JP2005229491A - Method for removing noises from digital images - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method for removing noises from a digital image which can obtain sufficient effect of noise removal even from an image with high contrast, does not diminish sharpness by suppressing the effect of noise removal on the contours of the image, and does not change the central position of pixel data before the processing even after noise removing processing. <P>SOLUTION: An addition average of respective paired pixels located symmetrically with a pixel to be processed p in between is calculated for peripheral pixels in respective top, bottom, left, right, and diagonal directions of the pixel to be precessed p, and the absolute value of a difference between each addition average and the pixel to be processed p is obtained. When the value is larger than a predetermined threshold th, both of the paired pixels are determined as having no correlation between them and excluded from the addition targets when calculating noise removal pixel data. When the absolute value of the difference is the predetermined threshold th or less, both of the paired pixels are determined as having correlation between them, and regarded as addition targets when calculating the noise removal pixel data. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a method for removing a noise component contained in a digital image signal.

  Noise removal from a digital image is generally realized by averaging the pixels having high spatial or temporal correlation. In other words, the pixel to be processed in the image signal is often highly correlated with its surrounding pixels and pixels in the same position in the spatial direction, such as one frame before, whereas the noise component has almost no correlation. By using and averaging these pixels, the pixel to be processed is approximated by pixel data generated so as to cancel out noise components while suppressing deterioration of the image signal as much as possible.

  When noise removal is realized by such processing, the higher the correlation of the pixels used for addition averaging, the higher the accuracy of approximation, and conversely, when pixels with low spatial correlation are added to the addition average, If the contour portion of the image becomes unclear and the spatial resolution deteriorates, and pixels with low temporal correlation are added to the addition average, an afterimage is generated with respect to a moving image and the dynamic resolution is deteriorated.

  A conventional noise removal method will be described below.

Conventionally, the noise removal method described in Patent Document 1 is known. FIG. 23 is a diagram for explaining a conventional noise removal method. FIG. 23 shows a two-dimensional image where H represents the horizontal direction, V represents the vertical direction, p represents the pixel to be processed, n (1), n (2),..., N (8 ) Is the surrounding pixel. In the conventional noise removal method, first, an absolute value of a difference between the processing target pixel p and each of its surrounding pixels n (1), n (2),. Is greater than a predetermined threshold value th (| p−n (x) |> th, x = 1, 2,..., 8), the surrounding pixels are determined to have no correlation. When the absolute value of the difference is equal to or smaller than a predetermined threshold th (| p−n (x) | ≦ th, x = 1, 2,..., 8) Neighboring pixels are determined to be correlated and are left as they are. Thereafter, an average of 8 pixels of n (1), n (2),..., N (8) is calculated, and the processing target pixel p is replaced with the calculated value.
JP 2000-106630 A

  However, in the conventional noise removal method, the absolute value of the difference between the pixel to be processed p and the peripheral pixel n (x) is larger than a predetermined threshold value, for example, at a portion where the contrast of the image signal is large, and the noise removal effect is lost. become. Hereinafter, a description will be given with reference to FIG. For simplicity, FIG. 24 refers to the horizontal direction including the pixel to be processed. The horizontal axis represents the pixel position in the horizontal direction, and the vertical axis represents the signal level of each pixel. In FIG. 24, the processing target pixel p is a pixel on the horizontal slope with a large contrast, and the absolute value of the difference between the peripheral pixels n (4) and n (5) located on the left and right of the processing target pixel p is Since it is larger than the threshold th, it is determined that there is no correlation between the pixel to be processed and the surrounding pixels. Therefore, the peripheral pixels n (4) and n (5) are replaced with p, and with the conventional noise removal method, the pixel to be processed remains p, that is, no noise removal effect is obtained.

  In the above conventional noise removal method, as the number of correlated pixels among the peripheral pixels decreases, the original processing target pixel p at the time of addition averaging when generating pixel data that replaces the processing target pixel p. Since the weighting is increased, the noise removal effect is reduced. Hereinafter, a description will be given with reference to FIG. FIG. 25 shows a case where the processing target pixel p is a pixel on a vertical line having a width of one pixel. In this case, the peripheral pixels n (2) and n (7) positioned above and below the processing target pixel p are correlated. There is no correlation with the other six surrounding pixels. Therefore, the processing target pixel p is replaced with pixel data of p = {n (2) + n (7) + 6p} / 8, and the weight of the original processing target pixel p is increased to 6/8. That is, the noise removal effect is reduced.

  In the conventional noise removal method, there are cases where the contour portion of the original image signal cannot be stored. Hereinafter, a description will be given with reference to FIG. For simplicity, FIG. 26 refers to the horizontal direction including the pixel to be processed. The horizontal axis represents the pixel position in the horizontal direction, and the vertical axis represents the signal level of each pixel. In (a) and (b) of FIG. 26, the processing target pixel p is a pixel located at the contour portion of the image, and the absolute value of the difference from the peripheral pixel n (4) located to the left of the processing target pixel p is Since the absolute value of the difference from the peripheral pixel n (5) that is equal to or less than the threshold th and to the right is larger than the threshold th, the peripheral pixel n (4) is correlated, and the peripheral pixel n (5) Is determined to have no correlation, and the processing target pixel p is replaced with pixel data calculated by averaging the processing target pixel p and the peripheral pixel n (4). Therefore, in (a) of FIG. 26, since the signal level of the original pixel p to be processed changes from the point A to the point B, the contour portion of the original image becomes dull and unclear. Conversely, in FIG. 26B, the signal level of the original pixel p to be processed changes from point C to point D, so that the contour portion of the original image is emphasized.

  Further, in the above conventional noise removal method, the peripheral pixels added to the addition average when generating the pixel data from which noise has been removed are not necessarily symmetrical in the vertical, horizontal, and diagonal directions, but are asymmetrical. Is replaced with pixel data having a center of gravity at a spatially different position from the original pixel to be processed. Hereinafter, a description will be given with reference to FIG. FIG. 27 shows a case in which only n (2), n (3) and n (5) are correlated with the processing target pixel p. In FIG. 27, the center of gravity of the processing target pixel p is at point A before the noise removal processing, whereas the center of gravity after noise removal processing moves to point B. In this way, if the spatial center of gravity position differs depending on the subject, or if the temporal center of gravity position is moved by the motion of the subject, the motion of the subject becomes awkward in the case of a moving image. For example, if the camera is shaken in the horizontal direction with respect to a subject with vertical stripes, the image moves in the horizontal direction while the interval between the vertical stripes changes.

  The present invention solves the above-mentioned conventional problems, preserves the sharpness of the contour portion of the original image, and can obtain a noise removal effect even when the image has a high contrast or the number of highly correlated pixels is small. It is another object of the present invention to provide a noise removal method in which the position of the center of gravity does not shift spatially and temporally after the noise removal processing.

  The invention according to claim 1 of the present invention is an eight-piece set in which two pixels located in one direction spatially continuously from the pixel to be processed in each of the vertical, left-right, and diagonal directions of the pixel to be processed are one set. , A digital image noise removal method for removing noise components of the processing target pixel using a total of 16 peripheral pixels, each of two peripheral pixels positioned continuously in one direction from the processing target pixel Only when the absolute value of the difference between the pixel and the processing target pixel is equal to or less than a predetermined threshold value, both of the two peripheral pixels are to be added, and the 2M to be added from among the 16 peripheral pixels. A method for removing noise from a digital image, characterized in that the pixel to be processed is replaced with an average value after adding pixels (M is a positive number equal to or less than 8) under a predetermined weight. Spatially unidirectionally In order to determine whether or not there is a correlation with the pixel to be processed and to make the correlated pixel an addition target for noise removal pixel data generation, the contour portion of the image is stored as it is, In addition, since the processing target pixel itself is not set as an addition target for generating the noise removal pixel data, the noise removal effect can be obtained even when the number of highly correlated pixels is small.

  In the invention according to claim 2 of the present invention, eight sets each including two consecutively located in one direction spatially from the processing target pixel in each of the upper, lower, left, and right directions of the processing target pixel. , A noise removal method for a digital image that removes noise components of the processing target pixel using a total of 16 peripheral pixels, wherein two of the peripheral pixels continuously located in one direction from the processing target pixel A pixel adjacent to the processing target pixel is X (X represents a signal level), a pixel adjacent to the pixel X is Y (Y represents a signal level), and the pixels X and Y are outside the processing target pixel side. The signal level at the outer dividing point is to be added only when the absolute value of the difference between the signal level 2X-Y at the dividing point and the processing target pixel is equal to or less than a predetermined threshold, and is calculated by the eight sets of peripheral pixels. From the signal level at each outer dividing point A digital image characterized in that the pixel to be processed is replaced with an average value after adding signal levels of M outer division points (M is a positive number of 8 or less) to be calculated under a predetermined weight. This is a noise removal method, and there is a correlation by determining whether or not there is a correlation between the outer division point on the processing target pixel side of the pixel that is spatially continuously located in one direction from the processing target pixel and the processing target pixel. Since the outer division point is used as the addition target for noise removal pixel data generation, the outline portion of the image is stored as it is, and the outer division point is used to determine the presence or absence of correlation. An effect can be obtained, and since the processing target pixel itself is not set as an addition target for noise removal pixel data generation, the noise removal effect can be obtained even when the number of highly correlated pixels is small.

  The invention according to claim 3 of the present invention is a total of four sets, each including two sets positioned spatially symmetrically across the processing target pixel in each of the vertical, horizontal, and diagonal directions of the processing target pixel. A digital image noise removal method for removing noise components of the processing target pixel using eight peripheral pixels, each of two peripheral pixels positioned symmetrically across the processing target pixel and the processing target Only when the absolute value of the difference from the pixel is equal to or less than a predetermined threshold, both of the two peripheral pixels are to be added, and 2M (M is the target of addition) from among the eight peripheral pixels. A method for removing noise from a digital image, wherein the pixel to be processed is replaced by an average value obtained by adding pixels of a positive number of 4 or less) under a predetermined weight. Pixels located symmetrically In order to determine whether or not there is a correlation with the pixel to be processed and to set the correlated pixel as an addition target for noise removal pixel data generation, the contour portion of the image is stored as it is, and the processing target pixel itself is denoised. Since it is not included in the pixel data generation target, noise reduction can be achieved even when there are few highly correlated pixels. In addition, pixels that are located spatially symmetrically across the processing target pixel must be paired with noise. Since it is a target for generating the removal pixel data, it has an effect that the position of the center of gravity does not shift spatially and temporally after the noise removal processing.

  The invention according to claim 4 of the present invention is a total of N sets, each of which is a set of two that are spatially symmetrical with the process target pixel sandwiched in each of the upper, lower, left, and right directions of the process target pixel. A method of removing noise of a digital image using 2N (N is a positive number) peripheral pixels to remove noise components of the processing target pixel, wherein the two peripheral pixels are symmetrically positioned across the processing target pixel Only when the absolute value of the difference between each of the pixel and the pixel to be processed is equal to or less than a predetermined threshold, both of the two peripheral pixels are to be added, and the addition target is selected from the 2N peripheral pixels. A noise removal method for a digital image, characterized in that the processing target pixel is replaced by an average value after adding 2M pixels (M is a positive number equal to or less than N) under a predetermined weight. Spatial symmetry across the target pixel Since the pixel to be placed is paired, the presence or absence of correlation with the processing target pixel is judged, and the correlated pixel is used as the addition target for noise removal pixel data generation. Since the pixels themselves are not included in the noise removal pixel data generation target, it is possible to obtain a noise removal effect even when there are few highly correlated pixels, and pixels that are always spatially symmetrical across the processing target pixel As a pair, the target of noise removal pixel data generation has an effect that the position of the center of gravity does not shift spatially and temporally after the noise removal processing.

  The invention according to claim 5 of the present invention is a total of four sets, each of which has two sets positioned symmetrically across the processing target pixel in the vertical and horizontal directions and the diagonal direction of the processing target pixel. A noise removal method for a digital image that uses eight peripheral pixels to remove a noise component of the processing target pixel, and is an average of two peripheral pixels located spatially symmetrically across the processing target pixel Only when the absolute value of the difference between the value and the processing target pixel is equal to or smaller than a predetermined threshold value, both of the two peripheral pixels are to be added, and 2M to be added from the eight peripheral pixels. A noise removal method for a digital image, characterized in that the processing target pixel is replaced with an average value after adding pixels (M is a positive number of 4 or less) under a predetermined weight. Spatially symmetrical across Pair the pixels to determine whether or not there is correlation with the pixel to be processed, and add the correlated pixels to the noise removal pixel data generation target. Since the addition average value of two pixels located spatially symmetrically across the processing target pixel is used for the determination, a noise removal effect can be obtained even in the case of an image with a large contrast, and the processing target pixel itself is Since it is not included in the removal pixel data generation target, it is possible to obtain a noise removal effect even when there are few highly correlated pixels. Since it is a target of noise removal pixel data generation, it has an effect that the position of the center of gravity does not shift spatially and temporally after the noise removal processing.

  The invention according to claim 6 of the present invention is a total of N sets, each of which has two sets positioned symmetrically across the processing target pixel in each of the upper, lower, left, and right directions of the processing target pixel. A method of removing noise of a digital image using 2N (N is a positive number) peripheral pixels to remove noise components of the processing target pixel, wherein the two peripheral pixels are symmetrically positioned across the processing target pixel Only when the absolute value of the difference between the addition average value of the pixel and the processing target pixel is equal to or less than a predetermined threshold, both of the two peripheral pixels are to be added, and the addition target is selected from the 2N peripheral pixels. A noise removal method for a digital image, wherein the pixel to be processed is replaced with an average value after adding 2M pixels (M is a positive number equal to or less than N) under a predetermined weight. Spatial symmetry across the target pixel In order to determine whether or not there is a correlation with the pixel to be processed by pairing the located pixels and make the correlated pixel as the addition target for noise removal pixel data generation, the contour part of the image is saved as it is, and the correlation Since the addition average value of two pixels located spatially symmetrically with the processing target pixel in between is used for the determination of the presence or absence of noise, a noise removal effect can be obtained even in the case of an image with a large contrast, and the processing target pixel itself Is not included in the noise removal pixel data generation target, so that even when there are few highly correlated pixels, noise removal effects can be obtained, and pixels that are located spatially symmetrically across the processing target pixel must be paired. Thus, since it is a target of noise removal pixel data generation, there is an effect that the position of the center of gravity does not shift spatially and temporally after the noise removal processing.

  The invention according to claim 7 of the present invention is a set including two pixels that are spatially the same position and symmetrically in time as the processing target pixel in one frame before and after the frame including the processing target pixel. A noise removal method for a digital image that uses a pixel to remove a noise component of the processing target pixel, and is an absolute value of a difference between an addition average value of two pixels of the set of pixels and the processing target pixel Only when the two pixels are equal to or less than a predetermined threshold, both of the two pixels are to be added, and the pixel to be processed is replaced by an average value after adding the two peripheral pixels. This is a noise removal method that pairs pixels that are located symmetrically in time with the pixel to be processed in between, and determines whether or not there is a correlation with the pixel to be processed. To be added The outline of the image is preserved as it is, the deterioration of the dynamic resolution due to the afterimage is prevented with respect to the moving image, and the addition of two pixels located symmetrically in time with the processing target pixel sandwiched in determining the presence or absence of correlation Since the average value is used, a noise removal effect can be obtained even in the case of an image whose signal level changes greatly with time, and the processing target pixel itself is not included in the addition of noise removal pixel data generation. A noise removal effect can be obtained even if there are two high pixels, and since pixels that are symmetrically located temporally across the pixel to be processed are paired to generate noise removal pixel data, noise Even after the removal process, the position of the center of gravity does not shift spatially and temporally.

  In the invention according to claim 8 of the present invention, two frames that are spatially the same position and symmetrically in time as the processing target pixel in about one frame with respect to the frame including the processing target pixel are set as one set. The pixel to be processed using a total of 2N pixels from N pixels to two pixels located in a spatially the same position and symmetrically in time with the pixel to be processed before and after N frames. The digital image noise removing method for removing the noise component of the N sets of pixels, wherein an absolute value of a difference between an addition average value of two pixels of each of the N sets of pixels and the processing target pixel is equal to or less than a predetermined threshold value Only when the two pixels are to be added, 2M pixels (M is a positive number less than N) of the 2N neighboring pixels are added to the 2N peripheral pixels under a predetermined weight. The pixel to be processed is replaced by the average value after addition in step A method for removing noise from a digital image characterized by the above, wherein a pair of pixels located symmetrically with respect to each other with respect to a processing target pixel is correlated to determine whether or not there is a correlation with the processing target pixel. In order to make a pixel to be an addition target for noise removal pixel data generation, the outline portion of the image is stored as it is, and an addition average value of two pixels located symmetrically in time with the processing target pixel sandwiched in determining the presence or absence of correlation. Therefore, it is possible to obtain a noise removal effect even in the case of an image whose signal level changes greatly with time, and since the processing target pixel itself is not an addition target for noise removal pixel data generation, a highly correlated pixel can be obtained. Noise reduction effect can be obtained even when there are few, and pixels that are symmetrically located in time with respect to the target pixel are always paired to generate noise removal pixel data. After also processed spatially have the effect of never deviate temporal gravity center position.

  The invention according to claim 9 of the present invention is a set including two pixels that are spatially the same position and symmetrically in time as the processing target pixel in one frame before and after the frame including the processing target pixel. A noise removal method for a digital image in which a noise component of the processing target pixel is removed using a pixel to be processed, wherein an absolute value of a difference between each of the two pixels of the set of pixels and the processing target pixel is both The digital image noise characterized in that both of the two pixels are subject to addition only when the threshold value is equal to or less than a predetermined threshold value, and the processing subject pixel is replaced by an average value after the addition of the two peripheral pixels. This is a removal method. Pair pixels that are located symmetrically with respect to the pixel to be processed, determine whether there is correlation with the pixel to be processed, and add the correlated pixels to generate noise-removed pixel data. To target The image outline is preserved as it is, and the processing target pixel itself is not included in the noise removal pixel data generation target. Therefore, even if there are two highly correlated pixels, a noise removal effect can be obtained, and processing is always performed. Since the pixels located symmetrically with respect to the target pixel are paired as a target for noise removal pixel data generation, the spatial and temporal barycentric positions do not shift even after the noise removal processing. .

  According to a tenth aspect of the present invention, a set including two pixels that are spatially the same position and symmetrically in time with the pixel to be processed in one frame before and after a frame including the pixel to be processed. The pixel to be processed using a total of 2N pixels from N pixels to two pixels located in a spatially the same position and symmetrically in time with the pixel to be processed before and after N frames. The digital image noise removing method for removing the noise component of the N sets of pixels, wherein the absolute value of the difference between each of the two pixels of each of the N sets of pixels and the processing target pixel is equal to or less than a predetermined threshold value. Only when both of the two pixels are to be added, 2M pixels (M is a positive number less than N) of the 2N peripheral pixels are added with a predetermined weight. Replace the pixel to be processed with the average value after addition A method for removing noise from a digital image characterized by the above, wherein a pair of pixels located symmetrically with respect to each other with respect to a processing target pixel is correlated to determine whether or not there is a correlation with the processing target pixel. Since the pixels are subject to addition for noise removal pixel data generation, the image outline is preserved as is, and the processing target pixels themselves are not subject to addition for noise removal pixel data generation, so even if there are few highly correlated pixels, noise A removal effect can be obtained, and pixels that are symmetrically located temporally across the pixel to be processed are paired as a target for noise removal pixel data generation. It has the effect that the position of the center of gravity does not shift.

  According to an eleventh aspect of the present invention, the frame including the pixel to be processed is located 2 in time and continuously in the same position spatially as the pixel to be processed one frame before and two frames before. A noise removal method for a digital image that uses a set of pixels to remove noise components of the processing target pixel, wherein a pixel located one frame before the processing target pixel is selected from the two pixels. A frame including the processing target pixel calculated from the pixels X and Y, where X (X represents a signal level) and a pixel located two frames before the processing target pixel is Y (Y represents a signal level) The signal level at the outer dividing point is added only when the absolute value of the difference between the signal level 2X-Y at the outer dividing point and the pixel to be processed is equal to or smaller than a predetermined threshold, and the outer level to be added Signal level at the minute point A method for removing noise from a digital image, wherein the processing target pixel is replaced by a correlation between an outer dividing point on a processing target pixel side of a pixel located temporally continuously from the processing target pixel and the processing target pixel In order to determine whether or not there is a characteristic, and to add a correlated external dividing point as an addition target for generation of noise-removed pixel data, the outline portion of the image is preserved as it is, and the dynamic resolution is deteriorated due to the afterimage of the moving image. In order to prevent and use the external dividing point on the processing target pixel side of the pixel located temporally continuously from the processing target pixel for the determination of the presence / absence of correlation, even in the case of an image whose signal level greatly changes over time Since the noise removal effect can be obtained and the processing target pixel itself is not set as the addition target for noise removal pixel data generation, the noise removal effect can be obtained even when two highly correlated pixels are present. It is having an effect.

  As described above, the present invention preserves the sharpness of the contour portion of the original image, suppresses the degradation of the dynamic resolution due to the afterimage with respect to the moving image, and has a pixel having high correlation with an image having a high contrast and peripheral pixels. There is an effect that it is possible to provide a noise removal method for a digital image that can ensure a noise removal effect when there are few, and can store a spatial and temporal barycentric position after the noise removal process.

  Hereinafter, embodiments of the present invention will be described with reference to FIGS.

(Embodiment 1)
FIG. 8 is a diagram for explaining a noise removal method for a digital image according to the first embodiment of the present invention. FIG. 8 is an image of a two-dimensional image, where H represents the horizontal direction and V represents the vertical direction, p is a pixel to be processed, n (x, y) is a peripheral pixel of the processing target pixel p (x = 1) , 2,..., 8; y = 1, 2).

  The digital image noise removal method of Embodiment 1 of the present invention will be described below with reference to FIG.

  In the digital image noise removal method according to the first embodiment of the present invention, first, the peripheral pixels in the vertical, horizontal, and diagonal directions of the processing target pixel p are successively located in one direction from the processing target pixel p. Each pair of n (1,1) and n (1,2), n (2,1) and n (2,2), ..., n (8,1) and n (8,2) When the absolute value of the difference between the pixel to be processed and the pixel to be processed p is obtained, and the absolute value of the difference between the two neighboring pixels that are paired is smaller than a predetermined threshold th (| p−n ( x, 1) | ≦ th and | pn (x, 2) | ≦ th, x = 1, 2,..., 8), the two pixels of the pair are judged to be correlated. In other cases, it is determined that there is no correlation between the two pixels in the pair. Remove from the addition target of the time's removal pixel data calculation. Thereafter, the paired pixels to be added are appropriately weighted to calculate an addition average, and the processing target pixel p is replaced with the calculated value. In this case, all the weights may be added equally and divided by the added number. To simplify the circuit configuration, for example, n (1,1) and n (1,2) are provided for each direction. ), N (2,1) and n (2,2) are correlated, S1 (1,2) = (n (1,1) + n (1,2) + n (2,1) ) + N (2,2)) / 4, and when there is no correlation only in the pair of n (2,1) and n (2,2), S1 (1,2) = (n (1,1) + n (1 , 2)) / 2, and when there is no correlation only in the pair of n (1,1) and n (1,2), S1 (1,2) = (n (2,1) + n (2,2)) / 2, n (1,1) and n (1,2), n (2,1), and n (2,2) pairs are not correlated, S1 (1,2) = 0 In the same way, the vertical and horizontal directions and diagonal directions are added And, finally averaging the two may be calculated noise reduction pixel data.

  In this noise removal method, the number of pixels to be added is increased or decreased in pairs, so that the circuit configuration can be simplified relatively easily.

  In the present noise removal method, the contour portion of the image can be stored even under the conditions as shown in FIG. Hereinafter, a description will be given with reference to FIG. 9 (a) and 9 (b), the processing target pixel p is a pixel located in the contour portion of the image, and in the case of FIG. 9 (a), a peripheral pixel located two pixels to the left of the processing target pixel p. The absolute value of the difference from n (1,2) is less than or equal to the threshold th, and the absolute value of the difference from the surrounding pixel n (1,1) located one pixel to the left is greater than the threshold th. In the case of FIG. 9B, the absolute value of the difference from the surrounding pixel n (1,2) located 2 pixels to the left of the processing target pixel p is larger than the threshold th and located 1 pixel to the left. The absolute value of the difference from the peripheral pixel n (1, 1) is equal to or less than the threshold th. However, in both cases (a) and (b) of FIG. 9, one of n (1,1) and n (1,2), which are continuously located in one direction from the processing target pixel p, is a processing target. The absolute value of the difference from the pixel p does not fall within the threshold th. Therefore, both n (1,1) and n (1,2) are determined to have no correlation, and are excluded from the addition target pixels at the time of noise removal pixel data calculation. That is, the noise removal effect is suppressed for the pixels located in the contour portion of such an image, the contour portion of the original image is preserved as it is, and deterioration of sharpness is suppressed. In addition, since intra-frame processing is performed, there is no deterioration in dynamic resolution and there is no temporal movement of the center of gravity.

  In addition, in the present noise removal method, the original pixel to be processed p is not used when calculating the noise removal pixel data. Therefore, even when the number of correlated pixels in the surrounding pixels is reduced, the noise removal effect is not reduced. It is not affected by the processing target pixel p.

  In this noise removal method, the original processing target pixel p is not used when calculating the noise removal pixel data. However, in order to remove a higher-frequency noise component, the addition target pixels at the time of calculating the noise removal pixel data are used. In addition, the original processing target pixel p may be added.

  As described above, according to the present embodiment, the sharpness of the contour portion of the original image is preserved, the deterioration of the dynamic resolution due to the afterimage of the moving image is prevented, and the peripheral pixels have few highly correlated pixels. In this case, it is also possible to realize a digital image noise removal method in which the position of the center of gravity does not shift even after the noise removal processing.

(Embodiment 2)
FIG. 10 is a diagram for explaining a noise removal method for a digital image according to the second embodiment of the present invention. FIG. 10 shows an image of a two-dimensional image, and its arrangement is the same as that of the digital image noise removing method according to the first embodiment of the present invention.

  Hereinafter, a digital image noise removing method according to Embodiment 2 of the present invention will be described with reference to FIG.

  In the noise removal method for a digital image according to the second embodiment of the present invention, first, peripheral pixels in each of the upper, lower, left, and right directions of the processing target pixel p are continuously located in one direction from the processing target pixel p. Each pair of n (1,1) and n (1,2), n (2,1) and n (2,2), ..., n (8,1) and n (8,2) , N (8) (n (x) = 2 * n (x, 1) −n) at the signal level n (1), n (2),. (X, 2), x = 1, 2,..., 8) are calculated, the absolute value of the difference between the signal level of each outer dividing point and the processing target pixel p is obtained, and the value is set in advance. If it is greater than the predetermined threshold th (| p−n (x) |> th, x = 1, 2,..., 8), it is determined that the outer dividing point of the pair has no correlation. , Noise removal If the absolute value of the difference is not greater than the predetermined threshold th (| p−n (x) | ≦ th, x = 1, 2,..., 8) The signal level at the outer dividing point is set as an addition target when calculating the noise-removed pixel data. Thereafter, an appropriate weight is applied to the signal level of the outer dividing point to be added, an addition average is calculated, and the processing target pixel p is replaced with the calculated value. In this case, all the weights may be added equally and divided by the added number. To simplify the circuit configuration, for example, n (1,1) and n (1,2) are provided for each direction. ), And when the outer dividing points of n (2,1) and n (2,2) are correlated, S1 (1,2) = (2 * n (1,1) -n ( 1,2) + 2 * n (2,1) -n (2,2)) / 2, and S1 (1 when there is no correlation only at the outer dividing points of n (2,1) and n (2,2) , 2) = (2 * n (1,1) -n (1,2)), and if there is no correlation only in the outer dividing points of n (1,1) and n (1,2), S1 (1, 2) = (2 * n (2,1) -n (2,2)), outside of n (1,1) and n (1,2), n (2,1) and n (2,2) When the minute points are not correlated, after addition averaging as in S1 (1,2) = 0, the vertical and horizontal directions, diagonal And averaging similarly direction each other, finally averaging the two may be calculated noise reduction pixel data.

  In this noise removal method, the number of target pixels for noise removal pixel data calculation is increased or decreased in pairs, so that it is relatively easy to simplify the circuit configuration. In addition, since the correlation is determined not at each pixel but at the outer dividing point on the processing target pixel p side of the two pixels, even when the signal level difference between each pixel and the processing target pixel p is large. Since the signal level of the outer dividing point may be within a threshold value with respect to the pixel p to be processed, if the same threshold value is set, more peripheral pixels should be added compared to the conventional noise removal method. Can do. In other words, in order to obtain the same noise removal effect, the present noise removal method has a feature that the threshold can be set smaller than the conventional noise removal method.

  FIG. 11 is an explanatory diagram for the case of the same conditions as FIG. 24 listed as problems in the conventional noise removal method. In FIG. 11, as in the case of FIG. 24, the processing target pixel p is a pixel on the horizontal slope with a large contrast, and the peripheral pixel n (1, 1) and the neighboring pixels n (1, 1) continuously located on the left side of the processing target pixel p. The absolute value of the difference from n (1,2) is larger than the threshold th. However, the signal level of the outer dividing point on the processing target pixel p side of n (1,1) and n (1,2) is the level of point A in the figure, and the absolute value of the difference from the processing target pixel p is equal to or less than the threshold th. It becomes. Therefore, the outer dividing points of n (1,1) and n (1,2) are determined to have a correlation, and become pixels to be added when noise removal pixel data is calculated. That is, a noise removal effect can be obtained even for pixels on a slope with a large contrast.

  Next, the operation of the present noise removal method under the conditions as shown in FIG. 12 will be described. 12 (a) and 12 (b), the processing target pixel p is a pixel located at the contour portion of the image. In the case of FIG. 12 (a), the peripheral pixel located two pixels to the left of the processing target pixel p. The absolute value of the difference from n (1,2) is less than or equal to the threshold th, and the absolute value of the difference from the surrounding pixel n (1,1) located one pixel to the left is greater than the threshold th. In the case of FIG. 12B, the absolute value of the difference from the peripheral pixel n (1, 2) located 2 pixels to the left of the processing target pixel p is larger than the threshold th and located 1 pixel to the left. The absolute value of the difference from the peripheral pixel n (1, 1) is equal to or less than the threshold th. However, in the case of FIG. 12A, the signal levels at the outer dividing points on the processing target pixel p side of n (1,1) and n (1,2), which are continuously located in one direction from the processing target pixel p, are The absolute value of the difference from the processing target pixel p is not within the threshold th. Therefore, both n (1,1) and n (1,2) are determined to have no correlation, and are excluded from the target pixels at the time of noise removal pixel data calculation. On the other hand, in the case of FIG. 12B, the signal levels at the outer dividing points on the processing target pixel p side of n (1,1) and n (1,2), which are continuously located in one direction from the processing target pixel p. Becomes point B, and the absolute value of the difference from the processing target pixel p falls within the threshold th. Therefore, the outer dividing points of n (1,1) and n (1,2) are determined to have a correlation, and become target pixels at the time of noise removal pixel data calculation. That is, with respect to the pixels located in the contour portion of such an image, the noise removal effect is suppressed for the sharp contour portion, the contour portion of the original image is preserved as it is, and deterioration of the sharpness is suppressed. In addition, since intra-frame processing is performed, there is no deterioration in dynamic resolution and there is no temporal movement of the center of gravity.

  In addition, in the present noise removal method, the original pixel to be processed p is not used when calculating the noise removal pixel data. Therefore, even when the number of correlated pixels in the surrounding pixels is reduced, the noise removal effect is not reduced. It is not affected by the processing target pixel p.

  In this noise removal method, the original processing target pixel p is not used when calculating the noise removal pixel data. However, in order to remove a higher-frequency noise component, the addition target pixels at the time of calculating the noise removal pixel data are used. In addition, the original processing target pixel p may be added.

  As described above, according to the present embodiment, the sharpness of the contour portion of the original image is preserved to some extent, the deterioration of the dynamic resolution due to the afterimage is prevented with respect to the moving image, and the slope portion where the processing target pixel has a large contrast A digital image in which sufficient noise removal effect can be obtained even if it exists in the image, and it is effective even when there are few highly correlated pixels in the surrounding pixels, and the temporal center of gravity position does not shift even after noise removal processing The noise removal method can be realized.

(Embodiment 3)
FIG. 5 is a diagram for explaining a noise removal method for a digital image according to the third embodiment of the present invention. FIG. 5 shows a two-dimensional image, where H represents the horizontal direction, V represents the vertical direction, p is the pixel to be processed, and n (x, y) is the peripheral pixel of the pixel to be processed p (x = 1). , 3, 5, 7; y = 1, 2).

  Hereinafter, a digital image noise removing method according to Embodiment 3 of the present invention will be described with reference to FIG.

  In the noise removal method for a digital image according to the third embodiment of the present invention, first, n (positioned symmetrically across the processing target pixel p with respect to peripheral pixels in the vertical, horizontal, and diagonal directions of the processing target pixel p. 1,1) and n (1,2), n (3,1) and n (3,2), n (5,1) and n (5,2), n (7,1) and n (7 , 2) The absolute value of the difference between each paired pixel and the processing target pixel p is obtained, and the absolute value of the difference between the two neighboring pixels to be paired is both greater than a predetermined threshold th. If it is small (| p−n (x, 1) | ≦ th and | p−n (x, 2) | ≦ th, x = 1,3,5,7) It is determined that there is a correlation, and is included in the addition when calculating the noise-removed pixel data. Otherwise, it is determined that the two pixels in the pair are not correlated. To, remove from the addition subject during denoising pixel data calculated. Thereafter, the paired pixels to be added are appropriately weighted to calculate an addition average, and the processing target pixel p is replaced with the calculated value. In this case, all the weights may be added equally and divided by the added number. To simplify the circuit configuration, for example, n (1,1) and n (1,2) are provided for each direction. ) Pair is correlated, S1 (1) = (n (1,1) + n (1,2)) / 2, n (1,1) and n (1,2) pair are correlated In the case of none, even if the averaging is performed as in S1 (1) = 0, the vertical direction, the horizontal direction, and the diagonal direction are similarly averaged, and finally both are averaged to calculate noise removal pixel data. good.

  In this noise removal method, the number of pixels to be added is increased or decreased in pairs, so that the circuit configuration can be simplified relatively easily.

  FIG. 7 is an explanatory diagram for the case of the same conditions as FIG. In (a) and (b) of FIG. 7, the processing target pixel p is a pixel located in the contour portion of the image as in the case of (a) and (b) of FIG. The absolute value of the difference with the surrounding pixel n (1, 1) located is equal to or smaller than the threshold th, and the absolute value of the difference with the surrounding pixel n (1, 2) located on the right is larger than the threshold th. . However, in both cases (a) and (b) of FIG. 7, of n (1,1) and n (1,2) located symmetrically with respect to the processing target pixel p, n (1,2) is processing. The absolute value of the difference from the target pixel p does not fall within the threshold th. Therefore, both n (1,1) and n (1,2) are determined to have no correlation, and are excluded from the addition target pixels at the time of noise removal pixel data calculation. That is, the noise removal effect is suppressed for the pixels located in the contour portion of such an image, the contour portion of the original image is preserved as it is, and deterioration of sharpness is suppressed. In addition, since intra-frame processing is performed, there is no deterioration in dynamic resolution.

  In addition, in the present noise removal method, the original pixel to be processed p is not used when calculating the noise removal pixel data. Therefore, even when the number of correlated pixels in the surrounding pixels is reduced, the noise removal effect is not reduced. It is not affected by the processing target pixel p.

  Furthermore, in this noise removal method, the peripheral pixels added to the averaging when generating the noise removal pixel data always exist symmetrically in the vertical, horizontal, and diagonal directions, so that they are spatially the same as the original processing target pixel. The pixel data having the center of gravity at the position is replaced. Further, since intra-frame processing is performed, there is no temporal movement of the center of gravity.

  In this noise removal method, the original processing target pixel p is not used when calculating the noise removal pixel data. However, in order to remove a higher-frequency noise component, the addition target pixels at the time of calculating the noise removal pixel data are used. In addition, the original processing target pixel p may be added.

  As described above, according to the present embodiment, the sharpness of the contour portion of the original image is preserved, the deterioration of the dynamic resolution due to the afterimage of the moving image is prevented, and the peripheral pixels have few highly correlated pixels. In this case, a noise removal method for a digital image can be realized in which the position of the center of gravity is not shifted spatially and temporally after the noise removal processing.

(Embodiment 4)
FIG. 6 is a diagram for explaining a noise removal method for a digital image according to the fourth embodiment of the present invention. FIG. 6 is a two-dimensional image, which is obtained by further increasing the number of pixels subject to noise removal pixel data calculation in the radiation direction with respect to the noise removal method of Embodiment 3 of FIG. The arrangement is the same as that of the digital image noise removal method according to the first and second embodiments of the present invention, but the number of pixels is changed because the combination of surrounding pixels to be paired is different.

  Hereinafter, a digital image noise removing method according to Embodiment 4 of the present invention will be described with reference to FIG.

  In the digital image noise removal method according to the fourth embodiment of the present invention, first, n (positioned symmetrically with respect to the processing target pixel p with respect to peripheral pixels in the vertical, horizontal, and diagonal directions of the processing target pixel p. 1,1) and n (1,2), n (2,1) and n (2,2),..., N (8,1) and n (8,2) in pairs When the absolute value of the difference between the two neighboring pixels forming a pair is smaller than a predetermined threshold th (| p−n (x, 1) If | ≦ th and | pn (x, 2) | ≦ th, x = 1, 2,..., 8), it is determined that the two pixels of the pair are correlated. In the other cases, it is determined that there is no correlation between the two pixels in the pair. Remove from the addition target at the time of the pixel data calculation. Thereafter, the paired pixels to be added are appropriately weighted to calculate an addition average, and the processing target pixel p is replaced with the calculated value. In this case, all the weights may be added equally and divided by the added number. To simplify the circuit configuration, for example, n (1,1) and n (1,2) are provided for each direction. ), N (2,1) and n (2,2) are correlated, S1 (1,2) = (n (1,1) + n (1,2) + n (2,1) ) + N (2,2)) / 4, and when there is no correlation only in the pair of n (2,1) and n (2,2), S1 (1,2) = (n (1,1) + n (1 , 2)) / 2, and when there is no correlation only in the pair of n (1,1) and n (1,2), S1 (1,2) = (n (2,1) + n (2,2)) / 2, n (1,1) and n (1,2), n (2,1), and n (2,2) pairs are not correlated, S1 (1,2) = 0 In the same way, the vertical and horizontal directions and diagonal directions are added And, finally averaging the two may be calculated noise reduction pixel data.

  Even in this noise removal method, since the number of pixels to be added increases or decreases in pairs by two, simplification of the circuit configuration is relatively easy.

  Similar to the digital image noise removing method according to the third embodiment of the present invention, the present noise removing method can preserve the contour portion of the image even under the conditions shown in FIG. Since this is done, there is no degradation of dynamic resolution.

  Further, even in the present noise removal method, the original processing target pixel p is not used when the noise removal pixel data is calculated. Therefore, even when the number of correlated pixels in the peripheral pixels is reduced, the noise removal effect is not reduced. It is not affected by the processing target pixel p.

  Furthermore, even in this noise removal method, the peripheral pixels added to the averaging when generating the noise removal pixel data always exist symmetrically in the vertical, horizontal, and diagonal directions, so that they are spatially the same as the original pixel to be processed. The pixel data having the center of gravity at the position is replaced. Further, since intra-frame processing is performed, there is no temporal movement of the center of gravity.

  Even in the present noise removal method, the original processing target pixel p is not used when calculating the noise removal pixel data. However, in order to remove higher-frequency noise components, the addition target pixels when calculating the noise removal pixel data are used. In addition, the original processing target pixel p may be added.

  As described above, according to the present embodiment, the sharpness of the contour portion of the original image is preserved, the deterioration of the dynamic resolution due to the afterimage of the moving image is prevented, and the peripheral pixels have few highly correlated pixels. In this case, a noise removal method for a digital image can be realized in which the position of the center of gravity is not shifted spatially and temporally after the noise removal processing.

(Embodiment 5)
FIG. 1 is a diagram for explaining a digital image noise removal method according to a fifth embodiment of the present invention. FIG. 1 shows an image of a two-dimensional image, and its arrangement is the same as that of the digital image noise removing method according to the third embodiment of the present invention.

  Hereinafter, a digital image noise removing method according to Embodiment 5 of the present invention will be described with reference to FIG.

  In the digital image noise removal method according to the fifth embodiment of the present invention, first, n (positioned symmetrically across the processing target pixel p with respect to peripheral pixels in each of the vertical, horizontal, and diagonal directions of the processing target pixel p. 1,1) and n (1,2), n (3,1) and n (3,2), n (5,1) and n (5,2), n (7,1) and n (7 , 2), the average value n (1), n (3), n (5), n (7) (n (x) = (n (x, 1) + n (x, 2)) / 2, x = 1, 3, 5, 7), the absolute value of the difference between each addition average value and the processing target pixel p is obtained, and the value is set to a predetermined threshold th. Is greater than (| p−n (x) |> th, x = 1, 3, 5, 7), it is determined that the two pixels in the pair are not correlated, and noise-removed pixel data is calculated. Addition of time vs. If the absolute value of the difference is less than or equal to the predetermined threshold th (| p−n (x) | ≦ th, x = 1, 3, 5, 7), the two pixels in the pair are both correlated. It is determined that the pixel is present, and is used as an addition target when calculating the noise-removed pixel data. Thereafter, the paired pixels to be added are appropriately weighted to calculate an addition average, and the processing target pixel p is replaced with the calculated value. A specific method is the same as that of the digital image noise removing method according to the third embodiment of the present invention.

  Since the number of pixels to be added is increased or decreased in pairs, it is relatively easy to simplify the circuit configuration. Further, since the correlation is determined not by individual pixels but by an average value of two pixels located symmetrically with respect to the processing target pixel p, each pixel has a signal level difference from the processing target pixel p. Even when the average value of two pixels is large, the difference from the pixel to be processed p may be within the threshold value by taking the average value of the two. Therefore, if the same threshold value is set, many peripheral pixels are compared with the conventional noise removal method. Can be added. In other words, in order to obtain the same noise removal effect, the present noise removal method has a feature that the threshold can be set smaller than the conventional noise removal method.

  FIG. 3 is an explanatory diagram for the case of the same conditions as FIG. In FIG. 3, the processing target pixel p is a pixel on the horizontal slope having a large contrast as in the case of FIG. 24, and the peripheral pixels n (1, 1) and n (1 , 2) is larger than the threshold value th. However, the average value of n (1,1) and n (1,2) is the level at point A in the figure, and the absolute value of the difference from the processing target pixel p is equal to or less than the threshold th. Therefore, both n (1,1) and n (1,2) are determined to have a correlation, and become pixels to be added when noise removal pixel data is calculated. That is, a noise removal effect can be obtained even for pixels on a slope with a large contrast.

  FIG. 4 is an explanatory diagram for the case of the same conditions as FIG. 4 (a) and 4 (b), the processing target pixel p is a pixel located at the contour portion of the image, as in FIGS. 23 (a) and 23 (b), and to the left of the processing target pixel p. The absolute value of the difference with the surrounding pixel n (1, 1) located is equal to or smaller than the threshold th, and the absolute value of the difference with the surrounding pixel n (1, 2) located on the right is larger than the threshold th. . However, in FIG. 4A, the average value of n (1,1) and n (1,2) is the level at point A in the figure, and the absolute value of the difference from the processing target pixel p is greater than the threshold th. Therefore, both n (1,1) and n (1,2) are determined to have no correlation, and are excluded from the addition target pixels at the time of noise removal pixel data calculation. That is, the noise removal effect is suppressed for the pixels located in the contour portion of such an image, the contour portion of the original image is preserved as it is, and deterioration of sharpness is suppressed. In FIG. 4B, the average value of n (1,1) and n (1,2) is the level at point B in the figure. If this is the same threshold value as in FIG. 23B, it is considered that there is a correlation. However, as described above, in order to obtain a noise removal effect of the same level in this noise removal method, it is compared with the conventional noise removal method. The threshold can be set small. Therefore, in this case, if the threshold is set small, both n (1,1) and n (1,2) can be excluded from the addition target pixels and the contour portion can be saved. In addition, since intra-frame processing is performed, there is no deterioration in dynamic resolution.

  In addition, in the present noise removal method, the original pixel to be processed p is not used when calculating the noise removal pixel data. Therefore, even when the number of correlated pixels in the surrounding pixels is reduced, the noise removal effect is not reduced. It is not affected by the processing target pixel p.

  Furthermore, in this noise removal method, the peripheral pixels added to the averaging when generating the noise removal pixel data always exist symmetrically in the vertical, horizontal, and diagonal directions, so that they are spatially the same as the original processing target pixel. The pixel data having the center of gravity at the position is replaced. Further, since intra-frame processing is performed, there is no temporal movement of the center of gravity.

  In this noise removal method, the original processing target pixel p is not used when calculating the noise removal pixel data. However, in order to remove a higher-frequency noise component, the addition target pixels at the time of calculating the noise removal pixel data are used. In addition, the original processing target pixel p may be added.

  As described above, according to the present embodiment, the sharpness of the contour portion of the original image is preserved, the deterioration of the dynamic resolution due to the afterimage is prevented with respect to the moving image, and the processing target pixel is changed to the slope portion having a large contrast. A sufficient noise removal effect can be obtained even when it exists, and it is also effective when there are few highly correlated pixels in the surrounding pixels, and the spatial and temporal center of gravity position does not shift after noise removal processing. A noise removal method for digital images can be realized.

(Embodiment 6)
FIG. 2 is a diagram for explaining a noise removal method for a digital image according to the sixth embodiment of the present invention. FIG. 2 is a two-dimensional image, which is obtained by further increasing the pixels to be subjected to noise removal pixel data calculation in the radiation direction with respect to the noise removal method of the fifth embodiment of FIG. The arrangement is the same as the digital image noise removal method of the fourth embodiment of the present invention.

  The digital image noise removing method according to the sixth embodiment of the present invention will be described below with reference to FIG.

  In the digital image noise removal method according to the sixth embodiment of the present invention, first, n (positioned symmetrically across the processing target pixel p with respect to the peripheral pixels in the vertical, horizontal, and diagonal directions of the processing target pixel p. 1,1) and n (1,2), n (2,1) and n (2,2),..., N (8,1) and n (8,2) in pairs N (1), n (2),..., N (8) (n (x) = (n (x, 1) + n (x, 2)) / 2, x = 1, 2 ,..., 8) are calculated, the absolute value of the difference between each addition average value and the processing target pixel p is obtained, and the value is greater than a predetermined threshold th (| p−n) (X) |> th, x = 1, 2,..., 8), the two pixels of the pair are judged to have no correlation, and are excluded from the addition target at the time of noise removal pixel data calculation. difference Is equal to or less than a predetermined threshold th (| p−n (x) | ≦ th, x = 1, 2,..., 8), the two pixels in the pair are correlated. Judgment is made as an addition target at the time of noise removal pixel data calculation. Thereafter, the paired pixels to be added are appropriately weighted to calculate an addition average, and the processing target pixel p is replaced with the calculated value. The specific method is the same as that of the digital image noise removing method according to the fourth embodiment of the present invention.

  Even in this noise removal method, since the number of pixels to be added increases or decreases in pairs by two, simplification of the circuit configuration is relatively easy.

  Similar to the noise removal method of the digital image according to the fifth embodiment of the present invention, the present noise removal method determines the correlation between the average of two pixels positioned symmetrically with respect to the pixel to be processed p, rather than each pixel individually. Therefore, even if the pixel has a large contrast as shown in FIG. 3, a noise removal effect can be obtained.

  Similar to the digital image noise removal method according to the fifth embodiment of the present invention, the present noise removal method can also preserve the contour portion of the image even under the conditions shown in FIG. Since this is done, there is no degradation of dynamic resolution.

  Further, even in the present noise removal method, the original processing target pixel p is not used when the noise removal pixel data is calculated. Therefore, even when the number of correlated pixels in the peripheral pixels is reduced, the noise removal effect is not reduced. It is not affected by the processing target pixel p.

  Furthermore, even in this noise removal method, the peripheral pixels added to the averaging when generating the noise removal pixel data always exist symmetrically in the vertical, horizontal, and diagonal directions, so that they are spatially the same as the original pixel to be processed. The pixel data having the center of gravity at the position is replaced. Further, since intra-frame processing is performed, there is no temporal movement of the center of gravity.

  Even in the present noise removal method, the original processing target pixel p is not used when calculating the noise removal pixel data. However, in order to remove higher-frequency noise components, the addition target pixels when calculating the noise removal pixel data are used. In addition, the original processing target pixel p may be added.

  As described above, according to the present embodiment, the sharpness of the contour portion of the original image is preserved, the deterioration of the dynamic resolution due to the afterimage is prevented with respect to the moving image, and the processing target pixel is changed to the slope portion having a large contrast. A sufficient noise removal effect can be obtained even when it exists, and it is also effective when there are few highly correlated pixels in the surrounding pixels, and the spatial and temporal center of gravity position does not shift after noise removal processing. A noise removal method for digital images can be realized.

(Embodiment 7)
FIG. 13 is a diagram for explaining a digital image noise removing method according to the seventh embodiment of the present invention. FIG. 13 shows how the image of a two-dimensional image in the horizontal direction H and the vertical direction V changes with time (k−1, k, k + 1), and p is a pixel to be processed at time k, n (x, y) is a pixel (x = 1; y = 1, 2) that appears at the same time as the processing target pixel p at times k−1 and k + 1 before and after it.

  Hereinafter, a digital image noise removing method according to Embodiment 7 of the present invention will be described with reference to FIG.

  In the digital image noise removing method according to the seventh embodiment of the present invention, first, a pixel that appears before and after the processing target pixel p in the same position is positioned symmetrically with respect to the processing target pixel p. The average value n (1) (n (1,1) + (n (1,1) + n (1,2)) / 2) of the pixels forming a pair of n (1,1) and n (1,2) Is calculated, the absolute value of the difference from the processing target pixel p is obtained, and when the value is larger than a predetermined threshold value th (| p−n (1) |> th), When two pixels are determined to be uncorrelated and excluded from the addition target when calculating the noise-removed pixel data, and the absolute value of the difference is equal to or smaller than a predetermined threshold th (| p−n (1) | ≦ th) The two pixels in the pair are determined to be correlated, and are added at the time of noise removal pixel data calculation. Thereafter, the average of the addition is calculated from the pair of pixels to be added, and the processing target pixel p is replaced by the calculated value. Since the number of pixels to be added is increased or decreased in pairs, it is relatively easy to simplify the circuit configuration. Further, since the correlation is determined not by individual pixels but by the average value of two pixels located symmetrically in time with the processing target pixel p in between, the signal with the processing target pixel p is used as each pixel. Even when the level difference is large, there is a case where the difference from the processing target pixel p may be within the threshold value by taking the average value of the two, and the threshold value can be set small.

  FIG. 15 is an explanatory diagram for a case where the signal level changes greatly with time as in a fast-moving subject. When attention is paid to the position of the processing target pixel p in FIG. 15, the signal level linearly changes as n (1, 1), p, n (1, 2) as the time k-1, k, k + 1 changes. The absolute value of the difference between the pixel n (1,1) and n (1,2) located before and after the target pixel p is larger than the threshold th. However, the average value of n (1,1) and n (1,2) is the level at point A in the figure, and the absolute value of the difference from the processing target pixel p is equal to or less than the threshold th. Therefore, both n (1,1) and n (1,2) are determined to have a correlation, and become pixels to be added when noise removal pixel data is calculated. That is, even a fast-moving subject can obtain a noise removal effect if its signal level change is close to linear.

  FIG. 16 is an explanatory diagram for a case where the signal level changes greatly with time like a fast-moving subject, but the signal level change is irregular. In FIG. 16, paying attention to the position of the processing target pixel p, the signal level irregularly changes to n (1,1), p, n (1,2) as time k-1, k, k + 1 changes. The absolute value of the difference from the pixel n (1,1) located before the processing target pixel p in time is equal to or less than the threshold th and the absolute value of the difference from the pixel n (1,2) located after Becomes larger than the threshold th. However, the average value of n (1,1) and n (1,2) is the level at point A in the figure, and the absolute value of the difference from the processing target pixel p is greater than the threshold th. Therefore, both n (1,1) and n (1,2) are determined to have no correlation, and are excluded from the addition target pixels at the time of noise removal pixel data calculation. That is, if the subject is a fast moving subject and the signal level changes irregularly, the noise removal effect is suppressed, and the deterioration of the dynamic resolution of the original image due to the afterimage is suppressed.

  In the present noise removal method, since the original processing target pixel p is not used when calculating the noise removal pixel data, the noise removal effect is not affected by the original processing target pixel p.

  Furthermore, in this noise removal method, since the pixels added to the addition average when generating the noise removal pixel data always exist symmetrically in time, pixel data having a centroid at the same position in time as the original pixel to be processed Will be replaced. In addition, since no intra-frame processing is performed, the contour portion of the original image is preserved as it is, deterioration of sharpness is suppressed, and there is no spatial movement of the center of gravity.

  In this noise removal method, the original processing target pixel p is not used when calculating the noise removal pixel data, but the original processing target pixel p may be added to the addition target pixel when calculating the noise removal pixel data. .

  As described above, according to the present embodiment, the sharpness of the contour portion of the original image is preserved, the deterioration of the dynamic resolution due to the afterimage is prevented with respect to the moving image, and the temporal signal level of the processing target pixel is reduced. Sufficient noise removal effect can be obtained even when the change is large, and it is also effective when there are few highly correlated pixels in the temporally preceding and following pixels, and also the spatial and temporal center of gravity position after noise removal processing It is possible to realize a noise removal method for digital images that does not shift.

(Embodiment 8)
FIG. 14 is a diagram for explaining a digital image noise removing method according to the eighth embodiment of the present invention. FIG. 14 is a graph in which the number of pixels targeted for noise removal pixel data calculation is increased in the time axis direction with respect to the noise removal method of the seventh embodiment in FIG. 13, and p is a pixel to be processed at time k, n (x, y) is a pixel (x = 1, 2; y = 1, 2) appearing at times k−1 and k + 1, k−2 and k + 2 before and after the processing target pixel p in the same spatial position. is there.

  The digital image noise removing method according to the eighth embodiment of the present invention will be described below with reference to FIG.

  In the noise removal method for a digital image according to the eighth embodiment of the present invention, first, pixels that appear before and after the processing target pixel p in the same position are positioned symmetrically with respect to the processing target pixel p. N (1,1) and n (1,2), and n (2,1) and n (2,2), the paired average value n (1), n (2) (n (X) = (n (x, 1) + n (x, 2)) / 2, x = 1, 2) is calculated, and an absolute value of a difference between each addition average value and the processing target pixel p is obtained. When the value is larger than a predetermined threshold value th set in advance (| p−n (x) |> th, x = 1, 2), it is determined that the two pixels of the pair are not correlated. Therefore, when the absolute value of the difference is equal to or less than a predetermined threshold th (| p−n (x) | ≦ th, x = 1, 2) The pixel of that pair is determined that there correlation two both the addition target when the noise removing pixel data calculated. Thereafter, the paired pixels to be added are appropriately weighted to calculate an addition average, and the processing target pixel p is replaced with the calculated value. Since the number of pixels to be added is increased or decreased in pairs, it is relatively easy to simplify the circuit configuration. Further, since the correlation is determined not by individual pixels but by the average value of two pixels located symmetrically in time with the processing target pixel p in between, the signal with the processing target pixel p is used as each pixel. Even when the level difference is large, there is a case where the difference from the processing target pixel p may be within the threshold value by taking the average value of the two, and the threshold value can be set small.

  Similar to the digital image noise removal method of the seventh embodiment of the present invention, the noise removal method also changes the signal level over time as in a fast-moving subject as shown in FIG. Is irregular, the noise removal effect is suppressed, and the degradation of the dynamic resolution of the original image due to the afterimage is suppressed.

  On the other hand, even in the case where the signal level changes greatly with time like a fast-moving subject, a noise removal effect can be obtained in the case of FIG. When attention is paid to the position of the processing target pixel p in FIG. 17, the signal levels are changed to n (2,1), n (1,1), p, as time k-2, k-1, k, k + 1, k + 2. The pixels n (1,1) and n (1,2), which are irregularly changed to n (1,2) and n (2,2) and are positioned before and after the pixel p to be processed in time, and n ( 2,1) and n (2,2) have an absolute value greater than the threshold th. However, the average value of n (1,1) and n (1,2) is the level at point A in the figure, and the average value of n (2,1) and n (2,2) is the level at point B in the figure. Both absolute values of differences from the processing target pixel p are equal to or less than the threshold th. Therefore, n (1,1) and n (1,2) and n (2,1) and n (2,2) are both determined to be correlated and become pixels to be added when noise-removed pixel data is calculated. . That is, a noise removal effect can be obtained even when the subject moves quickly and moves randomly.

  In addition, even in the present noise removal method, the original pixel to be processed p is not used when calculating the noise removal pixel data, so even if the number of correlated pixels in the temporally preceding and following pixels decreases, noise removal is performed. The effect is not affected by the original processing target pixel p.

  Furthermore, even in this noise removal method, the pixels added to the averaging when generating the noise removal pixel data always exist symmetrically in terms of time, so pixel data having a centroid at the same position in time as the original processing target pixel. Will be replaced. In addition, since no intra-frame processing is performed, the contour portion of the original image is preserved as it is, deterioration of sharpness is suppressed, and there is no spatial movement of the center of gravity.

  In this noise removal method, the original processing target pixel p is not used when calculating the noise removal pixel data, but the original processing target pixel p may be added to the addition target pixel when calculating the noise removal pixel data. .

  As described above, according to the present embodiment, the sharpness of the contour portion of the original image is preserved, the deterioration of the dynamic resolution due to the afterimage is prevented with respect to the moving image, and the temporal signal level of the processing target pixel is reduced. Sufficient noise removal effect can be obtained even when the change is large, and it is also effective when there are few highly correlated pixels in the temporally preceding and following pixels, and also the spatial and temporal center of gravity position after noise removal processing It is possible to realize a noise removal method for digital images that does not shift.

(Embodiment 9)
FIG. 18 is a diagram for explaining a noise removal method for a digital image according to the ninth embodiment of the present invention. FIG. 18 shows a state in which the image of a two-dimensional image in the horizontal direction H and the vertical direction V changes with time (k−1, k, k + 1), and p is a pixel to be processed at time k, n (x, y) is a pixel (x = 1; y = 1, 2) appearing at times k−1 and k + 1 before and after the processing target pixel p in the same spatial position.

  The digital image noise removal method according to the ninth embodiment of the present invention will be described below with reference to FIG.

  In the digital image noise removal method according to the ninth embodiment of the present invention, first, pixels that appear before and after the processing target pixel p in the same position are positioned symmetrically with respect to the processing target pixel p. The absolute value of the difference between the paired pixel of n (1,1) and n (1,2) and the processing target pixel p is obtained, and the absolute value of the difference between the two pixels that are paired is set in advance. If the threshold value is smaller than the predetermined threshold th (| pn (1,1) | ≦ th and | pn (1,2) | ≦ th), the two pixels in the pair are correlated. It is determined that there is an addition target at the time of noise removal pixel data calculation. Otherwise, it is determined that the two pixels in the pair are not correlated, and the addition target at the time of noise removal pixel data calculation is determined. remove. Thereafter, the average of the addition is calculated from the pair of pixels to be added, and the processing target pixel p is replaced by the calculated value. Since the number of pixels to be added is increased or decreased in pairs, it is relatively easy to simplify the circuit configuration.

  Even in the present noise removal method, the signal level greatly changes with time as in a fast-moving subject as shown in FIG. 16, but when the signal level change is irregular, the dynamic resolution of the original image due to the afterimage Deterioration of is suppressed. Hereinafter, a description will be given with reference to FIG. In FIG. 16, paying attention to the position of the processing target pixel p, the signal level irregularly changes to n (1,1), p, n (1,2) as time k-1, k, k + 1 changes. The absolute value of the difference from the pixel n (1,1) located before the processing target pixel p in time is equal to or less than the threshold th and the absolute value of the difference from the pixel n (1,2) located after Becomes larger than the threshold th. However, of n (1,1) and n (1,2) located symmetrically with respect to the processing target pixel p, n (1,2) is the absolute value of the difference from the processing target pixel p. It is not within the threshold th. Therefore, both n (1,1) and n (1,2) are determined to have no correlation, and are excluded from the addition target pixels at the time of noise removal pixel data calculation. That is, if the subject is a fast moving subject and the signal level changes irregularly, the noise removal effect is suppressed, and the deterioration of the dynamic resolution of the original image due to the afterimage is suppressed.

  In the present noise removal method, since the original processing target pixel p is not used when calculating the noise removal pixel data, the noise removal effect is not affected by the original processing target pixel p.

  Furthermore, in this noise removal method, since the pixels added to the addition average when generating the noise removal pixel data always exist symmetrically in time, pixel data having a centroid at the same position in time as the original pixel to be processed Will be replaced. In addition, since no intra-frame processing is performed, the contour portion of the original image is preserved as it is, deterioration of sharpness is suppressed, and there is no spatial movement of the center of gravity.

  In this noise removal method, the original processing target pixel p is not used when calculating the noise removal pixel data, but the original processing target pixel p may be added to the addition target pixel when calculating the noise removal pixel data. .

  As described above, according to the present embodiment, the sharpness of the contour portion of the original image is preserved, the deterioration of the dynamic resolution due to the afterimage is prevented with respect to the moving image, and the spatial and time after the noise removal processing are reduced. It is possible to realize a digital image noise removal method in which the center of gravity position does not shift.

(Embodiment 10)
FIG. 19 is a diagram for explaining a noise removal method for a digital image according to the tenth embodiment of the present invention. FIG. 19 shows how the two-dimensional images in the horizontal direction H and the vertical direction V change with time, and is the same as the digital image noise removal method of the eighth embodiment of the present invention.

  The digital image noise removal method according to the tenth embodiment of the present invention will be described below with reference to FIG.

  In the digital image noise removal method according to the tenth embodiment of the present invention, first, a pixel that appears before and after the processing target pixel p at the same spatial position is temporally sandwiched by the processing target pixel p. The absolute value of the difference between the paired pixels of n (1,1) and n (1,2), n (2,1) and n (2,2), which are symmetrically located, and the pixel to be processed p When the absolute values of the differences for the two pixels that are paired are both smaller than a predetermined threshold th (| p−n (x, 1) | ≦ th and | p−n (x, 2) When | ≦ th, x = 1, 2), the two pixels in the pair are both judged to be correlated, and are added at the time of calculating the noise-removed pixel data. Two pairs of pixels are judged to have no correlation, and are excluded from the addition target when calculating the noise removal pixel data.Thereafter, the paired pixels to be added are appropriately weighted to calculate an addition average, and the processing target pixel p is replaced with the calculated value. Since the number of pixels to be added is increased or decreased in pairs, it is relatively easy to simplify the circuit configuration.

  Similar to the digital image noise removal method of the ninth embodiment of the present invention, the noise removal method also changes the signal level greatly over time, as in a fast-moving subject such as that shown in FIG. Is irregular, the noise removal effect is suppressed, and the degradation of the dynamic resolution of the original image due to the afterimage is suppressed.

  Further, even in the present noise removal method, the original processing target pixel p is not used when the noise removal pixel data is calculated. Therefore, even when the number of correlated pixels in the peripheral pixels is reduced, the noise removal effect is not reduced. It is not affected by the processing target pixel p.

  Furthermore, even in this noise removal method, the pixels added to the averaging when generating the noise removal pixel data always exist symmetrically in terms of time, so pixel data having a centroid at the same position in time as the original processing target pixel. Will be replaced. In addition, since no intra-frame processing is performed, the contour portion of the original image is preserved as it is, deterioration of sharpness is suppressed, and there is no spatial movement of the center of gravity.

  In this noise removal method, the original processing target pixel p is not used when calculating the noise removal pixel data, but the original processing target pixel p may be added to the addition target pixel when calculating the noise removal pixel data. .

  As described above, according to the present embodiment, the sharpness of the contour portion of the original image is preserved, the deterioration of the dynamic resolution due to the afterimage is prevented with respect to the moving image, and the temporally correlated pixels are correlated. It is also effective when there are few high pixels, and it is possible to realize a noise removal method for a digital image in which the position of the center of gravity is not shifted spatially and temporally after the noise removal processing.

(Embodiment 11)
FIG. 20 is a diagram for explaining a noise removal method for a digital image according to the eleventh embodiment of the present invention. FIG. 20 shows a state in which the image of a two-dimensional image in the horizontal direction H and the vertical direction V changes with time (k−2, k−1, k), and p is a processing target at time k. A pixel, n (x, y) is a pixel (x = 1; y = 1, 2) appearing at times k−1 and k-2 one frame and two frames before the processing target pixel p in the same position. It is.

  Hereinafter, a digital image noise removing method according to Embodiment 11 of the present invention will be described with reference to FIG.

  In the noise removal method for a digital image according to the eleventh embodiment of the present invention, first, for a pixel that is a pair of n (1,1) and n (1,2) located temporally continuously from the processing target pixel p, The signal level n (1) (n (1) = 2 * n (1,1) −n (1,2)) at the outer dividing point on the processing target pixel p side is calculated, and the signal level and processing at the outer dividing point are calculated. When the absolute value of the difference from the target pixel p is obtained and the value is larger than a predetermined threshold value th (| pn (1) |> th), the outer dividing point of the pair is correlated. If the absolute value of the difference is equal to or less than a predetermined threshold th (| p−n (1) | ≦ th), it is excluded from the addition target at the time of noise removal pixel data calculation. Is the addition target when calculating the noise-removed pixel data. Thereafter, the processing target pixel p is replaced by the signal level of the outer dividing point that is the addition target.

  In this noise removal method, the number of target pixels for noise removal pixel data calculation is increased or decreased in pairs, so that it is relatively easy to simplify the circuit configuration. In addition, since the correlation is determined not at each pixel but at the outer dividing point on the processing target pixel p side of the two pixels, even when the signal level difference between each pixel and the processing target pixel p is large. Since the signal level at the outer dividing point may be within a threshold value with respect to the processing target pixel p, a relatively large number of peripheral pixels can be added.

  FIG. 21 is an explanatory diagram of a case where the signal level changes greatly with time as in a fast-moving subject. When attention is paid to the position of the processing target pixel p in FIG. 21, the signal level linearly changes as n (1, 2), n (1, 1), p as time k-2, k-1, k changes. The absolute value of the difference between the pixels n (1,1) and n (1,2) located one frame and two frames before the processing target pixel p in time is larger than the threshold th. However, the outer dividing points on the processing target pixel p side of n (1,1) and n (1,2) are at the level of point A in the figure, and the absolute value of the difference from the processing target pixel p is equal to or less than the threshold th. Therefore, both n (1,1) and n (1,2) are determined to have a correlation, and become pixels to be added when noise removal pixel data is calculated. That is, even a fast-moving subject can obtain a noise removal effect if its signal level change is close to linear.

  FIG. 22 is an explanatory diagram for a case where the signal level changes greatly with time like a fast-moving subject, but the signal level change is irregular. In FIG. 22, paying attention to the position of the processing target pixel p, the signal level irregularly changes to n (1, 2), n (1, 1), p as time k-2, k-1, k changes. The absolute value of the difference between the pixel to be processed p and the pixel n (1,2) positioned two frames before in time is equal to or less than the threshold th, and the pixel n (1,1) positioned one frame before ) Is larger than the threshold th. However, the outer dividing points on the processing target pixel p side of n (1,1) and n (1,2) are at the level of point A in the figure, and the absolute value of the difference from the processing target pixel p is larger than the threshold th. Therefore, both n (1,1) and n (1,2) are determined to have no correlation, and are excluded from the addition target pixels at the time of noise removal pixel data calculation. That is, if the subject is a fast moving subject and the signal level changes irregularly, the noise removal effect is suppressed, and the deterioration of the dynamic resolution of the original image due to the afterimage is suppressed.

  In the present noise removal method, since the original processing target pixel p is not used when calculating the noise removal pixel data, the noise removal effect is not affected by the original processing target pixel p.

  In addition, since no intra-frame processing is performed, the contour portion of the original image is preserved as it is, deterioration of sharpness is suppressed, and there is no spatial movement of the center of gravity.

  In this noise removal method, the original processing target pixel p is not used when calculating the noise removal pixel data, but the original processing target pixel p may be added to the addition target pixel when calculating the noise removal pixel data. .

  As described above, according to the present embodiment, the sharpness of the contour portion of the original image is preserved, the deterioration of the dynamic resolution due to the afterimage is prevented with respect to the moving image, and the temporal signal level of the processing target pixel is reduced. Sufficient noise removal effect can be obtained even when the change is large, and it is also effective when there are few highly correlated pixels in the temporally previous pixel, and spatial and temporal center of gravity position after noise removal processing It is possible to realize a noise removal method for digital images that does not shift.

  The noise removal method for a digital image according to the present invention preserves the sharpness of the contour portion of the original image, prevents deterioration of the dynamic resolution due to the afterimage of the moving image, and makes the processing target pixel a slope portion having a large contrast. A sufficient noise removal effect can be obtained even when it exists, and it is also effective when there are few highly correlated pixels in the surrounding pixels, and the spatial and temporal center of gravity position does not shift after noise removal processing. It is useful for removing noise components in equipment that handles moving images and still images that include a wide range of frequency components from low to high frequencies.

The figure explaining the concept of the noise removal method of the digital image in Embodiment 5 of this invention. The figure explaining the concept of the noise removal method of the digital image in Embodiment 6 of this invention. The figure explaining operation | movement of the noise removal method of the digital image in Embodiment 5 and 6 of this invention. The figure explaining operation | movement of the noise removal method of the digital image in Embodiment 5 and 6 of this invention. The figure explaining the concept of the noise removal method of the digital image in Embodiment 3 of this invention. The figure explaining the concept of the noise removal method of the digital image in Embodiment 4 of this invention. The figure explaining operation | movement of the noise removal method of the digital image in Embodiment 3 and 4 of this invention. The figure explaining the concept of the noise removal method of the digital image in Embodiment 1 of this invention. The figure explaining operation | movement of the noise removal method of the digital image in Embodiment 1 and 2 of this invention. The figure explaining the concept of the noise removal method of the digital image in Embodiment 2 of this invention. The figure explaining operation | movement of the noise removal method of the digital image in Embodiment 2 of this invention. The figure explaining operation | movement of the noise removal method of the digital image in Embodiment 2 of this invention. The figure explaining the concept of the noise removal method of the digital image in Embodiment 7 of this invention. The figure explaining the concept of the noise removal method of the digital image in Embodiment 8 of this invention. The figure explaining operation | movement of the noise removal method of the digital image in Embodiment 7 of this invention. The figure explaining operation | movement of the noise removal method of the digital image in Embodiment 7, 8, 9, and 10 of this invention. The figure explaining operation | movement of the noise removal method of the digital image in Embodiment 8 of this invention. The figure explaining the concept of the noise removal method of the digital image in Embodiment 9 of this invention. The figure explaining the concept of the noise removal method of the digital image in Embodiment 10 of this invention. The figure explaining the concept of the noise removal method of the digital image in Embodiment 11 of this invention. The figure explaining operation | movement of the noise removal method of the digital image in Embodiment 11 of this invention. The figure explaining operation | movement of the noise removal method of the digital image in Embodiment 11 of this invention. The figure explaining the concept of the noise removal method of the conventional digital image The figure explaining operation | movement of the noise removal method of the conventional digital image. The figure explaining operation | movement of the noise removal method of the conventional digital image. The figure explaining operation | movement of the noise removal method of the conventional digital image. The figure explaining operation | movement of the noise removal method of the conventional digital image.

Claims (11)

The above processing is performed using a total of 16 peripheral pixels, 8 groups each including two pixels that are spatially continuously located in one direction from the processing target pixel in each of the vertical, horizontal, and diagonal directions of the processing target pixel. A digital image noise removing method for removing a noise component of a target pixel, wherein an absolute value of a difference between each of two peripheral pixels continuously located in one direction from the processing target pixel and the processing target pixel is Only when both are equal to or less than a predetermined threshold value, both of the two peripheral pixels are to be added, and 2M pixels (M is a positive number of 8 or less) to be added from the 16 peripheral pixels. A method for removing noise from a digital image, wherein the pixel to be processed is replaced with an average value obtained by adding the values under a predetermined weight. The above processing is performed using a total of 16 peripheral pixels, 8 groups each including two pixels that are spatially continuously located in one direction from the processing target pixel in each of the vertical, horizontal, and diagonal directions of the processing target pixel. A digital image noise removal method for removing a noise component of a target pixel, wherein X (X is a pixel adjacent to a processing target pixel among two peripheral pixels continuously located in one direction from the processing target pixel Signal level) and a pixel adjacent to the pixel X is Y (Y is a signal level), and the signal level 2X-Y at the outer dividing point on the processing target pixel side of the pixels X and Y and the processing target Only when the absolute value of the difference from the pixel is equal to or less than a predetermined threshold, the signal level of the outer dividing point is to be added, and the signal level of each outer dividing point calculated by the eight sets of surrounding pixels is M groups to be added (M is 8 or more Method for removing noise of a digital image, characterized in that to replace the target pixel by an average value of the after adding under a signal level of the outer equinox predetermined weighting of the positive number). The pixel to be processed using a total of eight peripheral pixels, with four groups each including two pixels located symmetrically across the pixel to be processed in each of the vertical, horizontal, and diagonal directions of the pixel to be processed. A noise removal method for a digital image that removes the noise component of the digital image, wherein the absolute value of the difference between each of two peripheral pixels positioned symmetrically across the processing target pixel and the processing target pixel is a predetermined threshold value. Only when the following is true, both of the two peripheral pixels are subject to addition, and 2M pixels (M is a positive number less than or equal to 4) from among the eight peripheral pixels are given a predetermined weight. A method for removing noise from a digital image, wherein the pixel to be processed is replaced with an average value after addition under. A total of 2N (N is a positive number) peripheral pixels, where N sets are two sets that are spatially symmetrical across the processing target pixel in each of the upper, lower, left, and right directions of the processing target pixel. A digital image noise removing method that uses a noise component of the processing target pixel to remove the absolute difference between each of the two neighboring pixels located symmetrically across the processing target pixel and the processing target pixel. Only when both values are equal to or less than a predetermined threshold, both of the two peripheral pixels are to be added, and 2M of the 2N peripheral pixels to be added (M is a positive number less than N). A method for removing noise from a digital image, wherein the pixel to be processed is replaced with an average value after the pixels are added under a predetermined weight. The pixel to be processed using a total of eight peripheral pixels, with four groups each including two pixels located symmetrically across the pixel to be processed in each of the vertical, horizontal, and diagonal directions of the pixel to be processed. A noise removal method for a digital image that removes the noise component of the image, wherein the absolute value of the difference between the addition average value of two neighboring pixels that are spatially symmetrical across the processing target pixel and the processing target pixel 2M pixels (M is a positive number of 4 or less) to be added out of the 8 peripheral pixels only when both are equal to or less than a predetermined threshold A method for removing noise from a digital image, wherein the pixel to be processed is replaced with an average value obtained by adding the values under a predetermined weight. A total of 2N (N is a positive number) peripheral pixels, where N sets are two sets that are spatially symmetrical across the processing target pixel in each of the upper, lower, left, and right directions of the processing target pixel. A digital image noise removing method that uses a noise component of the processing target pixel to remove the sum of an average value of two peripheral pixels that are spatially symmetrical across the processing target pixel and the processing target pixel Only when the absolute value of the difference between the two and the neighboring pixels is equal to or less than a predetermined threshold, both of the two neighboring pixels are to be added, and 2M of the 2N neighboring pixels to be added (M is N or less) A method of removing noise from a digital image, wherein the pixel to be processed is replaced with an average value after addition of a positive number of pixels under a predetermined weight. A noise component of the processing target pixel using a pair of two pixels that are spatially the same position and temporally symmetrical with the processing target pixel in a frame including the processing target pixel. The digital image noise removing method is for removing the 2 only when the absolute value of the difference between the addition average value of two pixels of the one set of pixels and the processing target pixel is equal to or less than a predetermined threshold. A method for removing noise from a digital image, characterized in that both of the pixels are to be added, and the pixel to be processed is replaced by an average value after adding the two peripheral pixels. The pixel to be processed in the vicinity of N frames from a set of two pixels that are spatially the same position and symmetrical in time with the pixel to be processed in one frame before and after the frame including the pixel to be processed And digital image noise removing method for removing noise components of the pixel to be processed by using a total of 2N pixels, up to N sets of two pixels located in the same spatial position and symmetrically in time And both of the two pixels are determined only when the absolute value of the difference between the addition average value of the two pixels of each set of the N sets of pixels and the processing target pixel is equal to or less than a predetermined threshold value. The pixel to be processed is determined by an average value obtained by adding 2M pixels (M is a positive number equal to or less than N) from the 2N peripheral pixels to be added under a predetermined weight. A digital image characterized by 'S removal method. A noise component of the processing target pixel using a pair of two pixels that are spatially the same position and temporally symmetrical with the processing target pixel in a frame including the processing target pixel. A method for removing noise from a digital image, wherein the two pixels only when the absolute value of the difference between each of the two pixels of the set of pixels and the pixel to be processed is equal to or less than a predetermined threshold value. A method for removing noise from a digital image, wherein both of the pixels are added and the processing target pixel is replaced with an average value after adding the two peripheral pixels. The pixel to be processed in the vicinity of N frames from a set of two pixels that are spatially the same position and symmetrical in time with the pixel to be processed in one frame before and after the frame including the pixel to be processed And digital image noise removing method for removing noise components of the pixel to be processed by using a total of 2N pixels, up to N sets of two pixels located in the same spatial position and symmetrically in time The two pixels are added only when the absolute value of the difference between each of the two pixels of each of the N groups of pixels and the pixel to be processed is equal to or less than a predetermined threshold value. The processing target pixel is determined by an average value obtained by adding 2M pixels (M is a positive number equal to or less than N) from the 2N peripheral pixels as a target, and adding them under a predetermined weight. Digital image noise characterized by replacement Removal method. The processing is performed using a set of two pixels that are located in the same spatial position and temporally consecutively as the processing target pixel one frame before and two frames before the frame including the processing target pixel. A digital image noise removing method for removing a noise component of a target pixel, wherein X is a pixel located one frame before the processing target pixel among the two pixels (X represents a signal level), and the processing A pixel located two frames before the target pixel is defined as Y (Y represents a signal level), and a signal level 2X-Y at an outer dividing point to the frame including the processing target pixel calculated from the pixels X and Y The signal level at the outer dividing point is added only when the absolute value of the difference from the processing target pixel is equal to or less than a predetermined threshold, and the processing target pixel is replaced by the signal level at the outer dividing point to be added. Method for removing noise of a digital image, characterized in that.

Images