CN110136086A - Interval threshold image de-noising method based on BEMD - Google Patents
Interval threshold image de-noising method based on BEMD Download PDFInfo
- Publication number
- CN110136086A CN110136086A CN201910414811.9A CN201910414811A CN110136086A CN 110136086 A CN110136086 A CN 110136086A CN 201910414811 A CN201910414811 A CN 201910414811A CN 110136086 A CN110136086 A CN 110136086A
- Authority
- CN
- China
- Prior art keywords
- image
- component
- bemd
- bimf
- denoising
- 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
- 238000000034 method Methods 0.000 title claims abstract description 32
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 12
- 230000000694 effects Effects 0.000 claims abstract description 11
- 238000012545 processing Methods 0.000 claims description 13
- 230000008569 process Effects 0.000 claims description 5
- 230000015556 catabolic process Effects 0.000 claims description 4
- 239000004615 ingredient Substances 0.000 claims description 4
- 238000001514 detection method Methods 0.000 claims description 2
- 238000001914 filtration Methods 0.000 claims description 2
- 230000008901 benefit Effects 0.000 abstract description 4
- 230000009466 transformation Effects 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 230000006872 improvement Effects 0.000 description 3
- 239000000654 additive Substances 0.000 description 1
- 230000000996 additive effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 230000009022 nonlinear effect Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
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
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The present invention is based on the interval threshold image de-noising methods of BEMD, are a kind of interval threshold Image denoising algorithms based on Bidimensional Empirical Mode Decomposition.This method treats denoising image using the self-similarity of image and carries out boundary extension, effectively inhibits end effect existing for BEMD algorithm, carries out multi-resolution decomposition to continuation image using BEMD algorithm and obtains a series of two-dimentional intrinsic mode functions.Then the BIMF component for accounting for main component to noise carries out interval threshold denoising, and each rank BIMF component is finally added reconstruct and achievees the purpose that denoising.Experiment shows that the denoising effect of the method for the present invention better than existing BEMD Denoising Algorithm, also has advantage compared with traditional Denoising Algorithm.
Description
Technical field
The present invention relates to image denoising fields, more particularly to the interval threshold image de-noising method based on BEMD.
Background technique
The progress of modern computer technology has pushed the research and development of image processing techniques significantly, high quality, clearly
Image is the basis that field of image processing is researched and analysed.And in practice, image in formation, acquisition or transmission process Chang Yinwei from
Body or extraneous factor such as device, illumination etc. are by a degree of interference.The noise that interference generates, which can obscure, even covers image
In target to be studied edge, Texture eigenvalue, affect the accurate expression of image original information, reduce the accurate of processing result
Property.Therefore, how from effectively removed in distorted signal noise restore original image be in Digital Image Processing it is vital
One pretreatment link.
Image generally comprises a large amount of texture informations as two-dimensional random signal, along the picture of texture vertical direction in image
The plain gray value usually with mutation, it is this to be mutated the feature for making image that there is non-stationary signal.Therefore image can be counted as
It is two-dimensional non-stationary signal, denoising can be realized by extending one-dimensional Non-stationary Signal Analysis method to two dimension.It is existing
The processing method for non-stationary signal mainly have Short Time Fourier Transform (Short Time Fourier
Translation, STFT), Gabor transformation, Wigner-Ville distribution and wavelet transformation etc..Compared to Fourier transformation, with
Upper nonstationary random response method has progress to the processing of non-stationary nonlinear properties, but relies in Fu after all
The thought of leaf transformation, and need rule of thumb to preset basic function, do not have adaptivity.
Ensemble empirical mode decomposition method is to show a kind of good self-adaptive processing in Non-stationary Signal Analysis field in recent years
Method can avoid drawback present in traditional Non-stationary Signal Analysis method, two-dimensional expansion to a certain extent --- and two
Empirical mode decomposition is tieed up, also shows good performance in field of image processing.In image denoising field, existing BEMD denoising is calculated
Although method has certain denoising effect, but largely lose there are still texture information and do not bring in view of BEMD resolution characteristic
Error.
Summary of the invention
In order to solve BEMD Threshold Filter Algorithms, the present invention provides the interval threshold figure based on BEMD
As denoising method, this method treats denoising image using the self-similarity of image and carries out boundary extension, effectively BEMD is inhibited to calculate
End effect existing for method carries out multi-resolution decomposition to continuation image using BEMD algorithm and obtains a series of two-dimentional natural mode of vibration letters
Number.Then the BIMF component for accounting for main component to noise carries out interval threshold denoising, is finally added each rank BIMF component and reconstructs
Achieve the purpose that denoising.The denoising effect of the method for the present invention is better than existing BEMD Denoising Algorithm, compared with traditional Denoising Algorithm
With advantage, for this purpose, the present invention provides the interval threshold image de-noising method based on BEMD:
(1) picture signal to be denoised is inputted；
(2) it treats denoised signal and carries out boundary extension, obtain widened image to be processed after continuation；
(3) BEMD decomposition is carried out to the image to be processed after continuation, is obtained according to the sequence of frequency from high to low a series of
BIMF component, and according to certain rule by BIMF component be divided into noise account for leading BIMF component and image information account for it is leading
BIMF component；
(4) the BIMF component for accounting for main component to noise carries out interval threshold filtering, the BIMF component that obtains that treated；
(5) by treated, BIMF component accounts for leading BIMF component with image information is added, and reconstruct obtains final
Signal.
Further improvement of the present invention treats denoising image in the step (2) and carries out boundary extension, after obtaining continuation
Image, to inhibit the end effect of BEMD, the step of boundary extension are as follows:
Continuation is carried out to boundary using the non local self-similarity of image, it is assumed that image size is N × N, is carried out to it
Piecemeal, the size of each small image block is M × M, to the image block block for being in image borderedge, in a certain size search
Its similar block is found in window to be extended original image edge block using the adjacent block of similar block corresponding direction as continuation pixel,
Wherein the judgment criterion of similitude is based on average absolute value difference.
The BIMF component that BEMD is decomposed is divided into noise in the step (3) and accounts for master by further improvement of the present invention
The BIMF component and useful signal for leading ingredient account for the BIMF component two large divisions of main component, the steps include:
Step 3.1: to image to be processed, after BEMD is decomposed, image can be indicated are as follows:
In formula, IMFi(m, n) indicates that intrinsic mode function component, res (m, n) indicate residual components；
Step 3.2: for the IMF in decomposition resulti(m, n) component calculates the ENERGY E of its intrinsic mode function componenti,
Expression formula is expressed as
Step 3.3: more each rank ENERGY EiSize, take EiIndex j when reaching local minimum for the first time, which is used as, to be divided
Picture breakdown is by boundary
Wherein j is that the boundary obtained according to energy indexes, and preceding j-1 rank BIMF component is that noise accounts for leading ingredient, after j rank
BIMF component be image information account for main component.
Further improvement of the present invention, the BIMF component image for accounting for main component to noise in the step (4) carry out area
Between threshold denoising, the specific steps of which are as follows:
Step 4.1: extreme point detection being carried out to BIMF component image, obtains the extreme value comprising image maximum and minimum
Point distribution map；
Step 4.2: Delaunay Triangulation being done to the extreme value distribution figure, which is many with image by picture breakdown
Extreme point be endpoint triangle；
Step 4.3: small triangle each of is obtained as a processing unit with subdivision, in the same processing unit picture
Vegetarian refreshments uniformly carries out threshold process: if in the range of triangle pair is answered, there are the amplitudes of some extreme value to be absolutely greater than setting threshold
Value, then do soft-threshold denoising for the pixel value in the region；If triangle pair is answered in range, the amplitude absolute value of extreme point is whole
Less than given threshold, then the gray value of corresponding pixel points within the scope of this is set as zero.
Interval threshold image de-noising method of the application based on BEMD, has the feature that
The present invention provides a kind of interval threshold Image denoising algorithm based on BEMD, improves existing BEMD threshold denoising
Principle drawback existing for algorithm can remove multiplicative noise and additive noise well, obtain the denoising performance of algorithm
Very big promotion is arrived.Compared with traditional Denoising Algorithm, also there is certain advantage in low signal-to-noise ratio.
Detailed description of the invention
Fig. 1 Lena image is based on the partial schematic diagram after self-similarity continuation；
Fig. 2 BEMD decomposition process figure；
Each rank BIMF energy variation figure of Fig. 3；
Fig. 4 threshold denoising treated BIMF component and result figure.
Specific embodiment
Present invention is further described in detail with specific embodiment with reference to the accompanying drawing:
The present invention provides the interval threshold image de-noising method based on BEMD, and this method is treated using the self-similarity of image
It denoises image and carries out boundary extension, end effect existing for BEMD algorithm is effectively inhibited, using BEMD algorithm to continuation image
It carries out multi-resolution decomposition and obtains a series of two-dimentional intrinsic mode functions.Then the BIMF component for accounting for main component to noise carries out area
Between threshold denoising, finally by each rank BIMF component be added reconstruct achieve the purpose that denoising.The denoising effect of the method for the present invention is better than
Existing BEMD Denoising Algorithm also has advantage compared with traditional Denoising Algorithm.
Below based on 2018 work of MATLAB on PC machine (Intel (R) Core (TM) i5-7200U CPU 2.70GHz)
Tool, by taking the noisy Lena image of noise variance Sigma=0.01 as an example, in conjunction with attached drawing to the present invention is based on the interval thresholds of BEMD
The specific embodiment of image denoising scheme is described in further detail.
Step 1: treating denoising image and carry out boundary extension, part screenshot such as Fig. 1 after Lena image boundary continuation.Due to
In the present invention continuation be take the adjacent block of similar block as continuation, therefore continuation length cannot be arranged it is excessive, herein outward
The continuation length of four pixels.
Step 2: continuation image being decomposed with BEMD, the flow chart of BEMD algorithm such as Fig. 2.Image after decomposition can be with
It is expressed as
Step 3: calculating the ENERGY E for decomposing obtained BIMF componenti, the energy variation of each order component such as Fig. 3.Wherein horizontal axis
Indicate that index i, the longitudinal axis indicate the energy value of each rank BIMF component.See from Fig. 3 and obtains local pole for the first time at index i=2
Small value.Therefore previous BIMF component accounts for leading for noise, and image information accounts for BIMF component since second-order BIMF component
The operation of main component, i.e. threshold denoising is carried out for the first rank BIMF component.
Step 4: interval threshold denoising being carried out to preceding single order BIMF component, Threshold Denoising Method uses soft-threshold denoising.It is soft
Threshold denoising formula expression is
Wherein A is the set by delta-shaped region internal coordinate point each in Delaunay triangulation figure,For
IMFi(A) regional ensemble after threshold process, IMFi(mextre, nextre) be region A in pixel amplitude absolute value most
Big value.Threshold value is set as in formula
midi=median (var (IMFi(m,n)))
In definitionFor the noise estimation value of the i-th rank intrinsic mode function component, midiFor the i-th rank intrinsic mode function
The intermediate value of component variance.The single order BIMF component map such as Fig. 4 (a) of Lena image after threshold process.
Step 5: by treated, single order BIMF component is added reconstruct with residue BIMF component, obtains final processing knot
Fruit, noisy Lena scheme final denoising effect such as Fig. 4 (b).From result, it can be seen that, the mentioned method of the present invention can make an uproar in removal
The edge of image is effectively maintained while sound.
Table one for the present invention mentioned interval threshold BEMD Denoising Algorithm and tradition BEMD Denoising Algorithm performance comparison, from table
Middle data can see, and the present invention suggests plans with good denoising performance, be better than existing BEMD Denoising Algorithm.
1 interval threshold BEMD of table and tradition BEMD algorithm denoise Contrast on effect
| Noisy image | The BEMD of partial reconfiguration | Single-point threshold value BEMD | Interval threshold BEMD | |
| MSE | 640.21 | 213.71 | 187.49 | 163.80 |
| PSNR (unit: dB) | 20.06 | 24.83 | 25.40 | 25.98 |
The above described is only a preferred embodiment of the present invention, being not the limit for making any other form to the present invention
System, and made any modification or equivalent variations according to the technical essence of the invention, still fall within present invention model claimed
It encloses.
Claims (4)
1. the interval threshold image de-noising method based on BEMD, it is characterised in that:
(1) picture signal to be denoised is inputted；
(2) it treats denoised signal and carries out boundary extension, obtain widened image to be processed after continuation；
(3) BEMD decomposition is carried out to the image to be processed after continuation, obtains a series of BIMF according to the sequence of frequency from high to low
Component, and according to certain rule by BIMF component be divided into noise account for leading BIMF component and image information account for it is leading
BIMF component；
(4) the BIMF component for accounting for main component to noise carries out interval threshold filtering, the BIMF component that obtains that treated；
(5) by treated, BIMF component accounts for leading BIMF component with image information is added, and reconstruct obtains final letter
Number.
2. the interval threshold image de-noising method according to claim 1 based on BEMD, it is characterised in that: the step
(2) denoising image is treated in carries out boundary extension, the image after obtaining continuation, to inhibit the end effect of BEMD, boundary extension
The step of are as follows:
Continuation is carried out to boundary using the non local self-similarity of image, it is assumed that image size is N × N, carries out piecemeal to it,
The size of each small image block is M × M, to the image block block for being in image borderedge, in a certain size search window
Its similar block is found, using the adjacent block of similar block corresponding direction as continuation pixel, original image edge block is extended, wherein
The judgment criterion of similitude is based on average absolute value difference.
3. the interval threshold image de-noising method according to claim 1 based on BEMD, it is characterised in that: the step
(3) in by the BIMF component that BEMD is decomposed be divided into noise account for the BIMF component of leading ingredient and useful signal account for mainly at
The BIMF component two large divisions divided, the steps include:
Step 3.1: the image I (m, n) to be processed for being N × N to size, after BEMD is decomposed, image can be indicated are as follows:
In formula, l indicates the total order decomposed, IMFi(m, n) indicates that intrinsic mode function component, res (m, n) indicate residual components；
Step 3.2: for the IMF in decomposition resulti(m, n) component calculates the ENERGY E of its intrinsic mode function componenti, expression
Formula is expressed as
Step 3.3: more each rank ENERGY EiSize, take EiIndex j when reaching local minimum for the first time, will as demarcating
Picture breakdown is
Wherein j is that the boundary obtained according to energy indexes, and preceding j-1 rank BIMF component is that noise accounts for leading ingredient, after j rank
BIMF component is that image information accounts for main component.
4. the interval threshold image de-noising method according to claim 1 based on BEMD, it is characterised in that: the step
(4) the BIMF component image for accounting for main component to noise in carries out processing unit i.e. area in interval threshold denoising and BIMF component
Between judgement, the specific steps of which are as follows:
Step 4.1: extreme point detection being carried out to BIMF component image, obtains the extreme point comprising image maximum and minimum point
Butut；
Step 4.2: Delaunay Triangulation being done to the extreme value distribution figure, which is many pole with image by picture breakdown
Value point is the triangle of endpoint；
Step 4.3: small triangle each of is obtained as a processing unit with subdivision, in the same processing unit pixel
It is unified to carry out threshold process: if in the range of triangle pair is answered, there are the amplitudes of some extreme value to be absolutely greater than given threshold, then
Pixel value in the region is done into soft-threshold denoising；If triangle pair is answered in range, the amplitude absolute value of extreme point all less than
The gray value of corresponding pixel points within the scope of this is then set as zero by given threshold.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910414811.9A CN110136086A (en) | 2019-05-17 | 2019-05-17 | Interval threshold image de-noising method based on BEMD |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910414811.9A CN110136086A (en) | 2019-05-17 | 2019-05-17 | Interval threshold image de-noising method based on BEMD |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110136086A true CN110136086A (en) | 2019-08-16 |
Family
ID=67571232
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910414811.9A Pending CN110136086A (en) | 2019-05-17 | 2019-05-17 | Interval threshold image de-noising method based on BEMD |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110136086A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111680548A (en) * | 2020-04-27 | 2020-09-18 | 哈尔滨工程大学 | Distortion-free boundary continuation method for wavelet online denoising |
| CN113592782A (en) * | 2021-07-07 | 2021-11-02 | 山东大学 | Method and system for extracting X-ray image defects of composite carbon fiber core rod |
| CN117874421A (en) * | 2023-12-06 | 2024-04-12 | 广州海洋地质调查局 | Ocean gravity data noise reduction method and device, electronic equipment and storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103020916A (en) * | 2012-12-28 | 2013-04-03 | 北方工业大学 | Image denoising method combining two-dimensional Hilbert transform and BEMD |
| CN107464226A (en) * | 2017-07-31 | 2017-12-12 | 东南大学 | A kind of image de-noising method based on improvement two-dimensional empirical mode decomposition algorithm |
-
2019
- 2019-05-17 CN CN201910414811.9A patent/CN110136086A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103020916A (en) * | 2012-12-28 | 2013-04-03 | 北方工业大学 | Image denoising method combining two-dimensional Hilbert transform and BEMD |
| CN107464226A (en) * | 2017-07-31 | 2017-12-12 | 东南大学 | A kind of image de-noising method based on improvement two-dimensional empirical mode decomposition algorithm |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111680548A (en) * | 2020-04-27 | 2020-09-18 | 哈尔滨工程大学 | Distortion-free boundary continuation method for wavelet online denoising |
| CN113592782A (en) * | 2021-07-07 | 2021-11-02 | 山东大学 | Method and system for extracting X-ray image defects of composite carbon fiber core rod |
| CN113592782B (en) * | 2021-07-07 | 2023-07-28 | 山东大学 | Method and system for extracting X-ray image defects of composite material carbon fiber core rod |
| CN117874421A (en) * | 2023-12-06 | 2024-04-12 | 广州海洋地质调查局 | Ocean gravity data noise reduction method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107464226B (en) | A kind of image de-noising method based on improvement two-dimensional empirical mode decomposition algorithm | |
| CN103077508B (en) | Transform domain non local and minimum mean square error-based SAR (Synthetic Aperture Radar) image denoising method | |
| CN101950414B (en) | Non-local mean de-noising method for natural image | |
| CN101944230B (en) | Multi-scale-based natural image non-local mean noise reduction method | |
| CN110136086A (en) | Interval threshold image de-noising method based on BEMD | |
| CN108932699B (en) | Three-dimensional matching harmonic filtering image denoising method based on transform domain | |
| CN102509269A (en) | Image denoising method combined with curvelet and based on image sub-block similarity | |
| Weizheng et al. | Digital image denoising method based on mean filter | |
| CN105550998B (en) | Image enhancement method and image enhancement system based on second-generation wavelet integer transform | |
| Tang et al. | Adaptive threshold shearlet transform for surface microseismic data denoising | |
| Chui et al. | Image denoising method with adaptive Bayes threshold in nonsubsampled contourlet domain | |
| CN104732498A (en) | Thresholding image denoising method based on nonsubsampled Contourlet transformation | |
| CN110060220A (en) | Based on the image de-noising method and system for improving BM3D algorithm | |
| CN108876711A (en) | A kind of sketch generation method, server and system based on image characteristic point | |
| Susrutha et al. | Analysis on FFT and DWT transformations in image processing | |
| Zhang et al. | Bayesian-based speckle suppression for SAR image using contourlet transform | |
| Zhang et al. | A New Image Denoising Method Based on Wave Atoms and Cycle Spinning. | |
| Villegas et al. | A comparison of the bandelet, wavelet and contourlet transforms for image denoising | |
| Deng et al. | A CT Image denoise method using curvelet transform | |
| Chen et al. | Research on sonar image denoising method based on fixed water area noise model | |
| Butt et al. | Multilateral filtering: A novel framework for generic similarity-based image denoising | |
| Joseph et al. | A Novel Denoising Algorithm Based on Superpixel Clustering and Dictionary Learning Approach. | |
| Tomic et al. | Adaptive edge-preserving denoising by point-wise wavelet basis selection | |
| CN1173291C (en) | Method for using computer to recreate image containing spot noise | |
| CN103747268A (en) | Layered self-adaptive threshold video denoising method |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190816 |
|
| RJ01 | Rejection of invention patent application after publication |