the low-pass filter during image enlargement.
FIG. 10 is a diagram illustrating a specific example of processing by the low-pass filter during image enlargement.
FIG. 11 is a diagram illustrating details of a filter target determination unit (for left direction).
FIG. 12 is a diagram illustrating details of a filter target determination unit (for left direction).
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 100 Luminance part data processing part 100b Color difference part (BY) Data processing part 100r Color difference part (RY) Data processing part 101 Luminance part flag generation part 102 Luminance part enable signal generation part 103 Delay part 104 Filter part (LPF)
115 Edge detection unit 116 Edge left pixel filter range setting unit 117 High level detection unit 118 Edge right direction pixel filter range setting unit 1161, 1181 Edge interval measurement counter 1162 Filter target determination unit (for left direction)
1182 Filter target determination unit (for right direction)
1163, 1183 Threshold setting unit

Claims (3)

When image data is input, the edges of the image are detected from the input image data, and the pixels included between the detected edges are equal to or greater than a set value, the pixels included between the edges are filtered by a low-pass filter. A flag generation unit that is designated as a filtering target pixel to be applied;
Based on the designation of the flag generation unit, an enable signal generation unit that specifies a low-pass filter calculation pixel to be performed for each filter processing target pixel;
A low-pass filter that filters the input image based on the enable signal generated by the enable signal generation unit;
A noise suppression apparatus comprising: an enlargement ratio determining unit that determines an enlargement ratio of the image and changes the set value based on the enlargement ratio. When image data is input, the edges of the image are detected from the input image data, and when the pixels included between the detected edges are equal to or greater than a set value, the pixels included between the edges are filtered by a low-pass filter. A flag generation unit that is designated as a filter processing target pixel, and is designated as a filter non-processing target pixel if less than a set value;
Based on the designation of the flag generation unit, an enable signal generation unit that specifies a low-pass filter calculation pixel to be performed for each filter processing target pixel;
A low-pass filter that performs filtering on the input image based on the enable signal generated by the enable signal generation unit;
A noise suppression apparatus comprising: an enlargement ratio determining unit that determines an enlargement ratio of the image and enlarges the set value substantially in proportion to the enlargement ratio. Inputting image data and detecting each edge of the image from the input image data;
If the pixel included between the detected edges is greater than or equal to the set value, the pixel included between the edges is specified as a filter processing target pixel to be filtered by a low-pass filter, and if it is less than the set value, specified as a non-filter processing target pixel And the process of
Generating an enable signal for designating a low-pass filter calculation pixel to be applied to each filter processing target pixel based on the designation in the step;
Applying low pass filter processing to the input image based on the enable signal;
A noise suppression method comprising: determining an enlargement ratio of the image, and determining an enlargement ratio for enlarging the set value according to the enlargement ratio.

Images