CN111340726A - Image auxiliary denoising method based on supervised machine learning - Google Patents
Image auxiliary denoising method based on supervised machine learning Download PDFInfo
- Publication number
- CN111340726A CN111340726A CN202010119565.7A CN202010119565A CN111340726A CN 111340726 A CN111340726 A CN 111340726A CN 202010119565 A CN202010119565 A CN 202010119565A CN 111340726 A CN111340726 A CN 111340726A
- Authority
- CN
- China
- Prior art keywords
- factor
- denoising
- image
- noise
- denoised
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000010801 machine learning Methods 0.000 title claims abstract description 16
- 230000006870 function Effects 0.000 claims description 18
- 230000009466 transformation Effects 0.000 claims description 7
- 238000000354 decomposition reaction Methods 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 3
- 238000012843 least square support vector machine Methods 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 claims description 3
- 238000012706 support-vector machine Methods 0.000 abstract description 3
- 230000000694 effects Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 238000011426 transformation method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Medical Informatics (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Image Processing (AREA)
Abstract
The invention discloses an image auxiliary denoising method based on supervised machine learning, which adopts a non-subsampled contourlet transform method to convert a noise-containing image into a multi-scale and multi-directional transform sub-band in a Shearlet domain so as to divide a low-frequency sub-band factor without noise and a high-frequency sub-band factor with noise; the method comprises the steps of classifying factors needing to reserve edges and textures and to-be-denoised factors needing to be denoised in high-frequency sub-band factors through a support vector machine algorithm of a least square mode under supervised machine learning, finally removing noise through a soft threshold method to obtain a final auxiliary denoised image, and therefore the method has good denoising capacity, effectively removes noise parts, screens data needing to be denoised in the image layer by layer, ensures image details, saves running time and has high efficiency in the field of image processing.
Description
Technical Field
The invention relates to the field of image signal processing, in particular to an image auxiliary denoising method based on supervised machine learning.
Background
Images are affected by noise, blurring, and the like during recording, transmission, processing, and synthesis, resulting in degradation of image quality. Therefore, in the image processing flow, denoising is the core content of image processing. The noise part and the signal detail of the noise image are dispersed in a high-frequency area and are not easy to distinguish. However, the traditional image-assisted denoising method has the problems of poor denoising effect, less image detail information, long time consumption and the like. Therefore, finding a way to effectively denoise and ensure the image details remains the focus of current academic research.
Disclosure of Invention
Aiming at the defects in the prior art, the image auxiliary denoising method based on the supervised machine learning solves the problems of poor denoising effect, low detail restoration degree and long operation time in the traditional method.
In order to achieve the purpose of the invention, the invention adopts the technical scheme that: an image auxiliary denoising method based on supervised machine learning comprises the following steps:
s1, performing non-subsampled contourlet transformation on the image containing the noise to obtain a high-frequency sub-band factor and a low-frequency sub-band factor;
s2, constructing a least square support vector machine through a least square linear system by using a quadratic programming method, and classifying the high-frequency sub-band factors to obtain edge texture factors and to-be-denoised factors;
s3, carrying out denoising operation on the factor to be denoised by a soft threshold method to obtain a denoising factor;
and S4, splicing the denoising factor, the edge texture factor and the low-frequency subband factor in the Shearlet domain according to coordinates, and performing non-downsampling Shearlet inverse transformation to obtain an auxiliary denoising image.
Further, the step S1 includes the following steps:
s11, carrying out non-downsampling pyramid decomposition on the image containing the noise to obtain multi-scale features;
and S12, performing frequency plane segmentation and directional decomposition on the multi-scale features through a non-downsampling direction filtering device group to obtain high-frequency subband factors and low-frequency subband factors.
Further, the step S3 includes the following steps:
s31, estimating the preliminary noise standard deviation sigma according to the high-frequency sub-band factor by the following formula:
wherein, yijFor a factor value to be denoised of an image containing noise in the ith layer scale and the jth direction of a Shearlet domain, the Median () is a Median function;
s32, according to the factor to be denoised and the preliminary noise standard deviation sigma, estimating a threshold coefficient sigma through the following formulasth:
Wherein σyIn order to be the standard deviation of the noise,is the noise standard deviation sigma under the current target scale KyThe value range of the front target scale K is in a closed interval [1, N ]]N is the maximum scale, and Max () is the maximum function;
s33, judgmentWhether or not it is greater than a threshold coefficient sigmathIf yes, go to step S24, otherwise, go to step S25;
s34, according to the ideal threshold value ToptCarrying out image-assisted denoising on the factor to be denoised through a soft threshold function to obtain a denoising factor;
s35, according to Bayes lower threshold TLBAnd carrying out image-assisted denoising on the factor to be denoised through a soft threshold function to obtain a denoising factor.
Further, the step S34 is the ideal threshold ToptThe expression of (a) is:
Topt=argmin(r(T)) (4)
wherein r () is the mean square error function of the denoising factor and the true value of the factor without being affected by noise, and argmin (r (T)) represents the computer iteration process of the value of the threshold T when the r () function is minimized.
Further, the step S34 is based on bayesian lower threshold TLBThe expression of (a) is:
the invention has the beneficial effects that: adopting a non-downsampling contourlet transformation method to convert the noise-containing image into a multi-scale multidirectional transformation sub-band in a Shearlet domain, so as to divide a low-frequency sub-band factor without noise and a high-frequency sub-band factor with noise; the method comprises the steps of classifying factors needing to reserve edges and textures and to-be-denoised factors needing to be denoised in high-frequency sub-band factors through a support vector machine algorithm of a least square mode under supervised machine learning, finally removing noise through a soft threshold method to obtain a final auxiliary denoised image, and therefore the method has good denoising capacity, effectively removes noise parts, screens data needing to be denoised in the image layer by layer, ensures image details, saves running time and has high efficiency in the field of image processing.
Drawings
Fig. 1 is a schematic flow chart of an image-assisted denoising method based on supervised machine learning.
Detailed Description
The following description of the embodiments of the present invention is provided to facilitate the understanding of the present invention by those skilled in the art, but it should be understood that the present invention is not limited to the scope of the embodiments, and it will be apparent to those skilled in the art that various changes may be made without departing from the spirit and scope of the invention as defined and defined in the appended claims, and all matters produced by the invention using the inventive concept are protected.
As shown in fig. 1, in an embodiment of the present invention, an image-assisted denoising method based on supervised machine learning includes the following steps:
s1, performing non-subsampled contourlet transformation on the image containing the noise to obtain a high-frequency sub-band factor and a low-frequency sub-band factor;
s2, constructing a least square support vector machine through a least square linear system by using a quadratic programming method, and classifying the high-frequency sub-band factors to obtain edge texture factors and to-be-denoised factors;
s3, carrying out denoising operation on the factor to be denoised by a soft threshold method to obtain a denoising factor;
and S4, splicing the denoising factor, the edge texture factor and the low-frequency subband factor in the Shearlet domain according to coordinates, and performing non-downsampling Shearlet inverse transformation to obtain an auxiliary denoising image.
Step S1 includes the following steps:
s11, carrying out non-downsampling pyramid decomposition on the image containing the noise to obtain multi-scale features;
and S12, performing frequency plane segmentation and directional decomposition on the multi-scale features through a non-downsampling direction filtering device group to obtain high-frequency subband factors and low-frequency subband factors.
Step S3 includes the following steps:
s31, estimating the preliminary noise standard deviation sigma according to the high-frequency sub-band factor by the following formula:
wherein, yijFor a factor value to be denoised of an image containing noise in the ith layer scale and the jth direction of a Shearlet domain, the Median () is a Median function;
s32, according to the factor to be denoised and the preliminary noise standard deviation sigma, estimating a threshold coefficient sigma through the following formulasth:
Wherein σyIn order to be the standard deviation of the noise,is the noise standard deviation sigma under the current target scale KyThe value range of the front target scale K is in a closed interval [1, N ]]N is the maximum scale, and Max () is the maximum function;
s33, judgmentWhether or not it is greater than a threshold coefficient sigmathIf yes, go to step S24, otherwise, go to step S25;
s34, according to the ideal threshold value ToptCarrying out image-assisted denoising on the factor to be denoised through a soft threshold function to obtain a denoising factor;
s35, according to Bayes lower threshold TLBAnd carrying out image-assisted denoising on the factor to be denoised through a soft threshold function to obtain a denoising factor.
Step S34 ideal threshold ToptThe expression of (a) is:
Topt=argmin(r(T)) (4)
wherein r () is the mean square error function of the denoising factor and the true value of the factor without being affected by noise, and argmin (r (T)) represents the computer iteration process of the value of the threshold T when the r () function is minimized.
Step S34 Bayesian lower threshold TLBThe expression of (a) is:
the method adopts a non-downsampling contourlet transform method to convert a noise-containing image into a multi-scale and multi-direction transform sub-band in a Shearlet domain, so as to divide a low-frequency sub-band factor without noise and a high-frequency sub-band factor with noise; the method comprises the steps of classifying factors needing to reserve edges and textures and to-be-denoised factors needing to be denoised in high-frequency sub-band factors through a support vector machine algorithm of a least square mode under supervised machine learning, finally removing noise through a soft threshold method to obtain a final auxiliary denoised image, and therefore the method has good denoising capacity, effectively removes noise parts, screens data needing to be denoised in the image layer by layer, ensures image details, saves running time and has high efficiency in the field of image processing.
Claims (5)
1. An image auxiliary denoising method based on supervised machine learning is characterized by comprising the following steps:
s1, performing non-subsampled contourlet transformation on the image containing the noise to obtain a high-frequency sub-band factor and a low-frequency sub-band factor;
s2, constructing a least square support vector machine through a least square linear system by using a quadratic programming method, and classifying the high-frequency sub-band factors to obtain edge texture factors and to-be-denoised factors;
s3, carrying out denoising operation on the factor to be denoised by a soft threshold method to obtain a denoising factor;
and S4, splicing the denoising factor, the edge texture factor and the low-frequency subband factor in the Shearlet domain according to coordinates, and performing non-downsampling Shearlet inverse transformation to obtain an auxiliary denoising image.
2. The supervised machine learning-based image-assisted denoising method of claim 1, wherein the step S1 comprises the steps of:
s11, carrying out non-downsampling pyramid decomposition on the image containing the noise to obtain multi-scale features;
and S12, performing frequency plane segmentation and directional decomposition on the multi-scale features through a non-downsampling direction filtering device group to obtain high-frequency subband factors and low-frequency subband factors.
3. The supervised machine learning-based image-assisted denoising method of claim 1, wherein the step S3 comprises the steps of:
s31, estimating the preliminary noise standard deviation sigma according to the high-frequency sub-band factor by the following formula:
wherein, yijFor a factor value to be denoised of an image containing noise in the jth direction of the ith layer of a Shearlet domain, the Median () is a Median function;
s32, according to the factor to be denoised and the preliminary noise standard deviation sigma, estimating a threshold coefficient sigma through the following formulasth:
Wherein σyIn order to be the standard deviation of the noise,is the noise standard deviation sigma under the current target scale KyThe value range of the front target scale K is in a closed interval [1, N ]]N is the maximum scale, and Max () is the maximum function;
s33, judgmentWhether or not it is greater than a threshold coefficient sigmathIf yes, go to step S24, otherwise, go to step S25;
s34, according to the ideal threshold value ToptCarrying out image-assisted denoising on the factor to be denoised through a soft threshold function to obtain a denoising factor;
s35, according to Bayes lower threshold TLBAnd carrying out image-assisted denoising on the factor to be denoised through a soft threshold function to obtain a denoising factor.
4. The supervised machine learning-based image-assisted denoising method of claim 3, wherein the step S34 is performed by using an ideal threshold ToptThe expression of (a) is:
Topt=argmin(r(T)) (4)
wherein r () is the mean square error function of the denoising factor and the true value of the factor without being affected by noise, and argmin (r (T)) represents the computer iteration process of the value of the threshold T when the r () function is minimized.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010119565.7A CN111340726B (en) | 2020-02-26 | 2020-02-26 | Image auxiliary denoising method based on supervised machine learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010119565.7A CN111340726B (en) | 2020-02-26 | 2020-02-26 | Image auxiliary denoising method based on supervised machine learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111340726A true CN111340726A (en) | 2020-06-26 |
CN111340726B CN111340726B (en) | 2022-08-02 |
Family
ID=71183684
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010119565.7A Active CN111340726B (en) | 2020-02-26 | 2020-02-26 | Image auxiliary denoising method based on supervised machine learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111340726B (en) |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105608705A (en) * | 2016-01-14 | 2016-05-25 | 辽宁师范大学 | Image segmentation method based on non-down-sampling shearlet conversion and vector C-V model |
CN105913392A (en) * | 2016-04-08 | 2016-08-31 | 西安电子科技大学昆山创新研究院 | Degraded image overall quality improving method in complex environment |
CN106127711A (en) * | 2016-06-23 | 2016-11-16 | 浙江工业大学之江学院 | Shearlet conversion and quick two-sided filter image de-noising method |
CN106157261A (en) * | 2016-06-23 | 2016-11-23 | 浙江工业大学之江学院 | The shearler of translation invariance converts Medical Image Denoising method |
US20170053398A1 (en) * | 2015-08-19 | 2017-02-23 | Colorado Seminary, Owner and Operator of University of Denver | Methods and Systems for Human Tissue Analysis using Shearlet Transforms |
CN107403416A (en) * | 2017-07-26 | 2017-11-28 | 温州大学 | Improvement filtering and the medical ultrasound image denoising method of threshold function table based on NSCT |
CN107610074A (en) * | 2017-09-12 | 2018-01-19 | 新疆大学 | A kind of method for improving Remote Sensing Image Quality |
CN108416738A (en) * | 2018-01-16 | 2018-08-17 | 辽宁师范大学 | Image de-noising method based on undecimated wavelet transform and improvement fourth order PDEs |
CN109003232A (en) * | 2018-06-15 | 2018-12-14 | 浙江医院 | Medical MRI image de-noising method based on the smooth Shearlet of frequency domain scale |
CN109961411A (en) * | 2019-03-12 | 2019-07-02 | 浙江工业大学之江学院 | Non-lower sampling shears wave conversion medicine CT image denoising method |
-
2020
- 2020-02-26 CN CN202010119565.7A patent/CN111340726B/en active Active
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20170053398A1 (en) * | 2015-08-19 | 2017-02-23 | Colorado Seminary, Owner and Operator of University of Denver | Methods and Systems for Human Tissue Analysis using Shearlet Transforms |
CN105608705A (en) * | 2016-01-14 | 2016-05-25 | 辽宁师范大学 | Image segmentation method based on non-down-sampling shearlet conversion and vector C-V model |
CN105913392A (en) * | 2016-04-08 | 2016-08-31 | 西安电子科技大学昆山创新研究院 | Degraded image overall quality improving method in complex environment |
CN106127711A (en) * | 2016-06-23 | 2016-11-16 | 浙江工业大学之江学院 | Shearlet conversion and quick two-sided filter image de-noising method |
CN106157261A (en) * | 2016-06-23 | 2016-11-23 | 浙江工业大学之江学院 | The shearler of translation invariance converts Medical Image Denoising method |
CN107403416A (en) * | 2017-07-26 | 2017-11-28 | 温州大学 | Improvement filtering and the medical ultrasound image denoising method of threshold function table based on NSCT |
CN107610074A (en) * | 2017-09-12 | 2018-01-19 | 新疆大学 | A kind of method for improving Remote Sensing Image Quality |
CN108416738A (en) * | 2018-01-16 | 2018-08-17 | 辽宁师范大学 | Image de-noising method based on undecimated wavelet transform and improvement fourth order PDEs |
CN109003232A (en) * | 2018-06-15 | 2018-12-14 | 浙江医院 | Medical MRI image de-noising method based on the smooth Shearlet of frequency domain scale |
CN109961411A (en) * | 2019-03-12 | 2019-07-02 | 浙江工业大学之江学院 | Non-lower sampling shears wave conversion medicine CT image denoising method |
Non-Patent Citations (3)
Title |
---|
刘巧红等: "非下采样Shearlet域多变量模型的图像去噪", 《西华大学学报(自然科学版)》 * |
石满红等: "基于非下采样剪切波变换域三变量模型图像去噪算法", 《红外技术》 * |
高国荣等: "利用非抽样Shearlet域GSM模型进行图像去噪", 《武汉大学学报(信息科学版)》 * |
Also Published As
Publication number | Publication date |
---|---|
CN111340726B (en) | 2022-08-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105335947B (en) | Image de-noising method and image denoising device | |
Xu et al. | Deep edge-aware filters | |
CN108416740B (en) | Iterative adaptive median filtering method for eliminating salt and pepper noise | |
CN111242854B (en) | Image denoising method | |
CN107610074B (en) | Method for improving quality of remote sensing image | |
Zhao et al. | Texture variation adaptive image denoising with nonlocal PCA | |
Ghosh et al. | Saliency guided image detail enhancement | |
CN109118440B (en) | Single image defogging method based on transmissivity fusion and adaptive atmospheric light estimation | |
CN112862716A (en) | Sonar image target enhancement method based on dual-tree complex wavelet and fuzzy theory | |
Biswas et al. | A new adaptive image denoising method based on neighboring coefficients | |
CN111091107A (en) | Face region edge detection method and device and storage medium | |
CN113034387B (en) | Image denoising method, device, equipment and medium | |
CN111340726B (en) | Image auxiliary denoising method based on supervised machine learning | |
CN114240990B (en) | SAR image point target segmentation method | |
CN112927169B (en) | Remote sensing image denoising method based on wavelet transformation and improved weighted kernel norm minimization | |
Zia et al. | Rician noise removal from MR images using novel adapted selective non-local means filter | |
CN109242797B (en) | Image denoising method, system and medium based on homogeneous and heterogeneous region fusion | |
Zhang et al. | Image denoising using local adaptive layered Wiener filter in the gradient domain | |
Xu et al. | Adaptive Bilateral Texture Filter for Image Smoothing | |
Qiao et al. | Joint blind super-resolution and shadow removing | |
Zhao et al. | An improved Roberts edge detection algorithm based on mean filter and wavelet denoising | |
CN111242856B (en) | Dual image denoising method based on shear wave | |
CN113487496B (en) | Image denoising method, system and device based on pixel type inference | |
Ma et al. | Image Restoration Method based on Partition and Regularization for Industrial X-ray images | |
CN111583292B (en) | Self-adaptive image segmentation method for two-photon calcium imaging video data |
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 |