JP5084804B2 - Noise removing device, noise removing method, and video signal display device - Google Patents

Description

  The present invention relates to a noise removing device, a noise removing method, and a video signal display device to which the noise removing device or method is applied to remove noise in a video signal, and more particularly to a frame recursive noise removing device, a noise removing method, and the like. The present invention relates to a video signal display device.

  In recent years, along with the increase in the size of display screens of television receivers and the increase in image quality of video signals, when video signals are processed and displayed in high image quality, they are not included in the input video signal (also called input image signal). Components, that is, noise (also referred to as noise) components are becoming conspicuous, and high signal reliability is required in video signal processing. In order to reduce the noise component contained in this video signal and obtain a high-quality video, there is a noise reduction (also called noise reduction, abbreviated as NR) device, for example, correlation between frames (frame correlation) on the time axis. Many three-dimensional noise removal apparatuses have been proposed for removing noise components without).

  The three-dimensional noise removing apparatus subtracts the signal of the previous frame and the signal of the current input frame to take a difference between frames, and multiplies the difference between the frames by a certain coefficient (cyclic coefficient). The noise component is removed by adding / subtracting to the signal of the current input frame. In such a three-dimensional noise removal apparatus, while a noise removal effect can be obtained, tailing and afterimages may occur in a moving part. In this case, the coefficient can be obtained from a difference between frames or the like. The calculation is performed using the motion detection information, and the effect of noise removal is weakened by reducing the coefficient in a portion where there is motion, thereby reducing tailing and afterimage.

  As a conventional noise removing device, the maximum value and the average value of the difference data for a predetermined period are detected for the difference data from the signal one frame before, and the correction data is generated from the detected maximum value and average value. Generating a control signal for controlling ON / OFF of NR and detecting an amount of noise with high accuracy and removing noise (for example, refer to Patent Document 1), and moving pixel signal differences between a plurality of frames There has been proposed a device (see, for example, Patent Document 2) that detects noise as a degree of noise and sets the gain of the circulating noise thereby to remove noise.

  There is also a proposal of a device that performs noise removal processing according to the brightness of an image by controlling a cyclic coefficient of noise according to a motion detection result from a frame difference and a level of an image signal (see, for example, Patent Document 3). .

  In addition, motion detection is performed based on the difference between frames, the determination result is checked by a majority processing circuit, variation in the determination result is corrected, and a noise cyclic coefficient is determined using this as a motion signal, so that motion can be detected well. There is also a proposal of a device that performs noise removal (see, for example, Patent Document 4).

  Further, the difference frequency band is divided into a low frequency band and a high frequency band for the difference from the image one field before the motion vector correction, and a motion detection signal is obtained for each of the divided bands to control the feedback amount. Therefore, there is also a proposal of an apparatus that improves the afterimage reduction effect in the high frequency area and performs noise removal (see, for example, Patent Document 5).

Japanese Patent Laying-Open No. 2003-219208 (FIGS. 1 and 4) JP 2002-33942 A (FIGS. 1 and 3) Japanese Patent Laying-Open No. 2005-347821 (FIGS. 1 and 2) JP-A-9-81754 (FIG. 1) Japanese Patent Laying-Open No. 2004-96628 (FIG. 3)

  As described above, in the above-described conventional apparatus, motion detection is performed based on a difference value between frames, and a cyclic coefficient for noise is controlled according to the result of motion detection, thereby reducing tailing and afterimage of a motion part. It was. On the other hand, the difference between frames includes a difference due to “motion” of the video signal between the previous and next frames and a “noise” component in the input video signal. Therefore, if the sensitivity of motion detection is increased, the sensitivity of the difference due to noise also increases, and the difference due to noise is erroneously detected as motion, and a good noise removal effect cannot be obtained. If a video signal with a large noise component is input, such as when the gain of the input video signal level is increased, for example, when tailing or afterimage occurs in the part, the difference due to noise is erroneously detected as motion. Therefore, there is a problem that a noise removal effect cannot be obtained and good noise removal cannot be performed.

  Therefore, the present invention has been made to solve the above-described problems, and an object of the present invention is to detect the motion and noise of a video signal included in the difference between frames even for a video signal having a large noise component value. A noise removal device, a noise removal method, and a video signal display device capable of distinguishing components and performing good motion detection without erroneous detection, reducing tailing and afterimages, and obtaining a good noise removal effect It is to provide.

The present invention
A three-dimensional noise removing apparatus that removes a noise component having no correlation between frames based on a correlation between frames of a video signal,
Obtains the difference between the video signal and the previous frame video signal of the current frame, or obtains a difference component of the video signal from which noise has been removed is the output of the video signal and the previous frame noise removal apparatus of the current frame, Subtracting means for outputting a frame difference for each pixel ;
To the frame difference for each pixel, from the frame difference in peripheral pixels centering on the pixel performs determining majority decision motion or stationary at each pixel, the judgment result value of the frame difference A motion detection means for correcting and outputting a signal indicating the corrected frame difference as a motion detection signal indicating the amount of motion for each pixel of the video signal included in the frame difference;
Extract the high frequency component for each pixel in the frame difference , perform the averaging using the high frequency component in the surrounding pixels for the extracted high frequency component, from the averaged results, A noise determination unit that determines a noise component for each pixel in the video signal included in the frame difference and outputs a noise determination result indicating a degree of noise included;
Subtracts the noise determination result for each pixel from the motion detection signal for each pixel, the noise component included in the frame difference, except from the motion detection signal is a value indicating the amount of motion, excluding noise components The degree of motion that is corrected to a value indicating the amount of motion, and the signal resulting from this subtraction is generated and output as a motion degree signal indicating the degree of motion due to the frame difference at each pixel in which the noise component included in the frame difference is distinguished Generating means;
Coefficient conversion means for obtaining and outputting a cyclic coefficient that changes in accordance with the motion degree signal;
Multiplication means for multiplying the cyclic coefficient and the frame difference to obtain a noise cyclic amount;
The noise cyclic amount by subtracting the video signal of the present frame, and an arithmetic means for obtaining a Film image signal removed of noise,
The cyclic coefficient decreases as the degree of movement indicated by the movement degree signal increases, and is determined to be 0 when the degree of movement indicated by the movement degree signal is equal to or greater than a predetermined value.
Providing noise removing apparatus characterized by.

  According to the present invention, three-dimensional noise removal is performed to remove a noise component having no frame correlation from the signal of the previous frame and the signal of the current input frame. From the difference between the frames, the “motion” and “noise” of the video signal are obtained. ”Is obtained, a signal indicating the degree of motion is obtained from the determination result, and noise removal is performed by setting a coefficient for noise according to the signal indicating the degree of motion. Therefore, even for a video signal with a large noise component value, it is possible to distinguish between the motion of the video signal included in the difference between frames and the noise component, so that motion detection can be performed without false detection, and tailing and afterimage As a result, it is possible to obtain a good noise removal effect.

It is a block diagram which shows an example of a structure of the noise removal apparatus 1 by Embodiment 1 of this invention. It is a figure which shows an example of the input-output characteristic in the difference sensitivity conversion means 22 in the motion detection means 20 in the noise removal apparatus 1 by Embodiment 1. FIG. 6 is a block diagram showing an example of the configuration of an isolated point removing unit 23 in the motion detecting unit 20 in the noise removing apparatus 1 according to Embodiment 1. FIG. It is a figure for demonstrating the surrounding pixel range which performs majority decision in the isolated point removal means 23 in the noise removal apparatus 1 by Embodiment 1. FIG. 4 is a block diagram showing an example of the configuration of majority decision means 108 in the noise removal apparatus 1 according to Embodiment 1. FIG. It is a block diagram which shows the other structural example of the majority decision means 108b in the noise removal apparatus 1 by Embodiment 1. FIG. 6 is a diagram for explaining an operation of a BPF 31 in a noise determination unit 30 in the noise removal apparatus 1 according to Embodiment 1. FIG. 3 is a block diagram showing an example of the configuration of coefficient conversion means 16 in the noise removal apparatus 1 according to Embodiment 1. FIG. It is a figure which shows an example of the input-output characteristic in the coefficient conversion means 16 in the noise removal apparatus 1 by Embodiment 1. 4 is a flowchart for explaining the operation of the noise removal apparatus 1 according to the first embodiment. It is a block diagram which shows one structural example at the time of making the noise removal apparatus by Embodiment 1 into a non-cyclic type. It is a block diagram which shows an example of a structure of the noise removal apparatus 2 by Embodiment 2 of this invention. 6 is a flowchart for explaining the operation of the noise removal apparatus 2 according to the second embodiment. It is a block diagram which shows an example of a structure of the video signal display apparatus by Embodiment 3 of this invention.

A noise removal device, a noise removal method, and a video signal display device in an embodiment of the present invention are:
A correlation with a field one frame before is detected and noise is removed.

Embodiment 1 FIG.
FIG. 1 is a block diagram showing a configuration of a noise removal apparatus according to Embodiment 1 of the present invention (that is, an apparatus that can implement the noise removal method according to Embodiment 1).
The illustrated noise removal apparatus is a three-dimensional noise removal apparatus that removes noise components from the signal of the previous frame and the signal of the current input frame. From the correlation between the frames of the video signal, the noise component having no correlation between frames is removed. Remove.
The noise removal apparatus shown in FIG. 1 processes the output signal after noise removal with a delay of one frame, and is called a cyclic noise removal apparatus (frame cyclic noise removal apparatus).

  In FIG. 1, reference numeral 1 denotes the noise removal apparatus according to the first embodiment. The noise removal apparatus 1 receives an input video signal (also referred to as an input signal, an input frame signal, or a current frame signal) Di0 sequentially, and outputs the video signal. A frame memory 11 that delays one frame, a subtracting means 12, an amplitude limiting means 13, and a motion that outputs a motion degree signal MDS indicating the degree of motion from the difference between the input frame signal Di0 and the signal Im1 one frame before. The noise detection unit 14, the coefficient conversion unit 16 that converts the noise to the cyclic coefficient Km according to the signal MDS from the motion / noise detection unit 14, and the difference Dfn from the amplitude limiting unit 13 and the cyclic coefficient Km are multiplied. The multiplication means 17 for outputting as the noise circulation amount Nd, and the noise circulation amount Nd from the multiplication means 17 and the input signal Di0 are added or subtracted. And a calculating means 18 for processing.

The motion / noise detection means 14 detects whether the difference between the input frame signal Di0 and the signal Im1 one frame before corresponds to “motion” or “noise”, and the degree of motion from the detection result A movement degree signal MDS indicating
Motion detection means 20 for detecting motion from the inter-frame difference Diff of the video signal and outputting a non-linearly converted difference signal Dm and a motion detection signal MDT indicating the amount of motion;
A noise determination means 30 for determining whether the difference signal is noise based on the difference signal Dm subjected to nonlinear transformation from the motion detection means 20;
A movement degree generation means 19 for generating a movement degree signal MDS indicating the degree of movement from the movement detection signal MDT and the noise determination result NDT;
The movement degree generation unit 19 includes a noise subtraction unit 24 that performs a subtraction process between the motion detection signal MDT and the noise determination result NDT,
A low-pass filter (referred to as LPF in FIG. 1 and hereinafter abbreviated as LPF) 25;
It comprises motion degree conversion means 26 for converting the signal from the LFP 25 into a motion degree signal MDS indicating the degree of movement.

Further, the motion detection means 20 in the motion / noise detection means 14 includes:
Absolute value calculating means 21 for calculating the absolute value Dabs of the inter-frame difference Diff from the subtracting means 12;
Differential sensitivity conversion means 22 for nonlinearly converting the absolute difference Dabs and detecting motion;
Isolated point removing means 23,
The noise determination means 30 in the motion / noise detection means 14 is a band-pass filter that extracts a high-frequency component of the input difference signal in order to determine whether or not the motion based on the difference signal Dm from the motion detection means 20 is noise. (In FIG. 1, indicated as BPF, hereinafter abbreviated as BPF) 31, absolute value calculating means 32 for calculating the absolute value Nabs of the output of the BPF 31, and an average value Nave of noise in the pixels around each pixel And a noise determination sensitivity conversion means 34 for adjusting the sensitivity of the noise determination result by nonlinearly converting the output Nave from the averaging means 33.

  The frame memory 11 is a memory for delaying and outputting the video signal by one frame. The frame memory 11 delays the output signal Do0 (video signal after noise removal) of the noise removing device 1 by one frame, and the input video signal Di0 A video signal one frame before (also referred to as a previous frame signal) Im1 is output. The frame memory 11 is called a field memory when the input video signal is a field unit signal such as an interlaced scanning (interlace) signal, and is called a field memory. The delay for one frame is a delay for two fields. The frame memory 11 outputs the pixel of the signal Im1 one frame before at the same position as the pixel of the input signal Di0 of the current frame.

  The subtraction means 12 receives the current frame signal Di0 and the previous frame signal Im1, subtracts the current frame signal Di0 from the previous frame signal Im1, and calculates a difference value (frame difference) between the current frame signal Di0 and the previous frame signal Im1. Diff is required. The frame difference Diff includes components of “motion” and “noise” in the video signal. When the frame difference Diff is 0, the frame difference Diff is a completely stationary part (more strictly, If there is a motion or noise component (and the two are not completely cancelled), the value will increase.

  The amplitude limiting unit 13 limits the amplitude within a predetermined value (for example, within ± dTh) with respect to the frame difference Diff from the subtracting unit 12, and the amplitude-limited difference value Dfn in the pixel of the input signal Di0. Output as a noise component.

The motion / noise detection unit 14 receives the frame difference Diff from the subtraction unit 12. The motion / noise detection means 14 detects the motion in the video signal from the frame difference Diff between the input frame signal Di0 and the signal Im1 of the previous frame, determines the noise mixed with the motion, From the amount of motion (MDT) as a result of detection and the result of noise determination (degree of noise (NDT)), a motion degree signal MDS indicating the degree of motion in the frame difference is output.
The motion / noise detection means 14 sets the degree of motion to the (MD1 + 1) stage (for example, MD1 + 1 = 9 stage, MD1 is MD1 = 8), and when the frame difference Diff is detected to be due to motion, When the value MD1 (for example, MDS = 8, complete motion pixel) is detected and a still image portion is detected or when the difference is detected as a noise component rather than due to motion, zero (MDS = 0, complete still pixel) ), A motion degree signal MDS indicating the degree of motion is generated with a value between 0 and MD1 (for example, 0 to 8). The motion degree signal MDS from the motion / noise detection means 14 is sent to the coefficient conversion means 16.

  Next, the configuration of the motion / noise detection means 14 will be described in detail with reference to FIGS.

  The frame difference Diff from the subtracting means 12 is input to the motion detecting means 20 in the motion / noise detecting means 14. The motion detection means 20 extracts a difference (Dm) corresponding to the motion from the frame difference Diff, and sends a non-linearly converted difference signal to the noise determination means 30 and outputs a motion detection signal MDT indicating the amount of motion. To do.

  The frame difference Diff from the subtracting means 12 is input to the absolute value calculating means 21 in the motion detecting means 20. The absolute value calculation means 21 calculates and outputs the absolute value of the frame difference Diff (difference absolute value Dabs).

  The difference absolute value Dabs from the absolute value calculation means 21 is input to the difference sensitivity conversion means 22 in the motion detection means 20. In the difference sensitivity conversion means 22, for example, the difference absolute value Dabs is multiplied by a predetermined sensitivity magnification, the offset value is subtracted, and the calculation result is set to a value between a predetermined value (for example, 0 to dM (described later)). By limiting, nonlinear conversion is performed, and the difference signal Dm is output to the isolated point removing unit 23 in the motion detecting unit 20 and also sent to the noise determining unit 30. This difference signal Dm includes motion and a noise component having a large value.

  FIG. 2 shows the relationship between the absolute difference value Dabs (input, that is, the horizontal axis), which is an example of the input / output characteristics in the differential sensitivity conversion means 22, and the output differential signal Dm (vertical axis). In this case, when the difference absolute value Dabs shown on the horizontal axis in FIG. 2 is equal to or greater than Tm, a difference signal Dm indicating movement is output as a constant value dM, and the value up to the offset value Tof becomes a minute noise component, and the difference signal It is output as Dm = 0. The difference signal Dm is output between 0 and a value dM, and the value indicates the amount of motion or noise. Increasing the sensitivity magnification makes it easy to detect the frame difference absolute value Dabs as a motion. If the offset value is increased, the difference value up to that value is detected as a minute noise component (or a stationary part). It becomes. Therefore, the sensitivity of motion detection is adjusted by the sensitivity magnification and the offset value.

The difference signal Dm from the difference sensitivity conversion means 22 is input to the isolated point removal means 23 in the motion detection means 20. The isolated point removing means 23 removes isolated points so that there is no isolated value by performing majority decision on the difference signal and value magnitude within a predetermined peripheral pixel range with respect to the difference signal Dm at the target pixel, The difference signal at the pixel is corrected. That is, a difference signal (motion detection signal MDT) that is corrected as a difference value indicating the amount of motion by unifying the values by determining variations from the values in the surrounding pixels of the difference signal Dm, correcting the motion detection omission. Is output.
FIG. 3 is a block diagram showing a configuration example of the isolated point removing unit 23. The isolated point removing unit 23 includes a moving pixel determining unit 101, a still pixel determining unit 102, an in-area moving pixel total calculating unit 103, and the like. In-area still pixel cumulative calculation means 104, first comparison means (hereinafter referred to as “condition A comparison means”) 105, second comparison means (condition C comparison means) 106, and third comparison means ( Condition B comparing means) 107 and majority decision judging means 108.

  Here, the range of peripheral pixels to be determined by the isolated point removing means 23 is, for example, 3 vertical lines and 5 horizontal pixels centered on the pixel of interest (pixels indicated by diagonal lines in the figure) as shown in FIG. The total range is 15 pixels. Note that the pixel range is not limited to this, and may be, for example, a range of 5 pixels in the horizontal direction (only one line in the vertical direction) centering on the target pixel.

  In FIG. 3, the difference signal Dm from the difference sensitivity conversion unit 22 input to the isolated point removal unit 23 is input to the moving pixel determination unit 101 and the still pixel determination unit 102 in the isolated point removal unit 23.

  The motion pixel determination unit 101 compares the difference signal Dm with a predetermined threshold (first threshold), determines a motion pixel for which the difference signal Dm is determined to be motion, and outputs a comparison result. For comparison, for example, if the difference signal Dm is larger than a predetermined threshold, it is determined as a moving pixel, and as a result, a binarized value is output. The binarized value is, for example, a value “1” when it is determined as a motion pixel, and a value “0” otherwise. Note that the binarized value is not limited to the above, and may be any other value as long as it indicates the result of each pixel determined as a moving pixel by comparison.

  On the other hand, the still pixel determination unit 102 in the isolated point removal unit 23 compares the difference signal Dm with a predetermined threshold (second threshold), determines a still pixel for which the difference signal Dm is determined to be still, and compares the difference signal Dm. Output the result. For comparison, for example, if the difference signal Dm is smaller than a predetermined threshold, it is determined as a still pixel, and as a result, a binarized value is output. The binarized value is, for example, a value “1” when it is determined as a still pixel, and a value “0” otherwise. The binarized value is not limited to the above value, and may be another value as long as it is a value indicating the result of each pixel determined as a still pixel by comparison.

  In addition, the isolated point removing unit 23 shown in FIG. 3 includes the moving pixel determining unit 101 and the still pixel determining unit 102 and determines the moving pixel and the still pixel in each of them. The signal Dm is compared with a predetermined threshold value (equal to the first threshold value and the second threshold value described above). If the difference signal Dm is smaller than the predetermined threshold value, the pixel is determined as a still pixel, and is larger than the predetermined threshold value. For example, it may be determined as a moving pixel and a binarized value may be output. As shown in FIG. 3, by providing the motion pixel determination unit 101 and the still pixel determination unit 102, an intermediate value between motion and stillness (difference that cannot be determined as motion or stillness) is set. It becomes possible to determine a still pixel.

  A binarized value indicating a pixel determined as a moving pixel from the moving pixel determining unit 101 is input to the in-area moving pixel total calculating unit 103 in the isolated point removing unit 23. The in-area moving pixel total calculation means 103 accumulates the number of pixels determined as moving pixels in the pixel area shown in FIG. 4 according to the binarized value indicating the pixel determined as the moving pixel. To do. That is, the in-area moving pixel total calculation means 103 accumulates the values “1” determined as the moving pixels from the moving pixel determination means 101 in the area of 15 pixels in FIG. The total pixel value is output as the number of moving pixels pmv.

  The binarized value indicating the pixel determined as the still pixel from the still pixel determining unit 102 is input to the in-area still pixel total calculating unit 104 in the isolated point removing unit 23. The in-area still pixel total calculation means 104 accumulates the number of pixels determined to be still pixels in the pixel area shown in FIG. 4 according to the binarized value indicating the pixels determined as still pixels. To do. That is, the in-area still pixel total calculation means 104 accumulates the values “1” determined as still pixels from the still pixel determination means 102 in the 15-pixel area of FIG. The total pixel value is output as the number of still pixels pst.

  In FIG. 3, the number of moving pixels pmv output from the in-area moving pixel total calculating means 103 in the isolated point removing means 23 is sent to the condition A comparing means 105 and the condition C comparing means 106, The number of still pixels pst output from the in-area still pixel cumulative calculation means 104 in the point removal means 23 is sent to the condition B comparison means 107 and is compared with the preset thresholds thA, thC, and thB respectively. Then, conditions A, C, and B for removing isolated points are determined.

  The condition A comparison unit 105 in the isolated point removal unit 23 compares the number of moving pixels pmv in 15 pixels in the pixel area shown in FIG. 4 with a threshold thA, and when the number of moving pixels pmv is larger than thA, the area around the target pixel moves. As the condition A determination result CA, the value of the determination result CA is set to “1” (CA = 1, indicating that the periphery is determined to be a motion). Otherwise, the determination result CA is The value is “0” (CA = 0, indicating intermediate or still). Here, the preset threshold thA is set within the range of 15 pixels in the pixel area. For example, when thA = 4, when the number of moving pixels pmv in 5 pixels is 5 or more, CA = 1

  The condition C comparison unit 106 in the isolated point removal unit 23 compares the number of moving pixels pmv in 15 pixels in the pixel area shown in FIG. 4 with a threshold thC. If the number of moving pixels pmv is smaller than thC, the area around the target pixel is stationary. As the condition C determination result CC, the value of the determination result CC is set to “1” (CC = 1, and the periphery is determined to be stationary). Otherwise, the value of the determination result CC is set to “0”. (CC = 0, indicating intermediate or motion). Here, the preset threshold thC is set within a range of 15 pixels in the pixel area. For example, when thC = 7, if the number of moving pixels pmv in 15 pixels is less than 7, CC = 1 It is also possible to set the same value as the threshold value thA in the condition A comparing means 105.

  Further, the condition B comparison unit 107 in the isolated point removal unit 23 compares the number of still pixels pst in 15 pixels in the pixel area shown in FIG. 4 with a threshold thB, and if the number of still pixels pst is greater than thB, Is determined to be still, and as the condition B determination result CB, the value of the determination result CB is set to “1” (CB = 1, indicating that the periphery is determined to be a still image). The value of the result CB is set to “0” (CB = 0, indicating intermediate or motion). Here, the preset threshold thB is set within a range of 15 pixels in the pixel area. For example, when thB = 10, when the number of still pixels pst exceeds 10 in 15 pixels, CB = 1

  The condition A determination result CA, the condition C determination result CC, and the condition B determination result CB from the condition A comparison unit 105, the condition C comparison unit 106, and the condition B comparison unit 107 are the majority decision conditions for removing isolated points. A, condition C, and condition B are output to majority decision means 108.

  The difference signal Dm from the difference sensitivity conversion unit 22, the condition A determination result CA, the condition C determination result CC, and the condition B determination result CB are input to the majority decision determination unit 108 in the isolated point removal unit 23. The majority decision judging means 108 corrects the variation of the motion indicated by the values in the surrounding pixels with respect to the difference signal Dm in the target pixel in accordance with the condition A judgment result CA, the condition C judgment result CC, and the condition B judgment result CB. , Unify the values. That is, a majority decision is made as to whether the difference signal from the difference sensitivity conversion means 22 at the target pixel is motion or stationary, and a corrected difference signal (motion detection signal MDT) is output as a difference corresponding to the motion. FIG. 5 is a block diagram showing an example of the configuration of the majority decision judging means 108 in the isolated point removing means 23. The majority decision judging means 108 includes a condition A adding means 111, a condition B subtracting means 112, and a condition C subtracting means. 113 and selection means 114.

  The difference signal Dm from the difference sensitivity conversion unit 22 is sent to the condition A addition unit 111, the condition B subtraction unit 112, and the condition C subtraction unit 113 in the majority decision determination unit 108, and the correction value AAD set in advance in each of them. , BAD and CAD are added to or subtracted from the differential signal Dm.

  The condition A addition means 111 in the majority decision determination means 108 receives the difference signal Dm from the difference sensitivity conversion means 22, adds the correction value AAD to the difference signal Dm, and sends the addition result (Dm + AAD) to the selection means 114. . By adding the correction value AAD, the value of the difference signal Dm is increased, and the difference value corresponding to the movement is corrected.

  The condition B subtraction unit 112 in the majority decision determination unit 108 receives the difference signal Dm from the difference sensitivity conversion unit 22, subtracts the correction value BAD from the difference signal Dm, and supplies the subtraction result (Dm−BAD) to the selection unit 114. And send. By subtracting the correction value BAD, the difference signal Dm is made smaller and close to 0, and the difference value corresponding to stillness is corrected. When the subtraction result is a negative (−) value, it may be clipped to 0 and the value 0 may be output.

  Similarly, the condition C subtraction unit 113 in the majority decision determination unit 108 receives the difference signal Dm from the difference sensitivity conversion unit 22, subtracts the correction value CAD from the difference signal Dm, and selects the subtraction result (Dm−CAD). Send to means 114. By subtracting the correction value CAD, the difference signal Dm is made smaller and close to 0, and is corrected toward the difference value corresponding to stillness. When the subtraction result is a negative (−) value, it may be clipped to 0 and the value 0 may be output.

  To the selection means 114 in the majority decision judgment means 108, the difference signal Dm from the difference sensitivity conversion means 22, the addition result (Dm + AAD) from the condition A addition means 111, and the subtraction result (Dm− from the condition B subtraction means 112). BAD), the subtraction result (Dm−CAD) and the fixed value FIXD from the condition C subtraction means 113 are input, and the condition A determination result CA, the condition C determination result CC, and the condition B determination result CB are input. . In the selection means 114, the difference signal Dm, the addition result (Dm + AAD), the subtraction result (Dm−BAD), and the subtraction result (Dm) according to the condition A determination result CA, the condition C determination result CC, and the condition B determination result CB. -CAD) and a fixed value FIXD, a majority decision for isolated point removal is performed under the following conditions, and a motion detection signal MDT is output.

  If the condition A determination result CA is the value “1” (CA = 1) and it is determined that the periphery is a motion, the addition result (Dm + AAD) is selected and MDT = (Dm + AAD) is output. That is, the value of the difference signal Dm is corrected to be closer to movement.

  When the condition B determination result CB is the value “1” (CB = 1) and the periphery is determined to be a still image, the subtraction result (Dm−BAD) is selected and MDT = (Dm−BAD) is output. . In other words, the value of the difference signal Dm is corrected closer to the still image.

  When the condition C determination result CC is “1” (CC = 1) and the periphery is determined to be a still image, the subtraction result (Dm−CAD) is selected and MDT = (Dm−CAD) is output. In other words, the value of the difference signal Dm is corrected closer to the still image.

  Furthermore, when the condition A determination result CA is a value “1” (CA = 1) and the condition B determination result CB is a value “1” (CB = 1), a fixed value is used as an intermediate value between motion and stillness. Select FIXD and output MDT = FIXD. This is a case where it cannot be determined whether the value of the difference signal Dm is a still image or a motion.

  If the result of the above conditions A, B, and C cannot be obtained as ‘1’, the difference signal Dm is selected as it is and MDT = Dm is output.

  The majority decision means 108 in the isolated point removing means 23 can also be configured as shown by the majority decision means 108b in FIG. In FIG. 6, the same reference numerals are given to the same or corresponding components as those in FIG. In FIG. 6, the majority decision determining means 108b is provided with a switching means B115, a switching means A116, a switching means F117, and a switching means C118 in addition to the configuration in FIG. 5, and the difference signal Dm from the difference sensitivity conversion means 22 is provided. On the other hand, the addition result (Dm + AAD), the subtraction result (Dm−BAD), the subtraction result (Dm−CAD), and the fixed value FIXD according to the condition A determination result CA, the condition C determination result CC, and the condition B determination result CB. The signal is switched, the majority decision for removing the isolated points is performed as described above, and the motion detection signal MDT is output. Since the configuration and operation are the same as the determination described above, detailed description thereof is omitted.

  The majority decision judging means 108b includes a first switching means (hereinafter referred to as “switching means B”) 115, a second switching means (switching means A) 116, a third switching means (switching means F) 117, and a fourth switching means. In the switching means (switching means C) 118, the addition result or subtraction result and the fixed value are sequentially switched according to the condition A determination result CA, the condition C determination result CC, and the condition B determination result CB. Therefore, the priority of each determination condition can be set.

  The majority decision conditions for removing isolated points in the majority decision judging means 108 or 108b are not limited to the above order and conditions. For example, motion detection is performed only by the condition A judgment result CA and the condition B judgment result CB. The signal MDT may be configured to be corrected to the same value as the surroundings, or may be corrected to a difference signal that has been subjected to addition or subtraction processing based only on the condition A determination result CA and the condition B determination result CB.

  As described above, the isolated point removal unit 23 in the motion detection unit 20 makes a majority decision on the difference signal Dm from the difference sensitivity conversion unit 22 and the magnitude of the difference signal and value in a predetermined peripheral pixel range, The isolated point is removed to unify the values, the motion detection omission is corrected, and the motion detection signal MDT corrected as a difference corresponding to the motion is output to the noise subtracting means 24 in the motion degree generating means 19. The motion detection signal MDT is a signal indicating a difference value obtained by extracting a motion in the video signal from the frame difference Diff and removing an isolated point. When the value is large, the motion detection signal MDT is a motion pixel. Then, it becomes a still pixel. That is, the value of the motion detection signal MDT indicates the amount of motion.

  Next, the configuration of the noise determination means 30 in the motion / noise detection means 14 will be described. In FIG. 1, the noise determination unit 30 in the motion / noise detection unit 14 is input with the non-linearly converted difference signal Dm from the difference sensitivity conversion unit 22 in the motion detection unit 20. The noise determination means 30 determines a noise component having a relatively large value included in the difference signal Dm from the difference signal Dm, and outputs a noise determination result NDT indicating the degree of noise.

  Here, the difference signal Dm based on the frame difference Diff includes motion and a noise component having a relatively large value. As described above, the isolated point removing unit 23 in the motion detecting unit 20 corrects the isolated point of the difference signal Dm by removing it to obtain a value obtained by correcting a motion detection omission. Therefore, even a difference signal due to a noise component may be erroneously detected as a motion after being corrected as a difference corresponding to the motion due to variations in surrounding pixels. Therefore, the noise determination unit 30 determines the degree of the high frequency component present in the difference signal Dm, that is, the degree of the noise component included in the difference signal Dm.

  The difference signal Dm from the difference sensitivity conversion means 22 in the motion detection means 20 is input to the BPF 31 in the noise determination means 30. The BPF 31 extracts a high frequency component, that is, a noise component in the difference signal Dm. The BPF 31 constitutes a BPF for a range of three vertical lines and three horizontal pixels centering on the position of a target pixel (a pixel indicated by hatching in the figure) as shown in FIG. 7A, for example. Note that the BPF 31 is not limited to this, and may have a configuration (see FIG. 7B) for three horizontal pixels (only one line in the vertical direction) centered on the target pixel.

  The output of the BPF 31 is input to the absolute value calculation means 32 in the noise determination means 30. The absolute value calculating means 32 calculates the absolute value of the output of the BPF 31 and outputs it as a differential noise component Nabs.

  The difference noise component Nabs from the absolute value calculation means 32 is input to the averaging means 33 in the noise determination means 30. The averaging means 33 averages the noise components in the pixels around each pixel with respect to the differential noise component Nabs. This is to prevent the influence when the differential noise component Nabs is considered to change rapidly, such as the edge portion of the motion. The range in which the average value is calculated includes, for example, five horizontal pixels centered on the target pixel. The average value of the difference noise component Nabs at the five horizontal pixels is calculated and output as the average noise component Nave at the target pixel. To do. The range of pixels to be averaged is not limited to 5 horizontal pixels, and may be set to 3 horizontal pixels or 7 horizontal pixels.

  The average noise component Nave from the averaging means 33 is input to the noise determination sensitivity conversion means 34 in the noise determination means 30. In the noise determination sensitivity conversion means 34, for example, the average noise component Nave is multiplied by a predetermined sensitivity magnification, the offset value is subtracted, and the calculation result is limited to a predetermined value to perform nonlinear conversion, and the noise determination result NDT. Output as. The average noise component Nave is a noise component extracted from the difference signal Dm. The larger the value of the average noise component Nave, the higher the degree that the difference signal is a noise component, and the sensitivity noise is increased to increase the average noise component. It becomes easy to determine Nave as noise, and indicates the degree of noise in the differential signal.

  Then, the noise determination sensitivity conversion unit 34 outputs the averaged noise component whose sensitivity has been adjusted to the noise subtraction unit 24 in the motion degree generation unit 19 as the noise determination result NDT. This noise determination result NDT indicates the degree of a relatively large noise component included in the frame difference.

  Next, in FIG. 1, the motion detection signal MDT from the motion detector 20 and the noise determination result NDT from the noise determiner 30 are input to the motion degree generator 19 in the motion / noise detector 14. The motion level generation means 19 generates a motion level signal MDS indicating the level of motion in the frame difference by subtracting the noise determination result NDT from the motion detection signal MDT, and is configured as follows.

  The noise subtraction unit 24 in the motion degree generation unit 19 receives the motion detection signal MDT from the motion detection unit 20 and the noise determination result NDT from the noise determination unit 30. The noise subtracting means 24 subtracts the noise determination result NDT from the motion detection signal MDT and outputs a motion signal MSG that is the subtraction result.

  Here, “motion” in the video signal and “noise” mixed with the motion exist in the frame difference Diff. Therefore, by subtracting the noise determination result NDT indicating the degree of the noise component from the motion detection signal MDT indicating the amount of motion detected from the frame difference, the amount of motion is reduced when noise is detected as motion. The noise, that is, the result of motion detection toward the stillness can be corrected, and the noise component included in the frame difference can be distinguished from the amount of motion.

  The motion signal MSG from the noise subtracting means 24 is input to the LPF 25 in the motion degree generating means 19. The LPF 25 performs horizontal or vertical / horizontal low-pass filter processing on the motion signal MSG, performs spatial expansion processing, and outputs the result to the motion degree conversion means 26.

  An output from the LPF 25 is input to the motion level conversion unit 26 in the motion level generation unit 19 to generate and output a motion level signal MDS indicating the level of motion. The motion level conversion means 26 performs conversion processing such as multiplication and amplitude limitation on the motion signal after the spatial expansion processing by the LPF 25 and converts it into a 9-level motion level signal MDS from 0 to 8, for example, indicating the level of motion. To do. That is, if the motion signal from the LPF 25 is 0 or less, MDS = 0 is set, and if the value is equal to or greater than a predetermined value, the conversion is performed so that MDS = 8. It is converted into a motion degree signal MDS indicating the degree. The motion degree signal MDS indicating the degree of motion has a large motion signal value, and when a frame difference is detected as motion, MDS = 8 (complete motion pixel), and the motion signal value is small (0 and 0 or less). ) If the still image portion or the difference is detected as a noise component, MDS = 0 (complete still pixel), and the value between 0 and 8 indicates the degree of movement. The setting of the movement degree signal MDS has been described as nine stages from 0 to 8 indicating the degree of movement. However, the setting is not limited to this, and any value indicating the degree of movement may be set in nine stages or more. Even lower than that is possible. Further, the conversion to the motion level signal MDS may be performed by multiplying the motion signal from the LPF 25 by a predetermined value and then limiting the amplitude to a value from 0 to 8, thereby converting the motion signal to a signal indicating the motion level. Good.

  Then, the motion level signal MDS from the motion level conversion unit 26 in the motion level generation unit 19 is sent to the coefficient conversion unit 16.

  The coefficient conversion means 16 receives the motion degree signal MDS from the motion degree generation means 19 in the motion / noise detection means 14 and obtains a cyclic coefficient Km for noise corresponding to the motion degree signal MDS. The cyclic coefficient Km decreases as the degree of movement indicated by the movement degree signal MDS increases, and becomes 0 when the degree of movement indicated by the movement degree signal MDS is equal to or greater than a predetermined value, and the value of the movement degree signal MDS. It changes between 0 and 1 according to the increase or decrease of. Specifically, the coefficient conversion means 16 sets the cyclic coefficient Km = 0 in accordance with the degree of movement (the value of the movement degree signal MDS), when the movement degree signal MDS indicates MDS = 8, and the degree of movement. Is low and MDS = 0, which is detected as a stationary part or noise component, is calculated as Km = Kmax (Kmax ≦ 1). At this time, the cyclic coefficient Km is 0 ≦ Km ≦ 1.

  FIG. 8 is a block diagram showing an example of the configuration of the coefficient conversion unit 16 in the noise removal apparatus 1. The coefficient conversion unit 16 includes a coefficient calculation unit 201, a slope setting unit 202, a coefficient limiting unit 203, and a coefficient maximum. And value setting means 204.

  The coefficient calculation means 201 in the coefficient conversion means 16 receives the motion degree signal MDS from the motion / noise detection means 14 and the inclination setting value Ef from the inclination setting means 202. The coefficient calculation unit 201 calculates a coefficient value pkm corresponding to a cyclic coefficient from the value of the motion degree signal MDS and the inclination setting value Ef by the inclination setting unit 202. As described above, for example, when the motion level signal MDS is set from 0 to 8, the coefficient value pkm is calculated by multiplying the value obtained by subtracting the motion level signal MDS from the maximum value 8 and the slope setting value Ef. That is, it is obtained as pkm = (8−MDS) × Ef. At this time, the coefficient value pkm is 0 ≦ pkm ≦ 8 × Ef. In addition, although it has been described that the coefficient value is obtained by multiplication, for example, the coefficient value pkm may be generated by using a ROM (Read Only Memory) or the like and using the inclination setting value Ef and the value of the motion degree signal MDS as addresses.

  The slope setting means 202 in the coefficient conversion means 16 sets the slope (multiplication value) Ef for the multiplication processing in which the coefficient calculation means 201 calculates the coefficient value pkm from the motion degree signal MDS. The change amount of the cyclic coefficient Km with respect to the change of the value of the motion degree signal MDS is set by the inclination setting value Ef set by the inclination setting means 202.

  The coefficient value pkm calculated by the coefficient calculating unit 201 is output to the coefficient limiting unit 203.

  The coefficient limiting unit 203 in the coefficient converting unit 16 receives the coefficient value pkm from the coefficient calculating unit 201 and the coefficient maximum value Kmax (0 <Kmax ≦ 1) from the coefficient maximum value setting unit 204. The coefficient limiting means 203 limits the value to the coefficient maximum value Kmax with respect to the coefficient value pkm from the coefficient calculating means 201 and outputs it as a cyclic coefficient Km. Therefore, the cyclic coefficient Km is 0 ≦ Km ≦ Kmax.

  The coefficient maximum value setting means 204 in the coefficient conversion means 16 sets the maximum value Kmax of the cyclic coefficient Km for noise. The cyclic coefficient Km output from the coefficient limiting means 203 is limited to the maximum value Kmax. The cyclic coefficient Km sets a cyclic value for noise removal, and the strength of the noise removal effect can be adjusted by the maximum value Kmax.

  The coefficient conversion means 16 is configured to obtain the cyclic coefficient Km according to the degree of motion by the arithmetic processing by the coefficient calculation means 201 and the coefficient restriction means 203 and the restriction of the maximum value, but for example, the motion degree signal MDS It is possible to obtain the cyclic coefficient Km by using a ROM with the address as the address. Furthermore, even if the movement degree signal MDS, the cyclic coefficient change degree with respect to the movement degree signal MDS (for example, the inclination setting), and the maximum value of the cyclic coefficient are set as addresses, the same effects as described above can be obtained.

  FIG. 9 shows the relationship between the motion degree signal MDS (horizontal axis), which is an example of the input / output characteristics in the processing of the coefficient conversion means 16, and the output cyclic coefficient Km (vertical axis). The case of the inclination setting Ef1 is indicated by a solid line, and the case of the inclination setting Ef2 is indicated by a broken line. When the movement degree signal MDS is MD1 (for example, 8) and indicates movement, the cyclic coefficient Km = 0, and when the movement degree is low and MDS = 0 detected as a noise component, Km = Kmax As (Kmax ≦ 1), as the degree of movement (the value of the movement degree signal MDS) increases, the cyclic coefficient for noise considering movement is obtained by changing the value of the cyclic coefficient Km to be smaller.

  Then, the cyclic coefficient Km limited to the maximum value Kmax by the coefficient limiting unit 203 in the coefficient converting unit 16 is output to the multiplying unit 17 in FIG.

  The multiplication unit 17 is supplied with the difference value Dfn from the amplitude limiting unit 13 and the cyclic coefficient Km from the coefficient conversion unit 16. The multiplying unit 17 multiplies the difference value Dfn from the amplitude limiting unit 13 by the cyclic coefficient Km to obtain a noise cyclic amount Nd as Km × Dfn. The obtained noise circulation amount Nd is output to the calculation means 18.

  Here, the cyclic coefficient Km is calculated according to the value of the motion level signal MDS, and when the motion level signal MDS indicates motion, it is sent as the cyclic coefficient Km = 0. The cyclic amount Nd of Nd is Nd = 0. On the other hand, when the degree of motion is low and the noise component is detected, Km = Kmax, so that the noise circulation amount Nd outside the motion portion can be set to the maximum value, and the noise removal effect can be achieved. The noise removal effect can be changed stepwise according to the degree of movement.

  The calculation means 18 receives the noise circulation amount Nd and the input signal Di0 from the multiplication means 17. The calculating means 18 adds or subtracts the noise cyclic amount Nd to or from the input signal Di0 to remove noise in the video signal and obtain an output signal Do0 after noise removal. Here, the processing of the calculating means 18 is performed by adding or subtracting depending on the sign (positive or negative) of the noise circulation amount Nd from the multiplying means 17 so as to remove noise.

  Next, in the noise removal apparatus 1 according to the first embodiment, the motion and noise of the video signal are determined from the difference between frames, and the motion degree signal MDS indicating the degree of motion is obtained from the determination result, and the motion degree signal An operation of performing noise removal by setting a coefficient for noise according to MDS will be specifically described.

  FIG. 10 is a flowchart for explaining the noise removal operation in the noise removal apparatus 1. Hereinafter, a description will be given with reference to FIG.

  The video signal Di0 input to the noise removal apparatus 1 and the previous frame signal Im1 after noise removal delayed by one frame by the frame memory 11 are input to the subtracting means 12, and are input between the current frame signal Di0 and the previous frame signal Im1. A frame difference Diff is obtained (step S1). The amplitude limiting means 13 limits the amplitude within a predetermined value with respect to the frame difference Diff, and outputs the amplitude-limited difference value Dfn as a noise component.

  The frame difference Diff is input to the absolute value calculating means 21 in the motion detecting means 20 in the motion / noise detecting means 14, and the absolute value is calculated to obtain the absolute value Dabs of the frame difference. The difference absolute value Dabs is nonlinearly converted in the difference sensitivity conversion means 22, and the difference signal Dm after the nonlinear conversion is sent to the isolated point removal means 23 and the noise determination means 30 in the motion detection means 20 (step S2). This difference signal Dm includes a motion and a noise component having a relatively large value.

  The isolated point removal means 23 in the motion detection means 20 makes a majority decision on the difference signal and value magnitude in the peripheral pixel range for the difference signal Dm in the target pixel, and unifies the difference signal value in the peripheral pixels. The difference corresponding to the motion is obtained by correcting the value of the difference signal Dm so as to obtain the motion detection signal MDT (step S3). The motion detection signal MDT after correcting the difference signal Dm becomes a difference signal detected as motion, indicates the amount of motion, and is sent to the noise subtracting means 24 in the motion degree generating means 19.

  Further, the differential signal Dm from the differential sensitivity conversion unit 22 in the motion detection unit 20 is input to the BPF 31 in the noise determination unit 30. The BPF 31 extracts a high frequency component, that is, a noise component in the difference signal Dm, and then calculates an absolute value in the absolute value calculation means 32 to obtain a difference noise component Nabs (step S4).

  The differential noise component Nabs from the absolute value calculation means 32 is input to the averaging means 33 in the noise determination means 30, and an average value of pixels around each pixel is obtained as the average noise component Nave. The average noise component Nave is input to the noise determination sensitivity conversion means 34, and the average noise component Nave is nonlinearly converted and output as a noise determination result NDT (step S5). This noise determination result NDT indicates the degree of a relatively large noise component included in the frame difference.

  Here, the difference signal Dm includes a motion and a noise component having a relatively large value, and the isolated point removing means 23 in the motion detecting means 20 removes and corrects the isolated point of the difference signal Dm. The motion detection signal MDT is obtained by correcting the motion detection omission. Therefore, even a difference signal due to a noise component may be erroneously detected by being corrected as a difference corresponding to motion depending on the variation of surrounding pixels. The noise determination unit 30 may detect a high frequency included in the difference signal Dm. By determining the level of the frequency component, that is, the level of the noise component included in the difference signal Dm, it is determined whether the frame difference is due to the noise component, and is distinguished from the motion.

  Next, the noise subtraction unit 24 in the motion level generation unit 19 receives the motion detection signal MDT and the noise determination result NDT, subtracts the noise determination result NDT from the motion detection signal MDT, and obtains a motion signal MSG as a subtraction result. Output (step S6). By subtracting the noise determination result NDT indicating the degree of the noise component from the motion detection signal MDT indicating the motion amount detected from the frame difference, when the noise is detected as a motion, the motion amount is reduced, and the noise That is, it is possible to correct the result of motion detection toward a still side, and to distinguish the noise component included in the frame difference from the motion amount.

  The motion signal MSG from the noise subtracting unit 24 is input to the LPF 25, subjected to low-pass filter processing, subjected to spatial expansion processing, and input to the motion degree converting unit 26. The movement degree conversion means 26 generates a movement degree signal MDS indicating the degree of movement according to the signal after the LPF 25 (step 7). The motion level signal MDS is set in 9 stages from 0 to 8, for example (not limited to this), and if the motion signal from the LPF 25 is 0 or less, MDS = 0 and a value greater than or equal to a predetermined value In this case, conversion is performed so that MDS = 8. Therefore, this motion degree signal MDS indicates the degree of motion with a value between 0 and 8, and when the value of the motion signal is large and the frame difference is detected as motion, MDS = 8 (complete motion pixel). When the value of the motion signal is small (0 and 0 or less) and the still image portion or the difference is detected as a noise component, MDS = 0 (complete still pixel), and the value between 0 and 8 It indicates the degree of movement.

  The motion degree signal MDS from the motion degree conversion means 26 in the motion / noise detection means 14 is sent to the coefficient conversion means 16.

  In the coefficient conversion means 16, a cyclic coefficient Km for noise corresponding to the motion degree signal MDS from the motion / noise detection means 14 is obtained. The cyclic coefficient Km is calculated according to the degree of movement indicated by the value of the movement degree signal MDS. For example, as shown in FIG. 9, the movement degree signal Km is calculated from the inclination setting value Ef and the maximum value Kmax (Kmax ≦ 1). When MDS indicates motion, the cyclic coefficient Km = 0, the degree of motion is low, and when MDS = 0 detected as a noise component, Km = Kmax, and the value of the motion degree signal MDS increases. Accordingly, by changing the value of the cyclic coefficient Km so as to be small, the cyclic coefficient Km for noise in consideration of motion is obtained (step S8).

  The cyclic coefficient Km obtained in accordance with the motion level signal MDS indicating the level of motion in this way is input to the multiplication unit 17. The multiplying unit 17 multiplies the difference value Dfn from the amplitude limiting unit 13 by the cyclic coefficient Km, obtains the noise cyclic amount Nd as Km × Dfn, and sends the obtained noise cyclic amount Nd to the computing unit 18 (step) S9).

  The arithmetic means 18 removes noise in the video signal by adding or subtracting the noise cyclic amount Nd from the multiplication means 17 to the input signal Di0 (by the sign of the noise cyclic amount Nd) (step S10). Thus, the output signal Do0 after noise removal is obtained (step S11).

  Here, the cyclic coefficient Km is calculated according to the value of the motion level signal MDS, and when the motion level signal MDS indicates motion (in the case of MDS = 8), the cyclic coefficient Km is sent as 0. Therefore, the noise circulation amount Nd in the moving part is Nd = 0. Therefore, in the motion portion where the motion level signal MDS indicating the motion level is MDS = 8, the noise circulation amount added to the input signal Di0 is 0, and noise is not removed, and the motion level signal MDS Since the value of the noise circulation amount Nd becomes small according to the value of, tailing and afterimage can be reduced. On the other hand, when the degree of motion is low and a noise component is detected (in the case of MDS = 0), Km = Kmax, so the noise circulation amount Nd other than the motion part is set to the maximum value. This can increase the noise removal effect.

  As described above, according to the noise removal apparatus 1 of the first embodiment, the motion detection unit 20 obtains the difference corresponding to the motion by the motion detection unit 20 with respect to the frame difference Diff obtained by the subtraction unit 12, and detects the motion. While obtaining the signal MDT, the noise determination means 30 extracts the noise component contained in the difference signal Dm, and obtains the noise determination result NDT which is the degree of noise. Then, a motion degree signal MDS indicating the degree of motion is obtained from the motion detection signal MDT and the noise determination result NDT, a cyclic coefficient Km for noise corresponding to the motion degree signal MDS is obtained, and noise removal of the input signal is performed. . Therefore, even for a video signal with a large noise component value, the noise component is distinguished from the motion of the video signal included in the difference between frames, and the degree of motion is detected without erroneously detecting noise as motion. Since the noise cyclic coefficient corresponding to the detected motion level is obtained, the noise cyclic amount can be increased for the stationary noise portion, and the noise cyclic amount is set to zero for the moving portion. Pulling and afterimages can be reduced, and a good noise removal effect can be obtained.

  Note that the noise removal apparatus 1 in the first embodiment is configured as a frame cyclic noise removal apparatus that processes the output signal after noise removal by delaying it by one frame, and the frame memory 11 outputs the output signal Do0 after noise removal. Although it has been described that the frame difference Diff is obtained by delaying one frame, the present invention is not limited to this, and as shown in FIG. 11, the input signal Di0 of the current frame is delayed by one frame by the frame memory 11 as shown in FIG. It is also possible to configure a non-cyclic type noise removing device (referred to as a non-cyclic type noise removing device) for obtaining the difference. FIG. 11 is a block diagram showing the configuration of the acyclic noise removal apparatus 10. In FIG. 11, the same or corresponding components as those in FIG. 1 are denoted by the same reference numerals. The configuration and operation are the same as those in FIG. 1 described above except that the frame memory 11 delays the input signal Di0 by one frame, and thus detailed description thereof will be omitted. The calculation is based on the frame correlation between the frame and the previous frame.

  On the other hand, in the case of a frame cyclic type like the noise removal apparatus 1 of FIG. 1 described above, since the video signal after noise removal is delayed by a frame, the calculation is performed not only between two frames but also over many frames. The processing of the frame cyclic noise removal device 1 and the non-cyclic noise removal device 10 is the same, but the noise removal effect is large in the case of the cyclic noise removal device 1, but in the case of a video signal with a lot of movement, The non-cyclic noise removal device 11 can reduce tailing and afterimage.

  Although the frame memory 11 has been described as delaying the video signal by one frame, the input video signal may be an interlaced signal or a progressive scanning signal. If the signal is an interlaced signal, a delay of two fields is performed, and if the signal is a progressive signal, the same configuration can be achieved by delaying by one frame.

  Further, in the coefficient conversion means 16, the characteristic of the cyclic coefficient Km obtained according to the motion degree signal MDS has been described as shown in FIG. 9, but the present invention is not limited to this and is obtained. In accordance with the movement degree signal MDS, for example, when the predetermined movement degree signal MDS is smaller than the predetermined value MD2, Km is set as a predetermined fixed value, and the degree of movement is from the predetermined value MD2 to the maximum value of the movement degree signal indicating movement. It may be configured to gradually decrease Km as it increases, and convert so that Km = 0 at the maximum of the motion level signal, and the cyclic coefficient Km changes according to the motion level signal MDS indicating the motion level. If a cyclic coefficient for noise considering movement is obtained, the same effect can be obtained.

  Further, in the above description, each configuration of the noise removal apparatus 1 is described as a hardware configuration, but the present invention may be configured to be realized by software processing in program control.

Embodiment 2. FIG.
In the noise removal apparatus 1 of the first embodiment, the noise determination result is obtained by the noise determination means 30 based on the non-linearly converted difference signal Dm obtained from the difference sensitivity conversion means 22 in the motion detection means 20 in the motion / noise detection means 14. However, it is also possible to perform motion detection and obtain a noise determination result by using the frame difference Diff obtained from the subtracting unit 12.

  FIG. 12 is a block diagram showing a configuration of a noise removal apparatus according to Embodiment 2 of the present invention (that is, an apparatus capable of performing the noise removal method of Embodiment 2). It is composed. In FIG. 12, the same or corresponding components as those in FIG. 1 are denoted by the same reference numerals.

  In FIG. 12, the noise removal apparatus 2 according to the second embodiment uses a movement / noise detection means 15 instead of the movement / noise detection means 14 in the configuration of the noise removal apparatus 1 (see FIG. 1) according to the first embodiment. ing. The movement / noise detection means 15 is the same as the movement / noise detection means 14 of FIG. 1, but includes a noise determination means 40 instead of the noise determination means 30 of FIG. The noise determination unit 40 determines whether the frame difference Diff is noise based on the frame difference Diff from the subtraction unit 12. Configurations and operations other than those relating to the noise determination means 40 in the movement / noise detection means 15 are the same as those shown in the first embodiment, and a detailed description thereof will be omitted.

  In the second embodiment, the motion / noise detection unit 15 corresponds to “motion” or the difference between the input frame signal and the signal one frame before corresponds to “noise”, similar to the motion / noise detection unit 14. And a motion level signal indicating the level of motion is output from the detection result. Whether the difference is noise or not by the motion detection means 20 for detecting motion from the frame difference Diff and the frame difference Diff. It is comprised from the noise determination means 40 and the movement degree production | generation means 19 which judge.

  The frame memory 11 delays the signal Do0 after noise removal by one frame, and the subtracting unit 12 obtains the frame difference Diff between the current frame signal Di0 and the previous frame signal Im1, and the amplitude limiting unit 13 and the motion / noise detecting unit. 15 is the same as that shown in the first embodiment, and a detailed description thereof is omitted.

  The configuration of the noise determination means 40 in the movement / noise detection means 15 will be described. The noise determination means 40 includes a sensitivity conversion means 41 for nonlinearly converting the frame difference Diff, a BPF 42 for extracting a high frequency component of the differential signal Dn after the nonlinear conversion, and an absolute value calculation means 43 for calculating an absolute value Nabs2 of the output of the BPF 42. Then, an averaging means 44 for obtaining an average value Nave2 in the peripheral pixels of each pixel, and a noise determination sensitivity conversion means 45 for nonlinearly converting the output Nave2 from the averaging means 44 and adjusting the sensitivity of the noise determination result And. In FIG. 12, the frame difference Diff from the subtraction unit 12 is input to the noise determination unit 40 in the motion / noise detection unit 15. The noise determination unit 40 determines a noise component included in the frame difference Diff from the frame difference Diff, and outputs a noise determination result NDT2 indicating the degree of noise.

  The frame difference Diff includes motion and noise components. As described in the first embodiment, the motion detecting unit 20 extracts the motion of the frame difference by the isolated point removing unit 23 to obtain the motion detection signal MDT indicating the amount of motion. Therefore, even a difference due to a noise component may be erroneously detected by correcting it as a difference corresponding to motion depending on the variation of surrounding pixels. Therefore, the noise determination unit 40 determines the degree of the high frequency component present in the frame difference Diff, that is, the degree of the noise component included in the frame difference Diff.

  The frame conversion Diff from the subtraction unit 12 input to the motion / noise detection unit 15 is input to the sensitivity conversion unit 41 in the noise determination unit 40. In the sensitivity conversion means 41, for example, the frame difference Diff is multiplied by a predetermined sensitivity magnification, the offset value is subtracted, and the calculation result is limited to a predetermined value (for example, 0 to dN (predetermined value)). As a result, nonlinear conversion is performed and the difference signal Dn is output to the BPF 42. This difference signal Dn includes motion and noise components. By providing this sensitivity conversion means 41, the sensitivity of the frame difference for determining noise can be adjusted separately from the difference sensitivity conversion means 20 in the motion detection means 20.

  In the sensitivity conversion means 41, when the frame difference Diff is nonlinearly converted, the absolute value of the frame difference Diff is obtained, and then the absolute value is nonlinearly converted to obtain the difference signal Dn. In this case, the input / output characteristics of the nonlinear conversion can be set, for example, in the same manner as in the case shown in FIG.

  The differential signal Dn from the sensitivity conversion unit 41 is input to the BPF 42 in the noise determination unit 40. The BPF 42 extracts a high frequency component, that is, a noise component in the difference signal Dn. The BPF 42 can be configured in the same manner as the BPF 31. For example, as shown in FIG. 7A, the BPF 42 forms a BPF for a range of 3 pixels in the vertical direction and 3 pixels in the horizontal direction centered on the target pixel position. Note that the BPF 42 is not limited to this, and for example, a configuration (see FIG. 7B) for three horizontal pixels (only one line in the vertical direction) centering on the target pixel may be used.

  The output of the BPF 42 is input to the absolute value calculation means 43 in the noise determination means 40. The absolute value calculation means 43 calculates the absolute value of the output of the BPF 42 and outputs it as a differential noise component Nabs2.

  The difference noise component Nabs2 from the absolute value calculation means 43 is input to the averaging means 44 in the noise determination means 40. The averaging means 44 obtains an average value in the peripheral pixels of each pixel for the differential noise component Nabs2, and performs averaging with the noise components in the peripheral pixels. This is to prevent the influence when the differential noise component Nabs2 is considered to change rapidly, such as the edge portion of the motion. The range in which the average value is calculated includes, for example, five horizontal pixels centered on the target pixel. The average value of the differential noise component Nabs at the five horizontal pixels is calculated and output as the average noise component Nave2 at the target pixel. To do. Note that the range of pixels to be averaged is not limited to this.

  The average noise component Nave2 from the averaging means 44 is input to the noise determination sensitivity conversion means 45 in the noise determination means 40. In the noise determination sensitivity conversion means 45, for example, the average noise component Nave2 is multiplied by a predetermined sensitivity magnification, the offset value is subtracted, and the calculation result is limited to a predetermined value to perform nonlinear conversion, and the noise determination result NDT2 Output as. The average noise component Nave2 is a noise component extracted from the frame difference Diff. The larger the average noise component Nave2, the higher the degree that the difference signal is a noise component. By increasing the sensitivity magnification, the average noise component Nave2 is increased. This makes it easier to determine Nave2 as noise.

  Then, the noise determination sensitivity conversion unit 45 outputs the averaged noise component whose sensitivity has been adjusted to the noise subtraction unit 24 in the motion degree generation unit 19 as the noise determination result NDT2. This noise determination result NDT2 indicates the degree of the noise component included in the frame difference.

  Next, the motion detection signal MDT from the motion detector 20 and the noise determination result NDT2 from the noise determiner 40 are input to the motion degree generator 19 in the motion / noise detector 15. The motion level generation means 19 generates a motion level signal MDS indicating the level of motion in the frame difference by subtracting the noise determination result NDT2 from the motion detection signal MDT.

  The frame difference Diff includes “motion” in the video signal and “noise” mixed with the motion. Therefore, by subtracting the noise determination result NDT2 indicating the degree of the noise component from the motion detection signal MDT indicating the amount of motion detected from the frame difference, the amount of motion is reduced when noise is detected as motion. The noise, that is, the result of motion detection toward the stillness can be corrected, and the noise component included in the frame difference can be distinguished from the amount of motion.

  Hereinafter, the configuration for generating the motion level signal MDS indicating the level of motion in the motion level generation unit 19, the configuration for the coefficient conversion unit 16, the multiplication unit 17, and the noise by the calculation unit 18 are described in the first embodiment. This is the same as the noise removing apparatus 1 of FIG.

  Next, in the noise removal apparatus 2 according to the second embodiment, the motion and noise of the video signal are determined from the difference between the frames, and the motion level signal MDS indicating the motion level is obtained from the determination result, and the motion level signal MDS. The operation of removing noise by setting a coefficient for noise according to the above will be specifically described.

  FIG. 13 is a flowchart for explaining the noise removal operation in the noise removal apparatus 2. Hereinafter, a description will be given with reference to FIG.

  The video signal Di0 inputted to the noise removing device 2 and the previous frame signal Im1 after noise removal delayed by one frame by the frame memory 11 are inputted to the subtracting means 12, and are inputted between the current frame signal Di0 and the previous frame signal Im1. A frame difference Diff is obtained (step S201). The process in step S201 is the same as step S1 in FIG. The amplitude limiting means 13 limits the amplitude within a predetermined value with respect to the frame difference Diff, and outputs the amplitude-limited difference value Dfn as a noise component.

  The frame difference Diff is input to the absolute value calculating means 21 in the motion detecting means 20 in the motion / noise detecting means 15 and the absolute value is calculated to obtain the absolute value Dabs of the frame difference. The difference absolute value Dabs is nonlinearly converted by the difference sensitivity conversion means 22 and sent to the isolated point removal means 23 in the motion detection means 20 as a difference signal Dm (step S202). The process in step S202 is the same as step S2 in FIG.

  The isolated point removal means 23 in the motion detection means 20 makes a majority decision on the difference signal and value magnitude in the peripheral pixel range for the difference signal Dm in the target pixel, and unifies the difference signal value in the peripheral pixels. The difference corresponding to the motion is obtained by correcting the value of the difference signal Dm so as to obtain the motion detection signal MDT (step S203). The process in step S203 is the same as step S3 in FIG. The motion detection signal MDT after correcting the difference signal Dm becomes a difference detected as motion, and the amount of motion is obtained and sent to the noise subtracting means 24.

  The frame difference Diff is also input to the sensitivity conversion means 41 in the noise determination means 40, and the sensitivity conversion means 41 sends the difference signal Dn obtained by nonlinearly converting the frame difference Diff to the BPF 42 (step S204). The BPF 42 extracts a high frequency component, that is, a noise component in the difference signal Dn, and then calculates an absolute value in the absolute value calculation means 43 to obtain the difference noise component Nabs2 (step S205). The process in step S205 is the same as step S4 in FIG.

  The differential noise component Nabs2 from the absolute value calculation means 43 is input to the averaging means 44 in the noise determination means 40, and an average value of pixels around each pixel is obtained as the average noise component Nave2. The average noise component Nave2 is input to the noise determination sensitivity conversion unit 45, and the average noise component Nave2 is nonlinearly converted and output as a noise determination result NDT2 (step S206). The process in step S206 is the same as step S5 in FIG. This noise determination result NDT2 indicates the degree of a relatively large noise component included in the frame difference.

  Here, the frame difference Diff includes a motion and a noise component. The isolated point removing unit 23 in the motion detecting unit 20 corrects the isolated point by removing the isolated point and correcting the motion detection omission. The detection signal MDT is used. Therefore, even a difference signal due to a noise component may be erroneously detected by correcting it as a difference corresponding to motion depending on the variation of surrounding pixels, and the noise determination means 40 has a high frequency included in the frame difference Diff. By determining the level of the frequency component, that is, the level of the noise component included in the difference Diff, it is determined whether the frame difference is due to the noise component, and is distinguished from the motion.

  The noise subtraction unit 24 in the motion degree generation unit 19 receives the motion detection signal MDT and the noise determination result NDT2, subtracts the noise determination result NDT2 from the motion detection signal MDT, and outputs a motion signal MSG as a subtraction result ( Step S207). The processing in step S207 is the same as step S6 in FIG. When the noise is detected as motion by subtracting the noise determination result NDT2 indicating the degree of the noise component from the motion detection signal MDT indicating the motion amount detected from the frame difference, the motion amount is reduced to reduce the noise. That is, it is possible to correct the result of motion detection toward a still side, and to distinguish the noise component included in the frame difference from the motion amount.

  Hereinafter, an operation for generating a motion degree signal MDS indicating the degree of motion (step 208), an operation for obtaining a cyclic coefficient Km (step S209), a noise circulation amount Nd (step S210), and removing noise in the video signal The operation until the output signal Do0 after noise removal (steps S211 and S212) is the same as the operation in the first embodiment (steps S7 to S11 in FIG. 1 and FIG. 10). The detailed explanation is omitted.

  As described above, according to the noise removal apparatus 2 of the second embodiment, the motion detection unit 20 obtains a difference corresponding to the motion with respect to the frame difference Diff obtained by the subtraction unit 12 and detects the motion amount. While obtaining the signal MDT, the noise determination means 40 extracts a noise component included in the frame difference Diff, and obtains a noise determination result NDT2 indicating the degree of noise. Then, a motion degree signal MDS indicating the degree of motion is obtained from the motion detection signal MDT and the noise determination result NDT2, and a cyclic coefficient Km for noise corresponding to the motion degree signal MDS is obtained to remove noise from the input signal. . Therefore, even for a video signal with a large noise component value, the noise component is distinguished from the motion of the video signal included in the difference between frames, and the degree of motion is detected without misdetecting it as motion and detected accurately. Since the noise cyclic coefficient corresponding to the degree of motion is obtained, the noise cyclic amount can be increased for the stationary noise portion, and the noise cyclic amount can be set to zero for the moving portion and An afterimage can be reduced and a good noise removal effect can be obtained.

  The noise removal apparatus 2 according to the second embodiment is configured as a frame cyclic noise removal apparatus that delays and processes the output signal after noise removal by one frame. However, the present invention is not limited to this, and the implementation described above is not limited thereto. As in the non-cyclic type noise removing apparatus 10 shown in FIG. 11 in the first embodiment, in the non-cyclic type noise removing apparatus for obtaining the frame difference by delaying the input signal Di0 of the current frame by one frame by the frame memory 11 as it is, Noise detecting means 15 may be used.

  In the above description, the configuration of the noise removal apparatus 2 is described as being a hardware configuration. However, the present invention may be configured to be realized by software processing in program control.

Embodiment 3 FIG.
In the first and second embodiments, the noise removal device and the device capable of performing the noise removal method have been described. However, the present invention performs processing by removing noise from the input video signal and displays it with high image quality. It can also be applied to a video signal display device. In the third embodiment shown below, a video signal input from a TV broadcast signal, a recording / playback device such as a DVD or VTR, a TV broadcast receiver, or the like is processed, and the noise removal device 1 of the first or second embodiment is processed. A video signal display device that displays a video signal using 10 or 2 will be described.

  FIG. 14 is a block diagram showing an example of a video signal display apparatus according to Embodiment 3 of the present invention. In FIG. 14, the video signal display device 3 according to the third embodiment includes the noise removal device 1 according to the first embodiment (or the noise removal device 10 or the noise removal device 2 according to the second embodiment). A terminal 301, an input signal processing unit 302, a display processing unit 303, and a display unit 304 are provided to display a video signal from which noise has been removed. The configurations and operations other than those related to the input terminal 301, the input signal processing means 302, the display processing means 303, and the display means 304 are the same as those shown in the first and second embodiments, and detailed description thereof is omitted. To do.

  The input terminal 301 receives a TV broadcast signal, a signal from a recording / playback device such as a DVD or VTR, a TV broadcast receiver, or the like. A signal input to the input terminal 301 is sent to the input signal processing means 302.

  For example, when an analog signal is input to a TV broadcast signal, a signal from a recording / playback device such as a DVD or VTR, or a TV broadcast receiver, the input signal processing unit 302 converts the signal into a digital signal and synchronizes the signal. Input signal processing such as signal separation, and processing such as decoding of MPEG data when MPEG data is received. Then, the video signal subjected to the input signal processing is output to the noise removing device 1 (2, 10). Here, the video signal subjected to the input signal processing may be an interlace signal or a progressive signal.

  The noise removal device 1 (2, 10) is configured to, with respect to the video signal from the input signal processing means 302, a frame difference Diff between the input frame and the video signal after noise removal one frame before or the video signal one frame before. The motion detection means 20 obtains the difference corresponding to the motion to obtain the motion detection signal MDT, and the noise judgment means 30 or 40 extracts the noise component of the difference to obtain the noise judgment result. Then, a motion degree signal MDS indicating the degree of motion is obtained from the motion detection signal MDT and the noise determination result, a cyclic coefficient Km for noise corresponding to the motion degree signal MDS is obtained, and noise removal of the input signal is performed. Then, the video signal after noise removal is output to the display processing means 303. Since the noise removal operation is the same as in the first and second embodiments, detailed description thereof is omitted.

  The display processing means 303 receives the video signal from which noise is removed from the noise removing device 1 (2, 10). When the input video signal is an interlace signal, the display processing unit 303 performs processing such as converting the interlace signal into a progressive signal, and performs signal processing for converting into a display signal such as scaling processing. A display signal for display by means 304 is output to display means 304.

  The display unit 304 displays an image based on the display signal from the display processing unit 303.

  As described above, according to the video signal display device of the third embodiment, since the noise removing device or the noise removing method of the first and second embodiments is used, tailing and afterimages are reduced with respect to the moving part. Thus, it is possible to display a high-quality video based on the video signal from which a good noise removal effect is obtained.

  In the above description of each embodiment, the configuration indicated as “... Means” may be realized by an electric circuit (hardware) or realized by software.

  1, 2, 10 Noise removing device, 11 frame memory, 12 subtracting means, 13 amplitude limiting means, 14, 15 motion / noise detecting means, 16 coefficient converting means, 17 multiplying means, 18 calculating means, 19 motion degree generating means, 24 noise subtracting means, 25 LPF, 26 motion degree converting means, 20 motion detecting means, 21 absolute value calculating means, 22 differential sensitivity converting means, 23 isolated point removing means, 30, 40 noise judging means, 31, 42 BPF, 32 , 43 Absolute value calculation means, 33, 44 Averaging means, 34, 45 Noise determination sensitivity conversion means, 41 Sensitivity conversion means, 101 Motion pixel determination means, 102 Still pixel determination means, 103 In-area motion pixel total calculation means, 104 Area still pixel total calculation means, 105 Condition A ratio Comparison means, 106 condition C comparison means, 107 condition B comparison means, 108, 108b majority decision determination means, 111 condition A addition means, 112 condition B subtraction means, 113 condition C subtraction means, 114 selection means, 115 switching means B, 116 Switching means A, 117 switching means F, 118 switching means C, 201 coefficient calculating means, 202 slope setting means, 203 coefficient limiting means, 204 coefficient maximum value setting means, 301 input terminal, 302 input signal processing means, 303 display processing means 304 Display means.

Claims (5)

A three-dimensional noise removing apparatus that removes a noise component having no correlation between frames based on a correlation between frames of a video signal,
Obtains the difference between the video signal and the previous frame video signal of the current frame, or obtains a difference between the video signal from which noise has been removed is the output of the video signal and the previous frame noise removal apparatus of the current frame, each Subtracting means for outputting a frame difference for the pixel ;
To the frame difference for each pixel, from the frame difference in peripheral pixels centering on the pixel performs determining majority decision motion or stationary at each pixel, the judgment result value of the frame difference A motion detection means for correcting and outputting a signal indicating the corrected frame difference as a motion detection signal indicating the amount of motion for each pixel of the video signal included in the frame difference;
Extract the high frequency component for each pixel in the frame difference , perform the averaging using the high frequency component in the surrounding pixels for the extracted high frequency component, from the averaged results, A noise determination unit that determines a noise component for each pixel in the video signal included in the frame difference and outputs a noise determination result indicating a degree of noise included;
Subtracts the noise determination result for each pixel from the motion detection signal for each pixel, the noise component included in the frame difference, except from the motion detection signal is a value indicating the amount of motion, excluding noise components The degree of motion that is corrected to a value indicating the amount of motion, and the signal resulting from this subtraction is generated and output as a motion degree signal indicating the degree of motion due to the frame difference at each pixel in which the noise component included in the frame difference is distinguished Generating means;
Coefficient conversion means for obtaining and outputting a cyclic coefficient that changes in accordance with the motion degree signal;
Multiplication means for multiplying the cyclic coefficient and the frame difference to obtain a noise cyclic amount;
The noise cyclic amount by subtracting the video signal of the present frame, and an arithmetic means for obtaining a Film image signal removed of noise,
The cyclic coefficient decreases as the degree of movement indicated by the movement degree signal increases, and is determined to be 0 when the degree of movement indicated by the movement degree signal is equal to or greater than a predetermined value.
Noise removal device, characterized in that. The noise removal device according to claim 1, wherein the noise determination unit extracts a high-frequency component for each pixel by a band pass filter with respect to a range of surrounding pixels centering on each pixel. The noise determination means is
For the noise component extracted for each pixel, the average value of the noise component in the peripheral pixels around each pixel is obtained and averaged to obtain the average noise component,
The noise removal apparatus according to claim 1, wherein the average noise component is output as a noise determination result indicating a degree of noise included in a frame difference. A three-dimensional noise removal method for removing a noise component having no correlation between frames based on a correlation between frames of a video signal,
It obtains the difference between the video signal and the previous frame video signal of the current frame, or obtains a difference between the video signal and the video signal after previous frame noise removal of the current frame, and outputs the frame difference for each pixel Steps,
To the frame difference for each pixel, from the frame difference in peripheral pixels centering on the pixel performs determining majority decision motion or stationary at each pixel, the judgment result value of the frame difference Correcting and outputting a signal indicating the corrected frame difference as a motion detection signal indicating the amount of motion for each pixel of the video signal included in the frame difference;
Extract the high frequency component for each pixel in the frame difference , perform the averaging using the high frequency component in the surrounding pixels for the extracted high frequency component, from the averaged results, A noise determination step of determining a noise component for each pixel in the video signal included in the frame difference and outputting a noise determination result indicating a degree of noise included;
Subtracts the noise determination result for each pixel from the motion detection signal for each pixel, the noise component included in the frame difference, except from the motion detection signal is a value indicating the amount of motion, excluding noise components Correcting to a value indicating the amount of motion, and generating and outputting a signal based on the subtraction result as a motion level signal indicating the degree of motion based on the frame difference at each pixel in which a noise component included in the frame difference is distinguished ; ,
Obtaining and outputting a cyclic coefficient that changes according to the motion degree signal;
Multiplying the cyclic coefficient and the frame difference to obtain a noise cyclic amount;
The noise cyclic amount by subtracting the video signal of the present frame, and a step of obtaining a Film image signal removed of noise,
The cyclic coefficient decreases as the degree of movement indicated by the movement degree signal increases, and is determined to be 0 when the degree of movement indicated by the movement degree signal is equal to or greater than a predetermined value.
Noise removing method characterized by. A noise removing device according to any one of claims 1 to 3,
Display means for displaying video;
A video signal display device, comprising: a display processing unit that causes the display unit to display a video based on the video signal from which noise has been output output from the noise removal device.

Images