CN112184583A - Image noise reduction method and device - Google Patents
Image noise reduction method and device Download PDFInfo
- Publication number
- CN112184583A CN112184583A CN202011043543.3A CN202011043543A CN112184583A CN 112184583 A CN112184583 A CN 112184583A CN 202011043543 A CN202011043543 A CN 202011043543A CN 112184583 A CN112184583 A CN 112184583A
- Authority
- CN
- China
- Prior art keywords
- noise reduction
- value
- coefficient
- pixel
- pixel point
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000009499 grossing Methods 0.000 claims abstract description 74
- 238000003708 edge detection Methods 0.000 claims abstract description 12
- 239000011159 matrix material Substances 0.000 claims description 90
- 230000004927 fusion Effects 0.000 claims description 30
- 238000002156 mixing Methods 0.000 claims description 16
- 238000004364 calculation method Methods 0.000 claims description 14
- 238000003384 imaging method Methods 0.000 abstract description 27
- 230000000694 effects Effects 0.000 abstract description 13
- 230000000875 corresponding effect Effects 0.000 description 13
- 230000001276 controlling effect Effects 0.000 description 7
- 230000002596 correlated effect Effects 0.000 description 3
- 238000011946 reduction process Methods 0.000 description 2
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Processing (AREA)
- Facsimile Image Signal Circuits (AREA)
Abstract
The invention provides an image noise reduction method and device, wherein the method comprises the following steps: acquiring image data, and determining pixel points to be denoised in the image data; then determining the noise coefficient of the pixel point to be denoised; determining a smoothing coefficient based on the noise coefficient, wherein the smoothing coefficient is used for controlling the noise reduction strength; determining a preset noise reduction value of the pixel point to be subjected to noise reduction based on the smoothing coefficient, the pixel value of the pixel point to be subjected to noise reduction and the pixel values of the pixel points around the pixel point to be subjected to noise reduction; then, calculating an edge coefficient of a pixel point to be subjected to noise reduction by using an edge detection operator algorithm, and determining a detail addition value based on the edge coefficient; and finally, determining the sum of the preset noise reduction value and the detail adding value as a final noise reduction value of the pixel point to be subjected to noise reduction and outputting the final noise reduction value. When the method provided by the invention is adopted to reduce the noise of the image, the image after noise reduction can be ensured to be smooth and clear, the loss of details can be prevented, the detail expression is improved, and the image after noise reduction has high imaging quality and good display effect.
Description
Technical Field
The present invention relates to the field of image processing, and in particular, to an image denoising method and apparatus.
Background
In the field of image processing, noise is easily generated on an image due to influences of an imaging device or external environment noise interference, and the like, so that the imaging quality of the image is affected, and therefore, it is generally required to reduce noise of the image to improve the imaging quality of the image.
In the related art, in view of the structure of a non-local mean (NLM) noise reduction algorithm that can better express an image, an NLM noise reduction algorithm is generally used to reduce noise of an image, and the NLM noise reduction algorithm mainly includes:
wherein, f (i) is the pixel value of the pixel point to be denoised after denoising; j is used for indicating a reference pixel point, wherein the reference pixel point is any pixel point in a first m multiplied by n matrix taking a pixel point to be subjected to noise reduction as a central pixel point; f (j) is the pixel value of the reference pixel point;is the standard deviation of the gaussian kernel function; n (I) and I are both used to indicate the first m x n matrix; n (j) a second mxn matrix indicating a pixel centered on the reference pixel; d (i, j) is used for indicating Euclidean distances of the first m multiplied by n matrix and the second m multiplied by n matrix; w (i, j) is used for indicating the fusion weight of the first m × n matrix and the second m × n matrix and representing the similarity degree between the first m × n matrix and the second m × n matrix; h is a smoothing parameter for controlling the noise reduction strength.
In the related art, when the noise intensity of a certain frame of image is large, a large smoothing coefficient is set to improve the noise reduction degree; when the noise intensity of a certain frame image is small, a small smoothing coefficient is set to reduce the noise reduction intensity, and in the process of reducing the noise of a single frame image, the noise reduction method in the related art does not have the capability of adjusting the size of the h value. Therefore, when the method in the related art is adopted to perform noise reduction on the single-frame image with large noise, the noise reduction intensity of the edge area and the noise reduction intensity of the flat area of the image are consistent and are large. At this time, the situation that the edge area of the image is blurred due to the fact that the detail information of the edge area of the image is lost due to excessive noise reduction of some edge areas of the image occurs, and the imaging quality of the image is reduced, so that the imaging effect of the image is affected.
Disclosure of Invention
The invention aims to provide an image noise reduction method and an image noise reduction device, which aim to solve the technical problems that the image noise reduction method in the related technology is easy to cause low imaging quality and poor display effect of the noise-reduced image.
In order to solve the above technical problem, the present invention provides an image denoising method, including:
acquiring image data, and determining pixel points to be denoised in the image data;
determining a noise coefficient corresponding to the pixel point to be subjected to noise reduction based on the pixel value of the pixel point to be subjected to noise reduction and the pixel values of the pixel points around the pixel point to be subjected to noise reduction;
determining an image area of the pixel point to be subjected to noise reduction in the image data based on the noise coefficient, and determining a smooth coefficient based on the image area, wherein the smooth coefficient is used for controlling noise reduction intensity; then, based on the smoothing coefficient, the pixel value of the pixel point to be subjected to noise reduction and the pixel values of the pixel points around the pixel point to be subjected to noise reduction, the pixel point to be subjected to noise reduction is subjected to noise reduction to obtain a preset noise reduction value;
calculating an edge coefficient of the pixel point to be denoised by using an edge detection operator algorithm, and determining an edge fusion proportion based on the edge coefficient, wherein the edge fusion proportion is in positive correlation with the edge coefficient; determining a detail adding value based on the edge fusion proportion, a preset noise reduction value and a pixel value of a pixel point to be subjected to noise reduction, wherein the detail adding value is (the pixel value of the pixel point to be subjected to noise reduction-the preset noise reduction value) multiplied by the edge fusion proportion, and the detail adding value is used for representing the intensity of adding details to the pixel point subjected to noise reduction;
and determining the sum of the preset noise reduction value and the detail adding value as a final noise reduction value of the pixel point to be subjected to noise reduction and outputting the final noise reduction value.
Optionally, the method for determining the noise coefficient corresponding to the pixel to be noise-reduced based on the pixel value of the pixel to be noise-reduced and the pixel values of the pixels around the pixel to be noise-reduced includes:
providing a formula I, and calculating a noise coefficient corresponding to the pixel point to be denoised based on the formula I;
the formula I is as follows:
wherein j is used for indicating a reference pixel point, the reference pixel point is any pixel point in a first m × n matrix which takes the pixel point to be denoised as a central pixel point in the image data, and m and n are both greater than 1; f (j) pixel values indicating reference pixels in the first m × n matrix; i is used to indicate the first mxn matrix; num is the number of reference pixels in the first mxn matrix; a is used for indicating the adjustment amplitude of the noise coefficient, and a is more than or equal to 0.7 and less than or equal to 1.4.
Optionally, the method for determining the image area of the pixel point to be denoised in the image data based on the noise coefficient and determining the smoothing coefficient based on the image area includes:
presetting a first reference value and a second reference value, wherein the first reference value is greater than or equal to the second reference value; and the first reference value and the second reference value satisfy a preset condition, wherein the preset condition comprises: when the smoothing coefficient is equal to the product of the first reference value and a preset value, the noise reduction intensity for reducing the noise of the pixel point to be reduced based on the smoothing coefficient is a first critical intensity; when the smoothing coefficient is equal to the product of the second reference value and a preset value, the noise reduction intensity for reducing the noise of the pixel point to be reduced based on the smoothing coefficient is a second critical intensity; wherein the first critical intensity and the second critical intensity satisfy: blurring the noise-reduced image data when the noise reduction intensity for the image data exceeds the first critical intensity or the noise reduction intensity for the image data is lower than the second critical intensity;
judging the magnitude relation between the noise coefficient and the first reference value and the second reference value, and correspondingly determining the value of the smooth coefficient based on the magnitude relation; when the noise coefficient is larger than the first reference value, determining that the pixel point to be subjected to noise reduction is located in the edge area of the image data, and enabling the smoothing coefficient to be equal to the product of the first reference value and the preset value; when the noise coefficient is smaller than the second reference value, determining that the pixel point to be subjected to noise reduction is located in a flat area of the image data, and enabling the smooth coefficient to be equal to the product of the second reference value and a preset value; and when the noise coefficient is greater than or equal to the second reference value and less than or equal to the first reference value, enabling the smoothing coefficient to be equal to the product of the noise coefficient and the preset value.
Optionally, the first reference value is less than or equal to 0.14 and greater than 0.04; the second reference value is greater than or equal to 0.04 and less than or equal to the first reference value; the preset value is 3.3.
Optionally, the method for denoising the pixel to be denoised based on the smoothing coefficient, the pixel value of the pixel to be denoised, and the pixel values of the pixels around the pixel to be denoised to obtain the predetermined denoising value includes:
providing a second formula, and calculating the preset noise reduction value based on the second formula;
the formula II is as follows:
wherein f (i) is the predetermined noise reduction value; j is used for indicating a reference pixel point, the reference pixel point is any pixel point in a first m multiplied by n matrix which takes the pixel point to be denoised as a central pixel point in the image data, and m and n are both larger than 1; f (j) a pixel value indicating the reference pixel point;is the standard deviation of the gaussian kernel function; n (I) and I are both used to indicate the first m x n matrix; n (j) a second mxn matrix indicating a pixel in the image data centered on the reference pixel; d (i, j) is used for indicating Euclidean distances of the first m multiplied by n matrix and the second m multiplied by n matrix; w (i, j) is used for indicating the fusion weight of the first m × n matrix and the second m × n matrix and representing the similarity degree between the first m × n matrix and the second m × n matrix; h is used to indicate the smoothing parameter.
Optionally, the edge detection operator algorithm includes a sobel operator algorithm;
and the method for calculating the edge coefficient corresponding to the pixel point to be denoised by utilizing the edge detection operator algorithm comprises the following steps:
providing first operator templates G respectivelyxAnd a second operator template Gy;Gx、GyAre all p × q matrixes, and both p and q are odd numbers;
selecting a qxx matrix A from the image data by taking the pixel point to be denoised as a central pixel point, wherein x is an odd number; and based on the first operator template GxThe second operator template GyThe matrix a calculates an edge strength, which is abs (G)x*A)+abs(Gy*A);
Determining a predetermined pixel point in the qxx matrix A, wherein the predetermined pixel point comprises: pixel points which are not multiplied by 0 in the qxx matrix A when the edge intensity is calculated; then, determining the average value of the pixel values of the preset pixel points;
and dividing the edge intensity by the average value of the pixel values of the preset pixel points to calculate the edge coefficient.
Optionally, the G isx、GyAre all 7 × 7 matrixes;
optionally, the matrix a is a 7 × 7 matrix.
Optionally, the method for determining the edge blending ratio based on the edge coefficient includes:
presetting a first reference edge coefficient and a second reference edge coefficient, wherein the first reference edge coefficient is larger than the second reference edge coefficient;
the edge blending ratio is calculated based on the edge coefficient, the first reference edge coefficient, and the second reference edge coefficient, and the edge blending ratio is (edge coefficient-second reference edge coefficient)/(first reference edge coefficient-second reference edge coefficient).
In addition, the present invention also provides an image noise reduction apparatus, comprising:
the acquisition module is used for acquiring image data and determining pixel points to be denoised in the image data;
the first calculation module is used for determining a noise coefficient corresponding to the pixel point to be subjected to noise reduction based on the pixel value of the pixel point to be subjected to noise reduction and the pixel values of the pixel points around the pixel point to be subjected to noise reduction;
the second calculation module is used for determining an image area of the pixel point to be subjected to noise reduction in the image data based on the noise coefficient, and determining a smooth coefficient based on the image area, wherein the smooth coefficient is used for controlling the noise reduction intensity; then, based on the smoothing coefficient, the pixel value of the pixel point to be subjected to noise reduction and the pixel values of the pixel points around the pixel point to be subjected to noise reduction, the pixel point to be subjected to noise reduction is subjected to noise reduction to obtain a preset noise reduction value;
the third calculation module is used for calculating an edge coefficient of the pixel point to be denoised by utilizing an edge detection operator algorithm, and determining an edge fusion proportion based on the edge coefficient, wherein the edge fusion proportion and the edge coefficient are in positive correlation; determining a detail adding value based on the edge fusion proportion, a preset noise reduction value and a pixel value of a pixel point to be subjected to noise reduction, wherein the detail adding value is (the pixel value of the pixel point to be subjected to noise reduction-the preset noise reduction value) multiplied by the edge fusion proportion, and the detail adding value is used for representing the intensity of adding details to the pixel point subjected to noise reduction;
and the fourth calculation module is used for determining the sum of the preset noise reduction value and the detail adding value as a final noise reduction value of the pixel point to be subjected to noise reduction and outputting the final noise reduction value.
In summary, in the image noise reduction method and apparatus provided by the present invention, different smoothing coefficients are determined based on the difference of the pixel point to be noise reduced in the image region of the image data, and then the predetermined noise reduction value of the pixel point to be noise reduced is further calculated based on the smoothing coefficients. In the invention, the smoothing coefficient is mainly used for controlling the noise reduction degree, so that the noise reduction of different degrees can be realized for the pixel points in different image areas, thereby avoiding the occurrence of the conditions of low imaging quality and poor imaging effect of the image due to consistent noise reduction strength of each area of the image, and ensuring the imaging quality and the imaging effect of the image.
And determining an edge coefficient of a pixel point to be denoised, further determining a detail adding value based on the edge coefficient, and then determining the sum of a preset denoising value and the detail adding value as a final denoising value so as to add detail information to the denoised pixel point. Wherein the edge coefficient is positively correlated with the detail addition value. Therefore, when the edge coefficient is larger (that is, the detail information of the pixel point to be subjected to noise reduction is more), the detail adding value is larger, the detail information added to the pixel point to be subjected to noise reduction is more, so that the loss of the detail information in the noise reduction process can be prevented, the edge detail information of finally obtained image data is ensured, and the imaging quality and the imaging effect of the image are further ensured.
Drawings
Fig. 1 is a schematic flowchart of an image denoising method according to an embodiment of the present invention.
Detailed Description
The following describes an image denoising method and apparatus according to the present invention in further detail with reference to the accompanying drawings and specific embodiments. The advantages and features of the present invention will become more apparent from the following description. It is to be noted that the drawings are in a very simplified form and are not to precise scale, which is merely for the purpose of facilitating and distinctly claiming the embodiments of the present invention.
Fig. 1 is a schematic flowchart of an image denoising method according to an embodiment of the present invention, and as shown in fig. 1, the denoising method may include:
The image data may be image data in any format, and may be, for example, an RGB Bayer format, an RGBW format, an rgbiir format, an RCCB format, or the like.
The method for determining the noise coefficient of the pixel point to be denoised may include:
providing a formula I, and calculating a noise coefficient corresponding to the pixel point to be denoised based on the formula I.
The formula I is as follows:
in the formula one, j is used for indicating a reference pixel point, the reference pixel point is any one pixel point in a first m × n matrix which takes the pixel point to be denoised as a central pixel point in the image data, and m and n are positive integers and are both greater than 1; f (j) a pixel value indicating a reference pixel point; i is used to indicate the first mxn matrix; num is the number of reference pixels in the first mxn matrix; a is the adjustment range of the noise coefficient, a is a preset value, and a is more than or equal to 0.7 and less than or equal to 1.4.
In this step, the method for determining the image area of the pixel point to be noise-reduced in the image data based on the noise coefficient and determining the smoothing coefficient based on the image area may specifically include:
step one, presetting a first reference value sigmamamax and a second reference value sigmamamin.
Wherein the first reference value sigmaMax is greater than or equal to the second reference value sigmamamin. And the first reference value sigmaMax and the second reference value sigmamamin satisfy a preset condition, the preset condition including: when the smoothing coefficient is equal to the product of the first reference value sigmaMax and a preset value, the noise reduction intensity for reducing the noise of the pixel point to be reduced based on the smoothing coefficient is a first critical intensity; and when the smoothing coefficient is equal to the product of the second reference value sigmaMin and a preset value, the noise reduction intensity for reducing the noise of the pixel point to be reduced based on the smoothing coefficient is a second critical intensity. Wherein the first critical intensity and the second critical intensity satisfy: when the noise reduction intensity for the image data exceeds the first critical intensity or the noise reduction intensity for the image data is lower than the second critical intensity, the noise-reduced image data is blurred.
In this embodiment, the first reference value may be less than or equal to 0.14 and greater than 0.04. The second reference value may be equal to or greater than 0.04 and equal to or less than the first reference value.
And step two, judging the magnitude relation between the noise coefficient and the first reference value sigmamamax and the second reference value sigmamamin, and correspondingly determining the value of the smoothing coefficient based on the magnitude relation.
When the noise coefficient is greater than the first reference value sigmamamax, determining that the noise coefficient of a local area including a pixel to be noise-reduced in the image data is larger, and determining that the pixel to be noise-reduced is located in an edge area of the image data. At this time, the noise reduction strength of the pixel to be noise-reduced should be stronger, so that the smooth coefficient is equal to the product of the first reference value sigmamamax and the preset value, when the subsequent noise reduction is performed on the pixel to be noise-reduced based on the smooth coefficient, the noise reduction strength of the pixel to be noise-reduced can be ensured to be the first critical strength, the noise reduction degree of the pixel to be noise-reduced can be ensured to be stronger, and the image data after noise reduction can be ensured not to be fuzzy.
When the noise coefficient is smaller than the second reference value sigmaMin, determining that the noise coefficient of a local region including a pixel to be noise-reduced in the image data is smaller, and determining that the pixel to be noise-reduced is located in a flat region of the image data. At this time, if the noise reduction strength of the pixel to be noise-reduced should be weak, the smoothing coefficient may be made equal to the product of the second reference value sigmaMin and a preset value, so that when noise reduction is subsequently performed on the pixel to be noise-reduced based on the smoothing coefficient, it may be ensured that the noise reduction strength of the pixel to be noise-reduced is the second critical strength, it may be ensured that the noise reduction degree of the pixel to be noise-reduced is weak, and it may be ensured that image data after noise reduction is not blurred.
And when the noise coefficient is greater than or equal to a second reference value sigmaMin and less than or equal to a first reference value sigmaMax, the pixel point to be subjected to noise reduction is not located in the edge area and is not located in the flat area (that is, the pixel point to be subjected to noise reduction is located in the conventional area of the image data). At this time, the smoothing coefficient may be equal to a product of the noise coefficient and a preset value, and when the noise reduction is performed on the pixel to be noise-reduced based on the smoothing coefficient, it may be ensured that the noise reduction intensity of the pixel to be noise-reduced is between the second critical intensity and the first critical intensity, so that the image data after noise reduction may be prevented from being blurred.
In this embodiment, the preset value may be preset, for example, the preset value may be 3.3.
And expressing the corresponding relation between the noise coefficient and the size relation between the noise coefficient and the first reference value sigmamamax and the second reference value sigmaMin by adopting a formula III.
The formula III is as follows:
where CoefNoise is used to indicate the noise coefficient and h is used to indicate the smoothing coefficient.
It can be seen from the above that, in the embodiment of the present invention, when the pixel points to be denoised are located in different image regions of the image data, different smoothing coefficients are determined, wherein the noise reduction strength is controlled based on the smoothing coefficients, so that the pixel points in different image regions can be denoised to different degrees, thereby avoiding the occurrence of "low imaging quality and poor imaging effect of the image due to consistent noise reduction strength of each region of the image", and ensuring the imaging quality and the imaging effect of the image.
Further, in this step 300, after the smoothing coefficient is determined, noise reduction is performed on the to-be-reduced pixel point based on the smoothing coefficient, the pixel value of the to-be-reduced pixel point, and the pixel values of the pixels around the to-be-reduced pixel point to obtain a predetermined noise reduction value, where the specific method includes:
providing a second formula, and calculating the preset noise reduction value by adopting the second formula;
the formula II is as follows:
wherein f (i) is the predetermined noise reduction value; j is used for indicating a reference pixel point, wherein the reference pixel point is any pixel point in a first m multiplied by n matrix taking a pixel point to be subjected to noise reduction as a central pixel point; f (j) a pixel value indicating the reference pixel point;is the standard deviation of the gaussian kernel function; n (I) and I are both used to indicate the first m x n matrix; n (j) a second mxn matrix indicating a pixel centered on the reference pixel; d (i, j) is used for indicating Euclidean distances of the first m multiplied by n matrix and the second m multiplied by n matrix; w (i, j) is used for indicating the fusion weight of the first m × n matrix and the second m × n matrix and representing the similarity degree between the first m × n matrix and the second m × n matrix; h is used to represent the smoothing parameter.
And, referring to formula two, the magnitude of the smoothing coefficient (i.e. h) can be used to control the noise reduction strength. Specifically, when the value h is large, the value of w (i, j) in the above formula two is made large, so that when the predetermined noise reduction value f (i) is calculated, the specific gravity of the second mxn matrix is large, and the calculated predetermined noise reduction value is made to be close to the pixel values of the surrounding pixels of the pixel to be noise reduced, so that the noise reduction degree is large. And when the value h is smaller, the value w (i, j) in the formula two is made smaller, and when the predetermined noise reduction value f (i) is calculated, the proportion of the second mxn matrix is made smaller, so that the difference between the calculated predetermined noise reduction value and the pixel values of the surrounding pixels of the pixel to be noise reduced is made larger, and the noise reduction degree is made smaller. Based on this it can be determined: the smoothing coefficient h is positively correlated with the noise reduction strength.
Further, referring to the third formula again, when the smoothing coefficient h is determined based on the noise coefficient, an upper limit and a lower limit exist in the smoothing coefficient h, where the upper limit is a product of the first reference value sigmaMax and the preset value, and the lower limit is a product of the first reference value sigmamamin and the preset value. When the smoothing coefficient h is equal to the upper limit value, the noise reduction intensity for reducing noise of the image data by using the smoothing coefficient h is a first critical intensity, and when the smoothing coefficient h is equal to the lower limit value, the noise reduction intensity for reducing noise of the image data by using the smoothing coefficient h is a second critical intensity. In view of the positive correlation between the smoothing coefficient h and the noise reduction strength, it can be considered that: when the noise reduction method of the embodiment is used to reduce noise of image data, the maximum value of the noise reduction intensity is the first critical intensity, and the minimum value of the noise reduction intensity is the second critical intensity, in other words, the noise reduction intensity of the image data is always within a reasonable range (that is, greater than or equal to the second critical intensity and less than or equal to the first critical intensity), so that the noise-reduced image data is clear and smooth without blurring.
Further, in this embodiment, the first reference value sigmaMax may be used to control the noise reduction intensity and the definition of the edge area of the image data, and the second reference value sigmamamin may be used to control the noise reduction intensity and the definition of the flat area of the image data. Specifically, when the first reference value sigmamamax is increased, the smoothing coefficient h corresponding to the pixel point in the edge region of the image data is increased, so that the noise reduction intensity (i.e., the first critical intensity) in the edge region of the image data is increased, but the noise reduction intensity does not blur the noise-reduced image data, and thus the noise-reduced image data is ensured to be smooth and clear. And when the second reference value sigmaMin is increased, the smoothing coefficient h corresponding to the pixel point of the flat region of the image data is increased, so that the noise reduction strength of the flat region of the image data is increased, and the flat region of the image data subjected to noise reduction is smoother and clearer.
It should be noted that the value of a in the above formula one may affect the high frequency region of the noise-reduced image data. Specifically, when the value of a is increased, the calculated noise coefficient CoefNoise of the pixel points is increased, so that the number of the pixel points with the noise coefficients larger than the first reference value in the image data is increased, and when the image is subjected to noise reduction, the number of the pixel points with the noise reduction strength being the first critical strength in the image data is increased, that is, the number of the pixel points with the noise reduction strength being larger in the image data is increased, so that the high-frequency region of the image data subjected to noise reduction is increased. Similarly, when the value of a is reduced, the calculated noise coefficient CoefNoise of the pixel points is reduced, so that the number of the pixel points with the noise coefficients smaller than the second reference value in the image data is increased, and when the noise reduction is performed on the image, the number of the pixel points with the noise reduction intensity being the second critical intensity in the image data is increased, that is, the number of the pixel points with the noise reduction intensity being smaller in the image data is increased, so that the high-frequency region of the image data after the noise reduction is decreased.
Therefore, in the specific implementation process, the first reference value, the second reference value and the value a can be made as large as possible, so that the image data after noise reduction can be ensured to be smoother and clearer. It should be noted, however, that the first reference value should not be greater than its maximum value (e.g., should not be greater than 0.14).
The edge detection operator algorithm may include a Canny operator algorithm or a sobel operator algorithm, and the sobel operator algorithm is mainly used as an example in the present embodiment for description.
And the method for calculating the edge coefficient corresponding to the pixel point to be denoised by using the sobel operator algorithm comprises the following steps:
first, respectively providing a first operator template GxAnd a second operator template Gy;Gx、GyAre all p × q matrices, and p and q are both odd numbers.
In the example, the Gx、GyMay be a 7 x 7 matrix. Then:
secondly, selecting a qxx matrix A from the image data by taking the pixel point to be denoised as a central pixel point, wherein x is an odd number, and the value of x can be equal to the value of p, for example, can be 7; then, based on the first operator template GxThe second operator template GyThe matrix a calculates an edge strength, which is abs (G)x*A)+abs(GyA). Wherein abs is used to denote the absolute value, abs (G)xA) is used to represent the gradient in the horizontal direction of the pixel points to be denoised, abs (G)yA) is used to represent the gradient in the vertical direction of the pixel points to be denoised.
And, it should be noted that, because of the first operator template GxExcept that the diagonal elements and the elements in the same row as the central element are not 0, the other elements are 0, and then when the first operator template G is axWhen multiplying with the matrix A, only part of the pixel points in the matrix A can not be phase 0Multiplication, so that only the pixel values of a part of the pixels in the matrix A will affect the first operator template GxThe multiplication result with the matrix a, that is, only some of the pixel points in the matrix a may affect the gradient in the horizontal direction of the pixel point to be denoised, so that it may be avoided that an excessive number of pixel points affect the gradient in the horizontal direction, so that the finally calculated gradient in the horizontal direction is inaccurate.
Similarly, the second operator template GyExcept that the diagonal elements and the elements in the same row as the central element are not 0, the other elements are 0, and then when the second operator template G is usedyWhen the matrix A is multiplied, only partial pixel points in the matrix A influence the gradient of the pixel points to be denoised in the vertical direction, so that the situation that the finally calculated gradient in the vertical direction is inaccurate due to the influence of an excessive number of pixel points on the gradient in the vertical direction can be avoided.
Thirdly, determining a preset pixel point in the qxx matrix A, wherein the preset pixel point comprises: pixel points of the qxx matrix a not multiplied by 0 when calculating edge strength. And then, determining the average value of the pixel values of the preset pixel points.
And fourthly, dividing the edge intensity by the average value of the pixel values of the preset pixel points to calculate the edge coefficient.
Wherein the edge coefficients are mainly used for representing the complexity of the detail information. When the edge coefficient of the pixel point to be subjected to noise reduction is larger, the detail information of the pixel point to be subjected to noise reduction is more, and the pixel point to be subjected to noise reduction can be considered to be located in the edge area of the image data; when the edge coefficient of the pixel point to be subjected to noise reduction is smaller, the detail information of the pixel point to be subjected to noise reduction is less, and the pixel point to be subjected to noise reduction can be regarded as being located in a flat area of the image data.
It can be seen from the above that the edge coefficients for characterizing the complexity of the detail information can be calculated by performing the first to fourth steps. Thereafter, an edge blending ratio may be calculated based on the edge coefficients, so that a detail addition value may be subsequently calculated based on the edge blending ratio.
The method for determining the edge fusion proportion based on the edge coefficient may include:
presetting a first reference edge coefficient and a second reference edge coefficient, wherein the first reference edge coefficient is larger than the second reference edge coefficient. And, the edge blending ratio ═ edge coefficient-second reference edge coefficient)/(first reference edge coefficient-second reference edge coefficient).
The first reference edge coefficient may be greater than or equal to a maximum edge coefficient of the image data and less than or equal to a minimum edge coefficient of the image data, for example, the first reference edge coefficient may be the maximum edge coefficient of the image data. The second reference edge coefficient may be smaller than the first edge coefficient and equal to or greater than a minimum edge coefficient of the image data, for example, the second reference edge coefficient may be the minimum edge coefficient of the image data.
And it can be known from the calculation formula of the edge blending ratio that, when the edge coefficient is larger (that is, the detail information of the pixel point to be noise-reduced is more), the calculated edge blending ratio is also larger, and the calculation formula based on the detail addition value (that is, the detail addition value (the pixel value of the pixel point to be noise-reduced-predetermined noise reduction value) × the edge blending ratio) is known, and when the edge blending ratio is larger, the calculated detail addition value is also larger, and the strength of the subsequent detail addition is also larger. And when the edge coefficient is smaller (that is, the detail information of the pixel point to be denoised is less), the calculated edge fusion proportion is smaller, so that the finally calculated detail adding value is smaller, and the strength of the subsequent detail adding is smaller.
And 500, determining the sum of the preset noise reduction value and the detail adding value as a final noise reduction value of the pixel point to be subjected to noise reduction and outputting the final noise reduction value.
Therefore, in this embodiment, when the noise of the pixel in the edge region is reduced, not only the noise of the pixel to be reduced is reduced, but also details with different strengths are added to the pixel to be reduced based on the edge coefficient of the pixel to be reduced. And when the edge coefficient of the pixel point to be denoised is larger and the detail information is more, the intensity of added details is larger, and conversely, the intensity of added details is smaller. Therefore, details can be prevented from being lost, the edge detail information of the image data after noise reduction is ensured to be complete, and the image data after noise reduction is ensured to be clear and smooth.
Meanwhile, in the embodiment, when the noise of the pixel point to be denoised is denoised, different smooth coefficients are determined based on the difference of the image areas where the pixel point to be denoised is located in the image data, so that the noise of the pixel point to be denoised in different degrees can be reduced, the situations of low imaging quality and poor imaging effect of the image due to the consistent denoising strength of each area of the image can be avoided, and the imaging quality and the imaging effect of the image are ensured.
In addition, in this embodiment, when determining the smoothing coefficient, the upper limit value and the lower limit value are provided based on the smoothing coefficient, so that the noise reduction strength of the pixel point of the image data is always between the second critical strength and the first critical strength, and is not greater than the first critical strength or less than the second critical strength, thereby ensuring that the noise-reduced image data is always clear and smooth without blurring.
Furthermore, the present invention also provides an image noise reduction apparatus for performing the image noise reduction method shown in fig. 1, the apparatus may include:
the acquisition module is used for acquiring image data and determining pixel points to be denoised in the image data;
the first calculation module is used for determining a noise coefficient corresponding to the pixel point to be subjected to noise reduction based on the pixel value of the pixel point to be subjected to noise reduction and the pixel values of the pixel points around the pixel point to be subjected to noise reduction;
the second calculation module is used for determining an image area of the pixel point to be subjected to noise reduction in the image data based on the noise coefficient, and determining a smooth coefficient based on the image area, wherein the smooth coefficient is used for controlling the noise reduction intensity; then, based on the smoothing coefficient, the pixel value of the pixel point to be subjected to noise reduction and the pixel values of the pixel points around the pixel point to be subjected to noise reduction, the pixel point to be subjected to noise reduction is subjected to noise reduction to obtain a preset noise reduction value;
the third calculation module is used for calculating an edge coefficient of the pixel point to be denoised by utilizing an edge detection operator algorithm, and determining an edge fusion proportion based on the edge coefficient, wherein the edge fusion proportion and the edge coefficient are in positive correlation; determining a detail adding value based on the edge fusion proportion, a preset noise reduction value and a pixel value of a pixel point to be subjected to noise reduction, wherein the detail adding value is (the pixel value of the pixel point to be subjected to noise reduction-the preset noise reduction value) multiplied by the edge fusion proportion, and the detail adding value is used for representing the intensity of adding details to the pixel point subjected to noise reduction;
and the fourth calculation module is used for determining the sum of the preset noise reduction value and the detail adding value as a final noise reduction value of the pixel point to be subjected to noise reduction and outputting the final noise reduction value.
Optionally, the first computing module is further configured to:
providing a formula I, and calculating a noise coefficient corresponding to the pixel point to be denoised based on the formula I;
the formula I is as follows:
wherein j is used for indicating a reference pixel point, the reference pixel point is any pixel point in a first m × n matrix which takes the pixel point to be denoised as a central pixel point in the image data, and m and n are both greater than 1; f (j) pixel values indicating reference pixels in the first m × n matrix; i is used to indicate the first mxn matrix; num is the number of reference pixels in the first mxn matrix; a is used for indicating the adjustment amplitude of the noise coefficient, and a is more than or equal to 0.7 and less than or equal to 1.4.
Optionally, the second computing module is further configured to:
presetting a first reference value and a second reference value, wherein the first reference value is greater than or equal to the second reference value; and the first reference value and the second reference value satisfy a preset condition, wherein the preset condition comprises: when the smoothing coefficient is equal to the product of the first reference value and a preset value, the noise reduction intensity for reducing the noise of the pixel point to be reduced based on the smoothing coefficient is a first critical intensity; when the smoothing coefficient is equal to the product of the second reference value and a preset value, the noise reduction intensity for reducing the noise of the pixel point to be reduced based on the smoothing coefficient is a second critical intensity; wherein the first critical intensity and the second critical intensity satisfy: blurring the noise-reduced image data when the noise reduction intensity for the image data exceeds the first critical intensity or the noise reduction intensity for the image data is lower than the second critical intensity;
judging the magnitude relation between the noise coefficient and the first reference value and the second reference value, and correspondingly determining the value of the smooth coefficient based on the magnitude relation; when the noise coefficient is larger than the first reference value, determining that the pixel point to be subjected to noise reduction is located in the edge area of the image data, and enabling the smoothing coefficient to be equal to the product of the first reference value and the preset value; when the noise coefficient is smaller than the second reference value, determining that the pixel point to be subjected to noise reduction is located in a flat area of the image data, and enabling the smooth coefficient to be equal to the product of the second reference value and a preset value; and when the noise coefficient is greater than or equal to the second reference value and less than or equal to the first reference value, enabling the smoothing coefficient to be equal to the product of the noise coefficient and the preset value.
Optionally, the first reference value is less than or equal to 0.14 and greater than 0.04; the second reference value is greater than or equal to 0.04 and less than or equal to the first reference value; the preset value is 3.3.
Optionally, the second computing module is further configured to:
providing a second formula, and calculating the preset noise reduction value based on the second formula;
the formula II is as follows:
wherein f (i) is the predetermined noise reduction value; j is used for indicating a reference pixel point, the reference pixel point is any pixel point in a first m multiplied by n matrix which takes the pixel point to be denoised as a central pixel point in the image data, and m and n are both larger than 1; f (j) a pixel value indicating the reference pixel point;is the standard deviation of the gaussian kernel function; n (I) and I are both used to indicate the first m x n matrix; n (j) a second mxn matrix indicating a pixel in the image data centered on the reference pixel; d (i, j) is used for indicating Euclidean distances of the first m multiplied by n matrix and the second m multiplied by n matrix; w (i, j) is used for indicating the fusion weight of the first m × n matrix and the second m × n matrix and representing the similarity degree between the first m × n matrix and the second m × n matrix; h is used to indicate the smoothing parameter.
Optionally, the third computing module is further configured to:
providing first operator templates G respectivelyxAnd a second operator template Gy;Gx、GyAre all p × q matrixes, and both p and q are odd numbers;
selecting a qxx matrix A from the image data by taking the pixel point to be denoised as a central pixel point, wherein x is an odd number; and based on the first operator template GxThe second operator template GyCalculating the edge strength by the matrix A, theEdge Strength abs (G)x*A)+abs(Gy*A);
Determining a predetermined pixel point in the qxx matrix A, wherein the predetermined pixel point comprises: pixel points which are not multiplied by 0 in the qxx matrix A when the edge intensity is calculated; then, determining the average value of the pixel values of the preset pixel points;
and dividing the edge intensity by the average value of the pixel values of the preset pixel points to calculate the edge coefficient.
Optionally, the G isx、GyAre all 7 × 7 matrixes;
optionally, the matrix a is a 7 × 7 matrix.
Optionally, the third computing module is further configured to:
presetting a first reference edge coefficient and a second reference edge coefficient, wherein the first reference edge coefficient is larger than the second reference edge coefficient;
the edge blending ratio is calculated based on the edge coefficient, the first reference edge coefficient, and the second reference edge coefficient, and the edge blending ratio is (edge coefficient-second reference edge coefficient)/(first reference edge coefficient-second reference edge coefficient).
In summary, in the image noise reduction method and apparatus provided by the present invention, different smoothing coefficients are determined based on the difference of the pixel point to be noise reduced in the image region of the image data, and then the predetermined noise reduction value of the pixel point to be noise reduced is further calculated based on the smoothing coefficients. In the invention, the smoothing coefficient is mainly used for controlling the noise reduction degree, so that the noise reduction of different degrees can be realized for the pixel points in different image areas, thereby avoiding the occurrence of the conditions of low imaging quality and poor imaging effect of the image due to consistent noise reduction strength of each area of the image, and ensuring the imaging quality and the imaging effect of the image.
And determining an edge coefficient of a pixel point to be denoised, further determining a detail adding value based on the edge coefficient, and then determining the sum of a preset denoising value and the detail adding value as a final denoising value and outputting the final denoising value so as to add detail information to the pixel point to be denoised after denoising. Wherein the edge coefficient is positively correlated with the detail addition value. Therefore, when the edge coefficient is larger (that is, the detail information of the pixel point to be subjected to noise reduction is more), the detail adding value is larger, the detail information added to the pixel point to be subjected to noise reduction is more, so that the loss of the detail information in the noise reduction process can be prevented, the edge detail information of finally obtained image data is ensured, and the imaging quality and the imaging effect of the image are further ensured.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The above description is only for the purpose of describing the preferred embodiments of the present invention, and is not intended to limit the scope of the present invention, and any variations and modifications made by those skilled in the art based on the above disclosure are within the scope of the appended claims.
Claims (10)
1. A method for image noise reduction, the method comprising:
acquiring image data, and determining pixel points to be denoised in the image data;
determining a noise coefficient corresponding to the pixel point to be subjected to noise reduction based on the pixel value of the pixel point to be subjected to noise reduction and the pixel values of the pixel points around the pixel point to be subjected to noise reduction;
determining an image area of the pixel point to be subjected to noise reduction in the image data based on the noise coefficient, and determining a smooth coefficient based on the image area, wherein the smooth coefficient is used for controlling noise reduction intensity; then, based on the smoothing coefficient, the pixel value of the pixel point to be subjected to noise reduction and the pixel values of the pixel points around the pixel point to be subjected to noise reduction, the pixel point to be subjected to noise reduction is subjected to noise reduction to obtain a preset noise reduction value;
calculating an edge coefficient of the pixel point to be denoised by using an edge detection operator algorithm, and determining an edge fusion proportion based on the edge coefficient, wherein the edge fusion proportion is in positive correlation with the edge coefficient; determining a detail adding value based on the edge fusion proportion, a preset noise reduction value and a pixel value of a pixel point to be subjected to noise reduction, wherein the detail adding value is (the pixel value of the pixel point to be subjected to noise reduction-the preset noise reduction value) multiplied by the edge fusion proportion, and the detail adding value is used for representing the intensity of adding details to the pixel point subjected to noise reduction;
and determining the sum of the preset noise reduction value and the detail adding value as a final noise reduction value of the pixel point to be subjected to noise reduction and outputting the final noise reduction value.
2. The image noise reduction method according to claim 1, wherein the method for determining the noise coefficient corresponding to the pixel to be noise reduced based on the pixel values of the pixel to be noise reduced and the pixel values of the pixels around the pixel to be noise reduced comprises:
providing a formula I, and calculating a noise coefficient corresponding to the pixel point to be denoised based on the formula I;
the formula I is as follows:
wherein j is used for indicating a reference pixel point, the reference pixel point is any pixel point in a first m × n matrix which takes the pixel point to be denoised as a central pixel point in the image data, and m and n are both greater than 1; f (j) pixel values indicating reference pixels in the first m × n matrix; i is used to indicate the first mxn matrix; num is the number of reference pixels in the first mxn matrix; a is used for indicating the adjustment amplitude of the noise coefficient, and a is more than or equal to 0.7 and less than or equal to 1.4.
3. The method for reducing image noise according to claim 1, wherein the method for determining the image region of the pixel point to be noise-reduced in the image data based on the noise coefficient and determining the smoothing coefficient based on the image region comprises:
presetting a first reference value and a second reference value, wherein the first reference value is greater than or equal to the second reference value; and the first reference value and the second reference value satisfy a preset condition, wherein the preset condition comprises: when the smoothing coefficient is equal to the product of the first reference value and a preset value, the noise reduction intensity for reducing the noise of the pixel point to be reduced based on the smoothing coefficient is a first critical intensity; when the smoothing coefficient is equal to the product of the second reference value and a preset value, the noise reduction intensity for reducing the noise of the pixel point to be reduced based on the smoothing coefficient is a second critical intensity; wherein the first critical intensity and the second critical intensity satisfy: blurring the noise-reduced image data when the noise reduction intensity for the image data exceeds the first critical intensity or the noise reduction intensity for the image data is lower than the second critical intensity;
judging the magnitude relation between the noise coefficient and the first reference value and the second reference value, and correspondingly determining the value of the smooth coefficient based on the magnitude relation; when the noise coefficient is larger than the first reference value, determining that the pixel point to be subjected to noise reduction is located in the edge area of the image data, and enabling the smoothing coefficient to be equal to the product of the first reference value and the preset value; when the noise coefficient is smaller than the second reference value, determining that the pixel point to be subjected to noise reduction is located in a flat area of the image data, and enabling the smooth coefficient to be equal to the product of the second reference value and a preset value; and when the noise coefficient is greater than or equal to the second reference value and less than or equal to the first reference value, enabling the smoothing coefficient to be equal to the product of the noise coefficient and the preset value.
4. The image noise reduction method according to claim 3, wherein the first reference value is 0.14 or less and 0.04 or more; the second reference value is greater than or equal to 0.04 and less than or equal to the first reference value; the preset value is 3.3.
5. The method for reducing noise of an image according to claim 1, wherein the method for reducing noise of the pixel to be noise reduced based on the smoothing coefficient, the pixel value of the pixel to be noise reduced, and the pixel values of the pixels around the pixel to be noise reduced to obtain the predetermined noise reduction value comprises:
providing a second formula, and calculating the preset noise reduction value based on the second formula;
the formula II is as follows:
wherein f (i) is the predetermined noise reduction value; j is used for indicating a reference pixel point, the reference pixel point is any pixel point in a first m multiplied by n matrix which takes the pixel point to be denoised as a central pixel point in the image data, and m and n are both larger than 1; f (j) a pixel value indicating the reference pixel point;is the standard deviation of the gaussian kernel function; n (I) and I are both used to indicate the first m x n matrix; n (j) a second mxn matrix indicating a pixel in the image data centered on the reference pixel; d (i, j) is used to indicate the Europe of the first and second m x n matricesA distance in degrees; w (i, j) is used for indicating the fusion weight of the first m × n matrix and the second m × n matrix and representing the similarity degree between the first m × n matrix and the second m × n matrix; h is used to indicate the smoothing parameter.
6. The image denoising method of claim 1, wherein the edge detection operator algorithm comprises a sobel operator algorithm;
and the method for calculating the edge coefficient corresponding to the pixel point to be denoised by utilizing the edge detection operator algorithm comprises the following steps:
providing first operator templates G respectivelyxAnd a second operator template Gy;Gx、GyAre all p × q matrixes, and both p and q are odd numbers;
selecting a qxx matrix A from the image data by taking the pixel point to be denoised as a central pixel point, wherein x is an odd number; and based on the first operator template GxThe second operator template GyThe matrix a calculates an edge strength, which is abs (G)x*A)+abs(Gy*A);
Determining a predetermined pixel point in the qxx matrix A, wherein the predetermined pixel point comprises: pixel points which are not multiplied by 0 in the qxx matrix A when the edge intensity is calculated; then, determining the average value of the pixel values of the preset pixel points;
and dividing the edge intensity by the average value of the pixel values of the preset pixel points to calculate the edge coefficient.
8. the image noise reduction method according to claim 7, wherein the matrix a is a 7 x 7 matrix.
9. The image noise reduction method according to claim 1, wherein the method of determining the edge blending ratio based on the edge coefficient comprises:
presetting a first reference edge coefficient and a second reference edge coefficient, wherein the first reference edge coefficient is larger than the second reference edge coefficient;
the edge blending ratio is calculated based on the edge coefficient, the first reference edge coefficient, and the second reference edge coefficient, and the edge blending ratio is (edge coefficient-second reference edge coefficient)/(first reference edge coefficient-second reference edge coefficient).
10. An image noise reduction apparatus, characterized in that the apparatus comprises:
the acquisition module is used for acquiring image data and determining pixel points to be denoised in the image data;
the first calculation module is used for determining a noise coefficient corresponding to the pixel point to be subjected to noise reduction based on the pixel value of the pixel point to be subjected to noise reduction and the pixel values of the pixel points around the pixel point to be subjected to noise reduction;
the second calculation module is used for determining an image area of the pixel point to be subjected to noise reduction in the image data based on the noise coefficient, and determining a smooth coefficient based on the image area, wherein the smooth coefficient is used for controlling the noise reduction intensity; then, based on the smoothing coefficient, the pixel value of the pixel point to be subjected to noise reduction and the pixel values of the pixel points around the pixel point to be subjected to noise reduction, the pixel point to be subjected to noise reduction is subjected to noise reduction to obtain a preset noise reduction value;
the third calculation module is used for calculating an edge coefficient of the pixel point to be denoised by utilizing an edge detection operator algorithm, and determining an edge fusion proportion based on the edge coefficient, wherein the edge fusion proportion and the edge coefficient are in positive correlation; determining a detail adding value based on the edge fusion proportion, a preset noise reduction value and a pixel value of a pixel point to be subjected to noise reduction, wherein the detail adding value is (the pixel value of the pixel point to be subjected to noise reduction-the preset noise reduction value) multiplied by the edge fusion proportion, and the detail adding value is used for representing the intensity of adding details to the pixel point subjected to noise reduction;
and the fourth calculation module is used for determining the sum of the preset noise reduction value and the detail adding value as a final noise reduction value of the pixel point to be subjected to noise reduction and outputting the final noise reduction value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011043543.3A CN112184583B (en) | 2020-09-28 | 2020-09-28 | Image noise reduction method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011043543.3A CN112184583B (en) | 2020-09-28 | 2020-09-28 | Image noise reduction method and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112184583A true CN112184583A (en) | 2021-01-05 |
CN112184583B CN112184583B (en) | 2023-11-17 |
Family
ID=73946851
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011043543.3A Active CN112184583B (en) | 2020-09-28 | 2020-09-28 | Image noise reduction method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112184583B (en) |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090154807A1 (en) * | 2005-12-30 | 2009-06-18 | Telecom Italia S.P.A. | Edge Comparison in Segmentation of Video Sequences |
CN103208104A (en) * | 2013-04-16 | 2013-07-17 | 浙江工业大学 | Non-local theory-based image denoising method |
CN104899888A (en) * | 2015-06-18 | 2015-09-09 | 大连理工大学 | Legemdre moment-based image subpixel edge detection method |
CN105139343A (en) * | 2014-05-30 | 2015-12-09 | 上海贝卓智能科技有限公司 | Image processing method and device |
CN105719274A (en) * | 2014-12-10 | 2016-06-29 | 全视技术有限公司 | Edge Detection System And Methods |
CN107122756A (en) * | 2017-05-11 | 2017-09-01 | 南宁市正祥科技有限公司 | A kind of complete non-structural road edge detection method |
CN108198196A (en) * | 2018-01-23 | 2018-06-22 | 华东交通大学 | The design and implementation methods of quantum Image Edge-Detection based on Sobel operators |
CN109712102A (en) * | 2017-10-25 | 2019-05-03 | 杭州海康威视数字技术股份有限公司 | A kind of image interfusion method, device and image capture device |
-
2020
- 2020-09-28 CN CN202011043543.3A patent/CN112184583B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090154807A1 (en) * | 2005-12-30 | 2009-06-18 | Telecom Italia S.P.A. | Edge Comparison in Segmentation of Video Sequences |
CN103208104A (en) * | 2013-04-16 | 2013-07-17 | 浙江工业大学 | Non-local theory-based image denoising method |
CN105139343A (en) * | 2014-05-30 | 2015-12-09 | 上海贝卓智能科技有限公司 | Image processing method and device |
CN105719274A (en) * | 2014-12-10 | 2016-06-29 | 全视技术有限公司 | Edge Detection System And Methods |
CN104899888A (en) * | 2015-06-18 | 2015-09-09 | 大连理工大学 | Legemdre moment-based image subpixel edge detection method |
CN107122756A (en) * | 2017-05-11 | 2017-09-01 | 南宁市正祥科技有限公司 | A kind of complete non-structural road edge detection method |
CN109712102A (en) * | 2017-10-25 | 2019-05-03 | 杭州海康威视数字技术股份有限公司 | A kind of image interfusion method, device and image capture device |
CN108198196A (en) * | 2018-01-23 | 2018-06-22 | 华东交通大学 | The design and implementation methods of quantum Image Edge-Detection based on Sobel operators |
Non-Patent Citations (1)
Title |
---|
王勇; 李赟晟: "一种基于Bayer型图像数据的自适应非局部均值滤波算法", 集成电路应用, pages 13 - 15 * |
Also Published As
Publication number | Publication date |
---|---|
CN112184583B (en) | 2023-11-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US7792384B2 (en) | Image processing apparatus, image processing method, program, and recording medium therefor | |
US7324701B2 (en) | Image noise reduction | |
US8224116B2 (en) | Image processing method and imaging apparatus | |
EP2075756B1 (en) | Block-based image blending for camera shake compensation | |
US10198801B2 (en) | Image enhancement using self-examples and external examples | |
US10628924B2 (en) | Method and device for deblurring out-of-focus blurred images | |
US8818126B2 (en) | Method and apparatus for block-based image denoising | |
US8831372B2 (en) | Image processing device, image processing method and storage medium storing image processing program | |
EP2164040B1 (en) | System and method for high quality image and video upscaling | |
CN103606132A (en) | Multiframe digital image denoising method based on space domain and time domain combination filtering | |
US8335396B2 (en) | Image output apparatus, edge enhancement method, and recording medium | |
US20080085061A1 (en) | Method and Apparatus for Adjusting the Contrast of an Input Image | |
CN110874827B (en) | Turbulent image restoration method and device, terminal equipment and computer readable medium | |
CN109214996B (en) | Image processing method and device | |
CN113592776A (en) | Image processing method and device, electronic device and storage medium | |
CN113450272B (en) | Image enhancement method based on sinusoidal variation and application thereof | |
WO2014102876A1 (en) | Image processing device and image processing method | |
US8300150B2 (en) | Image processing apparatus and method | |
CN112184583B (en) | Image noise reduction method and device | |
CN114662519B (en) | QR code blind deblurring method based on position detection graph gradient and intensity prior | |
US11495162B2 (en) | Panel display optimization method, display panel, and storage medium | |
WO2013011797A1 (en) | Degradation restoration system, degradation restoration method and program | |
CN110363723B (en) | Image processing method and device for improving image boundary effect | |
JP4104475B2 (en) | Contour correction device | |
RU2332716C2 (en) | Method and device for anisotropic filtering of dynamic video picture |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |