CN109064470A - A kind of image partition method and device based on adaptive fuzzy clustering - Google Patents
A kind of image partition method and device based on adaptive fuzzy clustering Download PDFInfo
- Publication number
- CN109064470A CN109064470A CN201810989431.3A CN201810989431A CN109064470A CN 109064470 A CN109064470 A CN 109064470A CN 201810989431 A CN201810989431 A CN 201810989431A CN 109064470 A CN109064470 A CN 109064470A
- Authority
- CN
- China
- Prior art keywords
- image
- pixel
- processed
- matrix
- objective function
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- 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
- G06T2207/10072—Tomographic images
- G06T2207/10081—Computed x-ray tomography [CT]
-
- 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
- G06T2207/20004—Adaptive image processing
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30061—Lung
- G06T2207/30064—Lung nodule
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Image Analysis (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The present invention relates to technical field of image processing, especially a kind of image partition method and device based on adaptive fuzzy clustering.By obtaining image to be processed, weight factor is defined further according to gray scale fluctuation situation in the feature and neighborhood of pixels of pixel, each pixel traversed in the image to be processed obtains weight matrix;Then objective function is constructed according to weight matrix, fuzzy item and bound term, and obtain objective function in local minimum, obtain corresponding subordinated-degree matrix and cluster centre, image to be processed is split according to the subordinated-degree matrix and cluster centre, it realizes the full-automatic and high-precision segmentation to image, solves the problems, such as that existing C mean cluster segmentation need to be operated by customized tonal gradation regulationization to select fuzzy factor to cause its noise robustness poor.
Description
Technical field
The present invention relates to technical field of image processing, especially a kind of image partition method based on adaptive fuzzy clustering
And device.
Background technique
Currently, fuzzy clustering is because its unsupervised characteristic and understandable logical language are in image procossing, pattern-recognition, meter
Calculation machine vision etc. is multi-field to be widely used, wherein the theory of fuzzy C-means clustering (FCM) algorithm based on objective function is most
To be perfect, using also the most extensive.Traditional FCM algorithm is real by the distance and degree of membership of iteration more new samples to cluster centre
It now clusters, uses value for the membership of a pattern in [0,1] section paste set statement sample and each cluster centre.
Lung cancer is that morbidity and mortality growth is most fast, to one of population health and the maximum malignant tumour of life threat.
Many countries all report that the morbidity and mortality of lung cancer obviously increase in the past 50 years.National Cancer Center announces system within 2015
It counts and shows lung cancer morbidity rate close to 17.09%, wherein male accounts for therein 70.3%, and women accounts for therein 29.7%, extremely
The rate of dying is up to 21.68%, ranks first place in various tumours, the biggest threat that will be faced as human health.Patients with lung cancer exists
If early stage obtains standardization operative treatment, 5 annual survival rates are up to 90% or more, treat in I phase of lung cancer, and 5 annual survival rates are 60%,
But II-IV phase, 5 annual survival rate is reduced to 5% from 40%.In order not to miss optimal treatment period, it is necessary to hair as early as possible
Existing and treated as soon as possible.Lung neoplasm is that the lung cancer early stage most common form of expression in order to improve the recall rate of Lung neoplasm utilizes CT
The auxiliary diagnosis of image progress Lung neoplasm.Currently, explosive growth tendency is presented in CT images data, doctor will definitely increase in this way
Workload, cause to fail to pinpoint a disease in diagnosis and mistaken diagnosis.Studies have shown that accurately and effectively lung CT image segmentation can reduce calculation amount,
The efficiency of entire diagnostic system is improved, reduction is failed to pinpoint a disease in diagnosis and mistaken diagnosis, is played a significant role during pulmonary disease and functional assessment.
Currently, domestic and foreign scholars are to Lung neoplasm dividing method mainly include the following types: the Lung neoplasm of (1) based on region growing
Dividing method, this method are typically all in conjunction with threshold method.(2) the Lung neoplasm dividing method based on variable model, mainly
It is be split object by the boundary profile using object.(3) the Lung neoplasm dividing method based on filter, mainly
It is relatively good to the segmentation effect of frosted glass type tubercle.(4) the Lung neoplasm dividing method based on cluster, mainly according to sample data
Self-characteristic sample is belonged in its corresponding classification.Wherein C mean cluster dividing method is suitable for the lung knot of multiple types
The segmentation of section, it is more significant compared to C mean cluster dividing method applicability and effect for other dividing methods, still, pass
System C mean cluster dividing method needs to be operated according to customized tonal gradation regulationization to select fuzzy factor, this results in the C
Mean cluster dividing method noise robustness is poor, and dividing accuracy rate and robustness all cannot preferably improve.
Summary of the invention
The object of the present invention is to provide a kind of image partition method and device based on adaptive fuzzy clustering, to solve
Existing C mean cluster segmentation need to be operated by customized tonal gradation regulationization to select fuzzy factor to lead to its noise robustness
Poor problem.
In order to realize the full-automatic and high-precision segmentation to image, solving existing C mean cluster segmentation need to be by making by oneself
The problem that adopted tonal gradation regulation operates that fuzzy factor is selected to cause its noise robustness poor.The present invention provides a kind of base
In the image partition method of adaptive fuzzy clustering, comprising the following steps:
1) image to be processed is obtained, weight factor is defined according to gray scale fluctuation situation in pixel point feature and neighborhood of pixels
ak, each pixel for traversing the image to be processed obtains weight matrix a=(ak|k∈N);
2) objective function J is constructed according to weight matrix, fuzzy item and bound termARFCM, and by obtaining objective function
Local minimum obtains corresponding subordinated-degree matrix UnAnd cluster centreObjective function JARFCMIt is as follows:
Wherein, aiFor pixel xiCorresponding weight matrix, uijFor subordinating degree function, vjFor the cluster centre of jth class, xi
For pixel, GijFor fixed fuzzy factor, PijTo obscure item, QijFor bound term, N is total pixel of image to be processed;
3) according to subordinated-degree matrix UnAnd cluster centreThe image to be processed is split.
Beneficial effect is, by the more accurately accurate segmentation of the selection realization image of fuzzy factor, to improve segmentation
The precision and noise robustness of method.
Further, in order to eliminate apparent noise, treating capacity is reduced, obtains figure to be processed in above-mentioned apparatus in step 1)
As after also to the image to be processed pass sequentially through median filtering, binaryzation and edge repair operation handle to obtain it is to be processed
Substantial image.
Further, judge automatically whether central pixel point and field pixel are noise spot or edge to realize
Point, weight factor described in above-mentioned apparatus are as follows:
Wherein,Indicate neighborhood territory pixel point xkWith neighborhood mean-square valueAbsolute distance, σiIndicate neighborhood territory pixel
Variance, σ={ σi|i∈N}。
Further, in order to keep image texture to rise and fall the area mitigated in image segmentation process, and in order to will be lonely
Standing statue vegetarian refreshments is correctly divided into the classification belonging to it, and item P is obscured in the above methodijWith bound term QijFormula difference it is as follows:
Wherein, b is the control parameter that realm information influences intensity;NRIt is the pixel number of field set;NiIt is field set
Pixel;xrIt is the gray value of field pixel.
Further, in order to be accurately split to the Lung neoplasm of lung CT image, degree of membership described in the above method
Matrix UnAnd cluster centreFor by Lagrange multiplier method in constraint condition uij∈ [0,1],Under the conditions of make
Objective function JARFCMIt is obtained when obtaining local minimum, subordinating degree function uijAnd vjFormula it is as follows:
Wherein, urjFor field pixel xrBelong to the degree of membership of jth class.
For the ease of realize the above method, a kind of image segmentation device based on adaptive fuzzy clustering, including memory,
Processor and storage in memory and the computer program that can run on a processor, the processor execution described program
When perform the steps of
1) image to be processed is obtained, weight factor is defined according to gray scale fluctuation situation in pixel point feature and neighborhood of pixels
ak, each pixel for traversing the image to be processed obtains weight matrix a=(ak|k∈N);
2) objective function J is constructed according to weight matrix, fuzzy item and bound termARFCM, and by obtaining objective function
Local minimum obtains corresponding subordinated-degree matrix UnAnd cluster centreObjective function JARFCMIt is as follows:
Wherein, aiFor pixel xiCorresponding weight matrix, uijFor subordinating degree function, vjFor the cluster centre of jth class, xi
For pixel, GijFor fixed fuzzy factor, PijTo obscure item, QijFor bound term, N is total pixel of image to be processed;
3) according to subordinated-degree matrix UnAnd cluster centreThe image to be processed is split.
Further, in order to eliminate apparent noise, treating capacity is reduced, obtains figure to be processed in above-mentioned apparatus in step 1)
As after also to the image to be processed pass sequentially through median filtering, binaryzation and edge repair operation handle to obtain it is to be processed
Substantial image.
Further, judge automatically whether central pixel point and field pixel are noise spot or edge to realize
Point, weight factor described in above-mentioned apparatus are as follows:
Wherein,Indicate neighborhood territory pixel point xkWith neighborhood mean-square valueAbsolute distance, σiIndicate neighborhood territory pixel
Variance, σ={ σi|i∈N}。
Further, in order to keep image texture to rise and fall the area mitigated in image segmentation process, and in order to will be lonely
Standing statue vegetarian refreshments is correctly divided into the classification belonging to it, and item P is obscured in above-mentioned apparatusijWith bound term QijFormula difference it is as follows:
Wherein, b is the control parameter that realm information influences intensity;NRIt is the pixel number of field set;NiIt is field set
Pixel;xrIt is the gray value of field pixel.
Further, in order to be accurately split to the Lung neoplasm of lung CT image, degree of membership described in above-mentioned apparatus
Matrix UnAnd cluster centreFor by Lagrange multiplier method in constraint condition uij∈ [0,1],Under the conditions of make mesh
Scalar functions JARFCMIt is obtained when obtaining local minimum, subordinating degree function uijAnd vjFormula it is as follows:
Wherein, urjFor field pixel xrBelong to the degree of membership of jth class.
Detailed description of the invention
Fig. 1 is the algorithm flow chart of the image partition method of the invention based on adaptive fuzzy clustering;
Fig. 2 (a) is the CT original image of stand alone Lung neoplasm split-run test;
Fig. 2 (b) is the pulmonary parenchyma figure of stand alone Lung neoplasm split-run test;
Fig. 2 (c) is the label result figure of stand alone Lung neoplasm split-run test;
Fig. 2 (d) is the segmentation result figure of the stand alone Lung neoplasm split-run test of the method for the present invention;
Fig. 2 (e) is the segmentation result figure of the stand alone Lung neoplasm split-run test based on FCM algorithm;
Fig. 2 (f) is the segmentation result figure of the stand alone Lung neoplasm split-run test based on FCM-S algorithm;
Fig. 2 (g) is the stand alone Lung neoplasm split-run test segmentation result figure based on FLICM algorithm;
Fig. 2 (h) is the segmentation result figure of the stand alone Lung neoplasm split-run test based on rFCM algorithm;
Fig. 3 (a) is the CT original image of ground glass type Lung neoplasm split-run test;
Fig. 3 (b) is the pulmonary parenchyma figure of ground glass type Lung neoplasm split-run test;
Fig. 3 (c) is the label result figure of ground glass type Lung neoplasm split-run test;
Fig. 3 (d) is the segmentation result figure of the ground glass type Lung neoplasm split-run test of the method for the present invention;
Fig. 3 (e) is the segmentation result figure of the ground glass type Lung neoplasm split-run test based on FCM algorithm;
Fig. 3 (f) is the segmentation result figure of the ground glass type Lung neoplasm split-run test based on FCM-S algorithm;
Fig. 3 (g) is the segmentation result figure of the ground glass type Lung neoplasm split-run test based on FLICM algorithm;
Fig. 3 (h) is the segmentation result figure of the ground glass type Lung neoplasm split-run test based on rFCM algorithm;
Fig. 4 is the accuracy rate comparison diagram of dividing method and other methods of the invention;
Fig. 5 is the recall rate comparison diagram of dividing method and other methods of the invention;
Fig. 6 is the F1 value comparison diagram of dividing method and other methods of the invention.
Specific embodiment
The present invention will be further described in detail with reference to the accompanying drawing.
The present invention provides a kind of image partition method based on adaptive fuzzy clustering, which can be used for figure
As the image procossing in the fields such as processing, pattern-recognition and computer vision, comprising the following steps:
1) image to be processed is obtained, weight factor is defined according to gray scale fluctuation situation in pixel point feature and neighborhood of pixels
ak, each pixel for traversing image to be processed obtains weight matrix a=(ak|k∈N)。
2) objective function J is constructed according to weight matrixARFCM, and by obtaining objective function in local minimum, it obtains
Corresponding subordinated-degree matrix UnAnd cluster centreObjective function JARFCMIt is as follows:
Wherein, aiFor pixel xiCorresponding weight matrix, uijFor subordinating degree function, vjFor the cluster centre of jth class, xi
For pixel, GijFor fixed fuzzy factor, PijTo obscure item, QijFor bound term, N is total pixel of image to be processed.
Objective function in this step, fuzzy item and bound term are selected by weight matrix, and the factor of fuzzy Xiang Suocheng is
Weight matrix ai, work as aiWhen being 0, obscure item be 0, therefore, select bound term, on the contrary it is then select obscure item as in objective function
One.
3) according to subordinated-degree matrix UnAnd cluster centreImage to be processed is split.
The present invention using lung CT image as image to be processed for be illustrated, adopt obtain with the aforedescribed process it is to be processed
Lung CT image, and median filtering, binaryzation and edge repairing behaviour also are passed sequentially through to the lung CT image to be processed
It deals with to obtain pulmonary parenchyma image.
The present invention is handled original image by median filtering, eliminates noise to the shadow of final image quality with this
It rings.Start a certain pixel selected in image, establishes local window centered on the point, the pixel for secondly including window
Point brightness value does sorting operation, it can obtains the brightness sequence { p of target point adjacent pixel1,p2…pm,…pn, wherein n is
Sum of all pixels in target pixel points adjacent window apertures, m=(n+1)/2, then pmBrightness value be assigned to pi, will be in image
Each pixel is carried out this operation.Binarization operation is carried out using gray threshold to the CT image after filtering processing,
Then it is repaired by the variation of edge, pulmonary parenchyma image can be obtained.
In step 1), weight factor a is defined according to gray scale fluctuation situation in the feature and neighborhood of pixels of pixelk, time
The each pixel gone through in pulmonary parenchyma image obtains weight matrix a=(ak|k∈N)。
The improved algorithm based on FCM algorithm that the present invention uses, wherein in order to improve precision and inhibit noise, in target
Introduced in function can judge automatically central pixel point and neighborhood territory pixel point whether be noise spot or marginal point weight factor,
Weight factor is defined according to gray scale fluctuation situation in the feature and neighborhood of pixels of pixel.
Wherein,
Indicate neighborhood territory pixel point xkWith neighborhood mean-square valueAbsolute distance, σiIndicate the variance of neighborhood territory pixel, N
For total pixel of image.
If the distance is greater than the variances sigma of neighborhood territory pixeli, then illustrate xkIt is the noise spot or marginal point of neighborhood territory pixel point,
Weight factor a at this timekIt is 0;Conversely, xkFor the pixel inside region, can be weighed according to gray scale fluctuation situation in neighborhood
Repeated factor ak, then by each pixel in traversal image, obtain weight matrix a=(ak| k ∈ N), σ=(σi|i∈N)。
In step 2), objective function J is constructed according to weight matrixARFCM, and by obtaining objective function in Local Minimum
Value, obtains corresponding subordinated-degree matrix UnAnd cluster centre
It is opposite to the result of image segmentation in the objective function of the variation of fuzzy set theory factor m value size in the present invention
It influences less, so the value of m takes 2, then JARFCMFormula is as follows:
Wherein, aiFor pixel xiCorresponding weight matrix, uijFor subordinating degree function, vjFor the cluster centre of jth class, xi
For pixel, GijFor fixed fuzzy factor, PijTo obscure item, QijFor bound term, i is the subscript of pixel, and j is class categories
Subscript.
Above-mentioned fixation fuzzy factor GijIncluding spatial information and gray-scale information, with neighborhood territory pixel to center pixel away from
From its influence is controlled, the robustness and noise sensitive of segmentation, and the selection without any parameter are improved,
Wherein, urjField pixel xrBelong to the degree of membership of jth class.
Bound term not only takes into account using target pixel points adjacent pixel as reference information in the above method, but also to mould
Paste is subordinate to angle value and is limited, so that the area for keeping image texture fluctuating to mitigate in image segmentation process, and can will isolate
Pixel is correctly divided into the classification belonging to it, obscures item PijWith bound term QijFormula difference it is as follows:
Wherein, b is the control parameter that realm information influences intensity;NRIt is the pixel number of field set;NiIt is field set
Pixel;xrIt is the gray value of field pixel.
Further, in order to be accurately split to the Lung neoplasm of lung CT image, pass through in the above method
Lagrange multiplier method is in constraint condition uij∈ [0,1],Under the conditions of make objective function JARFCMObtain Local Minimum
The subordinated-degree matrix U of valuenAnd cluster centre
The present invention is a kind of dividing method of algorithm based on adaptive fuzzy clustering, i.e. ARFCM (Adaptable
Robust Fuzzy C-mean Clustering), the basic procedure of ARFCM algorithm, as shown in Figure 1:
(1) input image data determines clusters number c, iteration stopping threshold epsilon and enables t=0;
(2) fuzzy clustering matrix is initialized using random method;
(3) weight factor, fixed fuzzy factor, fuzzy item and bound term are calculated;
(4) according to information such as weight factors, cluster centre is calculatedWith subordinated-degree matrix Un;
(5) it updates and calculates fuzzy membership matrix Un+1If { Un+1-Un< ε, algorithm stop iteration, otherwise enable t=t+
1 return step (3) continues operation.
Select the lung CT image of 40 width low dosages as original image from lung images database alliance (LIDC),
Including stand alone Lung neoplasm and ground glass type Lung neoplasm, respectively as shown in Fig. 2 (a) and Fig. 3 (a);After the processing such as binaryzation
Image such as Fig. 2 (b) and Fig. 3 (b) shown in;The comment file according to provided by database carries out Lung neoplasm boundary using curve
Label is as shown in Fig. 2 (c) and Fig. 3 (c);Segmentation result, that is, Fig. 2 (d) and Fig. 3 (d) of dividing method of the present invention, FCM method
Segmentation result such as Fig. 2 (e) and Fig. 3 (e), the segmentation result of FCM-S method such as Fig. 2 (f) and Fig. 3 (f), the segmentation of FLICM method
As a result such as Fig. 2 (g) and Fig. 3 (g), the segmentation result of rFCM method such as Fig. 2 (h) and Fig. 3 (h).It is of the invention effective in order to verify
Property, method of the invention is based respectively on the accuracy rate of pixel with four kinds of typical methods (FCM, FCM-S, FLICM, rFCM)
(Precision) it is compared with recall rate (Recall) and F1 value, first group is stand alone Lung neoplasm, and second group is mill glass
Glass type Lung neoplasm.
Accuracy rate (P) and recall rate (R) are defined as follows:
In formula, ArearealIt is that the comment file according to provided by database carries out Lung neoplasm boundary using red curve
Lung neoplasm region after label, as goldstandard, Area is the Lung neoplasm region obtained using partitioning algorithm.Work as parameter alpha2=1
When, it is exactly the most common evaluation index F1 value, combines the result of accuracy rate and recall rate.As F1 higher, then algorithm is proved
Segmentation effect it is better.
It can be seen that the accuracy rate of the method for the present invention is higher than other methods from Fig. 4, Fig. 5 and Fig. 6, and F1 value is above
85%, Lung neoplasm is divided validity and is improved.
Specific embodiment of the present invention is presented above, but the present invention is not limited to described embodiment.
Under the thinking that the present invention provides, to the skill in above-described embodiment by the way of being readily apparent that those skilled in the art
Art means are converted, are replaced, are modified, and play the role of with the present invention in relevant art means it is essentially identical, realize
Goal of the invention it is also essentially identical, the technical solution formed in this way is to be finely adjusted to be formed to above-described embodiment, this technology
Scheme is still fallen in protection scope of the present invention.
Claims (10)
1. a kind of image partition method based on adaptive fuzzy clustering, which comprises the following steps:
1) image to be processed is obtained, weight factor a is defined according to gray scale fluctuation situation in pixel point feature and neighborhood of pixelsk, time
The each pixel for going through the image to be processed obtains weight matrix a=(ak|k∈N);
2) objective function J is constructed according to weight matrix, fuzzy item and bound termARFCM, and by obtaining objective function in part
Minimum value obtains corresponding subordinated-degree matrix UnAnd cluster centreObjective function JARFCMIt is as follows:
Wherein, aiFor pixel xiCorresponding weight matrix, uijFor subordinating degree function, vjFor the cluster centre of jth class, xiFor picture
Vegetarian refreshments, GijFor fixed fuzzy factor, PijTo obscure item, QijFor bound term, N is total pixel of image to be processed;
3) according to subordinated-degree matrix UnAnd cluster centreThe image to be processed is split.
2. the image partition method according to claim 1 based on adaptive fuzzy clustering, which is characterized in that in step 1)
Also the image to be processed is passed sequentially through at median filtering, binaryzation and edge repair operation after obtaining image to be processed
Reason obtains substantial image to be processed.
3. the image partition method according to claim 1 or 2 based on adaptive fuzzy clustering, which is characterized in that described
Weight factor are as follows:
Wherein,Indicate neighborhood territory pixel point xkWith neighborhood mean-square valueAbsolute distance, σiIndicate the variance of neighborhood territory pixel,
σ={ σi|i∈N}。
4. the image partition method according to claim 3 based on adaptive fuzzy clustering, which is characterized in that fuzzy item Pij
With bound term QijFormula difference it is as follows:
Wherein, b is the control parameter that realm information influences intensity;NRIt is the pixel number of field set;NiIt is the picture of field set
Vegetarian refreshments;xrIt is the gray value of field pixel.
5. the image partition method according to claim 4 based on adaptive fuzzy clustering, which is characterized in that described to be subordinate to
Spend matrix UnAnd cluster centreFor by Lagrange multiplier method in constraint condition uij∈ [0,1],Under the conditions of
Make objective function JARFCMIt is obtained when obtaining local minimum, subordinating degree function uijAnd vjFormula it is as follows:
Wherein, urjFor field pixel xrBelong to the degree of membership of jth class.
6. a kind of image segmentation device based on adaptive fuzzy clustering, including memory, processor and it is stored in memory
In and the computer program that can run on a processor, which is characterized in that the processor is realized following when executing described program
Step:
1) image to be processed is obtained, weight factor a is defined according to gray scale fluctuation situation in pixel point feature and neighborhood of pixelsk, time
The each pixel for going through the image to be processed obtains weight matrix a=(ak|k∈N);
2) objective function J is constructed according to weight matrix, fuzzy item and bound termARFCM, and by obtaining objective function in part
Minimum value obtains corresponding subordinated-degree matrix UnAnd cluster centreObjective function JARFCMIt is as follows:
Wherein, aiFor pixel xiCorresponding weight matrix, uijFor subordinating degree function, vjFor the cluster centre of jth class, xiFor picture
Vegetarian refreshments, GijFor fixed fuzzy factor, PijTo obscure item, QijFor bound term, N is total pixel of image to be processed;
3) according to subordinated-degree matrix UnAnd cluster centreThe image to be processed is split.
7. the image segmentation device according to claim 6 based on adaptive fuzzy clustering, which is characterized in that in step 1)
Also the image to be processed is passed sequentially through at median filtering, binaryzation and edge repair operation after obtaining image to be processed
Reason obtains substantial image to be processed.
8. the image segmentation device according to claim 6 or 7 based on adaptive fuzzy clustering, which is characterized in that described
Weight factor are as follows:
Wherein,Indicate neighborhood territory pixel point xkWith neighborhood mean-square valueAbsolute distance, σiIndicate the variance of neighborhood territory pixel,
σ={ σi|i∈N}。
9. the image segmentation device according to claim 8 based on adaptive fuzzy clustering, which is characterized in that fuzzy item Pij
With bound term QijFormula difference it is as follows:
Wherein, b is the control parameter that realm information influences intensity;NRIt is the pixel number of field set;NiIt is the picture of field set
Vegetarian refreshments;xrIt is the gray value of field pixel.
10. the image segmentation device according to claim 9 based on adaptive fuzzy clustering, which is characterized in that the person in servitude
Category degree matrix UnAnd cluster centreFor by Lagrange multiplier method in constraint condition uij∈ [0,1],Under the conditions of
Make objective function JARFCMIt is obtained when obtaining local minimum, subordinating degree function uijAnd vjFormula it is as follows:
Wherein, urjFor field pixel xrBelong to the degree of membership of jth class.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810989431.3A CN109064470B (en) | 2018-08-28 | 2018-08-28 | Image segmentation method and device based on self-adaptive fuzzy clustering |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810989431.3A CN109064470B (en) | 2018-08-28 | 2018-08-28 | Image segmentation method and device based on self-adaptive fuzzy clustering |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109064470A true CN109064470A (en) | 2018-12-21 |
CN109064470B CN109064470B (en) | 2022-02-22 |
Family
ID=64757431
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810989431.3A Active CN109064470B (en) | 2018-08-28 | 2018-08-28 | Image segmentation method and device based on self-adaptive fuzzy clustering |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109064470B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110147816A (en) * | 2019-04-10 | 2019-08-20 | 中国科学院深圳先进技术研究院 | A kind of acquisition methods of color depth image, equipment, computer storage medium |
CN110223249A (en) * | 2019-05-31 | 2019-09-10 | 北京林业大学 | Soil aperture three-dimensional dividing method and system based on fuzzy clustering |
CN110349160A (en) * | 2019-06-25 | 2019-10-18 | 电子科技大学 | One kind is based on super-pixel and fuzzy C-means clustering SAR image segmentation method |
CN110428437A (en) * | 2019-08-14 | 2019-11-08 | 大连海事大学 | The GGO dividing method of SLIC and secondary Density Clustering based on edge sensitive |
CN111047593A (en) * | 2019-10-28 | 2020-04-21 | 东北大学 | Image segmentation method based on distance adaptive clustering |
CN111402272A (en) * | 2020-03-19 | 2020-07-10 | 西安石油大学 | Neighborhood weighted fuzzy clustering image segmentation method |
CN112651464A (en) * | 2021-01-12 | 2021-04-13 | 重庆大学 | Unsupervised or weakly supervised constrained fuzzy c-means clustering method |
CN115578388A (en) * | 2022-12-07 | 2023-01-06 | 滨州华然化纤绳网有限公司 | Data processing method for textile production |
Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8332346B1 (en) * | 2008-01-31 | 2012-12-11 | Quest Software, Inc. | Fuzzy-learning-based extraction of time-series behavior |
CN103761726A (en) * | 2013-12-25 | 2014-04-30 | 河海大学 | Partitioning self-adaptation image segmentation method based on FCM |
CN104123561A (en) * | 2014-07-10 | 2014-10-29 | 中国矿业大学 | Spatial gravity model based fuzzy c-means remote sensing image automatic classification method |
CN104751469A (en) * | 2015-04-02 | 2015-07-01 | 西安电子科技大学 | Image segmentation method based on Kernel Fuzzy C means clustering |
CN105261004A (en) * | 2015-09-10 | 2016-01-20 | 西安电子科技大学 | Mean shift and neighborhood information based fuzzy C-mean image segmentation method |
CN105654453A (en) * | 2014-11-10 | 2016-06-08 | 华东师范大学 | Robust FCM image segmentation method |
CN106600605A (en) * | 2016-12-14 | 2017-04-26 | 陕西科技大学 | Unsupervised fast image segmentation algorithm |
CN106846326A (en) * | 2017-01-17 | 2017-06-13 | 合肥工业大学 | Image partition method based on multinuclear local message FCM algorithms |
CN107316060A (en) * | 2017-06-16 | 2017-11-03 | 鲁东大学 | Fuzzy clustering image partition method based on pixel local message and non-local information |
CN107730516A (en) * | 2017-10-16 | 2018-02-23 | 江南大学 | A kind of brain MR image division methods based on fuzzy clustering |
CN108416380A (en) * | 2018-02-28 | 2018-08-17 | 北京理工大学 | A kind of big data clustering algorithm reducing customer churn risk |
-
2018
- 2018-08-28 CN CN201810989431.3A patent/CN109064470B/en active Active
Patent Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8332346B1 (en) * | 2008-01-31 | 2012-12-11 | Quest Software, Inc. | Fuzzy-learning-based extraction of time-series behavior |
CN103761726A (en) * | 2013-12-25 | 2014-04-30 | 河海大学 | Partitioning self-adaptation image segmentation method based on FCM |
CN104123561A (en) * | 2014-07-10 | 2014-10-29 | 中国矿业大学 | Spatial gravity model based fuzzy c-means remote sensing image automatic classification method |
CN105654453A (en) * | 2014-11-10 | 2016-06-08 | 华东师范大学 | Robust FCM image segmentation method |
CN104751469A (en) * | 2015-04-02 | 2015-07-01 | 西安电子科技大学 | Image segmentation method based on Kernel Fuzzy C means clustering |
CN105261004A (en) * | 2015-09-10 | 2016-01-20 | 西安电子科技大学 | Mean shift and neighborhood information based fuzzy C-mean image segmentation method |
CN106600605A (en) * | 2016-12-14 | 2017-04-26 | 陕西科技大学 | Unsupervised fast image segmentation algorithm |
CN106846326A (en) * | 2017-01-17 | 2017-06-13 | 合肥工业大学 | Image partition method based on multinuclear local message FCM algorithms |
CN107316060A (en) * | 2017-06-16 | 2017-11-03 | 鲁东大学 | Fuzzy clustering image partition method based on pixel local message and non-local information |
CN107730516A (en) * | 2017-10-16 | 2018-02-23 | 江南大学 | A kind of brain MR image division methods based on fuzzy clustering |
CN108416380A (en) * | 2018-02-28 | 2018-08-17 | 北京理工大学 | A kind of big data clustering algorithm reducing customer churn risk |
Non-Patent Citations (9)
Title |
---|
FENG ZHAO等: "A novel fuzzy clustering algorithm with non local adaptive spatial constraint for image segmentation", 《SIGNAL PROCESSING》 * |
STELIO KRINIDIS等: "A Robust Fuzzy Local Information C-Means Clustering Algorithm", 《IEEE TRANSACTIONS ON IMAGE PROCESSING》 * |
ZHAO ZAIXIN等: "Neighbourhood weighted fuzzy c-means clustering algorithm for image segmentation", 《IET IMAGE PROCESS》 * |
刘梦娇等: "一种改进的局部模糊C-均值聚类分割算法研究", 《计算机科学》 * |
朱占龙: "融入邻域距离与隶属度模糊聚类图像分割算法", 《微电子学与计算机》 * |
沈灏: "基于空间信息的核模糊C均值聚类图像分割", 《山西电子技术》 * |
田平平: "基于邻域信息和模糊聚类的图像分割算法研究", 《中国优秀博硕士学位论文全文数据库(硕士) 信息科技辑》 * |
霍冠英等: "空间约束FCM与MRF结合的侧扫声呐图像分割算法", 《仪器仪表学报》 * |
高国勇: "改进的模糊聚类遥感影像分割", 《测绘与空间地理信息》 * |
Cited By (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110147816A (en) * | 2019-04-10 | 2019-08-20 | 中国科学院深圳先进技术研究院 | A kind of acquisition methods of color depth image, equipment, computer storage medium |
CN110147816B (en) * | 2019-04-10 | 2023-04-07 | 中国科学院深圳先进技术研究院 | Method and device for acquiring color depth image and computer storage medium |
CN110223249B (en) * | 2019-05-31 | 2021-03-30 | 北京林业大学 | Fuzzy clustering-based soil pore three-dimensional segmentation method and system |
CN110223249A (en) * | 2019-05-31 | 2019-09-10 | 北京林业大学 | Soil aperture three-dimensional dividing method and system based on fuzzy clustering |
CN110349160A (en) * | 2019-06-25 | 2019-10-18 | 电子科技大学 | One kind is based on super-pixel and fuzzy C-means clustering SAR image segmentation method |
CN110349160B (en) * | 2019-06-25 | 2022-03-25 | 电子科技大学 | SAR image segmentation method based on super-pixel and fuzzy C-means clustering |
CN110428437A (en) * | 2019-08-14 | 2019-11-08 | 大连海事大学 | The GGO dividing method of SLIC and secondary Density Clustering based on edge sensitive |
CN110428437B (en) * | 2019-08-14 | 2022-11-29 | 大连海事大学 | GGO segmentation method based on edge-sensitive SLIC and quadratic density clustering |
CN111047593A (en) * | 2019-10-28 | 2020-04-21 | 东北大学 | Image segmentation method based on distance adaptive clustering |
CN111047593B (en) * | 2019-10-28 | 2023-06-02 | 东北大学 | Image segmentation method based on distance self-adaptive clustering |
CN111402272A (en) * | 2020-03-19 | 2020-07-10 | 西安石油大学 | Neighborhood weighted fuzzy clustering image segmentation method |
CN111402272B (en) * | 2020-03-19 | 2023-09-29 | 西安石油大学 | Neighborhood weighted fuzzy clustering image segmentation method |
CN112651464A (en) * | 2021-01-12 | 2021-04-13 | 重庆大学 | Unsupervised or weakly supervised constrained fuzzy c-means clustering method |
CN112651464B (en) * | 2021-01-12 | 2022-11-25 | 重庆大学 | Unsupervised or weakly supervised constrained fuzzy c-means clustering method |
CN115578388A (en) * | 2022-12-07 | 2023-01-06 | 滨州华然化纤绳网有限公司 | Data processing method for textile production |
Also Published As
Publication number | Publication date |
---|---|
CN109064470B (en) | 2022-02-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109064470A (en) | A kind of image partition method and device based on adaptive fuzzy clustering | |
Shenbagarajan et al. | Tumor diagnosis in MRI brain image using ACM segmentation and ANN-LM classification techniques | |
CN104933711B (en) | A kind of automatic fast partition method of cancer pathology image | |
Joshi et al. | Classification of brain cancer using artificial neural network | |
CN101169868B (en) | Method for segmenting a digital medical image | |
CN112101451B (en) | Breast cancer tissue pathological type classification method based on generation of antagonism network screening image block | |
CN109145921A (en) | A kind of image partition method based on improved intuitionistic fuzzy C mean cluster | |
CN104899926A (en) | Medical image segmentation method and device | |
CN109036547A (en) | A kind of lung CT image computer aided system and method based on clustering | |
CN107977952A (en) | Medical image cutting method and device | |
CN109978880A (en) | Lung tumors CT image is carried out sentencing method for distinguishing using high dimensional feature selection | |
CN101103924A (en) | Galactophore cancer computer auxiliary diagnosis method based on galactophore X-ray radiography and system thereof | |
CN105069818A (en) | Image-analysis-based skin pore identification method | |
EP3207521A1 (en) | Image analysis method supporting illness development prediction for a neoplasm in a human or animal body | |
CN109363699A (en) | A kind of method and device of breast image lesion identification | |
He et al. | Mammographic segmentation and risk classification using a novel binary model based bayes classifier | |
CN109472784A (en) | Based on the recognition methods for cascading full convolutional network pathological image mitotic cell | |
Koziarski et al. | DiagSet: a dataset for prostate cancer histopathological image classification | |
He et al. | Mammographic image segmentation and risk classification based on mammographic parenchymal patterns and geometric moments | |
CN114266729A (en) | Chest tumor radiotherapy-based radiation pneumonitis prediction method and system based on machine learning | |
Hassan et al. | A dilated residual hierarchically fashioned segmentation framework for extracting gleason tissues and grading prostate cancer from whole slide images | |
Song et al. | Automated segmentation of overlapping cytoplasm in cervical smear images via contour fragments | |
CN112508860B (en) | Artificial intelligence interpretation method and system for positive check of immunohistochemical image | |
Fan et al. | Research on abnormal target detection method in chest radiograph based on YOLO v5 algorithm | |
Cao et al. | 3D convolutional neural networks fusion model for lung nodule detection onclinical CT scans |
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 |