CN110610460B - Combined filtering module denoising method guided by spatial entropy rate - Google Patents
Combined filtering module denoising method guided by spatial entropy rate Download PDFInfo
- Publication number
- CN110610460B CN110610460B CN201910677319.0A CN201910677319A CN110610460B CN 110610460 B CN110610460 B CN 110610460B CN 201910677319 A CN201910677319 A CN 201910677319A CN 110610460 B CN110610460 B CN 110610460B
- Authority
- CN
- China
- Prior art keywords
- filter
- frequency noise
- image
- gaussian
- entropy rate
- 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.)
- Active
Links
- 238000001914 filtration Methods 0.000 title claims abstract description 44
- 238000000034 method Methods 0.000 title claims abstract description 30
- 230000002146 bilateral effect Effects 0.000 claims description 18
- 238000001035 drying Methods 0.000 claims description 16
- 238000007781 pre-processing Methods 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 3
- 239000000126 substance Substances 0.000 claims 1
- 230000006870 function Effects 0.000 description 22
- 230000000694 effects Effects 0.000 description 5
- 238000004321 preservation Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 230000002349 favourable 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
- G06T5/10—Image enhancement or restoration by non-spatial domain filtering
-
- G06T5/70—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
- G06T2207/20028—Bilateral filtering
Abstract
The invention provides a combined filtering module denoising method with space entropy as guidance, which is characterized in that a local entropy function is set for a high-frequency noise image, and then a filter is set by combining the local entropy function and a Gaussian space kernel to carry out filtering and denoising on the preprocessed high-frequency noise image. The combined filtering module denoising method using the spatial entropy as the guide provided by the invention is finally used for filtering and dehumidifying a filter, and the edge protection characteristic of the Gaussian spatial kernel and the high representation characteristic of the local entropy function on image information are fully utilized through the combination of the local entropy function and the Gaussian spatial kernel, so that the characteristic information of an original image can be fully reserved while the high-frequency noise image is dehumidified.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a combined filtering module denoising method guided by a space entropy rate.
Background
Image noise refers to unnecessary or redundant disturbance information present in an image. The presence of noise seriously affects the quality of the image and therefore the noise must be reduced before image processing. Image filtering strives to remove any noise or spurious information while maintaining true information at all frequencies. The existing denoising methods generally comprise two types, namely learning methods and non-learning methods. The learning-type method usually needs a certain priori knowledge as a guide, and obtains a specific denoising model through many model training times, and the method faces two difficulties. Firstly, many problems in real life do not have ready-made objects to provide training data for the model, and secondly, the training calculation complexity of the model is very high, and timely application cannot be achieved. Therefore, many image denoising problems still need to research an efficient non-learning method.
The non-learning denoising method generally refers to a classical denoising filter, and the image denoising can be achieved without filter iteration operation. In the method, a Gaussian filter and a double-Gaussian filter represented by a Gaussian function are most widely applied and have certain denoising and image edge protection effects, but a single Gaussian filter denoises high-frequency noise with poor effect. Later, many scholars propose a combined filtering idea, and combined filtering is performed by combining advanced filtering such as Gaussian filtering, wiener filtering and wavelet filtering, and in the existing combined filtering method, due to the fact that weight combination is adopted, the weight is difficult to obtain the optimum, meanwhile, the protection on the edge of an image is not fully considered, and the limitations cause poor removing effect on the high-frequency noise of the image in reality.
Disclosure of Invention
Based on the technical problems in the background art, the invention provides a combined filtering module denoising method using a space entropy rate as a guide.
The invention provides a combined filtering module denoising method with space entropy as guidance, which is characterized in that a local entropy function is set for a high-frequency noise image, and then a filter is set by combining the local entropy function and a Gaussian space kernel to carry out filtering and denoising on the preprocessed high-frequency noise image.
Preferably, the method specifically comprises the following steps:
s1, preprocessing a high-frequency noise image through a bilateral filter;
s2, defining a local entropy rate function, and setting a Gaussian entropy rate filter by combining a Gaussian space and the local entropy rate function;
s3, taking the output of the Gaussian entropy rate filter as the guide input of a bilateral range kernel, and constructing a combined filtering and drying module; the scale of the bilateral range kernel is larger than the scale of the bilateral filter;
and S4, filtering and drying the preprocessed image through a combined filtering and drying module.
Preferably, the bilateral filter in step S1 is:
wherein omega p Is a (2r + 1) × (2r + 1) region of radius r centered at the point p in the high-frequency noise image, and σ represents a region Ω p Q is the region omega p Q-p is the two-dimensional Euclidean distance between the pixel points p and q in the high-frequency noise image, I p And I q Pixel values of pixel points p and q in the high-frequency noise image are respectively; τ is a normalization function, g r ,g σ A gaussian spatial kernel and a range kernel function.
Preferably, step S1 specifically includes: and calculating and updating the pixel value of each pixel point in the high-frequency noise image through a bilateral filter.
Preferably, in step S2, the local entropy rate function is:
p i =1/n k the pixel point q has a region omega p Probability of other corresponding pixel points, n k Is an integer interval of pixel values [ a ] k ,b k ],a k ,b k ∈[0,255]。
Preferably, in step S2, the gaussian entropy filter is:
wherein, the first and the second end of the pipe are connected with each other,n is the region omega p The number of inner pixel points.
Preferably, in step S3, the combined filtering and drying module is:
preferably, step S4 specifically includes: and calculating and updating the pixel value of each pixel point in the preprocessed image through a combined filtering and drying module.
The combined filtering module denoising method using the spatial entropy as the guide provided by the invention is finally used for filtering and dehumidifying, and the edge-preserving characteristic of the Gaussian space kernel and the high-characterization characteristic of the local entropy function to the image information are fully utilized through the combination of the local entropy function and the Gaussian space kernel, so that the characteristic information of the original image can be fully preserved while the high-frequency noise image is dehumidified.
In the invention, firstly, the high-frequency noise image is preprocessed, then a filter with the function of reserving the characteristics of the original image is arranged, and then the image is dried by the arranged filter; the image dryness removal from the three stages of preprocessing to feature preservation to dryness removal is realized, and the problem of noise pollution of the real image can be timely and effectively processed. According to the method, the modularization processing in the image filtering and drying process is realized through the local entropy rate function. Therefore, the image high-frequency noise is smoothed through the image modularization denoising, and the optimal denoising effect is realized.
Drawings
FIG. 1 is a flow chart of a denoising method of a combined filter module guided by a spatial entropy rate according to the present invention;
fig. 2 is a detailed flowchart of a denoising method of a combined filtering module guided by a spatial entropy rate according to the present invention.
Detailed Description
Referring to fig. 1, the combined filtering module denoising method using spatial entropy as a guide provided by the present invention sets a local entropy function for a high frequency noise image, and then sets a filter to filter and remove the preprocessed high frequency noise image by combining the local entropy function and a gaussian spatial kernel.
In the embodiment, the filter finally used for filtering and dehumidifying makes full use of the edge-preserving characteristic of the gaussian space kernel and the characteristic of high representation of the local entropy rate function on the image information through the combination of the local entropy rate function and the gaussian space kernel, and is favorable for fully preserving the characteristic information of the original image while dehumidifying the high-frequency noise image.
In the embodiment, firstly, the high-frequency noise image is preprocessed, then a filter with the function of retaining the characteristics of the original image is arranged, and then the image is dried through the arranged filter; the image dryness removal from the three stages of preprocessing to feature preservation to dryness removal is realized, and the problem of noise pollution of the real image can be timely and effectively processed. Referring to fig. 2, the combined filtering module denoising method guided by the spatial entropy rate in this embodiment specifically includes the following steps:
s1, preprocessing a high-frequency noise image through a bilateral filter.
Specifically, the bilateral filter in this step is:
wherein omega p Is a (2r + 1) × (2r + 1) region of radius r centered at the point p in the high-frequency noise image, and σ represents a region Ω p Q is the region omega p In any pixel point, q-p is the two-dimensional Euclidean distance between the pixel points p and q in the high-frequency noise image, I p And I q Pixel values of pixel points p and q in the high-frequency noise image are respectively; τ is a normalization function, g r ,g σ A gaussian spatial kernel and a range kernel function.
When the step is specifically realized, the pixel value of each pixel point in the high-frequency noise image is calculated and updated through the bilateral filter, and the high-frequency noise image is preprocessed.
In this step, the region Ω is passed p The introduction of the method lays a foundation for image modularization processing in the image denoising process, so that the high-frequency noise of the image is smoothed through the image modularization denoising, and the optimal denoising effect is realized.
And S2, defining a local entropy rate function, and setting a Gaussian entropy rate filter by combining a Gaussian space and the local entropy rate function.
Specifically, in this step, the local entropy rate function is:
p i =1/n k the pixel point q has a region omega p Probability of other corresponding pixel points, n k Is an integer interval of pixel values [ a k ,b k ],a k ,b k ∈[0,255]。
In this step, the gaussian entropy rate filter is:
wherein, the first and the second end of the pipe are connected with each other,n is the region omega p The number of inner pixel points. In the present embodiment, the term "E" is used Ω Classify the pixel points so that the region omega p The internal pixel points are divided into two types to calculate the local Gaussian space entropy rate so as to reduce the calculation amount and be beneficial to improving the calculation efficiency.
S3, taking the output of the Gaussian entropy rate filter as the guide input of a bilateral range kernel, and constructing a combined filtering and drying module; the scale of the bilateral range kernel is larger than the scale of the bilateral filter.
Specifically, the combined filtering and drying module comprises:
and S4, filtering and drying the preprocessed image through the combined filtering and drying module, namely calculating and updating the pixel value of each pixel point in the preprocessed image through the combined filtering and drying module, and acquiring the image with the updated pixel value as the dried image.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art should be considered to be within the technical scope of the present invention, and the technical solutions and the inventive concepts thereof according to the present invention are equivalent to or changed within the technical scope of the present invention.
Claims (4)
1. A combined filtering module denoising method taking space entropy as guidance is characterized in that a local entropy function is set for a high-frequency noise image, and then a filter is set by combining the local entropy function and a Gaussian space kernel to carry out filtering and dryness removal on the preprocessed high-frequency noise image;
the method specifically comprises the following steps:
s1, preprocessing a high-frequency noise image through a bilateral filter;
s2, defining a local entropy rate function, and setting a Gaussian entropy rate filter by combining a Gaussian space and the local entropy rate function;
s3, taking the output of the Gaussian entropy rate filter as the guiding input of a bilateral range kernel, and constructing a combined filtering and drying module; the scale of the bilateral range kernel is larger than the scale of the bilateral filter;
s4, filtering and drying the preprocessed image through a combined filtering and drying module;
in step S2, the local entropy rate function is:
p i =1/n k the pixel point q has a region omega p Probability of other corresponding pixel points, n k Is an integer interval of pixel values [ a ] k ,b k ],a k ,b k ∈[0,255];
In step S2, the gaussian entropy rate filter is:
wherein the content of the first and second substances,n is the region omega p The number of intra pixel points;
in step S3, the combined filtering and drying module is:
2. the denoising method of the combined filtering module guided by the spatial entropy rate as claimed in claim 1, wherein the bilateral filter in step S1 is:
wherein omega p Is a (2r + 1) × (2r + 1) region of radius r centered at the point p in the high-frequency noise image, and σ represents a region Ω p Q is the region omega p In any pixel point, q-p is the two-dimensional Euclidean distance between the pixel points p and q in the high-frequency noise image, I p And I q Respectively representing pixel values of pixel points p and q in the high-frequency noise image; τ is a normalization function, g r ,g σ A gaussian spatial kernel and a range kernel function.
3. The denoising method of the combined filtering module guided by the spatial entropy rate as claimed in claim 2, wherein the step S1 specifically comprises: and calculating and updating the pixel value of each pixel point in the high-frequency noise image through a bilateral filter.
4. The denoising method of the combined filter module guided by the spatial entropy rate as claimed in claim 1, wherein the step S4 is specifically: and calculating and updating the pixel value of each pixel point in the preprocessed image through a combined filtering and drying module.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910677319.0A CN110610460B (en) | 2019-07-25 | 2019-07-25 | Combined filtering module denoising method guided by spatial entropy rate |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910677319.0A CN110610460B (en) | 2019-07-25 | 2019-07-25 | Combined filtering module denoising method guided by spatial entropy rate |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110610460A CN110610460A (en) | 2019-12-24 |
CN110610460B true CN110610460B (en) | 2022-10-14 |
Family
ID=68889675
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910677319.0A Active CN110610460B (en) | 2019-07-25 | 2019-07-25 | Combined filtering module denoising method guided by spatial entropy rate |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110610460B (en) |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109087266A (en) * | 2018-08-09 | 2018-12-25 | 苏州大学 | A kind of image speckle iteration reduction method of combination bilateral filtering |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101525251B1 (en) * | 2008-12-29 | 2015-06-03 | 주식회사 동부하이텍 | Noise filter |
-
2019
- 2019-07-25 CN CN201910677319.0A patent/CN110610460B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109087266A (en) * | 2018-08-09 | 2018-12-25 | 苏州大学 | A kind of image speckle iteration reduction method of combination bilateral filtering |
Non-Patent Citations (1)
Title |
---|
全变分引导的双边滤波图像去噪方法;芦碧波等;《光学技术》;20180315(第02期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN110610460A (en) | 2019-12-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Zhang et al. | Learning deep CNN denoiser prior for image restoration | |
CN109035163B (en) | Self-adaptive image denoising method based on deep learning | |
CN108921800B (en) | Non-local mean denoising method based on shape self-adaptive search window | |
CN111275643B (en) | Real noise blind denoising network system and method based on channel and space attention | |
CN112233026A (en) | SAR image denoising method based on multi-scale residual attention network | |
CN106548176B (en) | Finger vein image enhancement method based on self-adaptive guide filtering | |
CN109934826A (en) | A kind of characteristics of image dividing method based on figure convolutional network | |
CN103208097A (en) | Principal component analysis collaborative filtering method for image multi-direction morphological structure grouping | |
CN102063708A (en) | Image denoising method based on Treelet and non-local means | |
CN106204482A (en) | Based on the mixed noise minimizing technology that weighting is sparse | |
CN108961188A (en) | A kind of image quality Enhancement Method, system and device | |
CN114429151A (en) | Magnetotelluric signal identification and reconstruction method and system based on depth residual error network | |
CN113392728B (en) | Target detection method based on SSA sharpening attention mechanism | |
CN107292855A (en) | A kind of image de-noising method of the non local sample of combining adaptive and low-rank | |
CN110610460B (en) | Combined filtering module denoising method guided by spatial entropy rate | |
CN111639555B (en) | Finger vein image noise accurate extraction and adaptive filtering denoising method and device | |
CN110176021B (en) | Level set image segmentation method and system for saliency information combined with brightness correction | |
CN115994870B (en) | Image processing method for enhancing denoising | |
CN107085839A (en) | SAR image method for reducing speckle with sparse coding is strengthened based on texture | |
CN108876711B (en) | Sketch generation method, server and system based on image feature points | |
CN103955893A (en) | Image denoising method based on separable total variation model | |
CN104616266B (en) | A kind of noise variance estimation method based on broad sense autoregression heteroscedastic model | |
CN110930339A (en) | Aviation and remote sensing image defogging method based on NSCT domain | |
CN112927169B (en) | Remote sensing image denoising method based on wavelet transformation and improved weighted kernel norm minimization | |
CN111709962B (en) | Image contour and texture feature decomposition method based on anisotropic L0 gradient sparse representation and DCT (discrete cosine transform) |
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 |