JP2011029989A - Noise reduction apparatus, and noise reduction method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a noise reduction apparatus effectively reducing noise of video including noise caused by encoding. <P>SOLUTION: The noise reduction apparatus for reducing noise of respective frames in input video signal, includes: a cyclic noise reducing means for reducing noise included in a current frame using a preceding frame; an encoding noise reducing means for reducing noise caused by an encoding system of the input video signal; a generation means for acquiring a pixel value differential between the current frame and the preceding frame to generate a frequency distribution for the acquired value of the differential; and a control means for controlling degree of noise reduction by the cyclic noise reducing means and degree of noise reduction by the encoding noise reducing means based on the frequency distribution generated by the generation means. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a noise reduction device and a noise reduction method for reducing noise in a video signal.

  In recent years, video signals have been encoded by an encoding method such as MPEG. A video signal obtained by decoding an encoded video may include noise peculiar to an encoding method caused by encoding, and it is desired to effectively reduce this noise.

  Conventionally, as means for reducing noise in a video signal, a cyclic filter that reduces noise based on the correlation of the video signal between frames and a spatial filter that reduces noise based on the correlation of the video signal between pixels in the frame It has been known. The recursive filter tends to generate an afterimage in a video with a large motion. In addition, the spatial filter tends to reduce sharpness in an image with weak correlation between pixels such as a contour portion. In consideration of this point, a technique for controlling the strengths of the spatial filter and the recursive filter according to the characteristics of the video is known. For example, in Patent Document 1, the motion of a video is detected, the strength of a cyclic filter is increased for a video with small motion, and the strength of the cyclic filter is decreased for a video with large motion and a spatial filter. A technique for increasing the strength of the paper is disclosed.

JP 2002-10106 A

  A video encoded by MPEG may include random noise and block noise resulting from encoding. In order to effectively reduce noise for an image including random noise and block noise, it is desirable to increase the strength of both the recursive filter and the spatial filter. However, since the technique of Patent Document 1 can only increase the strength of either the cyclic filter or the spatial filter based on the determination result of whether or not the motion of the video is large, noise caused by encoding In some cases, it was not possible to sufficiently reduce the noise of images including

  The present invention has been made in view of such problems, and an object of the present invention is to provide a noise reduction device capable of effectively performing noise reduction on an image including noise caused by encoding. To do.

The noise reduction device of the present invention is
A noise reduction device for reducing noise of each frame of an input video signal,
A cyclic noise reduction means for reducing noise contained in the current frame using the previous frame;
Encoding noise reduction means for reducing noise caused by the encoding method of the input video signal;
Generating means for acquiring a pixel value difference between the current frame and the previous frame, and generating a frequency distribution of the acquired difference value;
Control means for controlling the noise reduction intensity by the cyclic noise reduction means and the noise reduction intensity by the coding noise reduction means based on the frequency distribution generated by the generation means;
It is characterized by providing.

The noise reduction method of the present invention includes:
A noise reduction method for reducing noise of each frame of an input video signal,
An input process for inputting a video signal;
A cyclic noise reduction process for reducing noise included in the current frame of the input video signal using the previous frame;
An encoding noise reduction step for reducing noise caused by the encoding method of the input video signal;
Generating a difference between pixel values of the current frame and the previous frame, and generating a frequency distribution of the acquired difference values;
Based on the frequency distribution generated by the generating step, controlling the noise reduction strength in the cyclic noise reduction step and the noise reduction strength in the coding noise reduction step;
It is characterized by providing.

  According to the present invention, it is possible to effectively perform noise reduction on an image including noise caused by encoding.

The block diagram which shows the structure of a noise reduction apparatus. The figure which shows the relationship between an inter-frame difference histogram and noise reduction intensity | strength. The block diagram which shows the structure of a control part. The figure which shows the relationship between MPEG noise reduction intensity and a histogram integrated value. The figure which shows the relationship between cyclic type noise reduction intensity | strength and a histogram integrated value. The block diagram which shows the structure of a control part. The figure which shows the example of the histogram in which a peak exists below 2nd threshold value Th2.

Example 1
A first embodiment of the present invention will be described with reference to the accompanying drawings. FIG. 1 is a block diagram illustrating the configuration of the noise reduction device according to the first embodiment. The noise reduction apparatus includes an MPEG noise reduction unit 1 as a coding noise reduction unit, a cyclic noise reduction unit 2 as a cyclic noise reduction unit, a histogram generation unit 3 as a generation unit, and a control unit 4 as a control unit. . The noise reduction device outputs a video signal 200 in which noise included in the input video signal 100 is reduced. This noise reduction device reduces random noise and block noise contained in decoded video in video equipment such as a digital television or a DVD player that decodes and displays and outputs video encoded by the MPEG encoding method. Can be used for

  The input video signal 100 includes a video signal encoded by the MPEG method. A video signal obtained by decoding an MPEG-encoded video signal may include block noise, which is noise caused by MPEG encoding. The MPEG noise reduction unit 1 is a spatial noise reduction that reduces block noise by performing smoothing processing on pixels around a block boundary in a frame based on block boundary information in MPEG encoding. It includes a filter.

The cyclic noise reduction unit 2 multiplies the difference between the current frame of the input video signal and the previous frame (previous frame) held in a frame memory (not shown) by a cyclic coefficient and adds the result to the current frame. As a result, noise in the current frame is reduced. Note that the previous frame held in the frame memory may be one frame before the video signal 200 output by the noise reduction device or one frame before the input video signal 100. good. Further, it may be two or more frames before. In short, any type of cyclic noise reduction unit may be used as long as the noise of the current frame can be reduced using the previous frame in an image having a strong correlation between frames.

  The histogram generation unit 3 acquires a pixel value difference between the current frame of the input video signal and a frame one frame previous (previous frame) held in a frame memory (not shown). A frequency distribution of absolute values is generated as a histogram. Note that the previous frame held in the frame memory may be one frame before the video signal 200 output by the noise reduction device or one frame before the input video signal 100. good. Further, it may be two or more frames before. In short, in order to generate a histogram, it suffices if the difference in pixel values between the current frame and the previous frame can be acquired for a plurality of pixels. The histogram generation unit 3 and the cyclic noise reduction unit 2 may share a frame memory, or may include separate frame memories.

  The control unit 4 controls the noise reduction strength by the MPEG noise reduction unit 1 and the noise reduction strength by the cyclic noise reduction unit 2 based on the histogram of the pixel value difference between frames generated by the histogram generation unit 3. . Control of the noise reduction intensity (MPEG noise reduction intensity) by the MPEG noise reduction unit 1 is performed, for example, by changing the range of pixels to be smoothed. When increasing the MPEG noise reduction strength, the smoothing target pixel range is set wide, and when reducing the MPEG noise reduction strength, the smoothing target pixel range is set narrow. Control of the noise reduction strength (cyclic noise reduction strength) by the cyclic noise reduction unit 2 is performed, for example, by changing a cyclic coefficient. When the cyclic noise reduction strength is increased, the cyclic coefficient is set large, and when the cyclic noise reduction strength is weakened, the cyclic coefficient is set small.

  The control of the MPEG noise reduction strength and the cyclic noise reduction strength by the control unit 4 will be described with reference to FIG. In the present embodiment, the control of the MPEG noise reduction strength and the cyclic noise reduction strength is roughly divided into three types based on the histogram generated by the histogram generation unit 3.

  Case 1 is a case where the frequency of the difference value equal to or higher than the first threshold (Th1 or higher) is equal to or lower than a predetermined first reference value (for example, 0 or lower) in the histogram. In this case, it can be determined that the motion of the input video signal is small. In the case of such a video with small motion, block noise hardly occurs even if it is encoded by the MPEG method. Therefore, in case 1, since it can be estimated that there is little noise due to MPEG encoding, the MPEG noise reduction strength is set weak. Further, in the case of an image in which only a small motion is detected, even if the cyclic noise reduction strength is increased, afterimage interference is unlikely to occur, so the cyclic noise reduction strength is set to be strong. Thereby, the random noise contained in the input video signal can be effectively reduced.

Case 2 is a case where the frequency of the difference value equal to or greater than the second threshold (Th2 (> Th1) or greater) in the histogram is equal to or greater than a predetermined second reference value. In this case, it can be determined that the motion of the input video signal is large. When such a large motion video is encoded by the MPEG method, block noise is likely to occur. Therefore, in case 2, since it can be estimated that there is a lot of noise due to MPEG encoding, the MPEG noise reduction strength is set strongly. Further, in the case of a video in which a large motion is detected, if the intensity of noise reduction by the cyclic noise reduction unit 2 is increased, afterimage interference is likely to occur. Therefore, the cyclic noise reduction intensity is set low. As a result, it is possible to effectively reduce block noise caused by MPEG encoding while suppressing occurrence of afterimage interference.

  In case 3, in the histogram, the frequency of the difference value equal to or higher than the second threshold value (Th2 or higher) is smaller than the second reference value, but is lower than the second threshold value after the first threshold value (Th1 or higher and lower than Th2). ) This is a case where the frequency of the difference value is greater than or equal to a predetermined third reference value. In this case, although the motion of the input video signal is not as great as in case 2, it can be determined that the video has many small changes. Alternatively, it can be determined that the video includes a lot of random noise. The encoding by the MPEG system has a characteristic that it is easy to generate block noise in a video in which many such small changes occur. Therefore, in case 3, since it can be estimated that there is a lot of noise due to MPEG encoding, the MPEG noise reduction strength is set strongly. Further, in the case of an image including many random noises, noise reduction by the cyclic noise reduction unit 2 is effective. In the case 3, since the large motion that causes the afterimage interference is not detected when the cyclic noise reduction strength is increased, the cyclic noise reduction strength is also set strong.

  The first reference value is a frequency that serves as a reference when determining whether there is almost no motion of the input video signal based on the frequency of the difference value equal to or greater than the first threshold Th1. The second reference value is a frequency serving as a reference when determining whether or not the motion of the input video signal is large based on the frequency of the difference value equal to or greater than the second threshold Th2. The third reference value is a reference for determining whether or not a large number of small changes have occurred in the input video signal based on the frequency of the difference value not less than the first threshold Th1 and less than the second threshold Th2. It is a frequency. The first threshold value Th1 and the second threshold value Th2 are values serving as a reference for the difference value for determining whether the video signal has almost no motion, is relatively small, or is relatively large. When the video signal is 8 bits, for example, the first threshold Th1 can be set to a value around 10 and the second threshold Th2 can be set to a value around 20. This value is a constant that should be determined according to the specifications of the MPEG noise reduction unit 1 and the cyclic noise reduction unit 2.

  FIG. 3 is a block diagram illustrating a configuration of the control unit 4. As shown in FIG. 3, the control unit 4 includes a first integration unit 5, a second integration unit 6, an MPEG noise reduction intensity setting unit 7, a cyclic noise reduction intensity setting unit 8, and an adder 9. The control unit 4 receives the histogram input from the histogram generation unit 3, outputs the MPEG noise reduction strength (for example, the set value of the target pixel range to be smoothed) to the MPEG noise reduction unit 1, and sends it to the cyclic noise reduction unit 2. A cyclic noise reduction strength (for example, a set value of a cyclic coefficient) is output. The first integrating unit 5 outputs an integrated value Sum1 of the frequency of the difference values that are not less than the first threshold Th1 and less than the second threshold Th2 in the histogram. The second integration unit 6 outputs the integration value Sum2 of the frequency of the difference value equal to or greater than the second threshold Th2 in the histogram. The sum (Sum1 + Sum2) of the integrated value Sum1 output from the first integrating unit 5 and the integrated value Sum2 output from the second integrating unit 6 is input to the MPEG noise reduction intensity setting unit 7. The cyclic noise reduction intensity setting unit 8 receives the integrated value Sum2 output from the second integrating unit 6.

  FIG. 4 shows an example of an intensity characteristic for determining the MPEG noise reduction intensity according to the sum (Sum1 + Sum2) of the integrated value Sum1 and the integrated value Sum2 used in the MPEG noise reduction intensity setting unit 7. As shown in FIG. 4, the MPEG noise reduction strength is set to be weak when the sum of sums (Sum1 + Sum2) is small, and is set to be strong as the sum of sums (Sum1 + Sum2) increases. The MPEG noise reduction intensity is set to the maximum intensity when the sum of the integrated values (Sum1 + Sum2) is greater than a certain value.

FIG. 5 shows an example of an intensity characteristic that is used in the cyclic noise reduction intensity setting unit 8 to determine the cyclic noise reduction intensity according to the integrated value Sum2. As shown in FIG. 5, the cyclic noise reduction intensity is set to the maximum intensity when the integrated value Sum2 is small, and is set to be weak as the integrated value Sum2 increases when the integrated value Sum2 is a certain value or more.

  In case 1, since the sum of the integrated values (Sum1 + Sum2) is small and the integrated value Sum2 is small, the control unit 4 sets the MPEG noise reduction strength to be weak and sets the cyclic noise reduction strength to be strong. In the case 2, since the sum of the integrated values (Sum1 + Sum2) is large and the integrated value Sum2 is large, the control unit 4 sets the MPEG noise reduction strength strong and sets the cyclic noise reduction strength weak. In case 3, since the sum (Sum1 + Sum2) of the integrated values is moderate to large and the integrated value Sum2 is small, the control unit 4 sets the MPEG noise reduction strength strongly and sets the cyclic noise reduction strength strongly. .

  As described above, according to the noise reduction apparatus according to the present embodiment, the MPEG noise reduction strength and the cyclic noise reduction strength can be controlled based on the histogram of the pixel value difference between frames. Therefore, it is possible to perform an optimal noise reduction process according to not only the magnitude of the motion of the input video signal but also the amount of noise caused by MPEG encoding. Therefore, when the input video signal is a video signal encoded by MPEG, it is possible to effectively reduce normal random noise included in the video signal and also due to MPEG encoding included in the video signal. It is also possible to suitably reduce block noise.

  In this embodiment, the configuration in which the cyclic noise reduction unit 2 is connected in series in the subsequent stage of the MPEG noise reduction unit 1 has been described. However, the order of connection may be reversed, or both are connected in parallel. It may be a configuration. In addition, an example of generating a histogram by obtaining pixel value differences between frames for all pixels constituting the frame has been described. However, pixel values difference between frames for pixels constituting a predetermined region in the frame are described. A histogram may be generated by acquiring the histogram. Examples of the predetermined area include a central area obtained by cropping the upper, lower, left and right sides of the frame with a certain width, and a peripheral area of the target pixel. Further, the frame may be divided into a plurality of regions, and a histogram of pixel value differences between frames may be generated for each region. As a result, it is possible to determine the characteristics of the video for each area and control the MPEG noise reduction intensity and the cyclic noise reduction intensity to be different for each area. For example, for a video that contains both video with low motion and video with high motion in a frame, optimal noise reduction strength control is performed for each region, such as increasing the MPEG noise reduction strength only for regions with large motion. Is possible.

  The MPEG noise reduction unit 1 may have a function of reducing block noise based on compression rate information in MPEG encoding. This is because block noise tends to occur when the compression rate is increased in MPEG encoding. In this case, the control unit 4 may set the MPEG noise reduction strength stronger as the compression rate is higher. When the input video signal is not an MPEG encoded video, the noise reduction device may be configured to invalidate the noise reduction by the MPEG noise reduction unit 1. In addition to the MPEG noise reduction unit 1 and the cyclic noise reduction unit 2, a noise reduction apparatus may be provided that further includes a known spatial noise reduction unit that reduces noise based on the correlation between pixels in a frame. .

  The present invention can also be applied to a noise reduction apparatus having a configuration that includes means for reducing coding noise caused by a coding method other than MPEG as coding noise reduction means. In that case, the control unit 4 is configured by appropriately setting the threshold value of the difference and the reference value of the frequency based on the characteristics of the encoding method so that the condition that the encoding noise is likely to be generated can be determined based on the histogram. It is preferable to do. The code further includes a plurality of encoding noise reduction units respectively corresponding to the plurality of encoding methods, and a determination unit that determines the encoding method of the input video signal, and is used according to the determination result by the determination unit The noise reduction unit may be switched.

(Example 2)
Next, a second embodiment of the present invention will be described. Constituent elements common to those in the first embodiment are given the same names and symbols as those in the first embodiment, and detailed description thereof is omitted. FIG. 6 is a block diagram illustrating the configuration of the control unit 4 in the noise reduction apparatus according to the second embodiment. As shown in FIG. 6, in addition to the configuration of the control unit 4 of the noise reduction apparatus according to the first embodiment, the control unit 4 of the present embodiment further includes a peak detection unit 10 and a multiplier 11 that detect the peak of the histogram. I have. Other configurations are the same as those of the first embodiment. The peak detection unit 10 detects whether or not there is a peak in the frequency of the difference value equal to or less than the second threshold Th <b> 2 in the histogram of the pixel value difference between frames input from the histogram generation unit 3. Specifically, when there is a difference value that is equal to or less than the second threshold Th2 and the frequency exceeds a predetermined fourth reference value, it is considered that a peak exists.

  FIG. 7 shows an example of a histogram of pixel value differences between frames in the case where a peak exists in the range of the second threshold value or less (Th2 or less). In the example of FIG. 7, there is a peak at the first threshold Th1. When the presence of a peak is detected, it is likely that the input video signal has a moving edge or a change throughout the image such as fade-in or fade-out. If noise reduction by the cyclic noise reduction unit 2 is strongly applied to such a video signal, interference may occur.

  Therefore, when the peak detection unit 10 detects the presence of such a peak, the control unit 4 according to the present embodiment performs control to weaken the cyclic noise reduction intensity. Specifically, the peak detection unit 10 outputs a coefficient k such that k = 1 when no peak is detected in a range equal to or less than the second threshold Th2, and 0 ≦ k <1 when a peak is detected. To do. The coefficient k is multiplied by the output from the cyclic noise reduction intensity setting unit 8 in the multiplier 11. Thereby, when a peak is detected, the cyclic noise reduction intensity can be set weak. The coefficient k may be a constant or a variable value corresponding to the peak height. When the coefficient k is a variable value, the peak height can be acquired, and the coefficient k can be reduced as the peak height increases.

  According to the noise reduction apparatus according to the present embodiment, by controlling the noise reduction intensity based on the histogram of the pixel value difference between frames, not only the magnitude of the motion of the input video but also the movement edge, fade, etc. Optimal noise reduction processing according to the feature can be performed. Therefore, it is possible to more effectively reduce the noise of the input video and to output a high-quality video signal.

1: MPEG noise reduction unit 2: Cyclic noise reduction unit 3: Histogram generation unit 4: Control unit

Claims (6)

A noise reduction device for reducing noise of each frame of an input video signal,
A cyclic noise reduction means for reducing noise contained in the current frame using the previous frame;
Encoding noise reduction means for reducing noise caused by the encoding method of the input video signal;
Generating means for acquiring a pixel value difference between the current frame and the previous frame, and generating a frequency distribution of the acquired difference value;
Control means for controlling the noise reduction intensity by the cyclic noise reduction means and the noise reduction intensity by the coding noise reduction means based on the frequency distribution generated by the generation means;
A noise reduction device comprising:   The noise reduction apparatus according to claim 1, wherein the generation unit divides the frame into a plurality of regions, and generates a frequency distribution of the acquired difference values for each region. In the frequency distribution generated by the generation unit, the control unit
When the frequency of the difference value equal to or greater than the predetermined first threshold is equal to or less than the predetermined first reference value, it is determined that the motion of the input video signal is small and the noise due to the encoding method is small,
When the frequency of the difference value greater than or equal to the predetermined second threshold value greater than the first threshold value is equal to or greater than the predetermined second reference value, the motion of the input video signal is large and noise caused by the encoding method It is judged that there are many,
The frequency of the difference value greater than or equal to the second threshold is less than the second reference value, and the frequency of the difference value greater than or equal to the first threshold and less than the second threshold is greater than or equal to a predetermined third reference value. If it is, it is determined that the motion of the input video signal is small and there is a lot of noise due to the encoding method,
When it is determined that the motion of the input video signal is small, the strength of noise reduction by the cyclic noise reduction means is set stronger than when it is determined that the motion is large,
When it is determined that there is a lot of noise due to the encoding method of the input video signal, the strength of noise reduction by the encoding noise reduction means is set stronger than when it is determined that the noise is low The noise reduction device according to claim 1 or 2. In the frequency distribution generated by the generation unit, the control unit
4. The noise reduction device according to claim 3, wherein the intensity of noise reduction by the cyclic noise reduction unit is set to be weaker when there is a peak in the range of the second threshold value or less than when there is no peak. . The encoding method is MPEG,
The coding noise reduction unit is a spatial noise reduction unit that performs smoothing on pixels at a block boundary in a frame based on block boundary information related to encoding by MPEG. The noise reduction device according to any one of 1 to 4. A noise reduction method for reducing noise of each frame of an input video signal,
An input process for inputting a video signal;
A cyclic noise reduction process for reducing noise included in the current frame of the input video signal using the previous frame;
An encoding noise reduction step for reducing noise caused by the encoding method of the input video signal;
Generating a difference between pixel values of the current frame and the previous frame, and generating a frequency distribution of the acquired difference values;
Based on the frequency distribution generated by the generating step, controlling the noise reduction strength in the cyclic noise reduction step and the noise reduction strength in the coding noise reduction step;
A noise reduction method comprising:

Images