CN106952272B - A kind of anti-noise Otsu image segmentation method based on two generation wavelet transformations - Google Patents
A kind of anti-noise Otsu image segmentation method based on two generation wavelet transformations Download PDFInfo
- Publication number
- CN106952272B CN106952272B CN201710134347.9A CN201710134347A CN106952272B CN 106952272 B CN106952272 B CN 106952272B CN 201710134347 A CN201710134347 A CN 201710134347A CN 106952272 B CN106952272 B CN 106952272B
- Authority
- CN
- China
- Prior art keywords
- noise
- image
- segmentation
- otsu
- generation
- 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
Classifications
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
-
- 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
- G06T2207/10048—Infrared 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
- G06T2207/20048—Transform domain processing
- G06T2207/20064—Wavelet transform [DWT]
Landscapes
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention belongs to field of image processings, a kind of more particularly to anti-noise Otsu image segmentation method based on two generation wavelet transformations, bis- generation of db1 small echo integer transform is carried out by treating segmented image, calculates maximum between-cluster variance again after inhibiting in wavelet field to the noise of target and background.The present invention is the image partition method that a kind of noise immunity is strong, segmentation effect is good, and two generation wavelet methods are conducive to hardware realization, is applied to that noise jamming is serious and the higher system of requirement of real-time convenient for algorithm, has important researching value and broad prospect of application.
Description
Technical field
The invention belongs to field of image processing more particularly to a kind of anti-noise Otsu image segmentations based on two generation wavelet transformations
Method.
Background technique
Image segmentation is the basic fundamental of Image Information Processing and the premise of image understanding and pattern-recognition, image point
It cuts and has a wide range of applications.In processing medical image, three-dimensional of the image segmentation for the organ of generation lesion in people's body
Display all plays effectively guidance effect to the determination and analysis of lesion locations；In the analysis application of road traffic image
In, usable image cutting techniques separate the target vehicle to be extracted from the fuzzy complex background such as monitor or take photo by plane；Remote sensing images
The application for being segmented in military field is also very extensive, such as the investigation of strategy and tactics, the mapping of military marine field, high-resolution
Remote Sensing Image Segmentation data can for nature calamity condition monitoring and evaluation, map drafting and update, the forest reserves and environment
Monitoring and management etc., therefore, the segmentation of image all plays a crucial role.
In numerous dividing methods, Threshold segmentation is one of image partition method the most simple and effective, and key exists
In the selection of threshold value, common several classical threshold segmentation methods mainly have Otsu algorithm, information maximum entropy algorithm, minimum intersection
Entropy algorithm etc., for example, document 1: Qiao Wanbo, " a kind of improved Binary Sketch of Grey Scale Image method " of Cao Yinjie, document 2: Ding Xiao
Peak, " the improvement image segmentation algorithm based on maximum between-cluster variance " of He Kailin.One-dimensional maximum variance between clusters, with it because calculating
Simply, it is widely used the advantages that real-time height, strong robustness.But one-dimensional Otsu method does not consider the space correlation between pixel
Property, therefore when image includes noise, the segmentation effect of one-dimensional Otsu method is bad.It is right because inevitably there is noise jamming in image
This, proposes two-dimentional Otsu RAPIDLY RECURSIVE METHOD FOR, this method considered while improving operation efficiency pixel grayscale information and
The spatial coherence of its neighborhood improves the noise immunity of one-dimensional Otsu method.For the noise immunity for further increasing algorithm, document 3: scape
Know army, " a kind of image segmentation algorithm based on Two-dimensional Maximum inter-class variance " of Cai Anni, Sun Jingao are average in gray level-field
On the basis of gray level, a kind of image segmentation algorithm of Two-dimensional Maximum inter-class variance is proposed, which can be preferably to noisy figure
As being split.Although two-dimentional Otsu method improves the anti-noise ability of algorithm, but when noise jamming is serious, their segmentation effect
Fruit is still not ideal enough.
A kind of image fast segmentation method based on least square method curve matching of Chinese patent CN201610510826.1,
The image segmentation algorithm using least square method curve matching is proposed, but this method, when noise jamming is serious, there are figures
As dividing undesirable problem.
Image enchancing method and image enhancement of the Chinese patent CN201510903464.8 based on two generation small echo integer transforms
System is described and is handled by two generation small echo integer transforms image, can be good at for target being partitioned into from background
Come, there is good noiseproof feature.
Summary of the invention
The technical problem to be solved in the present invention is to provide a kind of anti-noise Otsu image segmentation sides based on two generation wavelet transformations
Method can split target image from background, have good noise robustness.
In order to solve the above technical problems, a kind of anti-noise Otsu image segmentation method based on two generation wavelet transformations of the present invention,
Include the following steps:
Step S1: treating segmented image and carry out the decomposition of bis- generation of db1 small echo integer transform single layer, obtains integer low frequency
Decomposition coefficient with ca and three high-frequency sub-band ch, cv and cd；
Step S2: negative in low frequency sub-band ca is become into the new low frequency sub-band ca1 of 0 composition, and 3 × 3 mean values are carried out to ca1
It is rounded after filtering and constitutes ca2；
Step S3: the coefficient maximum value M of ca2 is sought, in the inter-class variance of [0, M] interval computation ca2It is calculated using Otsu
Method obtains segmentation threshold T；
Step S4: maximum value M is changed to greater than T to the coefficient of ca2, the coefficient of ca2 is changed to 0 less than T, is formed after change
New low frequency sub-band ca3；
Step S5: the decomposition coefficient of three high-frequency sub-bands ch, cv and cd are all changed to 0, constitute new high-frequency sub-band
Ch1, cv1 and cd1；
Step S6: ca3, ch1, cv1 and cd1 are reconstructed using bis- generation of db1 small echo integer, the image W after constituting segmentation.
As prioritization scheme of the invention, in step s3, so that inter-class varianceMaximum value is point of Otsu algorithm
Cut threshold value T.
As prioritization scheme of the invention, the range of segmentation threshold T is [0, M].
The present invention has the effect of positive: the present invention overcomes one-dimensional Otsu method be difficult to be partitioned into it is satisfied as a result, and two
Although dimension Otsu method improves the noise immunity of algorithm, but effect is little compared with one-dimensional Otsu method, but also greatly increases operation
The problem of time.Preferable segmentation effect can be obtained to the image containing different noise types and different noise intensity, have compared with
Strong noise immunity and anti-noise robustness；Objectively the segmentation effect of invention and antinoise are verified from quantizating index.
Qualitative and quantitative analysis has absolutely proved that the present invention is the image partition method that a kind of noise immunity is strong, segmentation effect is good, and two
Be conducive to hardware realization for wavelet method, be applied to that noise jamming is serious and the higher system of requirement of real-time convenient for algorithm, tool
There are important researching value and broad prospect of application.
Detailed description of the invention
The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.
Fig. 1 is the operation block diagram of bis- generation of db1 small echo integer transform of the present invention；
Fig. 2 (a) is standard Lena gray level image, and Fig. 2 (b) is low-light (level) infrared image, and Fig. 2 (c) is remote sensing images；
Fig. 3 (a) is plus the standard Lena gray level image of Gaussian noise, Fig. 3 (b) are the infrared figure of low-light (level) for adding Gaussian noise
Picture, Fig. 3 (c) are the remote sensing images for adding Gaussian noise；
Fig. 4 (a) is to add the standard Lena gray level image of Gaussian noise through classical Otsu algorithm image segmentation as a result, Fig. 4
It (b) is to add the low-light (level) infrared image of Gaussian noise through classical Otsu algorithm image segmentation as a result, Fig. 4 (c) is to increase this to make an uproar
Result of the remote sensing images of sound through classical Otsu algorithm image segmentation；
Fig. 5 (a) is to add the standard Lena gray level image of Gaussian noise through the algorithm image segmentation of document 1 as a result, Fig. 5
It (b) is to add the low-light (level) infrared image of Gaussian noise through the algorithm image segmentation of document 1 as a result, Fig. 5 (c) is to add Gaussian noise
Algorithm image segmentation of the remote sensing images through document 1 result；
Fig. 6 (a) is to add the standard Lena gray level image of Gaussian noise through the algorithm image segmentation of document 2 as a result, Fig. 6
It (b) is to add the low-light (level) infrared image of Gaussian noise through the algorithm image segmentation of document 2 as a result, Fig. 6 (c) is to add Gaussian noise
Algorithm image segmentation of the remote sensing images through document 2 result；
Fig. 7 (a) is to add the standard Lena gray level image of Gaussian noise through the algorithm image segmentation of document 3 as a result, Fig. 7
It (b) is to add the low-light (level) infrared image of Gaussian noise through the algorithm image segmentation of document 3 as a result, Fig. 7 (c) is to add Gaussian noise
Algorithm image segmentation of the remote sensing images through document 3 result；
Fig. 8 (a) is to add the standard Lena gray level image of Gaussian noise through method image segmentation of the invention as a result, Fig. 8
It (b) is to add the low-light (level) infrared image of Gaussian noise through method image segmentation of the invention as a result, Fig. 8 (c) is to increase this to make an uproar
Result of the remote sensing images of sound through method image segmentation of the invention；
Fig. 9 (a) is the lena figure with salt-pepper noise, and Fig. 9 (b) is the lena figure with multiplicative noise, and Fig. 9 (c) is band Poisson
The lena of noise schemes；
Figure 10 (a) is the lena figure with salt-pepper noise through classical Otsu algorithm image segmentation as a result, Figure 10 (b) is to carry
Property noise lena figure through classical Otsu algorithm image segmentation as a result, Figure 10 (c) be the lena figure with poisson noise through classics
The result of Otsu algorithm image segmentation；
Figure 11 (a) is for the lena figure with salt-pepper noise through 1 algorithm image segmentation of document as a result, Figure 11 (b) is the property carried
The lena figure of noise is through 1 algorithm image segmentation of document as a result, Figure 11 (c) is the lena figure with poisson noise through 1 algorithm of document
The result of image segmentation；
Figure 12 (a) is for the lena figure with salt-pepper noise through 2 algorithm image segmentation of document as a result, Figure 12 (b) is the property carried
The lena figure of noise is through 2 algorithm image segmentation of document as a result, Figure 12 (c) is the lena figure with poisson noise through 2 algorithm of document
The result of image segmentation；
Figure 13 (a) is for the lena figure with salt-pepper noise through 3 algorithm image segmentation of document as a result, Figure 13 (b) is the property carried
The lena figure of noise is through 3 algorithm image segmentation of document as a result, Figure 13 (c) is the lena figure with poisson noise through 3 algorithm of document
The result of image segmentation；
Figure 14 (a) is the lena figure with salt-pepper noise through present invention progress image segmentation result, and Figure 14 (b) is that the property carried is made an uproar
The lena figure of sound carries out image segmentation result through the present invention, and Figure 14 (c) is that the lena figure with poisson noise carries out figure through the present invention
As segmentation result.
Specific embodiment
As shown in Figure 1, the invention discloses a kind of anti-noise Otsu image segmentation methods based on two generation wavelet transformations, including
Following steps:
Step S1: treating segmented image and carry out the decomposition of bis- generation of db1 small echo integer transform single layer, obtains integer low frequency
Decomposition coefficient with ca and three high-frequency sub-band ch, cv and cd；
Step S2: negative in low frequency sub-band ca is become into the new low frequency sub-band ca1 of 0 composition, and 3 × 3 mean values are carried out to ca1
It is rounded after filtering and constitutes ca2；Wherein, negative in low frequency sub-band ca is become 0 to calculate convenient for subsequent Otsu algorithm, because
Otsu algorithm is calculated within the scope of positive integer.
Step S3: the coefficient maximum value M of ca2 is sought, in the inter-class variance of [0, M] interval computation ca2It is calculated using Otsu
Method obtains segmentation threshold T；
Step S4: maximum value M is changed to greater than T to the coefficient of ca2, the coefficient of ca2 is changed to 0 less than T, is formed after change
New low frequency sub-band ca3；Wherein, maximum value M is changed to greater than T to the coefficient of ca2, the coefficient of ca2 is changed to 0 less than T, be for
In wavelet field image background is effectively distinguished with prospect.
Step S5: the decomposition coefficient of three high-frequency sub-bands ch, cv and cd are all changed to 0, constitute new high-frequency sub-band
Ch1, cv1 and cd1；Wherein, the decomposition coefficient of three high-frequency sub-bands ch, cv and cd are all changed to 0, are to remove high frequency
Noise.
Step S6: ca3, ch1, cv1 and cd1 are reconstructed using bis- generation of db1 small echo integer, the image W after constituting segmentation.
The pixel value of image is integer, and wavelet arithmetic, i.e. Second Generation Wavelet Transformation are a kind of new biorthogonal wavelets
Structural scheme determines high-frequency information by predictive operator, and primarily determines low-frequency information, then by update operator, to preliminary
Determining low-frequency information is modified, so that it is determined that low-frequency information, including 3 processing steps, specific implementation such as Fig. 1 institute
Show,
(1) it decomposes: by input signal SiEven order and odd numbered sequences are resolved into according to parity, decomposable process is expressed as
F(si)=(si-1,di-1)；Wherein, si-1Indicate low-frequency approximation component, di-1Indicate the high frequency detail component of signal, F (si) indicate
For decomposable process.
(2) it predicts: using the correlation between data, with the s of even orderi-1Predicted value P (si-1) go prediction (or interpolation)
Odd numbered sequences di-1, i.e., the predicted value of odd signals, the reality of odd signals will be used as after filter P dual numbers signal function
Value and predicted value subtract each other to obtain residual signals.It in practice, although can not be from subset si-1Middle Accurate Prediction subset di-1, but P
(si-1) very close to di-1, therefore P (s can be usedi-1) and di-1Difference replace di-1, the d that generates in this wayi-1Than original di-1Packet
Containing less information, d is then obtainedi-1=di-1-P(si-1), smaller subset s can be used herei-1With small echo subset di-1
Instead of original signal Si.Repetitive assignment and prediction process, { s can be used in original signal collection after n stepn dn ... s1 d1Indicate.
(3) it updates.In order to make certain global properties of original signal collection in its subset si-1Relaying continuation of insurance is held, it is necessary to be carried out more
Newly.To find better subset si-1, so that it keeps a certain scalar characteristic Q (x) (such as mean value, vanishing moment is constant) of original image,
Existing Q (si-1)=Q (si).A better subset s is generated by operator Ui-1, it is allowed to keep original signal siSome characteristics,
Renewal process expression formula is si-1=si-1+U(di-1)。
Bis- generation of db1 small echo integer transform is carried out by treating segmented image, to the noise of target and background in wavelet field
Good inhibition has been carried out, has then obtained segmentation threshold T using Otsu algorithm, basic process is as follows:
If image pixel number is N, tonal range is [0, L-1], and the pixel number of corresponding grey scale grade i is n, probability are as follows:
pi=ni/ N i=0,1,2 ..., L-1
Pixel in image is pressed gray value threshold value T0It is divided into two class C0And C1, C0By gray value [0, T0] between picture
Element composition, C1By gray value in [T0+ 1, L-1] between pixel composition, for intensity profile probability, the mean value of entire image are as follows:
Then C0And C1Mean value are as follows:
Wherein
It can be obtained by formula 1, formula 2 and 3 three formula of formula:
Inter-class variance is defined as:
Allow T0In [0, L-1] range successively value, make σB 2Maximum T0Value is the optimal threshold of Otsu method.
Experimental result and analysis
Experimental situation: Windows7 system Intel Pentinum CPU G860, dominant frequency 3.0GHZ, memory 4G become language
Say MATLAB7.0.
Experimental image chooses three width images:
First width: resolution ratio is the standard Lena gray level image of 512 × 512 normal illumination, as shown in Fig. 2 (a)；
Second width: resolution ratio is 269 × 350 low-light (level) infrared images, as shown in Fig. 2 (b)；
Third width: resolution ratio is 500 × 375 remote sensing images, as shown in Fig. 2 (c)；
It is 0 that mean value is separately added into three width images, the Gaussian noise that variance is 0.03, and the image after Fig. 2 (a) addition is such as
Shown in Fig. 3 (a), image such as Fig. 3 (b) after Fig. 2 (b) is added is shown, and image such as Fig. 3 (c) after Fig. 2 (c) is added is shown.
1) Fig. 3 (a), Fig. 3 (b) and Fig. 3 (c) are split respectively using classical Otsu method, the result point of segmentation
Not as shown in Fig. 4 (a), Fig. 4 (b) and Fig. 4 (c).
2) using document 1 " a kind of improved Binary Sketch of Grey Scale Image method " respectively to Fig. 3 (a), Fig. 3 (b) and Fig. 3 (c)
It is split, the result of segmentation is respectively as shown in Fig. 5 (a), Fig. 5 (b) and Fig. 5 (c).
3) using document 2 " the improvement image segmentation algorithm based on maximum between-cluster variance " respectively to Fig. 3 (a), Fig. 3 (b) and
Fig. 3 (c) is split, and the result of segmentation is respectively as shown in Fig. 6 (a), Fig. 6 (b) and Fig. 6 (c).
4) using document 3 " a kind of image segmentation algorithm based on Two-dimensional Maximum inter-class variance " respectively to Fig. 3 (a), Fig. 3
(b) it is split with Fig. 3 (c), the result of segmentation is respectively as shown in Fig. 7 (a), Fig. 7 (b) and Fig. 7 (c).
It can be seen that above-mentioned three kinds of methods from above-mentioned three kinds of methods all to fail effectively to be divided image, it is visually several
With add original image no significant difference of making an uproar, using the present invention Fig. 3 (a), Fig. 3 (b) and Fig. 3 (c) are split respectively, segmentation
As a result respectively as shown in Fig. 8 (a), Fig. 8 (b) and Fig. 8 (c), it can be seen that the present invention due to being added at noise in the algorithm in advance
Reason, effectively overcomes noise jamming, image segmentation is best.
In order to examine the versatility of the method for the present invention anti-noise, scheme to added with the lena that intensity is 0.08 salt-pepper noise, such as Fig. 9
(a) shown in；The lena of 0.08 multiplicative noise schemes, as shown in Fig. 9 (b)；The lena of poisson noise schemes, and is divided as shown in Fig. 9 (c)
It cuts:
(1) Fig. 9 (a), Fig. 9 (b) and Fig. 9 (c) are split respectively using classical Otsu method, the result point of segmentation
Not as shown in Figure 10 (a), Figure 10 (b) and Figure 10 (c)；
(2) using document 1 " a kind of improved Binary Sketch of Grey Scale Image method " respectively to Fig. 9 (a), Fig. 9 (b) and Fig. 9 (c)
It is split, the result of segmentation is respectively as shown in Figure 11 (a), Figure 11 (b) and Figure 11 (c)；
(3) using document 2 " the improvement image segmentation algorithm based on maximum between-cluster variance " respectively to Fig. 9 (a), Fig. 9 (b) and
Fig. 9 (c) is split, and the result of segmentation is respectively as shown in Figure 12 (a), Figure 12 (b) and Figure 12 (c)；
(4) using document 3 " a kind of image segmentation algorithm based on Two-dimensional Maximum inter-class variance " respectively to Fig. 9 (a), Fig. 9
(b) it is split with Fig. 9 (c), the result of segmentation is respectively as shown in Figure 13 (a), Figure 13 (b) and Figure 13 (c)；
It can be seen that above-mentioned three kinds of methods from above-mentioned three kinds of methods to be closer to the result that image is split, all hold
Vulnerable to influence of noise, noise immunity is poor.Fig. 9 (a), Fig. 9 (b) and Fig. 9 (c) are split using the present invention, the result of segmentation
Respectively as shown in Figure 14 (a), Figure 14 (b) and Figure 14 (c), it can be seen that method of the invention reduces the interference of noise, obtains
Preferable segmentation effect has stronger versatility.
Below using Y-PSNR (PSNR) as image segmentation quantizating index (figure when dividing with each method noiseless
As referring to image), the noiseproof feature of verifying several method is carried out from objective angle.Table 1 is three width picture strip the last 0.03
The segmentation data of Gaussian noise are spent, table 2 is the segmentation data of lena picture strip difference noise.
Three images of the table 1 with Gaussian noise divide quantized data
The segmentation data of 2 lena picture strip difference noise of table
In terms of the operation time of Tables 1 and 2, classical Otsu method, a kind of " the improved Binary Sketch of Grey Scale Image side of document 1
Method ", document 2 " the improvement image segmentation algorithm based on maximum between-cluster variance " three's arithmetic speed it is almost the same, be better than document 3
" a method of the image segmentation algorithm based on Two-dimensional Maximum inter-class variance " and it is of the invention.In terms of noise robustness, this hair
Bright method is apparently higher than preceding 4 in the Y-PSNR (PSNR) of the image segmentation of Gaussian noise, multiplicative noise and poisson noise
Kind method, it is more a bit weaker than preceding 4 kinds of methods to the inhibition of salt-pepper noise.Method of the invention is in processing low signal-to-noise ratio in table 2
When infrared image, remote sensing images, preceding 4 kinds of PSNR is generally 6~7 or so, and remote sensing images after method segmentation of the invention
PSNR is 14.2210 (being higher by first 2 times of 4 kinds of methods or so), and it is preceding 4 kinds of algorithms that the infrared image PSNR of segmentation, which is 18.4840,
3 times or so, show that method of the invention is optimal in the noiseproof feature of segmentation low signal-to-noise ratio (SNR) images.
Particular embodiments described above has carried out further in detail the purpose of the present invention, technical scheme and beneficial effects
It describes in detail bright, it should be understood that the above is only a specific embodiment of the present invention, is not intended to restrict the invention, it is all
Within the spirit and principles in the present invention, any modification, equivalent substitution, improvement and etc. done should be included in guarantor of the invention
Within the scope of shield.
Claims (3)
1. a kind of anti-noise Otsu image segmentation method based on two generation wavelet transformations, characterized by the following steps:
Step S1: it treats segmented image and carries out the decomposition of bis- generation of db1 small echo integer transform single layer, obtain an integer low frequency sub-band ca
With the decomposition coefficient of three high-frequency sub-bands ch, cv and cd；
Step S2: negative in low frequency sub-band ca is become into the new low frequency sub-band ca1 of 0 composition, and 3 × 3 mean filters are carried out to ca1
It is rounded afterwards and constitutes ca2；
Step S3: the coefficient maximum value M of ca2 is sought, in the inter-class variance of [0, M] interval computation ca2It is obtained using Otsu algorithm
Segmentation threshold T；
Step S4: maximum value M is changed to greater than T to the coefficient of ca2, the coefficient of ca2 is changed to 0 less than T, is formed newly after change
Low frequency sub-band ca3；
Step S5: being all changed to 0 for the decomposition coefficient of three high-frequency sub-bands ch, cv and cd, constitute new high-frequency sub-band ch1,
Cv1 and cd1；
Step S6: ca3, ch1, cv1 and cd1 are reconstructed using bis- generation of db1 small echo integer, the image W after constituting segmentation.
2. a kind of anti-noise Otsu image segmentation method based on two generation wavelet transformations according to claim 1, feature exist
In: in step s3, so that inter-class varianceMaximum value is the segmentation threshold T of Otsu algorithm.
3. a kind of anti-noise Otsu image segmentation method based on two generation wavelet transformations according to claim 2, feature exist
In: the range of segmentation threshold T is [0, M].
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710134347.9A CN106952272B (en) | 2017-03-08 | 2017-03-08 | A kind of anti-noise Otsu image segmentation method based on two generation wavelet transformations |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710134347.9A CN106952272B (en) | 2017-03-08 | 2017-03-08 | A kind of anti-noise Otsu image segmentation method based on two generation wavelet transformations |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106952272A CN106952272A (en) | 2017-07-14 |
| CN106952272B true CN106952272B (en) | 2019-09-10 |
Family
ID=59467914
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710134347.9A Active CN106952272B (en) | 2017-03-08 | 2017-03-08 | A kind of anti-noise Otsu image segmentation method based on two generation wavelet transformations |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106952272B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110211086B (en) * | 2018-02-28 | 2023-04-28 | 腾讯科技（深圳）有限公司 | Image segmentation method, device and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102968798A (en) * | 2012-12-12 | 2013-03-13 | 北京航空航天大学 | SAR (Synthetic Aperture Radar) image sea-land segmentation method based on wavelet transform and OTSU threshold |
| CN103914848A (en) * | 2014-04-16 | 2014-07-09 | 浙江理工大学 | Farmland image segmentation method based on wavelet transform and maximum between-cluster variance |
| CN104036455A (en) * | 2013-03-06 | 2014-09-10 | 江苏理工学院 | Infrared image detail enhancement method based on second-generation wavelet |
| CN104198184A (en) * | 2014-08-11 | 2014-12-10 | 中国人民解放军空军工程大学 | Bearing fault diagnostic method based on second generation wavelet transform and BP neural network |
| CN105184752A (en) * | 2015-09-23 | 2015-12-23 | 成都融创智谷科技有限公司 | Image processing method based on wavelet transform |
| CN105306098A (en) * | 2015-11-12 | 2016-02-03 | 北京锐安科技有限公司 | Method and device for constructing the second generation wavelet kernel function |
-
2017
- 2017-03-08 CN CN201710134347.9A patent/CN106952272B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102968798A (en) * | 2012-12-12 | 2013-03-13 | 北京航空航天大学 | SAR (Synthetic Aperture Radar) image sea-land segmentation method based on wavelet transform and OTSU threshold |
| CN104036455A (en) * | 2013-03-06 | 2014-09-10 | 江苏理工学院 | Infrared image detail enhancement method based on second-generation wavelet |
| CN103914848A (en) * | 2014-04-16 | 2014-07-09 | 浙江理工大学 | Farmland image segmentation method based on wavelet transform and maximum between-cluster variance |
| CN104198184A (en) * | 2014-08-11 | 2014-12-10 | 中国人民解放军空军工程大学 | Bearing fault diagnostic method based on second generation wavelet transform and BP neural network |
| CN105184752A (en) * | 2015-09-23 | 2015-12-23 | 成都融创智谷科技有限公司 | Image processing method based on wavelet transform |
| CN105306098A (en) * | 2015-11-12 | 2016-02-03 | 北京锐安科技有限公司 | Method and device for constructing the second generation wavelet kernel function |
Non-Patent Citations (5)
| Title |
|---|
| Otsu image segmentation algorithm based on morphology and wavelet transformation;Gui-mei Zhang etc;《2011 3rd International Conference on Computer Research and Development》;20110331;全文 * |
| 一种结合小波变换与二维最大熵法的图像阈值分割方法;刘耀辉 等;《现代计算机(专业版)》;20050731;全文 * |
| 基于最大类间方差的改进图像分割算法;丁晓峰 等;《计算机工程与设计》;20151031;第36卷(第10期);全文 * |
| 基于标记分水岭算法的高分辨率遥感图像分割方法;孙 颖 等;《科 学 技 术 与 工 程》;20080630;第8卷(第11期);全文 * |
| 灰度图像的二维Otsu曲线阈值分割法;范九伦 等;《电 子 学 报》;20070430;第35卷(第4期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106952272A (en) | 2017-07-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111681174B (en) | Joint filtering method for inhibiting salt-pepper/Gaussian noise mixing target image | |
| Merry | Wavelet theory and applications: a literature study | |
| CN101303764B (en) | Method for self-adaption amalgamation of multi-sensor image based on non-lower sampling profile wave | |
| CN110246106B (en) | NSST domain flotation froth image enhancement and denoising method based on quantum harmony search fuzzy set | |
| CN110675340A (en) | Single image defogging method and medium based on improved non-local prior | |
| CN104036455A (en) | Infrared image detail enhancement method based on second-generation wavelet | |
| CN104849757A (en) | System and method for eliminating random noise in seismic signals | |
| CN107818547A (en) | The minimizing technology of the spiced salt and Gaussian mixed noise in a kind of sequence towards twilight image | |
| Haddad et al. | Wave atoms based compression method for fingerprint images | |
| CN113808042A (en) | SAR image denoising method based on wavelet transformation and generation countermeasure network | |
| Zhang et al. | Fusion of infrared and visible light images based on nonsubsampled shearlet transform | |
| Liu et al. | An effective wavelet-based scheme for multi-focus image fusion | |
| CN106952272B (en) | A kind of anti-noise Otsu image segmentation method based on two generation wavelet transformations | |
| CN113592729A (en) | Infrared image enhancement method for electrical equipment based on NSCT domain | |
| CN115187688A (en) | Fog map reconstruction method based on atmospheric light polarization orthogonal blind separation and electronic equipment | |
| CN116343051B (en) | Geological environment monitoring method and system based on remote sensing image | |
| Pai et al. | Medical color image enhancement using wavelet transform and contrast stretching technique | |
| TWI460667B (en) | Rebuilding method for blur fingerprint images | |
| CN109584190A (en) | A kind of contrast image Enhancement Method being limited based on wavelet transformation low frequency histogram | |
| CN113780301B (en) | Self-adaptive denoising machine learning application method for defending against attack | |
| Li et al. | A novel remote sensing image enhancement method using unsharp masking in nsst domain | |
| CN115187855A (en) | Seabed substrate sonar image classification method | |
| Garg et al. | Image denoising using curvelet transformation using log gabour filter | |
| Guo et al. | An Image Denoising Algorithm based on Kuwahara Filter | |
| CN109300086B (en) | Image blocking method based on definition |
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 |