CN113536214A - Image noise reduction method and device and storage device - Google Patents
Image noise reduction method and device and storage device Download PDFInfo
- Publication number
- CN113536214A CN113536214A CN202010292363.2A CN202010292363A CN113536214A CN 113536214 A CN113536214 A CN 113536214A CN 202010292363 A CN202010292363 A CN 202010292363A CN 113536214 A CN113536214 A CN 113536214A
- Authority
- CN
- China
- Prior art keywords
- image
- noise
- pixel
- denoised
- initial
- 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.)
- Pending
Links
- 230000009467 reduction Effects 0.000 title claims abstract description 88
- 238000000034 method Methods 0.000 title claims abstract description 83
- 230000004927 fusion Effects 0.000 claims description 56
- 230000006870 function Effects 0.000 claims description 20
- 238000013507 mapping Methods 0.000 claims description 20
- 238000012545 processing Methods 0.000 claims description 19
- 230000009466 transformation Effects 0.000 claims description 16
- 238000004590 computer program Methods 0.000 claims description 13
- 230000004931 aggregating effect Effects 0.000 claims description 4
- 230000001131 transforming effect Effects 0.000 claims 1
- 230000000694 effects Effects 0.000 abstract description 9
- 230000008569 process Effects 0.000 description 16
- 238000010586 diagram Methods 0.000 description 13
- 238000004364 calculation method Methods 0.000 description 9
- 238000004422 calculation algorithm Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 238000000354 decomposition reaction Methods 0.000 description 3
- 238000011426 transformation method Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 238000003491 array Methods 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000007667 floating Methods 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000013138 pruning Methods 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- G06T5/70—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration by the use of more than one image, e.g. averaging, subtraction
Abstract
The application discloses a method, a device and a storage device for image noise reduction, comprising the following steps: acquiring an initial image, and determining a plurality of reference pixel points in the initial image; respectively calculating the noise value of the region where each reference pixel point is located in the initial image; selecting an image block to be denoised containing the reference pixel points from the initial image according to the noise value of the area where each reference pixel point is located, and acquiring a denoising threshold corresponding to the image block to be denoised; based on a noise reduction threshold corresponding to each image block to be subjected to noise reduction, performing noise reduction on the image block to be subjected to noise reduction to obtain a corresponding image block to be subjected to noise reduction; fusing the obtained noise-reduced image blocks to obtain a noise-reduced image; and the distance between adjacent reference pixel points is smaller than or equal to the size of the image block to be denoised. The technical scheme provided by the invention can improve the effect of image noise reduction.
Description
Technical Field
The present application relates to the field of image processing technologies, and in particular, to a method, an apparatus, and a storage apparatus for image noise reduction.
Background
In recent years, with the development of image sensing technology and the rapid development of computer software and hardware capabilities, research in the field of computer vision has been continuously developed and advanced. Image sensing technology is widely used in applications such as face recognition, vehicle detection, medical image analysis, etc., and the practical effects of these applications often depend heavily on the noise level of the image. Generally, when the image noise level is low, the confidence of the results output by these applications is high, and vice versa. However, in most cases, the image acquisition is often affected by various factors, and noise is difficult to avoid, so an image denoising technique with a good denoising effect is urgently needed to solve the problem of poor image denoising in the prior art.
Disclosure of Invention
The technical problem mainly solved by the application is to provide an image noise reduction method, an image noise reduction device and a storage device, which can improve the image noise reduction effect.
In order to solve the technical problem, the application adopts a technical scheme that: provided is a method for image noise reduction, comprising:
acquiring an initial image, and determining a plurality of reference pixel points in the initial image;
respectively calculating the noise value of the region where each reference pixel point is located in the initial image;
selecting an image block to be denoised containing the reference pixel points from the initial image according to the noise value of the area where each reference pixel point is located, and acquiring a denoising threshold corresponding to the image block to be denoised;
based on a noise reduction threshold corresponding to each image block to be subjected to noise reduction, performing noise reduction on the image block to be subjected to noise reduction to obtain a corresponding image block subjected to noise reduction;
fusing the obtained noise-reduced image blocks to obtain a noise-reduced image;
and the distance between the adjacent reference pixel points is smaller than or equal to the size of the image block to be denoised.
In order to solve the above technical problem, another technical solution adopted by the present application is: there is provided an apparatus for image noise reduction, the apparatus comprising a memory and a processor coupled, wherein,
the memory includes local storage and stores a computer program;
the processor is adapted to run the computer program to perform the method as described above.
In order to solve the above technical problem, the present application adopts another technical solution: there is provided a storage device storing a computer program executable by a processor for implementing the method as described above.
The technical scheme provided by the application comprises the steps of respectively calculating the noise value of each reference pixel point in an initial image after the initial image is obtained and a plurality of reference pixel points are determined in the initial image, then correspondingly selecting the image block to be denoised containing the reference pixel points in the initial image according to the obtained noise value of the area where each reference pixel point is located, obtaining the denoising threshold value corresponding to the current image block to be denoised according to the noise value of the area where each image block to be denoised is located, further realizing the denoising treatment of the image block to be denoised where the reference pixel points are located with different denoising threshold values according to the noise level of the area where the different reference pixel points are located, and avoiding the problems of excessive smoothness of a low-noise area or too low denoising strength of a high-noise area caused by denoising by using a global threshold value compared with the prior art, the technical scheme provided by the application can better reduce the noise of the image realization effect, and solves the technical problem which is urgently needed to be solved in the prior art.
Drawings
FIG. 1 is a schematic flowchart illustrating an embodiment of a method for image denoising according to the present application;
FIG. 2 is a schematic diagram illustrating selection of a reference pixel point in an embodiment of a method for image denoising according to the present application;
FIG. 3 is a schematic diagram illustrating selection of a reference pixel point in an embodiment of a method for image denoising according to the present application;
fig. 4 is a schematic diagram illustrating obtaining of an image block corresponding to a reference pixel point in an embodiment of an image denoising method according to the present application;
FIG. 5 is a flowchart illustrating another embodiment of a method for image denoising according to the present application;
FIG. 6 is a schematic flow chart illustrating a method for image denoising according to another embodiment of the present disclosure;
FIG. 7 is a schematic diagram illustrating an edge extension processing performed on an initial image according to an embodiment of an image denoising method according to the present application;
FIG. 8 is a flowchart illustrating a method of image denoising according to another embodiment of the present disclosure;
FIG. 9 is a diagram illustrating a mapping relationship between fusion weights and pixel values according to an embodiment of a method for image denoising according to the present application;
FIG. 10 is a flow chart illustrating a method of image denoising according to another embodiment of the present disclosure;
FIG. 11 is a flowchart illustrating a method of image denoising according to another embodiment of the present disclosure;
FIG. 12 is a flowchart illustrating a method of image denoising according to another embodiment of the present disclosure;
FIG. 13 is a schematic diagram of an embodiment of an apparatus for image noise reduction according to the present application;
fig. 14 is a schematic structural diagram of an embodiment of a memory device according to the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. It is to be understood that the specific embodiments described herein are merely illustrative of the application and are not limiting of the application. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
In the description of the present application, "plurality" means at least two, e.g., two, three, etc., unless explicitly specifically limited otherwise. Furthermore, the terms "include" and "have," as well as any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to only those steps or elements listed, but may alternatively include other steps or elements not listed, or inherent to such process, method, article, or apparatus.
Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. It is explicitly and implicitly understood by one skilled in the art that the embodiments described herein can be combined with other embodiments.
In the prior art, a method based on a transform domain is used for denoising an image, that is, the image is converted into the transform domain, so that high-frequency information containing noise is conveniently extracted, and then the high-frequency information is adjusted by setting a threshold value, so as to achieve image denoising.
Referring to fig. 1, fig. 1 is a schematic flowchart illustrating an embodiment of an image denoising method according to the present application. In the current embodiment, the method provided by the present application includes:
s110: an initial image is obtained, and a plurality of reference pixel points are determined in the initial image.
The initial image is an image which needs to be subjected to noise reduction processing. The reference pixel points are pixel points selected from the initial image according to a set reference pixel point selection rule. The number of reference pixels in each initial image is determined based on the size of the initial image and a set reference pixel selection rule, and is not limited herein. In addition, the set reference pixel point selection rule can be set and adjusted by a user according to actual requirements, and the set reference pixel point selection rule is not limited at all.
In an embodiment, after the initial image is obtained, the initial image may be further subjected to data decomposition to obtain YUV data, that is, Y channel data, U channel data, and V channel data may be obtained. After the data decomposition, the image data obtained by the decomposition is respectively subjected to the following related denoising methods to respectively obtain the denoised Y-channel data, the denoised U-channel data and the denoised V-channel data, and then the denoised Y-channel data, U-channel data and V-channel data are subjected to data synthesis to obtain a denoised image with a better denoising effect. It is to be understood that the type of initial image specifically described above is not limited thereto.
S120: and respectively calculating the noise value of the area where each reference pixel point is located in the initial image.
In step S110, after determining a plurality of reference pixels in the initial image, the noise value of the region in which each reference pixel is located in the initial image is calculated.
Further, step S120 calculates a noise value of a region where each reference pixel point in the initial image is located, respectively, including: the area with the reference pixel point as the vertex and the size as the preset area size is determined, and the noise value of the area where the reference point is located is obtained based on the pixel values of all the pixel points in the area.
The area where each reference pixel point is located is an image block with a preset size and with the reference pixel point as a vertex in the initial image, the image block with the preset size is used for determining the noise value of the area where the current reference pixel point is located, and the size of the image block to be denoised and the corresponding denoising threshold corresponding to the current reference pixel point are determined according to the noise value of the area where the reference pixel point is located. Specifically, in an embodiment, each reference pixel point is an upper left vertex of an image block with a preset size, and the preset size is 8 × 8. In another embodiment, each reference pixel point is a top right vertex of an image block with a predetermined size, and the predetermined size is 16 × 16.
The image block is an image block which is obtained by taking a reference pixel point as a vertex in an initial image and selecting according to a preset image block selection rule, wherein the image block comprises a plurality of pixel points, the image block to be denoised is an image block waiting for denoising, the image block to be denoised is an image block which takes the reference pixel point as the vertex and has a size which is reselected according to a noise value of an area where the reference vertex is located, and the specific direction of the reference pixel point in the image block to be denoised is not limited. In the same initial image, the sizes of the to-be-denoised image blocks corresponding to different reference pixel points can be different, the size of the to-be-denoised image block corresponding to each reference pixel point is determined according to a preset image block selection rule and the noise value of the region where the current reference pixel point is located, the to-be-denoised image blocks corresponding to adjacent reference pixel points can have mutually overlapped parts, namely, the same pixel point can be simultaneously located in a plurality of adjacent to-be-denoised image blocks. The preset image block selection rule is described in detail in the corresponding section below.
In an embodiment, step S120 may be to separately calculate a noise estimation value of a region where each reference pixel point in the initial image is located. That is, a noise estimation value in an image block of a preset size where the reference pixel point is located is calculated, and in the current embodiment, a pixel variance of the image block of the preset size is selected as the noise estimation value. The preset size is determined according to an empirical value, and is not limited herein, for example, the preset size may be 8 × 8.
S130: and selecting an image block to be denoised containing the reference pixel points from the initial image according to the noise value of the area where each reference pixel point is located, and acquiring a denoising threshold corresponding to the image block to be denoised.
And the distance between adjacent reference pixel points is smaller than or equal to the size of the image block to be denoised.
After the noise value of the region where each reference pixel point is located in the initial image is calculated and obtained in step S120, the image block to be denoised, which includes the reference pixel point, is selected from the initial image further according to the noise value of the region where each reference pixel point is located, which is calculated and obtained in step S120. The image blocks to be denoised are image blocks selected according to the noise values of the regions where the reference pixels are located and including the reference pixels, and because the noise values of the regions where different reference pixels are located in the same initial image are different, the sizes of the image blocks to be denoised corresponding to different reference pixels in the same initial image may be different, which may be specifically described in the following corresponding embodiments.
When the image block to be denoised containing the reference pixel points is selected in the initial image or after the image block to be denoised containing the reference pixel points is selected, the denoising threshold corresponding to the current image block to be denoised can be obtained according to the noise value of the area where each reference pixel point is located. The noise reduction threshold is a noise reduction experience value set by a user in advance according to noise reduction effects corresponding to different noise values, that is, when the noise values are in different preset noise ranges, the noise values correspond to different noise reduction thresholds.
Further, the step S130 selects an image block to be denoised containing the reference pixel point from the initial image according to the noise value of the region where each reference pixel point is located, and obtains a denoising threshold corresponding to the image block to be denoised, including: determining a preset noise range to which the noise value of the region where the reference pixel point is located belongs, selecting an image block to be denoised, which contains the reference pixel point and has the size of a preset block size corresponding to the preset noise range, from the initial image, and acquiring a denoising threshold corresponding to the preset noise range as a denoising threshold corresponding to the image block to be denoised. And when the noise in the preset noise range is larger, the corresponding preset block size is larger.
The determining of the preset noise range to which the noise value of the region where the reference pixel point is located belongs refers to determining of the preset noise range to which the noise value of the region where the current reference pixel point is located belongs, so as to determine the size of the to-be-denoised image block to be selected according to the determination result. The preset noise range is determined according to the bit width of the initial image, that is, the upper limit value and/or the lower limit value of the preset noise range is determined according to the bit width of the initial image, and the number of the preset noise ranges is greater than or equal to three.
And further acquiring a noise reduction threshold corresponding to the image block to be subjected to noise reduction corresponding to the preset noise range while or after selecting the image block to be subjected to noise reduction. Specifically, according to the preset noise range to which the noise value of the region where the current reference pixel point is located, which is obtained through judgment in the above steps, belongs, the noise reduction threshold of the image block to be noise-reduced where the current reference pixel point is located is determined. When the noise in the preset noise range is larger, the corresponding noise reduction threshold value is larger.
Specifically, in the process of image denoising, for a high-noise area, the purpose of image denoising is to eliminate noise as much as possible, and then the purpose is achieved by fully utilizing information of pixels around a reference pixel point in the process of denoising, so that a relatively large image block needs to be selected by taking the reference pixel point as a vertex. Correspondingly, a larger threshold needs to be selected for a high-noise area to effectively suppress the noise signal.
As in the present embodiment, to implement the above assumption, first, a noise value of the region where the reference pixel is located is calculated, that is, a noise level of the region where the reference pixel is located is estimated. Specifically, the variance of pixels in an image block with the current reference pixel point as the top left vertex and the preset size of 8 × 8 is calculated, and the noise level of the image block is reflected by using the variance, if the variance is large, the region where the current reference pixel point is located can be considered as a high-noise region, otherwise, the region where the current reference pixel point is located is considered as a low-noise region.
Specifically, a two-dimensional image a and a reference pixel point P are given, wherein the coordinate index of the reference pixel point P is (i, j), a pixel mean value μ of a region where the reference pixel point P is located is first calculated, and a calculation formula of the pixel mean value is as follows:
after the pixel mean value of the reference pixel point P is obtained, the noise value (which may also be referred to as a noise estimation value in other embodiments) of the region where the reference pixel point is located is obtained, that is, the reference pixel point is obtainedVariance σ of the region where P is located2. Wherein, the calculation formula of the variance is as follows:
wherein, (x, y) in the above formula refers to the pixel coordinates of each pixel in the region where the reference pixel is located.
After the noise estimation value of the region where the reference pixel point P is located is obtained, the image block to be denoised, which contains the reference pixel point and has the size of the preset block size, is selected from the initial image according to the noise value of the region where the reference pixel point P is located and the given preset noise range, and the denoising threshold of the image to be denoised is determined.
E.g. when the given predetermined noise range is defined by the first variance threshold t1And a second variance threshold t2Within a defined noise range, where t1＜t2Then the noise value sigma of the region where the reference pixel point P is located is determined2The predetermined noise range to which it belongs. Wherein, in the current embodiment, the preset noise range includes less than t1Is greater than or equal to t1And is less than or equal to t2And is greater than t2These three ranges. Specifically, in the current embodiment, the size of the image to be denoised, which needs to be selected and uses the reference pixel point P as a vertex, is determined according to the following formula. The following formula may also be understood as a rule for choosing the sizes of the image blocks corresponding to the high, medium and low noise regions and their different regions. Specifically, the definition formula of the block sizes of the different region images is as follows:
it should be noted that, the image block sizes corresponding to different preset noise ranges and the variance threshold used for defining the preset noise ranges may be adjusted according to actual needs, and are not limited herein.
Correspondingly, after determining the needAnd at the same time of or after the size of the image block to be subjected to noise reduction is selected, determining a noise reduction threshold T corresponding to the image block to be subjected to noise reduction according to the noise value of the area where the reference pixel point is located. What is σ2Less than t1When the noise reduction threshold T is th _ low, when σ is2Is greater than or equal to t1And is less than or equal to time t2Then, the noise reduction threshold T is th _ mid when σ is2Greater than t2Then, the denoising threshold T is th _ high, and specifically, the selection formula of the denoising threshold T is as follows:
in the above equation, th _ low, th _ mid, and th _ high determine the degree of noise reduction in the different noise level regions. Wherein th _ low ≦ th _ mid ≦ th _ high, i.e. the degree of noise reduction is greater when the noise reduction threshold T is larger, and conversely the degree of noise reduction is smaller when the noise reduction threshold T is smaller. It should be noted that th _ low, th _ mid, th _ high, t1, and t2 are all empirical parameters preset by the algorithm, and specific values of the parameters are not limited herein, and in different embodiments, a user may adjust values of the parameters to achieve the noise reduction effect of the adjustment algorithm.
The values of the five parameters, th _ low, th _ mid, th _ high, t1, and t2, all need to be adjusted according to the bit width of the initial image. For example, taking 8-bit width as an example, the default values of th _ low, th _ mid, and th _ high may be 100, 280, and 1200, respectively, and the default values of t1 and t2 are 5000 and 10000.
S140: and performing noise reduction processing on the image blocks to be subjected to noise reduction based on the noise reduction threshold corresponding to each image block to be subjected to noise reduction so as to obtain corresponding image blocks subjected to noise reduction.
After the denoising threshold corresponding to the image to be denoised is obtained in step S130, based on the denoising threshold corresponding to each image block to be denoised, denoising each image block to be denoised according to the denoising threshold corresponding to each image block to be denoised. In an embodiment, a set frequency domain transformation method may be adopted to perform noise reduction processing on an image block to be subjected to noise reduction processing, and for a detailed part of the technology of the noise reduction processing, reference may be made to the following description in an embodiment corresponding to fig. 5.
S150: and fusing the obtained noise-reduced image blocks to obtain a noise-reduced image.
After each image block to be denoised in the initial image is denoised and a denoised image block is obtained, fusing the obtained denoised image blocks according to a set rule to obtain a denoised image, and further completing the denoising treatment of the initial image.
In the embodiment corresponding to fig. 1 of the present application, after an initial image is obtained and a plurality of reference pixels are determined in the initial image, a noise value of each reference pixel in the initial image is calculated respectively, then a to-be-denoised image block including the reference pixel is selected correspondingly in the initial image according to the obtained noise value of the region where each reference pixel is located, and a denoising threshold corresponding to the current to-be-denoised image block is obtained according to the noise value of the region where each to-be-denoised image block is located, so as to implement denoising processing of different denoising thresholds on the to-be-denoised image block where the reference pixel is located according to the noise values of the regions where different reference pixels are located, thereby avoiding the problem of excessive smoothness of a low noise region or excessively low denoising strength of a high noise region caused by denoising using a global threshold in comparison with denoising in the prior art, and further, the noise of the image can be reduced better.
Further, the determining a plurality of reference pixel points in the initial image in step S110 includes: and searching a plurality of reference pixel points in the initial image according to a preset step length. The preset step length can be set and adjusted by a user according to an empirical value.
Furthermore, a value range of the preset step length is set, so that a user can set and adjust the preset step length in the set range. If the preset value range of the step length is 1,2 and 3, the user selects the required step length in the preset value range according to the requirement.
Referring to fig. 2 and fig. 3, fig. 2 and fig. 3 are schematic diagrams illustrating selection of reference pixels in an embodiment of an image denoising method according to the present application. In an embodiment, given a topogram image with size wb × hb (where wb is the topogram image width and hb is the topogram image height, and the specific topogram can be referred to below), taking the reference pixel search step s as 3 as an example, the calculation process of the reference pixel coordinate index array is as follows (d is the image topogram width):
firstly, defining two one-dimensional arrays which are respectively used for storing row and column coordinate index values of reference pixel points;
secondly, traversing the line coordinate range [0, hb-d ] by taking s as a step length and taking 0 as an initial point. For each traversed point i, the value is added to the row coordinate index array, and the specific process is as shown in fig. 2.
Thirdly, similarly, 0 is taken as an initial point, and s is taken as a step length to traverse the row coordinate range [0, wb-d ]. For each traversed point j, its value is added to the column coordinate index array, the process is shown in FIG. 3.
And traversing the initial image to obtain all reference pixel points according to the reference pixel point row coordinate index array and the reference pixel point column coordinate index array, and obtaining the image block to be denoised with the traversed current reference pixel point as a vertex. As mentioned above, the size of the obtained image block needs to be determined according to the noise value of the region where the current reference pixel is located. Fig. 4 may be referred to for an obtaining process of an image block corresponding to a reference pixel, where fig. 4 is a schematic diagram illustrating obtaining of an image block corresponding to a reference pixel in an embodiment of an image denoising method according to the present application. As illustrated in fig. 4, the current reference pixel point P is taken as a vertex, and it is determined that an 8 × 8 image block needs to be selected according to a given preset noise range in which the noise value of the region in which the reference pixel point is located, so that an 8 × 8 image block to be denoised is selected and obtained by taking P as a vertex as illustrated in fig. 4.
Referring to fig. 5, fig. 5 is a schematic flowchart illustrating another embodiment of an image denoising method according to the present application. The above step S140 performs noise reduction processing on the image block to be denoised based on the noise reduction threshold corresponding to each image block to be denoised to obtain a corresponding image block to be denoised, and in the current embodiment, the method includes:
s501: and carrying out frequency domain transformation on each image block to be denoised to obtain a corresponding frequency domain block.
Firstly, a preset frequency domain transformation method is adopted to carry out frequency domain transformation on each image block to be denoised so as to obtain a corresponding frequency domain block. The frequency domain transformation method at least comprises the following steps: discrete cosine transform, discrete sine transform, etc., without being limited thereto.
Further, step S501 includes: and performing frequency domain transformation on each image block to be denoised by adopting 2D Walsh Hadamard transformation to obtain a corresponding frequency domain block. In the current embodiment, the 2D walsh hadamard transform is selected to reduce the complexity of the algorithm operation and reduce the resource requirements for hardware, so that the algorithm can be applied to hardware that does not support floating point operations.
S502: and judging whether the absolute value of the pixel value of each pixel point in the frequency domain block is smaller than the noise reduction threshold corresponding to the image block to be subjected to noise reduction.
After frequency domain transformation is carried out on each image block to be denoised to obtain a corresponding frequency domain block, whether the absolute value of the pixel value of each pixel point in each frequency domain block is smaller than the denoising threshold value corresponding to each image block to be denoised is further judged.
S503: and adjusting the pixel value of the pixel point to be zero.
And if the absolute value of the pixel point is judged to be smaller than the noise reduction threshold corresponding to the image block to be subjected to noise reduction where the pixel point is located, adjusting the pixel value of the pixel point to be zero.
Otherwise, if the pixel value of the pixel point in the frequency domain block is judged to be larger than or equal to the noise reduction threshold corresponding to the image block to be noise reduced where the pixel point is located, the pixel value of the pixel point is kept unchanged. After each pixel point in the frequency domain block is determined to obtain the adjusted frequency domain block, the following step S504 is executed.
S504: and performing inverse transformation of frequency domain transformation on the adjusted frequency domain block to obtain a corresponding noise-reduced image block.
And performing corresponding inverse transformation of frequency domain transformation on the adjusted frequency domain block to obtain a corresponding noise-reduced image block.
As described above, the 2D walsh hadamard transform is applied to each image block in the frequency domain as applied in step S501, and then the inverse 2D walsh hadamard transform is applied in step S504 to obtain the corresponding noise-reduced image block.
Taking the frequency domain transform method as an example of 2D walsh hadamard transform, assume that the size of the current image block to be denoised is N × N (N is 2)nN is 2,3,4), the 2D walsh hadamard transform is calculated as follows:
in the above equation, the calculation formula of the transformation kernel g (x, u, y, v) is as follows:
where bi (z) denotes the value of the i +1 th bit of the z binary representation, u and v denote coordinates in the transformed frequency domain, and z is used to denote x, y, u and v.
For the 2D walsh inverse hadamard transform, which is similar to the 2D walsh hadamard forward transform, the calculation formula is as follows:
the definition of the transformation kernel g (x, u, y, v) in the above formula is consistent with the calculation method in the 2D walsh hadamard transformation, and the calculation formula will not be repeated here.
Referring to fig. 6, fig. 6 is a schematic flowchart illustrating a method for image denoising according to another embodiment of the present application. In the present embodiment, the step S150 of fusing the obtained noise-reduced image blocks to obtain a noise-reduced image includes:
s601: and assigning the pixel value of each pixel point in the noise-reduced image block to the corresponding pixel point in the initial image to obtain the initial noise-reduced image.
After the image block to be denoised is denoised, the pixel values of all the pixel points in the image block to be denoised are assigned to the corresponding pixel points in the initial image correspondingly, and then the initial denoised image is obtained. If there are 9 image blocks to be denoised in the initial image, after denoising is performed on the image blocks to be denoised, the pixel values of the pixel points in the denoised image obtained after denoising are assigned to the corresponding pixel points in the initial image, that is, the pixel values of the pixel points in the 9 denoised images are used to replace the pixel values of the corresponding pixel points in the initial image, so as to obtain the initial denoised image.
Further, in an embodiment, in the step S601, assigning the pixel value of each pixel point in the noise-reduced image block to a corresponding pixel point in the initial image to obtain the initial noise-reduced image, includes: and if the at least two noise-reduced image blocks contain the same pixel point, respectively acquiring the pixel values of the pixel point in the at least two noise-reduced image blocks.
After the pixel values of the pixel points in the at least two noise-reduced image blocks are respectively obtained, the final pixel values of the pixel points are determined based on the pixel values of the pixel points in the at least two noise-reduced image blocks, and the final pixel values are assigned to the corresponding pixel points in the initial image. For example, in an embodiment, the pixel point q is simultaneously located in the three noise-reduced image blocks a1, a2, and A3, the pixel values of the pixel point q in the noise-reduced image blocks a1, a2, and A3 are respectively obtained, and then the average value of the obtained three pixel values is calculated as the final pixel value of the pixel point q and is assigned to the pixel point q. In another embodiment, a pixel value of a pixel point in at least two of the noise-reduced image blocks with a high probability may also be selected as a final pixel value of the pixel point, and the obtained final pixel value is assigned to a corresponding pixel point in the initial image.
Further, please refer to fig. 7, where fig. 7 is a schematic diagram illustrating an edge widening process performed on an initial image according to an embodiment of the image denoising method of the present application. In an embodiment, before determining a plurality of reference pixel points in an initial image, the method provided by the present application further includes: and performing edge-widening processing on the initial image according to a set rule so as to increase an edge-widening area outwards on each edge of the initial image.
Specifically, for each channel of data, it is subjected to edge pruning. Taking an initial image a with an image size of w × h as an example, where w is an initial image width and h is an initial image height, and assuming that the width of the edge extension of the image is d, an edge extension image B with a size of wb × hb is calculated (where wb ═ w +2d is the edge extension image width and hb ═ h +2d is the edge extension image height), the following steps are performed:
for the region with the middle size of w × h of the edge-up image, the pixels of the initial image are assigned one by one, and the calculation process of the pixels in the central region of the edge-up image is shown in (a) of fig. 7. Therefore, the initial image is traversed, the row-column coordinate index value of the current pixel is recorded, the index position of the current pixel in the edge-up image is calculated according to the row-column coordinate index value, and the value of the current pixel is assigned to the pixel in the edge-up image. Where d is set according to the size of the initial image, and is not limited herein. E.g. d may take the width of 8 pixels.
For example, when the pixel values of the ith row (i is 0,1,2, 3.,. d-1) in the edge-extended image are assigned one by using the pixels of the 2d-i row in the edge-extended image, the process is shown in (b) in fig. 7.
Similarly, for the pixels in the hb-i (i ═ 1, 2.. times, d) th row in the edge-extended image, the pixels in the hb + i-2d-2 th row in the edge-extended image are assigned one by one, and the process is shown in (c) of fig. 7.
As shown in fig. 7 (d), for the pixels in the ith column (i ═ 0,1,2, 3.., d-1) in the edge-extended image, the pixels in the 2d-i columns in the edge-extended image are assigned one-to-one.
As illustrated in fig. 7 (f), for the pixels in the wb-i column (i ═ 1, 2.. times, d) in the edge-extended image, the pixels in the wb + i-2d-2 column in the edge-extended image are used for one-to-one assignment.
In the current embodiment, when the above edge extension processing is further performed on the initial image according to the set rule before determining the plurality of reference pixel points, the method correspondingly assigns the pixel value of each pixel point in the noise-reduced image block to the corresponding pixel point in the initial image in step S601 to obtain the initial noise-reduced image, and then the method provided in this application further includes: and cutting a corresponding edge-widening area in the initial noise-reduced image, and further reserving a part corresponding to the initial image in the initial noise-reduced image.
In the embodiment corresponding to fig. 7, by performing edge extension processing on the initial image, better noise reduction processing can be performed on the edge area of the initial image, and thus the overall effect of image noise reduction is better improved.
S602: and determining the fusion weight corresponding to each pixel point according to the pixel value of each pixel point in the assigned initial noise reduction image.
After the noise-reduced image is obtained, the fusion weight corresponding to each pixel point is further determined according to the pixel value of each pixel point in the assigned initial noise-reduced image, so that the initial noise-reduced image and the initial image are fused according to the pixel value of each pixel point, and the final noise-reduced image is obtained.
Further, please refer to fig. 8, where fig. 8 is a schematic flowchart of a further embodiment of the image denoising method according to the present application. The embodiment corresponding to fig. 8 mainly illustrates the process of determining the fusion weight. Specifically, step S602 includes:
s801: and acquiring the maximum pixel value in the initial noise reduction image.
After the initial noise-reduced image is obtained, the maximum pixel values of all pixel points in the initial noise-reduced image are obtained at first. In an embodiment, in the method provided by the present application, the maximum pixel value in the initial noise-reduced image may also be directly obtained according to the latest assignment result of each pixel point in the initial noise-reduced image block in the process of assigning the pixel value of each pixel point in the noise-reduced image block to the corresponding pixel point in the initial image.
S802: at least a first pixel threshold and a second pixel threshold are determined between zero and a maximum pixel value.
And the first pixel threshold value is smaller than the second pixel threshold value and smaller than the maximum pixel value in the initial noise-reduced image. The first pixel threshold and the second pixel threshold may be determined according to a maximum pixel value in the initial noise-reduced image, for example, a set proportion of the maximum pixel value in the initial noise-reduced image may be selected as the first pixel threshold and the second pixel threshold according to an empirical value. In another embodiment, the first pixel threshold and the second pixel threshold may also be determined according to the distribution probability of the pixel values corresponding to each pixel point in the noise-reduced image, that is, the specific value of each pixel threshold is determined according to the percentage of the interval in which each pixel value is located, for example, between the pixel values from zero to the maximum value, the end values corresponding to the first fifteen percent of the distribution with the smaller and larger pixel values may be respectively selected as the first pixel threshold, and the right end values corresponding to the fifteen percent to eighty percent of the distribution with the smaller and larger pixel values may be respectively selected as the second pixel threshold.
It is understood that, in different embodiments, a plurality of pixel thresholds may be set according to actual needs, for example, 4 pixel thresholds may be set, and correspondingly, when a plurality of pixel thresholds are set, a mapping function between each pixel threshold and a corresponding preset ratio may be further obtained according to the set pixel thresholds, which may be specifically described in the corresponding embodiments below.
S803: and determining a mapping function between the fusion weight and the pixel value according to the first pixel threshold, the first preset ratio, the second pixel value threshold and the second preset ratio.
Specifically, a mapping function between the fusion weight and the pixel value may be obtained by fitting or calculating according to the first pixel threshold, the first preset ratio, the second pixel value threshold and the second preset ratio to determine the mapping function between the fusion weight and the pixel value.
S804: and determining the fusion weight corresponding to the pixel point in the initial noise-reduced image according to the mapping function.
And determining the fusion weight corresponding to each pixel point according to the mapping function between the fusion weight and the pixel value determined in the step S803 and the pixel value corresponding to each pixel in the initial noise-reduced image.
For image fusion, the core idea is to compensate the loss of detail information by introducing initial image information. In one embodiment, assuming that the initial image is ori, the initial noise-reduced image is den, and the final noise-reduced image is res, then: res ═ α × den + (1- α) × ori.
In the above equation, α is a fusion weight, and the calculation method of the fusion weight α is shown in fig. 9.
Fig. 9 is a schematic diagram of a mapping relationship corresponding to a fusion weight and a pixel value in an embodiment of an image denoising method according to the present application. In fig. 9, α 0, α 1, α 2, p1, p2, p3, p4 and p5 are algorithm parameters, and p3 is a maximum pixel value in the initial noise-reduced image, and may specifically vary according to the input image. Wherein p3 may be the largest pixel value in the initial image in another embodiment.
In the method provided by the application, considering that the noise of the bright area and the dark area of the image is generally less, the information of the initial image is more concerned in the process of carrying out noise reduction processing on the image, and more initial images are taken during image fusion; for other areas, the information of the initial noise reduction image is concerned more, and more initial noise reduction images can be taken during image fusion, so that better noise reduction processing is performed on the image, and meanwhile, the loss of the details of image data due to noise reduction is avoided, and the accuracy of the image is ensured. In an actual scene, noise reduction processing for different scene images can be realized by regulating values of α 0, α 1, α 2 and corresponding pixel thresholds (in the current embodiment, the pixel thresholds include p1, p2, p3, p4 and p 5).
In the embodiment illustrated in fig. 9, after the maximum pixel value is obtained in the initial noise-reduced image, α 0, α 1, α 2, p1, p2, p3, p4, and p5 are further determined, then mapping functions between the fusion weight and the pixel value between pixel values 0 and p1 and mapping functions between the fusion weight and the pixel value between p1 and p4, between p4 and p5, between p5 and p2, between p2 and p3 are respectively obtained, and then fusion weights corresponding to respective pixel points in the initial noise-reduced image are respectively determined according to the respective mapping functions. In an embodiment, α 0, α 1, and α 2 are empirical values selected according to a noise value of the image, and may be adjusted according to actual needs, for example, in the current embodiment, values of α 0, α 1, and α 2 may be set to 0.1, 0.3, and 0.8 by default, values of p1, p2, p4, and p5 may be set according to a maximum pixel value p3 of the image, and specific selection manners may be, for example, p1 ═ 0.15 × p3, p2 ═ 0.85 × p3, p4 ═ 0.25 × p3, and p5 × 0.75 × p 3.
S603: and fusing the initial noise-reduced image and the initial image by utilizing the fusion weight so as to obtain the noise-reduced image.
After the fusion weight corresponding to each pixel point is determined, the initial noise-reduced image and the initial image are further fused according to the determined fusion weight to obtain the noise-reduced image.
Further, the step S603 fuses the initial noise-reduced image and the initial image by using the fusion weight to obtain a noise-reduced image, including: fusing the initial noise-reduced image and the initial image by using the following formula to obtain a noise-reduced image,
wherein, a is the fusion weight,
the pixel value of the ith pixel point in the noise reduction image,
the pixel value of the ith pixel point in the initial noise reduction image,
the pixel value of the ith pixel point in the initial image is obtained.
Referring to fig. 10, fig. 10 is a schematic flowchart illustrating a method for image denoising according to another embodiment of the present application.
Step S140 is to perform denoising processing on the image blocks to be denoised based on the denoising threshold corresponding to each image block to be denoised to obtain corresponding denoised image blocks, and the method further includes:
s1001: and acquiring pixel values of pixel points which belong to the plurality of noise-reduced image blocks in the plurality of noise-reduced image blocks simultaneously.
When the image block to be denoised is selected, different image blocks to be denoised may simultaneously include the same pixel points, and then, in the current embodiment, after the image block to be denoised is denoised to obtain the image block to be denoised, the pixel values of the pixel points belonging to the plurality of image blocks to be denoised in the plurality of image blocks to be denoised at the same time are further obtained.
S1002: and determining the final pixel value of the pixel point based on the pixel values of the pixel point in the plurality of noise-reduced image blocks, and assigning the final pixel value to the corresponding pixel point in the plurality of noise-reduced image blocks.
After the pixel values of the pixel point in the plurality of noise-reduced image blocks are obtained, the average value of the plurality of pixel values or most of the plurality of pixel values can be determined as the final pixel value of the pixel point, and the final pixel value is respectively assigned to the corresponding pixel points in the plurality of noise-reduced image blocks, so that when image block splicing and fusion are carried out, any value in the plurality of noise-reduced image blocks simultaneously including the pixel point can be selected as the pixel value of the pixel point.
Referring to fig. 11, fig. 11 is a schematic flowchart illustrating a method for image denoising according to another embodiment of the present application. In the present embodiment, the step S150 of fusing the obtained noise-reduced image blocks to obtain a noise-reduced image includes:
s1101: and determining fusion weights corresponding to the pixel points according to the pixel values of the pixel points in the noise-reduced image blocks, and fusing the noise-reduced image blocks and the image blocks to be noise-reduced according to the fusion weight ratio to obtain final noise-reduced image blocks.
In the current embodiment, the denoised image blocks and the image to be denoised are first fused to obtain final denoised image blocks, and then step S1102 is executed to aggregate the final denoised image blocks to obtain the denoised image. The determination of the fusion weight α is the same as that in the embodiment corresponding to fig. 8, and is not described herein again.
Further, please refer to fig. 12, fig. 12 is a schematic flowchart illustrating a method for image denoising according to another embodiment of the present application. In the present embodiment, the determining the fusion weight corresponding to the pixel point according to the pixel value of the pixel point in the noise-reduced image block in step S1101 further includes:
s1201: and acquiring the maximum pixel value in the noise-reduced image block.
And acquiring the maximum pixel value in the current denoised image block.
S1202: a first pixel threshold and a second pixel threshold are determined between zero and the maximum pixel value, the first pixel threshold being less than the second pixel threshold.
S1203: and determining a mapping function between the fusion weight and the pixel value according to the first pixel threshold, the first preset ratio, the second pixel threshold and the second preset ratio.
S1204: and determining the fusion weight corresponding to the pixel point in the denoised image block according to the mapping function. And determining fusion weights corresponding to all pixel points in the image block with the noise reduction function according to the obtained mapping function, and fusing the image block with the noise reduction function and the image block to be subjected to the noise reduction according to the fusion weights.
In the embodiment corresponding to fig. 12, an example of obtaining the fusion weight ratio in an embodiment where after the image blocks to be denoised are denoised to obtain the denoised image blocks, the image blocks are fused in units of image blocks is described. In the current embodiment, since the number of pixel points included in the image block is smaller than that of the complete denoised image, a smaller number of pixel threshold values may be selected, then a mapping function between the fusion weight and the pixel value is obtained, and finally the fusion weight of each pixel point may be determined according to the obtained mapping function. The relevant contents of steps S1201 to S1204 in the current embodiment may also be correspondingly described with reference to the corresponding parts of fig. 8, and are not repeated herein.
Furthermore, in an embodiment, if the image blocks are fused, for the image blocks with smaller size (i.e., the image blocks with fewer pixel points) that have been denoised, the image blocks to be denoised and the image blocks that have been denoised may be directly fused according to the preset fusion weight.
Furthermore, in another embodiment, when the image blocks are fused, for the image blocks smaller than or equal to the preset size, the corresponding preset fusion weight may be directly determined according to the average pixel value of the image block. The preset fusion weight is an empirical value obtained through testing.
Fusing the noise-reduced image block and the image block to be noise-reduced according to the fusion weight ratio in step S1101 to obtain a final noise-reduced image block, further comprising: and fusing the denoised image blocks and the image blocks to be denoised by using the following formula to obtain the final denoised image blocks.
Wherein, alpha is the fusion weight,
for the pixel value of the ith pixel point in the final noise-reduced image block,
for the pixel value of the ith pixel point in the noise-reduced image block,
the pixel value of the ith pixel point in the image block to be denoised is the same as that in the embodiment corresponding to fig. 8, and the determination of the fusion weight α is not described herein again.
S1102: and aggregating the final noise-reduced image blocks to obtain a noise-reduced image.
And aggregating the final noise-reduced image blocks, namely simply aggregating the final noise-reduced image blocks according to the pixel coordinates of the final noise-reduced image blocks to obtain the noise-reduced image.
Referring to fig. 13, fig. 13 is a schematic structural diagram of an embodiment of an image denoising device according to the present application. In the current embodiment, the apparatus 1300 for image denoising provided herein includes a processor 1301 and a memory 1302 coupled thereto. The apparatus 1300 for image noise reduction may perform the method for image noise reduction described in fig. 1 to 12 and any corresponding embodiment thereof.
The memory 1302 includes a local storage (not shown) and stores a computer program, and the computer program can implement the method described in any of the embodiments of fig. 1 to 12 and the corresponding embodiments.
A processor 1301 is coupled to the memory 1302, the processor 1301 being configured to run a computer program to perform the method of image denoising as described above with reference to fig. 1 to 12 and any corresponding embodiment thereof.
Further, in another embodiment, the apparatus 1300 for image noise reduction provided by the present application may further include a communication circuit (not shown), which is connected to the processor 1301 and is configured to perform data interaction with an external terminal device under the control of the processor 1301 to obtain initial image data or instruction data. The instruction data at least comprises a computer program upgrading instruction and a data packet required by computer program upgrading.
Referring to fig. 14, fig. 14 is a schematic structural diagram of an embodiment of a memory device according to the present application. The storage device 1400 stores a computer program 1401 capable of being executed by a processor, and the computer program 1401 is used for implementing the method for reducing noise of an image as described in any one of the embodiments of fig. 1 to 12 and corresponding embodiments. Specifically, the storage 1400 may be one of a memory, a personal computer, a server, a network device, or a usb disk, and is not limited herein.
The above description is only for the purpose of illustrating embodiments of the present application and is not intended to limit the scope of the present application, and all modifications of equivalent structures and equivalent processes, which are made by the contents of the specification and the drawings of the present application or are directly or indirectly applied to other related technical fields, are also included in the scope of the present application.
Claims (15)
1. A method of image noise reduction, comprising:
acquiring an initial image, and determining a plurality of reference pixel points in the initial image;
respectively calculating the noise value of the region where each reference pixel point is located in the initial image;
selecting an image block to be denoised containing the reference pixel points from the initial image according to the noise value of the area where each reference pixel point is located, and acquiring a denoising threshold corresponding to the image block to be denoised;
based on a noise reduction threshold corresponding to each image block to be subjected to noise reduction, performing noise reduction on the image block to be subjected to noise reduction to obtain a corresponding image block subjected to noise reduction;
fusing the obtained noise-reduced image blocks to obtain a noise-reduced image;
and the distance between the adjacent reference pixel points is smaller than or equal to the size of the image block to be denoised.
2. The method according to claim 1, wherein the selecting an image block to be denoised containing the reference pixel points in the initial image according to the noise value of the region where each reference pixel point is located, and obtaining the denoising threshold corresponding to the image block to be denoised comprises:
determining a preset noise range to which the noise value of the area where the reference pixel point is located belongs, selecting an image block to be denoised in the initial image, wherein the image block to be denoised comprises the reference pixel point and has a preset block size corresponding to the preset noise range, and acquiring a denoising threshold corresponding to the preset noise range as a denoising threshold corresponding to the image block to be denoised.
3. The method according to claim 2, wherein the larger the noise in the preset noise range, the larger the corresponding preset block size, and the larger the corresponding noise reduction threshold;
the upper limit value and/or the lower limit value of the preset noise range are/is determined according to the bit width of the initial image;
the number of the preset noise ranges is greater than or equal to three.
4. The method according to claim 1, wherein the image block to be denoised is an image block corresponding to the reference pixel point as a vertex;
the determining a plurality of reference pixel points in the initial image includes:
searching a plurality of reference pixel points in the initial image according to a preset step length;
the respectively calculating the noise value of the region where each reference pixel point is located in the initial image comprises:
determining a region with the reference pixel point as a vertex and the size as the size of a preset region, and obtaining the noise value of the region where the reference point is located based on the pixel values of the pixel points in the region.
5. The method according to claim 1, wherein the denoising the image blocks to be denoised based on the denoising threshold corresponding to each image block to be denoised to obtain corresponding denoised image blocks comprises:
performing frequency domain transformation on each image block to be denoised to obtain a corresponding frequency domain block;
judging whether the absolute value of the pixel value of each pixel point in the frequency domain block is smaller than the noise reduction threshold corresponding to the image block to be subjected to noise reduction;
if so, adjusting the pixel value of the pixel point to be zero;
and performing inverse transformation of the frequency domain transformation on the adjusted frequency domain block to obtain a corresponding noise-reduced image block.
6. The method according to claim 5, wherein said frequency-domain transforming each of the image blocks to obtain a corresponding frequency-domain block comprises:
and carrying out frequency domain transformation on each image block by adopting Walsh Hadamard transform to obtain a corresponding frequency domain block.
7. The method according to claim 1, wherein said fusing the obtained denoised image blocks to obtain a denoised image comprises:
assigning the pixel value of each pixel point in the noise-reduced image block to the corresponding pixel point in the initial image to obtain an initial noise-reduced image;
determining fusion weights corresponding to all pixel points according to the pixel values of all the pixel points in the initial noise reduction image after assignment;
and fusing the initial noise-reduced image and the initial image by using the fusion weight to obtain the noise-reduced image.
8. The method according to claim 7, wherein the determining the fusion weight corresponding to each pixel point according to the assigned pixel value of each pixel point in the initial noise-reduced image comprises:
acquiring a maximum pixel value in the initial noise-reduced image;
determining at least a first pixel threshold and a second pixel threshold between zero and the maximum pixel value, wherein the first pixel threshold is less than the second pixel threshold;
determining a mapping function between the fusion weight and the pixel value according to the first pixel threshold value, the first preset ratio, the second pixel value threshold value and the second preset ratio;
determining the fusion weight corresponding to a pixel point in the initial noise-reduced image according to the mapping function;
the fusing the initial noise-reduced image and the initial image by using the fusion weight to obtain the noise-reduced image includes:
fusing the initial noise-reduced image and the initial image using the following formula to obtain the noise-reduced image,
wherein, alpha is the fusion weight,
the pixel value of the ith pixel point in the noise reduction image,
the pixel value of the ith pixel point in the initial noise reduction image,
the pixel value of the ith pixel point in the initial image is obtained.
9. The method of claim 7, wherein prior to determining a number of reference pixels in the initial image, the method further comprises:
performing edge-widening processing on the initial image according to a set rule so as to increase an edge-widening area outwards on each edge of the initial image;
after assigning the pixel value of each pixel point in the noise-reduced image block to the corresponding pixel point in the initial image to obtain the initial noise-reduced image, the method further includes:
and cutting the corresponding edge-rubbing area in the initial noise-reduced image.
10. The method of claim 8, wherein assigning the pixel value of each pixel in the noise-reduced image block to a corresponding pixel in the initial image to obtain an initial noise-reduced image comprises:
if at least two of the noise-reduced image blocks contain the same pixel point, respectively acquiring pixel values of the pixel point in the at least two noise-reduced image blocks;
and determining the final pixel value of the pixel point based on the pixel values of the pixel point in at least two of the noise-reduced image blocks, and assigning the final pixel value to the corresponding pixel point in the initial image.
11. The method according to claim 1, wherein after performing denoising processing on the image blocks to be denoised based on the denoising threshold corresponding to each image block to be denoised to obtain corresponding denoised image blocks, the method further comprises:
acquiring pixel values of pixel points which belong to a plurality of noise-reduced image blocks in the plurality of noise-reduced image blocks simultaneously;
and determining the final pixel value of the pixel point based on the pixel values of the pixel point in the plurality of image blocks with the noise reduced, and assigning the final pixel value to the corresponding pixel point in the plurality of target image blocks.
12. The method according to claim 1, wherein said fusing the obtained denoised image blocks to obtain a denoised image comprises:
determining fusion weights corresponding to the pixels according to pixel values of the pixels in the noise-reduced image blocks, and fusing the noise-reduced image blocks and the image blocks to be noise-reduced according to the fusion weight ratio to obtain final noise-reduced image blocks;
and aggregating the final noise-reduced image blocks to obtain the noise-reduced image.
13. The method according to claim 12, wherein the determining the fusion weight corresponding to the pixel point according to the pixel value of the pixel point in the noise-reduced image block includes:
acquiring a maximum pixel value in the noise-reduced image block;
determining a first pixel threshold and a second pixel threshold between zero and the maximum pixel value, the first pixel threshold being less than the second pixel threshold;
determining a mapping function between the fusion weight and the pixel value according to the first pixel threshold value, the first preset ratio, the second pixel threshold value and the second preset ratio;
determining the fusion weight corresponding to the pixel point in the denoised image block according to the mapping function;
fusing the denoised image blocks and the image blocks to be denoised according to the fusion weight ratio to obtain the final denoised image blocks, comprising:
fusing the denoised image blocks and the image blocks to be denoised by using the following formula to obtain the final denoised image blocks,
wherein, alpha is the fusion weight,
for the pixel value of the ith pixel point in the final noise-reduced image block,
for the pixel value of the ith pixel point in the noise-reduced image block,
and the pixel value of the ith pixel point in the image block to be denoised is obtained.
14. An apparatus for image noise reduction, the apparatus comprising a memory and a processor coupled, wherein,
the memory includes local storage and stores a computer program;
the processor is configured to run the computer program to perform the method of any one of claims 1 to 13.
15. A storage device, characterized in that the storage device stores a computer program executable by a processor for implementing the method of any one of claims 1-13.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010292363.2A CN113536214A (en) | 2020-04-14 | 2020-04-14 | Image noise reduction method and device and storage device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010292363.2A CN113536214A (en) | 2020-04-14 | 2020-04-14 | Image noise reduction method and device and storage device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113536214A true CN113536214A (en) | 2021-10-22 |
Family
ID=78119993
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010292363.2A Pending CN113536214A (en) | 2020-04-14 | 2020-04-14 | Image noise reduction method and device and storage device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113536214A (en) |
-
2020
- 2020-04-14 CN CN202010292363.2A patent/CN113536214A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kim et al. | Optimized contrast enhancement for real-time image and video dehazing | |
| US9483835B2 (en) | Depth value restoration method and system | |
| CN109191395B (en) | Image contrast enhancement method, device, equipment and storage medium | |
| CN106331723B (en) | Video frame rate up-conversion method and system based on motion region segmentation | |
| WO2018082185A1 (en) | Image processing method and device | |
| EP2721583B1 (en) | Content-adaptive method and apparatus for determining optical flow | |
| CN109743473A (en) | Video image 3 D noise-reduction method, computer installation and computer readable storage medium | |
| US9449371B1 (en) | True motion based temporal-spatial IIR filter for video | |
| CN104376542A (en) | Image enhancement method | |
| CN112734654B (en) | Image processing method, device, equipment and storage medium | |
| CN109214996B (en) | Image processing method and device | |
| EP2659452A1 (en) | Method and apparatus for providing a mechanism for gesture recognition | |
| US8995755B2 (en) | Two-dimensional to stereoscopic conversion systems and methods | |
| US20170148212A1 (en) | Color-based dynamic sub-division to generate 3d mesh | |
| CN113011433B (en) | Filtering parameter adjusting method and device | |
| CN113536214A (en) | Image noise reduction method and device and storage device | |
| US9154671B2 (en) | Image processing apparatus, image processing method, and program | |
| EP2930687A2 (en) | Image segmentation using blur and color | |
| CN115471413A (en) | Image processing method and device, computer readable storage medium and electronic device | |
| CN110363723B (en) | Image processing method and device for improving image boundary effect | |
| CN112200730B (en) | Image filtering processing method, device, equipment and storage medium | |
| CN112150532A (en) | Image processing method and device, electronic equipment and computer readable medium | |
| CN111383183B (en) | Image edge enhancement method and device and computer storage medium | |
| CN112950515A (en) | Image processing method and device, computer readable storage medium and electronic device | |
| CN112950590A (en) | Terrain image adaptability analysis method and device and readable storage medium |
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 |