JP5039017B2 - Noise level detector - Google Patents

Description

  The present invention relates to a noise level detection device that detects the level of a noise component superimposed on a digital video signal.

  In general, in addition to the original video itself, noise added via a transmission path or the like is superimposed on the analog video signal. Conventionally, as a technique for reducing such noise, there is a technique called a noise reducer, and various methods have been proposed so far (see, for example, Patent Document 1).

  In order to operate this noise reducer, it is necessary to detect the level of the noise component in the screen (noise level) as a reference for reducing noise. For example, a noise reducer (noise reduction circuit) described in Patent Literature 1 detects a noise level from a signal in a blanking period of an analog video signal. Since the signal is flat (the voltage level is constant) during this blanking period, if even a slight signal component is superimposed during this period, it can be regarded as noise. Therefore, conventionally, noise reduction processing corresponding to the amount of noise is performed by regarding the noise superimposed on the signal in the blanking period as the noise superimposed on the entire video signal.

Digital video signals, on the other hand, are less likely to degrade than analog video signals on the transmission line because the video signal data is discrete values, and even when transmission noise is superimposed, noise is reduced by error correction etc. Can be made.
However, in recent years, noise (camera noise) superimposed on a signal output from a high-definition camera has become a problem in digital video signals. Conventionally, the noise mixed on the transmission path has surpassed the camera noise, so that the noise has not been a problem. However, as in Super Hi-Vision, the reduction in the area of high-definition solid-state imaging devices (CCD, CMOS sensor, etc.) reduces the light energy obtained per unit photoreceptor, so noise is mixed into the output image. In particular, a large amount of noise is superimposed on the video shot especially at night.
Japanese Patent No. 3299026 (paragraph 0031)

As described above, a recent digital video signal has a problem that camera noise is superimposed as the video becomes higher in definition.
This noise has fine sand particles and moves on the screen, so the image quality of the video is degraded. Further, the fact that the particles are fine means that they are high-frequency components, and this noise increases the amount of information during encoding. For this reason, when a video including a lot of noise is encoded, there is a problem that transmission bits are occupied to transmit the noise, and the decoded video becomes a video that is further deteriorated than the original video.
Further, since there is no blanking period in the digital video signal unlike the analog video signal, there is a problem that noise cannot be detected by a conventional noise reducer.

  The present invention has been made to solve the above-described problems, and is a noise level detection device capable of detecting a superimposed noise level even for a digital video signal having no blanking period. The purpose is to provide.

  The present invention was devised to achieve the above object. First, the noise level detection apparatus according to claim 1 detects a noise level which is a level of a noise component superimposed on a digital video signal. The noise level detection apparatus is configured to include block division means, orthogonal transform means, high frequency component detection means, high frequency component accumulation means, and noise level determination means.

  In such a configuration, the noise level detection apparatus divides the digital video signal into blocks having a predetermined size for each screen by the block dividing means. The block obtained by dividing the screen includes at least a flat portion with little change in pixel value in the screen. And a noise level detection apparatus orthogonally transforms the block divided | segmented by the block division means for every block by the orthogonal transformation means. As a result, the pixel value of the block is converted into the coefficient of the frequency component.

  The noise level detection device averages the coefficient of the high frequency component excluding the DC component of the block transformed by the orthogonal transform unit and the vicinity of the predetermined DC component by the high frequency component detection unit in the block, It is detected as the level of the high frequency component.

  Furthermore, the noise level detection apparatus accumulates the number of blocks corresponding to the level of the high frequency component detected by the high frequency component detection means for each level by the high frequency component accumulation means. Note that the high-frequency component of a block composed of a flat portion with little change in pixel value is almost “0” if there is no noise, so the high-frequency component of the flat block can be regarded as noise.

  Therefore, the noise level detection device determines that the noise level determination means determines, as a noise level, an average value corresponding to the number of blocks corresponding to a predetermined number from the low level value of the high frequency component level accumulated by the high frequency component accumulation means. To do. The low level value may be the minimum value of the high-frequency component level, or a value that is larger than the minimum value by a predetermined level may be used. Thereby, the high-frequency component of the flat block can be detected with high accuracy, and the noise level can be determined.

  Further, the noise level detection device according to claim 2 is the noise level detection device according to claim 1, further comprising a luminance detection means.

  In this configuration, the noise level detection device accumulates the number of blocks corresponding to the level of the high frequency component for each luminance detected by the luminance detection unit and for each level of the high frequency component by the high frequency component accumulation unit. Then, the noise level detection device corresponds to the luminance with an average value corresponding to the predetermined number of blocks from the low level value of the high frequency component level for each luminance accumulated by the high frequency component accumulating unit by the noise level determination unit. The noise level is judged and output. Thereby, the noise level detection device can determine the noise level for each luminance.

  According to a third aspect of the present invention, there is provided the noise level detecting device according to the second aspect, wherein the high-frequency component accumulating unit is a representative luminance for each predetermined luminance range of the luminance detected by the luminance detecting unit. The number of blocks for each level of the high frequency component is accumulated, and the noise level determination means includes a representative luminance noise determination means and a luminance-specific noise calculation means in order to detect the noise level from the accumulated value; did.

  In such a configuration, the noise level detection device calculates an average value corresponding to the number of blocks corresponding to a predetermined number from the low level value of the high-frequency component level accumulated for each representative luminance by the representative luminance noise determination unit. Judged as noise level. Then, the noise level detection device calculates the noise level for each luminance by linear interpolation from the noise level with respect to the representative luminance determined by the representative luminance noise determining unit by the luminance-specific noise calculating unit. As a result, the noise level detection device can obtain a noise level corresponding to the luminance that has not been accumulated from the noise level of the representative luminance by calculation (linear interpolation).

  Furthermore, the noise level detection device according to claim 4 is the noise level detection device according to claim 2, wherein the high-frequency component accumulating means has a representative luminance for each predetermined luminance range of the luminance detected by the luminance detection means. The number of blocks for each level of the high frequency component is accumulated, and the noise level determination means includes a representative luminance noise determination means and a luminance-specific noise determination means in order to detect the noise level from the accumulated value; did.

  In such a configuration, the noise level detection device determines the noise level for the representative luminance based on the level of the high-frequency component accumulated for each representative luminance by the representative luminance noise determination means. Then, the noise level detection apparatus sets the noise level for the representative luminance determined by the representative luminance noise determination unit by the luminance-specific noise determination unit as the noise level of the luminance corresponding to the luminance range to which the representative luminance belongs. Thereby, the noise level detection apparatus can determine the noise level corresponding to the luminance that has not been accumulated as the same noise level as the representative luminance for each luminance range to which the representative luminance belongs.

  The noise level detection apparatus according to claim 5 is a noise level detection apparatus that detects a noise level that is a level of a noise component superimposed on a digital video signal, and includes a block dividing unit, an orthogonal transform unit, The high frequency component detection means and the noise level determination means are provided.

  In such a configuration, the noise level detection apparatus divides the digital video signal into blocks having a predetermined size for each screen by the block dividing means. And a noise level detection apparatus orthogonally transforms the block divided | segmented by the block division means for every block by the orthogonal transformation means.

The noise level detection device averages the coefficient of the high frequency component excluding the DC component of the block transformed by the orthogonal transform unit and the vicinity of the predetermined DC component by the high frequency component detection unit in the block, It is detected as the level of the high frequency component.
The noise level detection device determines the noise level determination means to determine a low level value of the level of the high frequency component detected by the high frequency component detection means or an average value for a predetermined number from the low level value as a noise level and output To do. As a result, the noise level detection device can accurately detect the high-frequency component of the flat block and determine the noise level.

  A noise level detection device according to a sixth aspect of the present invention is the noise level detection device according to the fifth aspect, further comprising a luminance detection means.

  In such a configuration, the noise level detection device is configured to calculate, for each luminance, a low level value of a level of a high frequency component or an average value for a predetermined number from the low level value for each luminance detected by the luminance detection unit. The noise level is judged and output. Thereby, the noise level detection device can determine the noise level for each luminance.

  According to a seventh aspect of the present invention, there is provided a noise level detection apparatus according to the sixth aspect, wherein the noise level determination means includes representative luminance noise determination means and luminance-specific noise calculation means. did.

  In such a configuration, the noise level detection device is configured such that the representative luminance noise determination unit performs a low-level value of the high-frequency component level or the low-level value for each representative luminance in a predetermined luminance range of the luminance detected by the luminance detection unit. The average value for a predetermined number is determined as the noise level for each representative luminance. Then, the noise level detection device calculates the noise level for each luminance by linear interpolation from the noise level with respect to the representative luminance determined by the representative luminance noise determining unit by the luminance-specific noise calculating unit. As a result, the noise level detection device can obtain a noise level corresponding to a luminance that is not representative luminance by calculation (linear interpolation) from the noise level of representative luminance.

  Furthermore, the noise level detection apparatus according to claim 8 is the noise level detection apparatus according to claim 6, wherein the noise level determination means includes representative luminance noise determination means and luminance-specific noise determination means. did.

  In such a configuration, the noise level detection device determines the noise level for the representative luminance based on the level of the high-frequency component accumulated for each representative luminance by the representative luminance noise determination means. Then, the noise level detection apparatus sets the noise level for the representative luminance determined by the representative luminance noise determination unit by the luminance-specific noise determination unit as the noise level of the luminance corresponding to the luminance range to which the representative luminance belongs. Accordingly, the noise level detection apparatus can determine the noise level corresponding to the luminance that is not representative luminance as the same noise level as the representative luminance for each luminance range to which the representative luminance belongs.

The noise level detection device according to claim 9 is the noise level detection device according to any one of claims 3, 4, 7, and 8, wherein the brightness range increases as the brightness increases. It is characterized by a large setting.
In such a configuration, the noise level detection device can detect the noise level according to the visual accuracy while limiting the number of representative luminances by increasing the luminance range of high luminance in which the visual accuracy with which humans recognize noise is reduced. It becomes possible.

  The noise level detection device according to claim 10 is the noise level detection device according to any one of claims 1 to 9, further comprising a block selection unit.

In such a configuration, the noise level detection apparatus partially selects the blocks divided by the block dividing means by the block selecting means so that the ratio is equal to or higher than a predetermined ratio with respect to the entire screen in the screen. This ratio is desirably 50% or more (more preferably 70% or more) of the entire screen.
Then, the noise level detection apparatus performs orthogonal transformation on the block selected by the block selection unit by the orthogonal transformation unit. As a result, the noise level detection apparatus can select a block including at least a flat portion with little change in pixel value in the screen, and can be used as a block for detecting the noise level.

Furthermore, the noise level detection apparatus according to claim 11 is configured such that the noise level determination means determines the noise level based on a level for a predetermined time in a field or frame of the digital video signal.
In such a configuration, the noise level detection apparatus determines the noise level not only from one screen but also from a plurality of screens. As a result, it is possible to disperse the error in the case of determining on one screen.

The present invention has the following excellent effects.
According to the first and fifth aspects of the present invention, it is possible to detect a noise level superimposed on a digital video signal even if the digital video signal has no blanking period. Thus, by applying the present invention to a noise reducer, noise can be accurately removed from a digital video signal.

  According to the second and sixth aspects of the invention, in addition to the effect that the noise level can be detected from the digital video signal having no blanking period, the noise level can be detected for each luminance. Thus, the present invention can detect a noise level suitable for visual performance with respect to human brightness. Further, by applying the present invention to a noise reducer, it is possible to adjust the amount of noise removal according to the luminance.

  According to the third, fourth, seventh, and eighth aspects of the invention, the noise level is determined only for the representative luminance, and the other luminances can be obtained from the noise level of the representative luminance. The load can be reduced.

  According to the invention described in claim 9, by increasing the luminance range as the luminance increases, it is possible to detect the noise level for each luminance with less representative luminance while maintaining the visual performance for human luminance. become.

  According to the tenth aspect of the present invention, by selecting blocks partially at a predetermined ratio or more with respect to the screen, it is possible to detect a flat portion with little change in pixel value, and noise from the entire screen. Compared with the case of detecting the level, the noise level can be detected from the digital video signal with a smaller amount of calculation.

  According to the invention described in claim 11, since the noise level can be detected not only by one screen but also by a plurality of screens, the error is distributed as compared with the case of detecting by one screen, and the noise level detection accuracy is improved. Can be increased.

<Outline of the present invention>
The noise level detection apparatus according to the present invention detects a noise level that is a level of a noise component superimposed on a digital video signal (hereinafter simply referred to as a video signal).
Usually, a video screen (image) has a flat portion with little change in pixel value, for example, an infinite region such as the sky, a region where the camera is not in focus, or the like. These areas are areas where the frequency components of the image have few high frequency components. On the other hand, noise is a high-frequency component with fine particles. That is, in the screen, the high frequency component in the region with a low high frequency component can be regarded as being superimposed by noise.

  Therefore, the noise level detection apparatus according to the present invention detects the noise level from the high frequency component in the region where the high frequency component is small in the screen. Hereinafter, a noise level detection apparatus according to the present invention will be described with reference to the drawings.

<First Embodiment>
[Configuration of noise level detector]
First, the configuration of the noise level detection apparatus according to the first embodiment of the present invention will be described with reference to FIG. FIG. 1 is a block diagram showing the configuration of the noise level detection apparatus according to the first embodiment of the present invention. Here, the noise level detection apparatus 1 includes blocking means 10, orthogonal transform means 20, high frequency component detection means 30, high frequency component accumulation means 40, and noise level determination means 50.

  The blocking unit 10 converts the video signal into data for each block having a predetermined size. The blocking unit 10 receives a video signal from the outside, and outputs the divided blocks to the orthogonal transform unit 20. Here, the blocking unit 10 includes a block dividing unit 11 and a block selecting unit 12.

  The block dividing means 11 divides the video signal into blocks having a predetermined size for each screen, that is, for each field or frame constituting the screen. Although the size of this block is not particularly limited, for example, the block is divided into sizes of about 4 × 4 pixels to 128 × 128 pixels.

  The block selection unit 12 partially selects the blocks divided by the block division unit 11 so that the ratio is equal to or higher than a predetermined ratio with respect to the entire screen in the screen. The block selected by the block selecting unit 12 is output to the orthogonal transform unit 20. For example, as shown in FIG. 2A, the block selection unit 12 provides a gap for each block B in the screen D and covers an area of 50% or more (more preferably 70% or more) of the entire screen. A plurality of blocks B, B,. As a result, a block including a region that is a candidate for a flat portion with a small change in pixel value in the screen is selected.

  Here, the block selection unit 12 selects a block partially so that the ratio is equal to or greater than a predetermined ratio from the screen. However, as shown in FIG. It is good also as selecting. In this case, the block selecting unit 12 is omitted from the configuration, and all the blocks divided by the block dividing unit 11 are output to the orthogonal transform unit 20.

The orthogonal transform unit 20 performs orthogonal transform on the block output from the blocking unit 10 for each block. This orthogonal transform means 20 transforms each block into a frequency component by performing orthogonal transform on each block. This frequency component is output to the high frequency component detection means 30. For example, the orthogonal transform unit 20 transforms a block into a frequency component by a discrete cosine transform (DCT) which is a kind of orthogonal transform. Thereby, as shown in FIG. 3, for example, the pixel values (p _{11 to} p ₁₄ , p _{21 to} p ₂₄ , p _{31 to} p ₃₄ , p _{41 to} p ₄₄ ) of the block B of 4 × 4 pixels It is converted into a value (DCT coefficient; x _{11 to} x ₁₄ , x _{21 to} x ₂₄ , x _{31 to} x ₃₄ , x _{41 to} x ₄₄ ) indicating the same number of frequency components as the number. The DCT coefficients in the figure, the upper left coefficient x ₁₁ represents a DC component, lower right, it becomes higher frequency toward a lower left direction.

  In addition, as long as this orthogonal transformation transforms the discrete signal of a block into a frequency domain, you may use transformations other than DCT transformation. For example, discrete sine transform (DST), fast Fourier transform (FFT), Hadamard transform, or the like can be used. Thus, by converting the discrete signal of a block into the frequency domain, it becomes possible to analyze which frequency component each block has.

  The high frequency component detection means 30 averages the coefficients of the high frequency components excluding the DC component of the block converted by the orthogonal transform means 20 and the predetermined vicinity of the DC component in the block, and obtains the level of the high frequency component of the block. It is to detect.

  In the orthogonally transformed frequency component, the low frequency component including the direct current component is a component including a large amount of the frequency component of the image signal. Therefore, here, the high-frequency component detection means 30 sets the coefficient of the low-frequency component to “0” and averages the coefficient of the high-frequency component in the other region within the block, so that the level of the high-frequency component of the block Will be considered. This averaging can be calculated by the square root of the mean square sum of the coefficients of each high-frequency component, the absolute value average, or the like.

  Here, with reference to FIG. 4 (refer to FIG. 1 as appropriate), a method in which the high frequency component detection means 30 detects the level of the high frequency component from the frequency component of the block will be described. FIG. 4 is a schematic explanatory diagram for explaining a technique for detecting the level of the high frequency component from the frequency component of the block. 4A shows a frequency coefficient (corresponding to FIG. 3B) of a certain block obtained by the orthogonal transform means 20, and FIG. 4B shows a state in which only a high frequency component is extracted from the frequency component. Show. Here, in order to simplify the description, the block size is assumed to be 4 × 4 pixels.

As shown in FIG. 4 (a), it is assumed that a certain block is orthogonally transformed by the orthogonal transform means 20, and the same number of coefficients as the number of pixels in the block (here, 4 × 4) are obtained.
At this time, the high frequency component detecting means 30 regards the DC component and the coefficient of the horizontal frequency component and the vertical frequency component including the DC component as “0” as the coefficients in the vicinity of the DC component and the DC component (FIG. 4B). )), And extract other coefficients as high-frequency component coefficients. That is, the coefficients x ₁₁ , x ₁₂ , x ₁₃ , x ₁₄ , x ₂₁ , x ₃₁ , x ₄₁ are regarded as “0”, and the other coefficients x ₂₂ , x ₂₃ , x ₂₄ , x ₃₂ , x ₃₃ , x _{34 are considered.} , X ₄₂ , x ₄₃ , x ₄₄ are extracted as high frequency component coefficients.

  Then, the high frequency component detecting means 30 averages each coefficient extracted as the high frequency component coefficient within the block, thereby obtaining the level of the high frequency component of the block. Here, when averaging is performed using the square root of the mean square sum, the high frequency component detection means 30 calculates the level L of the high frequency component according to the following equation (1).

  Alternatively, in the case of averaging by absolute value averaging, the high frequency component detection means 30 calculates the level L of the high frequency component by the following equation (2).

  In the above equations (1) and (2), the coefficient 16/9 is a normalization coefficient, and is a coefficient for normalizing nine high frequency component coefficients to 16 coefficients of the entire block. It is. By normalizing in this way, the level of the high frequency component for the entire block can be obtained.

Here, as the DC component and the coefficient in the vicinity of the DC component, the horizontal and vertical frequency components including the DC component are regarded as “0”, and the level of the high frequency component of the block is calculated, but the coefficient regarded as “0” Various patterns can be used as long as they contain a DC component. For example, as shown in pattern A of FIG. 5 (a), the high frequency component detection means 30 positions the coefficient of the minimum vertical frequency component and the maximum horizontal frequency component, and the position of the coefficient of the maximum vertical frequency component and the minimum horizontal frequency component. Are divided into two regions R ₁ and R ₂ , the coefficient of the region R _{1 including} the DC component is regarded as “0”, and the level of the high frequency component is calculated by the coefficient of the other region R ₂ To do.

Further, for example, as shown in FIG. 5B, the diagonal line shown in FIG. 5A may be curved so as to include more high-frequency components and divided into two regions R ₁ and R _2. . As a result, more high-frequency components in the image can be extracted. Further, for example, as shown in FIG. 5C, a region R ₁ partially including the coefficient on the DC component side is compared to the region on the high frequency component side excluding the horizontal frequency component including the DC component and the vertical frequency component. it may be classified blocks in the region R ₂ otherwise. As a result, among the high frequency components in the image, it is possible to remove the components close to the DC component and extract the high frequency components.
Returning to FIG. 1, the description of the configuration of the noise level detection apparatus will be continued.

  The high frequency component accumulating unit 40 accumulates the number of blocks corresponding to the level of the high frequency component detected by the high frequency component detecting unit 30 for each level. The high-frequency component accumulating means 40 accumulates (histograms) the number of blocks corresponding to the level of the high-frequency component for each level or field unit constituting the screen (see FIG. 6). As a result, it is analyzed which level of which high-frequency component is distributed in the screen. The accumulated value for each level accumulated by the high frequency component accumulating means 40 is output to the noise level determining means 50.

The noise level determination means 50 determines a low level value of the high frequency component level accumulated by the high frequency component accumulation means 40 or an average value corresponding to a predetermined number of blocks from the low level value as a noise level. is there. As described with reference to FIG. 4, the coefficients detected by the high frequency component detection means 30 as the high frequency components (for example, the coefficients x ₂₂ , x ₂₃ , x ₂₄ , x ₃₂ , x ₃₃ , x ₃₄ , x ₄₂ , x ₄₃ , x ₄₄ ) is almost “0” if it is a flat part with little change in the pixel value of the screen, so if noise is superimposed on the flat part, It can be said that the high frequency component is noise. Therefore, the noise level determination means 50 has a low level value of the level of the high frequency component or a predetermined number of the low level value (for example, 2 to 5) in the total of the blocks having the high frequency component accumulated by the high frequency component accumulation means 40. The average value (weighted average value) according to the number of blocks is determined as the noise level.

  Here, the minimum value of the high frequency component level accumulated by the high frequency component accumulating means 40 is used as the low level value of the high frequency component level. Note that the noise level of an image captured by a CCD or the like may differ depending on the brightness of the screen. When the minimum value of the high-frequency component level is used, the noise level may be determined at a low level. Therefore, the low level value is not necessarily limited to the minimum value, and a value larger than the minimum value within a predetermined range from the minimum value may be used.

  Here, the noise level determination method of the noise level determination means 50 will be described with reference to FIG. 6 (refer to FIG. 1 as appropriate). FIG. 6 is a graph showing the cumulative value (frequency) of the blocks for each level of the high frequency component as a histogram. Note that this graph is a diagram for visually explaining the accumulated value of the block, and does not mean that the high-frequency component accumulating means 40 generates the graph.

Usually, the high frequency component includes the high frequency component of the image and noise at the same time. At this time, in the graph of FIG. 6, the histogram on the right from the center of the graph having a large high-frequency component is considered to be a high-frequency component of the video (including noise). On the other hand, several histograms at the left end of the graph are considered to be almost noise-only components that do not include high-frequency components of video.
Therefore, the noise level determination means 50 calculates an average value from the cumulative values corresponding to several histograms at the left end of the graph and determines it as a noise level.

For example, as shown in FIG. 6, when determining the noise level from three histograms, the noise level determination means 50 selects L ₁ , L ₂ , L ₃ with low levels of high frequency components, and the respective frequencies. An average value (weighted average value) is obtained from C ₁ , C ₂ , and C ₃ by the following equation (3), and set as a noise level N _L.

Here, the noise level determination means 50 calculates the noise level by the weighted average of the levels of a plurality of high frequency components, but the noise level may be determined by the arithmetic average of the levels of the high frequency components without considering the frequency. Good. Further, the noise level determination means 50 may simply set the minimum high frequency component level (L _{1 in} the case of FIG. 6) as the noise level.

  And the noise level determination means 50 outputs the determined noise level to a noise reducer (not shown). Moreover, the noise level determination means 50 is good also as converting a noise level into a DC voltage, and outputting it to a noise reducer. As this noise reducer, a general noise reducer can be used, and for example, a noise reduction circuit described in Japanese Patent No. 3299026 can be used.

  By configuring the noise level detection apparatus 1 as described above, the noise level detection apparatus 1 can detect the noise level superimposed on the video signal from the video itself. Thereby, the noise level detection apparatus 1 can detect the noise level even for a digital video signal having no blanking period.

  Here, the example in which the noise level detection apparatus 1 detects the noise level of the video signal in units of screens (in the spatial direction) has been described. However, as illustrated in FIG. 7, a plurality of screens (a plurality of fields or frames) are detected. ) (Space + time direction). That is, the noise level detection apparatus 1 accumulates the blocks corresponding to the level of the high frequency component for a predetermined time (a time corresponding to a plurality of fields and a plurality of frames) in the high frequency component accumulating means 40, and the noise level determining means 50. Thus, the noise level may be determined based on the accumulated result.

  Further, here, the noise level detection apparatus 1 uses the low level value of the level or the low level value in the histogram of the block for each level of the high frequency component accumulated by the high frequency component accumulation unit 40 by the noise level determination unit 50. An average value (such as a weighted average) according to the predetermined number of blocks is determined as the noise level. However, the noise level determination means 50 is simply a low level value (minimum value) or a predetermined number of times from the low level value among the high frequency component levels detected by the high frequency component detection means 30 regardless of the number of blocks. The average value of the level values may be determined as the noise level. In this case, the noise level detection apparatus 1 can be configured by omitting the high-frequency component accumulating means 40 from the configuration.

[Operation of noise level detector]
Next, the operation of the noise level detection apparatus 1 will be described with reference to FIG. FIG. 8 is a flowchart showing the operation of the noise level detection apparatus according to the first embodiment of the present invention.

First, the noise level detection apparatus 1 divides the video signal into blocks of a predetermined size (for example, 4 × 4 pixels to 128 × 128 pixels) by the block dividing unit 11 of the blocking unit 10 (step S1). .
Furthermore, the noise level detection apparatus 1 causes the block selected by the block selecting unit 12 of the blocking unit 10 to have the block divided in step S1 at a predetermined ratio (for example, 70%) or more with respect to the entire screen in the screen. Thus, a partial selection is made (step S2). The block selected in this way includes a block including a flat portion in which a change in pixel value is small in the screen.

  And the noise level detection apparatus 1 converts each block into a frequency component by orthogonal transformation for every block selected by step S2 by the orthogonal transformation means 20 (step S3). Thereafter, the noise level detection apparatus 1 uses the high frequency component detection means 30 to average the coefficients of the high frequency components excluding the direct current component and the predetermined vicinity of the direct current component in the frequency component obtained in step S3. The level of the high frequency component of the block is detected (step S4). As a result, the noise level detection apparatus 1 detects both the high-frequency component level of the image on which the noise is superimposed and the high-frequency component level generated only by the noise.

  Then, the noise level detection apparatus 1 accumulates the number of blocks corresponding to the level of the high frequency component detected in step S4 by the high frequency component accumulating unit 40 for each screen (step S5). As a result, the noise level detection apparatus 1 can analyze the distribution of the level of the high frequency component of each block. At this time, it can be said that the block having a low level of the high frequency component generates the high frequency component only by noise.

Therefore, the noise level detection device 1 uses the noise level determination means 50 to obtain a minimum value (low level value) of the high-frequency component level accumulated in step S5 or a predetermined number (for example, 2 to 5) from the minimum value. The average value corresponding to the number of blocks is determined as the noise level in the screen (step S6).
Thereby, the noise level detection apparatus 1 can detect the noise level superimposed on the video signal from the video itself.

Second Embodiment
[Configuration of noise level detector]
Next, the configuration of a noise level detection apparatus according to the second embodiment of the present invention will be described with reference to FIG. FIG. 9 is a block diagram showing the configuration of the noise level detection apparatus according to the second embodiment of the present invention.

  In general, the appearance frequency of noise varies depending on the brightness of the screen. This is due to the non-linearity of the imaging device that captures video. Even if the same level of noise is mixed in the video, the bright and dark portions of the screen look different. This is due to the non-linearity of human vision. That is, when noise reduction is performed, it is desirable to detect the noise level adaptively according to the brightness of the pixel (DC value [luminance value]).

  Therefore, the noise level detection device 1B adds a function of detecting the noise level for each screen and for each luminance, whereas the noise level detection device 1 detects the noise level for each screen of the video signal.

  Here, the noise level detection apparatus 1B includes a blocking unit 10, an orthogonal transformation unit 20, a luminance detection unit 25, a high frequency component detection unit 30, a high frequency component accumulation unit (high frequency component accumulation unit) 40B for each luminance, Brightness level noise level determination means (noise level determination means) 50B. Since the configuration other than the luminance detection unit 25, the luminance-specific high-frequency component accumulation unit 40B, and the luminance-specific noise level determination unit 50B is the same as that of the noise level detection device 1 described in FIG. Is omitted.

  The luminance detection means 25 detects the luminance of the block that is blocked (selected) by the blocking means 10. Here, the luminance detection means 25 extracts only the direct current component from the frequency components orthogonally transformed for each block by the orthogonal transformation means 20 to obtain a luminance value. As another method for detecting the luminance of a block, the luminance detecting unit 25 is a block before orthogonal transformation, that is, for each block output from the blocking unit 10 (more specifically, the block selecting unit 12). In addition, the luminance of the block may be detected by obtaining an average of pixel values of several pixels.

  Here, with reference to FIG. 10, a method in which the luminance detection unit 25 detects the luminance from the block will be described. FIG. 10 is an explanatory diagram for explaining a method in which the luminance detecting means detects the luminance from the block. Here, in order to simplify the description, the block size is assumed to be 4 × 4 pixels.

  First, as shown in FIG. 10A, the first method uses the DC component DC as the luminance value among the frequency components when the block B is orthogonally transformed. This value is one of the frequency components output from the orthogonal transform means 20, and the brightness detection means 25 can use the value as the brightness value.

  In the second method, as shown in FIG. 10B, several pixels are selected from the pixels P of the block B, and the average luminance value (average DC value) is used as the luminance value of the block B. calculate. In the example of (b-1), an example is shown in which the luminance average value is calculated using 4 pixels (shaded lines in the figure) out of 4 × 4 pixels. Further, in the example of (b-2), an example is shown in which the luminance average value is calculated using 5 pixels (shaded lines in the figure) out of 4 × 4 pixels.

Note that the number of pixels for calculating the luminance average value is predetermined according to the size of the block. For example, when a block having a size of about 4 × 4 pixels to 128 × 128 pixels is used, the luminance detection unit 25 selects a pixel of about 4 to 64 pixels according to the size of the block, and calculates the average luminance value. calculate. In this way, since a predetermined number of pixels are selected in the block and the average value is obtained, the extraction positions of the pixels are extracted from arbitrary positions as shown in (b-2) of FIG. May be.
Returning to FIG. 9, the description of the configuration of the noise level detection apparatus 1B will be continued.

  The luminance-specific high-frequency component accumulating means (high-frequency component accumulating means) 40B accumulates the number of blocks for each level of the high-frequency component detected by the high-frequency component detecting means 30 for each block luminance detected by the luminance detecting means 25. It is. The high-frequency component accumulating unit 40B by luminance has the same function as the high-frequency component accumulating unit 40 described with reference to FIG. 1 except that the number of blocks for each level of the high-frequency component is accumulated by luminance. That is, the high frequency component accumulating unit 40B for each luminance generates the histogram described in FIG. 6 for each luminance.

  The luminance level-specific noise level determination means (noise level determination means) 50B sets the low-level value of the high-frequency component level for each luminance accumulated by the luminance-specific high-frequency component accumulation means 40B or the number of blocks corresponding to a predetermined number from the low level value. The corresponding average value is determined as a noise level corresponding to the luminance (noise level for each luminance). The brightness-specific noise level determination means 50B has the same function as the noise level determination means 50 described with reference to FIG. 1 except that the noise level is determined for each brightness. That is, the noise level determination means 50B for each luminance obtains a low level value of the level of the high frequency component or an average value corresponding to a predetermined number of blocks from the low level value from the histogram for each luminance shown in FIG. The noise level for each luminance.

  By configuring the noise level detection apparatus 1B as described above, the noise level detection apparatus 1B can detect the noise level superimposed on the video signal from the video itself. Furthermore, since the noise level detection device 1B can detect the noise level adaptively according to the brightness of the screen, noise is removed according to the brightness of the screen in the subsequent noise reduction (not shown). It becomes possible to do.

  Here, the noise level detection device 1B is configured to detect the level in the histogram of the blocks for each luminance and each high-frequency component level accumulated by the luminance-specific high-frequency component accumulating unit 40B by the luminance-specific noise level determination unit 50B. A low level value (minimum value) or an average value (such as a weighted average) corresponding to a predetermined number of blocks from the low level value is determined as the noise level. However, the noise level determination means 50B for each luminance is simply the low level value (minimum value) for each luminance or the low level value among the high frequency component levels detected by the high frequency component detection means 30 regardless of the number of blocks. The average value of a predetermined number of level values may be determined as the noise level. In this case, the noise level detection apparatus 1B can be configured by omitting the luminance-specific high-frequency component accumulating means 40B from the configuration.

[Operation of noise level detector]
Next, referring to FIG. 11 (refer to FIG. 9 as appropriate for the configuration), the operation of the noise level detection apparatus 1B will be described. FIG. 11 is a flowchart showing the operation of the noise level detection apparatus according to the second embodiment of the present invention. In addition, about operation | movement of step S11-S14, since it is the same as operation | movement of step S1-S4 demonstrated in FIG. 8, description is abbreviate | omitted.

  After step S14, the noise level detection apparatus 1B detects the luminance of the block selected in step S12 by the luminance detection means 25 (step S15). The operation in step S15 may be performed before step S14 or in parallel with step S14.

  After that, the noise level detection apparatus 1B accumulates the number of blocks for each level of the high-frequency component detected in step S14 for each luminance of the block detected in step S15 by the luminance-specific high-frequency component accumulation means 40B (step S16). ). Thereby, the noise level detection apparatus 1B can analyze the distribution of the level of the high-frequency component of each block for each luminance.

Therefore, the noise level detection apparatus 1B uses a minimum value (low level value) or a predetermined number of high frequency component levels for each luminance accumulated in step S16 by the luminance level-specific noise level determination unit 50B (for example, 2). The average value of (˜5) is determined as the noise level (noise level for each luminance) at the luminance in the screen (step S17).
Thus, the noise level detection device 1B can detect the noise level superimposed on the video signal from the video itself in accordance with the brightness of the video.

<Second Embodiment (Modification 1)>
Next, a modified example of the noise level detection apparatus 1B according to the second embodiment described in FIG. 9 will be described with reference to FIG. FIG. 12 is a block diagram showing a configuration of a modification example (No. 1) of the noise level detection apparatus according to the second embodiment of the present invention.

  In the noise level detection apparatus 1B described with reference to FIG. 9, the number of blocks for each level of high frequency components is accumulated for all luminances, and the noise level is detected for each luminance. However, the noise level detection apparatus 1C classifies the luminance in advance, performs the total of the high-frequency component levels using one luminance in the segment as the representative luminance, and interpolates from the accumulated value of the representative luminance of each segment. A noise level is detected for each luminance by obtaining a cumulative value of other luminances by (linear interpolation).

  Here, the noise level detection apparatus 1C includes a blocking unit 10, an orthogonal transform unit 20, a luminance detection unit 25, a high frequency component detection unit 30, and a representative luminance-specific high frequency component accumulation unit (high frequency component accumulation unit) 40C. And a noise level determination means (noise level determination means) 50C for each brightness. Since the configuration other than the representative luminance-specific high-frequency component accumulating unit 40C and the luminance-specific noise level determining unit 50C is the same as that of the noise level detection device 1B described with reference to FIG.

  The representative luminance-specific high-frequency component accumulating means (high-frequency component accumulating means) 40 </ b> C is detected by the high-frequency component detecting means 30 for representative luminance in a predetermined luminance range among the luminances of the blocks detected by the luminance detecting means 25. Accumulate the number of blocks for each level of high frequency components. The representative luminance-specific high-frequency component accumulating unit 40C accumulates the number of blocks corresponding to the level of the high-frequency component for each field or frame constituting the screen for each level. The histogram formation in the representative luminance-specific high-frequency component accumulating means 40C can be performed in the same manner as the high-frequency component accumulating means 40 (see FIG. 1) (see FIG. 6). Note that the representative luminance-specific high-frequency component accumulating unit 40C is different from the high-frequency component accumulating unit 40 in that this histogram is generated for each representative luminance.

  The number of divisions of the luminance range is not particularly limited. For example, the number of luminance ranges is 4 to 5 with respect to the luminance values “0” to “255”. In addition, each luminance range may be equal, but it is desirable to increase the luminance range as the luminance is brighter because the luminance accuracy is lower when the luminance accuracy is higher.

  The luminance level-specific noise level determination means (noise level determination means) 50C is a noise level for each luminance (noise level for each luminance) based on the accumulated high-frequency component level for each representative luminance accumulated by the representative luminance-specific high-frequency component accumulation means 40C. Is determined. Here, the luminance level noise level determination unit 50C includes a representative luminance noise determination unit 51 and a luminance-specific noise calculation unit 52.

  The representative luminance noise determining unit 51 determines a noise level for the representative luminance based on the level of the high frequency component accumulated for each representative luminance. That is, the representative luminance noise determination unit 51, like the noise level determination unit 50 (see FIG. 1), calculates a low level value of the level of the high frequency component or an average value corresponding to a predetermined number of blocks from the low level value. The noise level is determined for each representative luminance.

  The noise-by-brightness calculation unit 52 calculates a noise level for each luminance from the noise level with respect to the representative luminance determined by the representative luminance noise determination unit 51 by linear interpolation. The luminance-specific noise calculation means 52 determines the luminance value of the representative luminance before and after the luminance (if there is no representative luminance before, the minimum luminance value is “0”, and if there is no representative luminance after And the maximum luminance value “255”) and the noise level corresponding to the representative luminance, the noise level of the target luminance (noise level by luminance) is calculated by linear interpolation.

  Here, with reference to FIG. 13 (refer to FIG. 12 as appropriate), a method of calculating the luminance level by luminance in the luminance level by noise level determination unit 50C will be described. FIG. 13 is an explanatory diagram for explaining a method of calculating noise by luminance by linear interpolation.

  As shown in FIG. 13, here, as the range of the luminance range, the luminance values are “0” to “31”, “32” to “63”, “64” to “127”, “128” to “255”. It is divided into four categories. In addition, the representative luminance of each luminance range is “16”, “48”, “96”, “192”, which is an intermediate value of each luminance range. The noise level for each representative luminance is determined by the luminance-specific noise level determination means 50C.

  Here, when calculating the noise level of the luminance x (for example, 160), the noise-by-luminance calculation means 52 calculates from the levels corresponding to “96” and “192”, which are representative luminances before and after the luminance x. Then, the noise level α for each luminance is calculated by linear interpolation.

  By configuring the noise level detection device 1C as described above, the noise level detection device 1C can suppress the data amount of the histogram in which the high-frequency components are accumulated compared to the noise level detection device 1B (see FIG. 9). . Thus, in addition to the effects of the noise level detection device 1B (see FIG. 9), the noise level detection device 1C can realize a reduction in the amount of data and an increase in the processing speed associated therewith.

<Second Embodiment (Modification 2)>
Next, another modification of the noise level detection apparatus 1B according to the second embodiment described in FIG. 9 will be described with reference to FIG. FIG. 14: is a block diagram which shows the structure of the modification (the 2) of the noise level detection apparatus concerning 2nd Embodiment of this invention.

  The noise level detection devices 1B and 1C (see FIGS. 9 and 12) detected the noise level for each luminance. However, the noise level detection device 1D detects the noise level for each predetermined luminance section. That is, the noise level detection device 1D detects the noise level as the same noise level for each luminance section. In addition, this brightness | luminance area can be made into the same area as the brightness | luminance area used in 1 C of noise level detection apparatuses demonstrated in FIG.

  Here, the noise level detection apparatus 1D includes a blocking unit 10, an orthogonal transformation unit 20, a luminance detection unit 25, a high frequency component detection unit 30, and a representative luminance-specific high frequency component accumulation unit (high frequency component accumulation unit) 40C. And a noise level determination means (noise level determination means) 50D for each brightness. The luminance-specific noise level determination unit 50D includes a representative luminance noise determination unit 51 and a luminance-specific noise determination unit 53. That is, the noise level detection device 1D is configured by replacing the noise-specific noise calculation unit 52 with the luminance-specific noise determination unit 53, as compared with the noise level detection device 1C (see FIG. 9). Since the configuration other than the luminance-specific noise determination unit 53 is the same as that of the noise level detection apparatus 1C described with reference to FIG. 12, the same reference numerals are given and description thereof is omitted.

  The luminance-specific noise determination unit 53 determines the noise level for the representative luminance determined by the representative luminance noise determination unit 51 as the luminance noise level corresponding to the luminance range to which the representative luminance belongs. That is, the luminance-specific noise determination unit 53 determines, for each arbitrary luminance, the noise level corresponding to the representative luminance of the luminance range including the luminance as the noise level of the luminance.

  Here, with reference to FIG. 15 (refer to FIG. 14 as appropriate), a method of determining the luminance-specific noise level in the luminance-specific noise determination unit 53 will be described. FIG. 15 is an explanatory diagram for explaining a method of determining noise by luminance for each luminance range.

  As shown in FIG. 15, here, as the range of the brightness range, the brightness values are “0” to “31”, “32” to “63”, “64” to “127”, “128” to “255”. It is divided into four categories. In addition, the representative luminance of each luminance range is “16”, “48”, “96”, “192”, which is an intermediate value of each luminance range. The level of the high frequency component with respect to each representative luminance is determined by the representative luminance noise determining unit 51.

  Here, when determining the noise level of the luminance x (for example, 160), the noise-specific noise determining unit 53 uses the representative luminance in the luminance range (here, “128” to “255”) to which the luminance x belongs. A level corresponding to a certain “192” is determined as a noise level α for each luminance.

  By configuring the noise level detection device 1D as described above, the noise level detection device 1D can suppress the amount of calculation by linear interpolation compared to the noise level detection device 1C (see FIG. 12). Thus, the noise level detection device 1D can realize higher speed than the noise level detection device 1C (see FIG. 12).

It is a block block diagram which shows the structure of the noise level detection apparatus which concerns on 1st Embodiment of this invention. It is a schematic diagram which shows the area | region which selects the block in a screen, Comprising: (a) is a figure which shows the case where it selects with the clearance gap between blocks, (b) selects the whole screen as a block. It is a figure which shows the relationship between the pixel value of a block, and the frequency component by orthogonal transformation. It is typical explanatory drawing for demonstrating the method to detect the level of a high frequency component from the frequency component of a block. It is a figure which shows the example of a pattern for demonstrating the area | region of the frequency component used as the object in order to detect the level of a high frequency component. It is the graph which represented the cumulative value of the block for every level of a high frequency component as a histogram. It is a figure which shows the example which extracts a block over a some field or frame. It is a flowchart which shows operation | movement of the noise level detection apparatus which concerns on 1st Embodiment of this invention. It is a block block diagram which shows the structure of the noise level detection apparatus which concerns on 2nd Embodiment of this invention. It is typical explanatory drawing for demonstrating the method for a brightness | luminance detection means to detect a brightness | luminance from a block. It is a flowchart which shows operation | movement of the noise level detection apparatus which concerns on 2nd Embodiment of this invention. It is a block diagram which shows the structure of the modification (the 1) of the noise level detection apparatus which concerns on 2nd Embodiment of this invention. It is explanatory drawing for demonstrating the method of calculating the noise classified by brightness | luminance by linear interpolation. It is a block diagram which shows the structure of the modification (the 2) of the noise level detection apparatus which concerns on 2nd Embodiment of this invention. It is explanatory drawing for demonstrating the method of determining the noise according to brightness | luminance for every brightness | luminance range.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1, 1B, 1C, 1D Noise level detection apparatus 10 Blocking means 11 Block division means 12 Block selection means 20 Orthogonal transformation means 25 Luminance detection means 30 High frequency component detection means 40 High frequency component accumulation means 40B, 40C High frequency component accumulation by representative luminance Means (high-frequency component cumulative means)
50 Noise level determination means 50B, 50C Brightness-specific noise level determination means (noise level determination means)
51 representative luminance noise determining means 52 luminance calculating means 53 luminance determining means for luminance

Claims (11)

A noise level detection device for detecting a noise level that is a level of a noise component superimposed on a digital video signal,
Block dividing means for dividing the digital video signal into blocks of a predetermined size for each screen;
Orthogonal transform means for orthogonally transforming the blocks divided by the block dividing means for each block;
High-frequency component detection means for averaging the coefficient of the high-frequency component excluding the DC component of the block transformed by the orthogonal transform means and the predetermined vicinity of the DC component in the block and detecting it as the level of the high-frequency component of the block; ,
High-frequency component accumulating means for accumulating the number of blocks corresponding to the level of the high-frequency component detected by the high-frequency component detecting means for each level;
Noise level determination means for determining and outputting an average value corresponding to the number of blocks corresponding to a predetermined number of times from the low level value of the high frequency component level accumulated by the high frequency component accumulation means,
A noise level detection apparatus comprising: Further comprising a luminance detecting means for detecting the luminance for each block;
The high-frequency component accumulating means accumulates the number of blocks corresponding to the level of the high-frequency component for each luminance and for each level,
The noise level determination means determines an average value corresponding to a predetermined number of blocks from the low level value of the high frequency component level for each luminance accumulated by the high frequency component accumulation means as the noise level for each luminance. The noise level detection apparatus according to claim 1, wherein the noise level detection apparatus outputs the noise level. The high-frequency component accumulating unit accumulates the number of blocks for each level of the high-frequency component for the representative luminance for each predetermined luminance range of the luminance detected by the luminance detecting unit,
The noise level determination means includes
Representative luminance noise determining means for determining an average value corresponding to a predetermined number of blocks from a low level value of the high-frequency component level accumulated for each representative luminance as a noise level for each representative luminance;
From the noise level with respect to the representative luminance determined by the representative luminance noise determining unit, noise-by-brightness calculating unit for calculating the noise level for each luminance by linear interpolation;
The noise level detection apparatus according to claim 2, further comprising: The high-frequency component accumulating unit accumulates the number of blocks for each level of the high-frequency component for the representative luminance for each predetermined luminance range of the luminance detected by the luminance detecting unit,
The noise level determination means includes
Representative luminance noise determining means for determining an average value corresponding to a predetermined number of blocks from a low level value of the high-frequency component level accumulated for each representative luminance as a noise level for each representative luminance;
A noise-by-brightness determination unit that sets a noise level for the representative luminance determined by the representative luminance noise determination unit as a noise level of a luminance corresponding to the luminance category to which the representative luminance belongs;
The noise level detection apparatus according to claim 2, further comprising: A noise level detection device for detecting a noise level that is a level of a noise component superimposed on a digital video signal,
Block dividing means for dividing the digital video signal into blocks of a predetermined size for each screen;
Orthogonal transform means for orthogonally transforming the blocks divided by the block dividing means for each block;
High-frequency component detection means for averaging the coefficient of the high-frequency component excluding the DC component of the block transformed by the orthogonal transform means and the predetermined vicinity of the DC component in the block and detecting it as the level of the high-frequency component of the block; ,
A noise level determination means for determining and outputting a low level value of the level of the high frequency component detected by the high frequency component detection means or an average value for a predetermined number from the low level value as the noise level;
A noise level detection apparatus comprising: Further comprising a luminance detecting means for detecting the luminance for each block;
The noise level determining means determines a low level value of the level of the high frequency component for each luminance detected by the luminance detecting means or an average value for a predetermined number from the low level value as a noise level for each luminance. 6. The noise level detection apparatus according to claim 5, wherein the noise level detection apparatus outputs the noise level. The noise level determination means includes
For each representative luminance in a predetermined luminance range of the luminance detected by the luminance detecting means, a low level value of the level of the high-frequency component or an average value for a predetermined number from the low level value is calculated as noise for each representative luminance. Representative luminance noise determining means for determining a level;
From the noise level with respect to the representative luminance determined by the representative luminance noise determining unit, noise-by-brightness calculating unit for calculating the noise level for each luminance by linear interpolation;
The noise level detection apparatus according to claim 6, further comprising: The noise level determination means includes
For each representative luminance in a predetermined luminance range of the luminance detected by the luminance detecting means, a low level value of the level of the high-frequency component or an average value for a predetermined number from the low level value is calculated as noise for each representative luminance. Representative luminance noise determining means for determining a level;
A noise-by-brightness determination unit that sets a noise level for the representative luminance determined by the representative luminance noise determination unit as a noise level of a luminance corresponding to the luminance category to which the representative luminance belongs;
The noise level detection apparatus according to claim 6, further comprising:   The noise level detection device according to any one of claims 3, 4, 7, and 8, wherein the luminance range is set to be larger as the luminance is higher. Block selection means for partially selecting the blocks divided by the block division means so as to be a predetermined ratio or more with respect to the entire screen in the screen;
The noise level detection apparatus according to claim 1, wherein the orthogonal transform unit performs orthogonal transform on the block selected by the block selection unit.   11. The noise level determination unit determines the noise level based on a level for a predetermined time in a field or frame of the digital video signal. The noise level detection apparatus described in 1.

Images