CN106919950A - Probability density weights the brain MR image segmentation of geodesic distance - Google Patents
Probability density weights the brain MR image segmentation of geodesic distance Download PDFInfo
- Publication number
- CN106919950A CN106919950A CN201710053148.5A CN201710053148A CN106919950A CN 106919950 A CN106919950 A CN 106919950A CN 201710053148 A CN201710053148 A CN 201710053148A CN 106919950 A CN106919950 A CN 106919950A
- Authority
- CN
- China
- Prior art keywords
- pixel
- super
- brain
- segmentation
- image
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- 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
- G06T2207/10088—Magnetic resonance imaging [MRI]
-
- 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
- G06T2207/20076—Probabilistic 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
- G06T2207/30016—Brain
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The invention discloses the brain MR image segmentation that a kind of probability density weights geodesic distance, belong to technical field of image processing.Including:Some width simulation brain database images are read in, statistics with histogram is carried out, the interval sample value of most integrated distribution is drawn;Each pixel on pending image using the sample value for obtaining as priori to selection carries out Multilayer networks;Super-pixel segmentation is carried out to pending image based on probability density function;Super-pixel after segmentation is scanned, filter out non-compliant super-pixel and enter line splitting, it is two classes to be gathered all pixels in super-pixel again with FCM algorithms, connected region is found according to classification results, and using the pixel in each connected region as a new class, update super-pixel classification results matrix;Clustered with the basis of super-pixel of the FCM algorithms after all renewals, obtained pending image brain tissue segmentation result.The present invention improves the degree of accuracy of super-pixel segmentation and brain tissue's segmentation.
Description
Technical field
The present invention relates to technical field of image processing, a kind of brain MR figures of probability density weighting geodesic distance are particularly related to
As dividing method.
Background technology
Image segmentation is one of most classical research topic in the fields such as image procossing, graphical analysis and computer vision,
It is one of maximum difficult point, image Segmentation Technology plays pivotal player in many medical image applications, is also each in image
The further basis analyzed of the pathology of tissue and organ is planted, by using image segmentation, region interested in image is carried
Take out, be that clinical diagnosis and treatment etc. provide foundation, and brain is the vitals of human body, therefore study dividing for brain area
Cut diagnosis equal important in inhibiting of the technology for brain three-dimensional reconstruction, the research of neural circuitry and clinical brain diseases.
Super-pixel segmentation is a kind of image over-segmentation algorithm, can be as the pretreatment work in some image applications, example
Such as segmentation, conspicuousness detection, recognition of face.Super-pixel can greatly reduce successive image treatment with the redundancy in capture images
The complexity of task.A kind of existing effective superpixel segmentation method is based on geodesic distance (Geodesic Distance)
Super-pixel, with geodesic distance rather than Euclidean distance measure pixel between similarity, for natural image split
Effect is pretty good, but for brain MR (Magnetic Resonance, abbreviation MR) image, the method can not be separated often exactly
One tiny brain tissue's region super-pixel block.
Fuzzy C-Means Cluster Algorithm (Fuzzy C-Means, abbreviation FCM) is the fuzzy clustering image being most widely used
Partitioning algorithm.Relative to other dividing methods, FCM can retain more information of initial pictures.However, traditional FCM is calculated
Method fail in image segmentation consider each point gray feature and its neighborhood territory pixel correlation degree, result in the algorithm for
Noise and the uneven comparing of gray scale are sensitive, regarding to the issue above, it has been proposed that many improved FCM algorithms, although improved
Method is improved to some extent at the aspect such as anti-noise or efficiency, but due to the high complexity of brain image, can not still obtain
Gratifying segmentation result, therefore actual requirement can not be met using traditional single method segmentation.
The content of the invention
The present invention provides the brain MR image segmentation that a kind of probability density weights geodesic distance, which raises super-pixel
Segmentation and the degree of accuracy of brain tissue's segmentation.
In order to solve the above technical problems, present invention offer technical scheme is as follows:
A kind of probability density weights the brain MR image segmentation of geodesic distance, including:
Step 1:Read in some width simulation brain database images, statistics with histogram carried out to it, draw white matter, grey matter or
The interval sample value of cerebrospinal fluid most integrated distribution;
Step 2:Piece image is randomly selected as pending image from the simulation brain database images, using obtaining
The interval sample value of the most integrated distribution to carry out probability to each pixel on pending image as priori close
Degree estimation, obtains probability density function;
Step 3:Super-pixel segmentation is carried out to pending image based on the probability density function for obtaining, and is recorded super
Pixel classifications matrix of consequence;
Step 4:Super-pixel after segmentation is scanned, is filtered out according to super-pixel color standard difference non-compliant
Super-pixel enters line splitting, and during division, it is two classes to be gathered all pixels in super-pixel again with FCM algorithms, is tied according to classification afterwards
Fruit finds connected region, and using the pixel in each connected region as a new class, updates the super-pixel classification results square
Battle array;
Step 5:According to the super-pixel classification results matrix after updating, the super-pixel base with FCM algorithms after all renewals
Clustered on plinth, obtained pending image brain tissue segmentation result.
The invention has the advantages that:
Probability density of the invention weights the brain MR image segmentation of geodesic distance, first draws statistics with histogram
Sample value Multilayer networks are carried out to each pixel on image as priori value, then weighted with based on probability density
The super-pixel method of geodesic distance carries out super-pixel segmentation to image, then scans super-pixel and filters out non-compliant super picture
Element enters line splitting, and merokinesis optimization is carried out using super-pixel internal feature, and super-pixel classification results square is updated after the completion of division
Battle array, is clustered again on the basis of being finally based on the super-pixel split with FCM algorithms, completes the segmentation of brain image.The method
At least have the following advantages that:(1) design the new weights influence factor to define geodesic distance, incorporated probability density function, make
Contrast more obvious between brain different tissues, gradient calculation is more reasonable;(2) process of local segmentation post processing is increased, it is right
Super-pixel carries out merokinesis, and pixel is more accurately sorted out, and further increases the degree of accuracy of super-pixel segmentation, that is, use
Most simple most traditional FCM carries out last cluster, as a result still fine;(3) by based on probability density weighting geodesic distance
The methods such as super-pixel segmentation technology, FCM are connected, and brain tissue's segmentation result is obtained on the basis of super-pixel.
Brief description of the drawings
Fig. 1 is the schematic flow sheet of the brain MR image segmentation of probability density weighting geodesic distance of the invention;
Fig. 2 is the principle schematic of the brain MR image segmentation of probability density weighting geodesic distance of the invention;
Fig. 3 be probability density of the invention weighting geodesic distance brain MR image segmentation in the flow of step 3 show
It is intended to;
Fig. 4 is that the idiographic flow of the brain MR image segmentation of probability density weighting geodesic distance of the invention is illustrated
Figure;
Fig. 5 (a):It is pending image of the invention:Synthesis brain MR example images;
Fig. 5 (b):It is the super-pixel segmentation figure for synthesizing brain MR example images corresponding to Fig. 5 (a) of the invention;
Fig. 5 (c):It is according to regions such as the brain MR example images of present invention realization final white matter, grey matter, cerebrospinal fluid
Segmentation result figure;
Fig. 6 (a):It is pending image of the invention:True brain MR example images;
Fig. 6 (b):It is the super-pixel segmentation figure corresponding to the true brain MR example images of Fig. 6 (a) of the invention;
Fig. 6 (c):It is according to regions such as the brain MR example images of present invention realization final white matter, grey matter, cerebrospinal fluid
Segmentation result figure.
Specific embodiment
To make the technical problem to be solved in the present invention, technical scheme and advantage clearer, below in conjunction with accompanying drawing and tool
Body embodiment is described in detail.
The present invention provides the brain MR image segmentation that a kind of probability density weights geodesic distance, as shown in figures 1 to 6, bag
Include:
Step 1:Read in some width simulation brain database images, statistics with histogram carried out to it, draw white matter, grey matter or
The interval sample value of cerebrospinal fluid most integrated distribution;
In this step, the simulation brain database images of reading are the synthesis brain MR image of bmp forms, read in its gray scale
Value.
Step 2:Piece image is randomly selected as pending image from simulation brain database images, using obtaining most
The interval sample value of integrated distribution carries out Multilayer networks as priori to each pixel on pending image, obtains
To probability density function;
Step 3:Super-pixel segmentation is carried out to pending image based on the probability density function for obtaining, and records super-pixel
Classification results matrix;
Step 4:Super-pixel after segmentation is scanned, is filtered out according to super-pixel color standard difference non-compliant
Super-pixel enters line splitting, and during division, it is two classes to be gathered all pixels in super-pixel again with FCM algorithms, is tied according to classification afterwards
Fruit finds connected region, and using the pixel in each connected region as a new class, updates super-pixel classification results matrix;
Step 5:According to the super-pixel classification results matrix after updating, the super-pixel base with FCM algorithms after all renewals
Clustered on plinth, obtained pending image brain tissue segmentation result.
Probability density of the invention weights the brain MR image segmentation of geodesic distance, first draws statistics with histogram
Sample value Multilayer networks are carried out to each pixel on image as priori value, then weighted with based on probability density
The super-pixel method of geodesic distance carries out super-pixel segmentation to image, then scans super-pixel and filters out non-compliant super picture
Element enters line splitting, and merokinesis optimization is carried out using super-pixel internal feature, and super-pixel classification results square is updated after the completion of division
Battle array, is clustered again on the basis of being finally based on the super-pixel split with FCM algorithms, completes the segmentation of brain image.The method
At least have the following advantages that:(1) design the new weights influence factor to define geodesic distance, incorporated probability density function, make
Contrast more obvious between brain different tissues, gradient calculation is more reasonable;(2) process of local segmentation post processing is increased, it is right
Super-pixel carries out merokinesis, and pixel is more accurately sorted out, and further increases the degree of accuracy of super-pixel segmentation, that is, use
Most simple most traditional FCM carries out last cluster, as a result still fine;(3) by based on probability density weighting geodesic distance
The methods such as super-pixel segmentation technology, FCM are connected, and brain tissue's segmentation result is obtained on the basis of super-pixel.
Preferably, step 1 is further:
Some width simulation brain MR (being obtained from BrainWeb databases) images are read in, K-means algorithms are first used it
Preliminary classification is carried out, N number of interval is divided into after then gray value is normalized, to the gray value of its white matter, grey matter or cerebrospinal fluid
Scope is counted, and count white matter, grey matter or cerebrospinal fluid most integrated distribution K is interval, in this K interval, each
M gray value is chosen as sample in interval, and the numerical value of the sample is the interval sample value of most integrated distribution.In this step, N,
K, m are the integer more than 0.
In the present invention, due to having coincidence between brain image grey matter, white matter, the grey scale pixel value of cerebrospinal fluid different tissues
Place, between them the definite division border of neither one, and the classification that each pixel belongs to is fuzzy, directly to it
Segmentation is difficult.So, the present invention carries out statistics with histogram first, the sample value that will be obtained as priori, for
Multilayer networks are carried out to each pixel on image in next step.
Further, step 2 includes:
Step 21:Calculate K probability Estimation model of each pixel in pending image respectively according to formula (1):
Wherein, x is the gray value of each pixel in image, xiIt is the individual interval of kth (k=1,2 ..., K) as priori
M gray value sample of value, h is bandwidth (control parameter);
Step 22:The K probability Estimation model that will be obtained calculates mixing probability density function by formula (2), obtains every
The Multilayer networks value of individual pixel, i.e. each pixel belong to the possibility of white matter, grey matter or cerebrospinal fluid:
Wherein, p (k) is factor of influence of each probability Estimation model to data point, between 0~1;
Step 23:Mixing probability density function to obtaining is normalized, and obtains probability density function:
Wherein, wmax and wmin are the maximum and minimum value of P (x).
In the present invention, probability density function is incorporated, can made between the brain different tissues such as grey matter, white matter, cerebrospinal fluid
Contrast it is more obvious, and make our follow-up gradient calculations more accurate.
As a modification of the present invention, as shown in figure 3, step 3 includes:
Step 31:Initialization seed point, is first uniformly distributed n/2 seed point on image, then uses self adaptation hexagon
(Adaptive Hexagonal) method inserts other seed points, and the complexity of each seed point hexagon is calculated with formula (4), looks for
To the hexagon that complexity is maximum, it is divided into six small hexagons of overlap, is inserted in the maximum small hexagon of complexity
One new seed point, iteration successively, until n cluster centre of sampling, complexity is defined as:
Wherein, HiIt is seed point siHexagonal area, N is the pixel quantity of image I, and M is to meet condition in braces
Pixel p quantity, α is control parameter, It is the ladder of pixel p on image
Degree, GσIt is the Gaussian function with standard deviation sigma, ω is a regulation parameter, is preventedIt is zero situation;
Step 32:Seed point is upset, each seed point is moved to the minimum gradient locations in its local 3*3 region, and remember
Its X, Y-coordinate information are recorded as new seed point;
Step 33:The seed point sensitising gradient based on probability density is calculated with formula (5):
PSSG(si, G (t))=| | Sp(si,G(t))|| (5)
Wherein,SpIt is the P calculated with sobel operators on geodetic pathwThe gradient of (x);
Step 34:Calculated using FMM (fast marching method) algorithm and geodesic distance weighted based on probability density,
Row bound of going forward side by side spreads, and produces a series of pixel composition super-pixel block with like attribute;
In this step, A:Geodesic distance is calculated by the following method:
From seed point siOne to any one pixel p probability density weighting geodesic distance is described as:From seed point siOpen
Begin to reach pixel p along a most short path, every is multiplied by a weighting function W (s on pathi, G (t)) minimum arc
Integration long, it is defined as:
Wherein, G (t) is from seed point siTo a geodetic path between pixel, t is continually changing, and takes 0~1
Between value, weight W is set to the gradient that a pixel belongs to white matter, grey matter or cerebrospinal fluid possibility, for defining from kind
Sub- point siDistance increment onto certain pixel path G:
Geodesic distance is calculated by the expanding policy of the Fast Marching Method (FMM) with appropriate velocity field iterative diffusion
, velocity function is defined based on formula (7), its computing formula is:
B:Boundary diffusion, produces super-pixel block:
In diffusion process, the new speed of each pixel is no longer static, depends on the seed point nearest from it
(with it there is most short probability density to weight the pixel of geodesic distance), since given seed point, is counted with formula (5)
The gradient of its neighborhood territory pixel point is calculated, is extended along the pixel with maximal rate formula (8), the pixel color value after extension
To be replaced by the color value of seed point, the probability seed point sensitising gradient of next pixel can constantly pass through diffusing through for FMM
Journey is calculated with formula (5) with the value of current neighborhood territory pixel point and updated, and is spread forward every time from the minimum (speed of seed point geodesic distance
Degree function is maximum) pixel, finished until all of pixel all spreads, obtain the classification results matrix of super-pixel.
Step 35:Position and the color of seed point are updated with publicity (9);
Wherein, SlIt is l-th super-pixel, slIt is super-pixel SlSeed point, xl', cl' respectively be update after seed point position
Put and color value,Measurement pixel is subordinate to the weighting function of seed point degree;
Step 36:Repeat step 33, step 34, step 35, algorithm are intended to optimize an energy function, are defined as formula
(10), when the change of energy function in subsequent iteration twice just stops less than a specific threshold, preliminary super-pixel
Segmentation is completed,
In the present invention, for the high complexity of brain image, design the new weights influence factor to define geodesic distance, make
With new gradient calculation method, a more obvious border is had at the fuzzy region edge of brain MR image, can be with calibrated
True Ground Split goes out each tiny brain tissue's region super-pixel block.
Further, step 4 includes:
Step 41:Super-pixel after scanning segmentation, if the color standard difference of super-pixel is more than certain threshold value Tc, then to full
The super-pixel of sufficient this condition enters line splitting, and its calculating is defined as:
C(sl)=λ Stdl> Tc (11)
Wherein, StdlIt is the standard deviation of pixel color in each super-pixel, TcIt is threshold value, λ is control parameter;
Step 42:It is two classes, local FCM to be gathered all pixels in the super-pixel of each needs division again with FCM algorithms
Object function be:
Wherein, C is the number of cluster, QnIt is the number of pixel in the super-pixel of Current Scan, μijIt is j-th pixel category
In the fuzzy membership function of ith cluster, constraint μ is metij∈[0,1],M is acted on fuzzy membership
Weighting function, viIt is ith cluster center, xjIt is the color value of pixel in the super-pixel of Current Scan;
Step 43:Connected region is found according to the super-pixel classification results for obtaining, and the pixel in each connected region
Used as a new class, i.e., one new super-pixel block updates the matrix of consequence of super-pixel classification;
Step 44:To needing the super-pixel repeat step 42, step 43 of division until all super pictures for meeting splitting condition
Element all divisions are completed.
In the present invention, local segmentation post processing is carried out after the completion of primary segmentation again, office is carried out using super-pixel internal feature
Optimization is split in part, pixel is more accurately sorted out, and improves the segmentation accuracy of super-pixel, allows the division of super-pixel more
Brain different tissues are separated by further cluster exactly.
Preferably, step 5 is further:
Using super-pixel (rather than pixel) as the input of FCM algorithms, gathered on the basis of the super-pixel split
Class, completes the segmentation of brain image, and the FCM object functions based on super-pixel are:
Wherein, C is the number of cluster, and Q is the number of super-pixel in image, μijIt is that j-th super-pixel belongs to ith cluster
Fuzzy membership function, meet constraint μij∈[0,1],M is to act on the weighting function on fuzzy membership,
viIt is ith cluster center, ξ j are super-pixel SjColor average.
In the present invention, the super-pixel segmentation technology based on probability density weighting geodesic distance, FCM algorithms are connected,
Brain tissue's segmentation result is obtained on the basis of super-pixel, segmentation efficiency and the degree of accuracy is improve.
Used as a modification of the present invention, the image of reading can also be the true brain MR image of dcm forms, true brain
The processing mode of portion's MR images is similar with the processing mode of synthesis brain MR image, and difference is that synthesis brain MR image reads in
Be gray value, and it is the density value limited by window width and window level that true brain MR image reads in.
Processing result image is analyzed below, is such as directed to Fig. 5 (a), the picture that Fig. 6 (a) is provided, using side above
Method is processed, and shown in result such as Fig. 5 (c), Fig. 6 (c), Fig. 5 (b), Fig. 6 (b) are super-pixel segmentation figure.
In Fig. 5 (b) and Fig. 6 (b), it can be seen that super-pixel segmentation of the present invention is accurate, border from visual effect
Laminating edge, can accurately separate each tiny region, be subsequently to separate white matter (WM), grey matter (GM), spinal fluid
(CSF) the cluster work of brain tissue such as provides solid foundation.Even our effect is clustered with most simple most traditional FCM
Fruit is still fine.
In Fig. 5 (c) and Fig. 6 (c), it can be seen that the present invention can exactly be partitioned into brain tissue's image, maximum journey
Remain the raw information of image degree.
The above is the preferred embodiment of the present invention, it is noted that for those skilled in the art
For, on the premise of principle of the present invention is not departed from, some improvements and modifications can also be made, these improvements and modifications
Should be regarded as protection scope of the present invention.
Claims (6)
1. a kind of probability density weights the brain MR image segmentation of geodesic distance, it is characterised in that including:
Step 1:Some width simulation brain database images are read in, statistics with histogram is carried out to it, draw white matter, grey matter or brain ridge
The interval sample value of liquid most integrated distribution;
Step 2:Piece image is randomly selected as pending image from the simulation brain database images, using the institute for obtaining
State the interval sample value of most integrated distribution probability density is carried out to each pixel on pending image as priori and estimate
Meter, obtains probability density function;
Step 3:Super-pixel segmentation is carried out to pending image based on the probability density function for obtaining, and records super-pixel
Classification results matrix;
Step 4:Super-pixel after segmentation is scanned, non-compliant super picture is filtered out according to super-pixel color standard difference
Element enters line splitting, and during division, it is two classes to be gathered all pixels in super-pixel again with FCM algorithms, is sought according to classification results afterwards
Connected region is looked for, and using the pixel in each connected region as a new class, updates the super-pixel classification results matrix;
Step 5:According to the super-pixel classification results matrix after updating, on the basis of super-pixel of the FCM algorithms after all renewals
Clustered, obtained pending image brain tissue segmentation result.
2. probability density according to claim 1 weights the brain MR image segmentation of geodesic distance, it is characterised in that
The step 1 is further:
Some width simulation brain MR images are read in, preliminary classification first is carried out with K-means algorithms to it, then by gray value normalizing
Be divided into N number of interval after change, the intensity value ranges to its white matter, grey matter or cerebrospinal fluid are counted, count white matter, grey matter or
K of person's cerebrospinal fluid most integrated distribution is interval, and in this K interval, m gray value is chosen as sample in each interval.
3. probability density according to claim 2 weights the brain MR image segmentation of geodesic distance, it is characterised in that
The step 2 includes:
Step 21:Calculate K probability Estimation model of each pixel in pending image respectively according to formula (1):
Wherein, x is the gray value of each pixel in image, xiIt is the individual interval m as priori value of kth (k=1,2 ..., K)
Individual gray value sample, h is bandwidth (control parameter);
Step 22:The K probability Estimation model that will be obtained calculates mixing probability density function by formula (2), obtains each picture
The Multilayer networks value of vegetarian refreshments, i.e. each pixel belong to the possibility of white matter, grey matter or cerebrospinal fluid:
Wherein, p (k) is factor of influence of each probability Estimation model to data point, between 0~1;
Step 23:Mixing probability density function to obtaining is normalized, and obtains probability density function:
Wherein, wmax and wmin are the maximum and minimum value of P (x).
4. probability density according to claim 3 weights the brain MR image segmentation of geodesic distance, it is characterised in that
The step 3 includes:
Step 31:Initialization seed point, is first uniformly distributed n/2 seed point on image, is then inserted with self adaptation hexagon method
Enter other seed points, the complexity of each seed point hexagon is calculated with formula (4), find the maximum hexagon of complexity,
It is divided into six small hexagons of overlap, inserts a new seed point in the maximum small hexagon of complexity, successively iteration,
Until n cluster centre of sampling, complexity is defined as:
Wherein, HiIt is seed point siHexagonal area, N is the pixel quantity of image I, and M is the picture that condition is met in braces
The quantity of vegetarian refreshments p, α is control parameter, It is the gradient of pixel p on image, Gσ
It is the Gaussian function with standard deviation sigma, ω is a regulation parameter, is preventedIt is zero situation;
Step 32:Seed point is upset, each seed point is moved to the minimum gradient locations in its local 3*3 region, and record it
X, Y-coordinate information are used as new seed point;
Step 33:The seed point sensitising gradient based on probability density is calculated with formula (5):
PSSG(si, G (t))=| | Sp(si,G(t))|| (5)
Wherein,SpIt is the P calculated with sobel operators on geodetic pathwThe gradient of (x);
Step 34:Calculated using FMM algorithms and geodesic distance is weighted based on probability density, row bound of going forward side by side diffusion is produced a series of
Pixel composition super-pixel block with like attribute;
In this step, A:Geodesic distance is calculated by the following method:
From seed point siOne to any one pixel p probability density weighting geodesic distance is described as:From seed point siStart edge
A most short path and reach pixel p, every is multiplied by a weighting function W (s on pathi, G (t)) minimum arc length product
Point, it is defined as:
Wherein, G (t) is from seed point siTo a geodetic path between pixel, t is continually changing, takes between 0~1
Value, weight W is set to the gradient that a pixel belongs to white matter, grey matter or cerebrospinal fluid possibility, for defining from seed point si
Distance increment onto certain pixel path G:
Geodesic distance is calculated by the expanding policy of the Fast Marching Method FMM with appropriate velocity field iterative diffusion, base
Velocity function is defined in formula (7), its computing formula is:
B:Boundary diffusion, produces super-pixel block:
In diffusion process, the new speed of each pixel is no longer static, depends on the seed point nearest from it, i.e., with
There is most short probability density to weight the pixel of geodesic distance for it, since given seed point, calculate it with formula (5) adjacent
The gradient of domain pixel, extends along the pixel with maximal rate formula (8), and the pixel color value after extension will be by planting
The color value substitution of son point, the probability seed point sensitising gradient of next pixel can be used constantly by the diffusion process of FMM works as
The value of preceding neighborhood territory pixel point is calculated with formula (5) and updated, and is spread forward every time from the minimum i.e. speed letter of seed point geodesic distance
The maximum pixel of number, finishes until all of pixel all spreads, and obtains super-pixel classification results matrix;
Step 35:Position and the color of seed point are updated with publicity (9);
Wherein, SlIt is l-th super-pixel, slIt is super-pixel SlSeed point, xl', cl' respectively be update after seed point position and
Color value,Measurement pixel is subordinate to the weighting function of seed point degree;
Step 36:Repeat step 33, step 34, step 35, algorithm are intended to optimize an energy function, are defined as formula (10),
When the change of energy function in subsequent iteration twice just stops less than a specific threshold, preliminary super-pixel segmentation
Complete,
5. probability density according to claim 4 weights the brain MR image segmentation of geodesic distance, it is characterised in that
The step 4 includes:
Step 41:Super-pixel after scanning segmentation, if the color standard difference of super-pixel is more than certain threshold value Tc, then to meeting this
The super-pixel of condition enters line splitting, and its calculating is defined as:
C(sl)=λ Stdl> Tc (11)
Wherein, StdlIt is the standard deviation of pixel color in each super-pixel, TcIt is threshold value, λ is control parameter;
Step 42:It is two classes, the mesh of local FCM to be gathered all pixels in the super-pixel of each needs division again with FCM algorithms
Scalar functions are:
Wherein, C is the number of cluster, QnIt is the number of pixel in the super-pixel of Current Scan, μijIt is that j-th pixel belongs to
The i fuzzy membership function of cluster, meets constraint μij∈[0,1],M is to act on the power on fuzzy membership
Weight function, viIt is ith cluster center, xjIt is the color value of pixel in the super-pixel of Current Scan;
Step 43:Find connected region according to the super-pixel classification results that obtain, and using the pixel in each connected region as
A new class, i.e., one new super-pixel block updates the super-pixel classification results matrix;
Step 44:To need division super-pixel repeat step 42, step 43 until all super-pixel for meeting splitting condition all
Division is completed.
6. probability density according to claim 5 weights the brain MR image segmentation of geodesic distance, it is characterised in that
The step 5 is further:
Using super-pixel as the input of FCM algorithms, clustered on the basis of the super-pixel split, completed brain image
Split, the FCM object functions based on super-pixel are:
Wherein, C is the number of cluster, and Q is the number of super-pixel in image, μijIt is mould that j-th super-pixel belongs to ith cluster
Paste membership function, meets constraint μij∈[0,1],M is to act on the weighting function on fuzzy membership, viIt is
Ith cluster center, ξjIt is super-pixel SjColor average.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710053148.5A CN106919950B (en) | 2017-01-22 | 2017-01-22 | The brain MR image segmentation of probability density weighting geodesic distance |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710053148.5A CN106919950B (en) | 2017-01-22 | 2017-01-22 | The brain MR image segmentation of probability density weighting geodesic distance |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106919950A true CN106919950A (en) | 2017-07-04 |
CN106919950B CN106919950B (en) | 2019-10-25 |
Family
ID=59454078
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710053148.5A Active CN106919950B (en) | 2017-01-22 | 2017-01-22 | The brain MR image segmentation of probability density weighting geodesic distance |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106919950B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107516314A (en) * | 2017-08-25 | 2017-12-26 | 山东大学 | The super voxel dividing method of medical image and device |
CN108596236A (en) * | 2018-04-18 | 2018-09-28 | 东南大学 | It is a kind of to roll into a ball partition method based on the thalamic nuclei of global connection features and geodesic distance |
CN112132842A (en) * | 2020-09-28 | 2020-12-25 | 华东师范大学 | Brain image segmentation method based on SEEDS algorithm and GRU network |
CN113052859A (en) * | 2021-04-20 | 2021-06-29 | 哈尔滨理工大学 | Super-pixel segmentation method based on self-adaptive seed point density clustering |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103353986A (en) * | 2013-05-30 | 2013-10-16 | 山东大学 | Brain MR image segmentation method based on superpixel fuzzy clustering |
CN104463870A (en) * | 2014-12-05 | 2015-03-25 | 中国科学院大学 | Image salient region detection method |
CN105678797A (en) * | 2016-03-04 | 2016-06-15 | 广东顺德中山大学卡内基梅隆大学国际联合研究院 | Image segmentation method based on visual saliency model |
CN106296654A (en) * | 2016-07-26 | 2017-01-04 | 中国科学技术大学 | A kind of image superpixel dividing method keeping edge |
-
2017
- 2017-01-22 CN CN201710053148.5A patent/CN106919950B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103353986A (en) * | 2013-05-30 | 2013-10-16 | 山东大学 | Brain MR image segmentation method based on superpixel fuzzy clustering |
CN104463870A (en) * | 2014-12-05 | 2015-03-25 | 中国科学院大学 | Image salient region detection method |
CN105678797A (en) * | 2016-03-04 | 2016-06-15 | 广东顺德中山大学卡内基梅隆大学国际联合研究院 | Image segmentation method based on visual saliency model |
CN106296654A (en) * | 2016-07-26 | 2017-01-04 | 中国科学技术大学 | A kind of image superpixel dividing method keeping edge |
Non-Patent Citations (8)
Title |
---|
ALEX LEVINSHTEIN等: "《TurboPixels: Fast Superpixels Using Geometric Flows》", 《IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE》 * |
XUE BAI等: "《Geodesic Matting: A Framework for Fast Interactive Image and Video Segmentation and Matting》", 《INTERNATIONAL JOURNAL OF COMPUTER VISION》 * |
***: "《几种新超像素算法的研究》", 《控制工程》 * |
沈洋等: "《交互式前景抠图技术综述》", 《计算机辅助设计与图形学学报》 * |
王爱齐等: "《基于测地距离的超像素生成方法》", 《大连理工大学学报》 * |
陈放等: "《基于超像素和模糊聚类的医学超声图像分割算法》", 《半导体光电》 * |
韩守东等: "《基于高斯超像素的快速Graph Cuts图像分割方法》", 《自动化学报》 * |
高珊珊等: "《近似测地距离度量下的图像抗噪分割方法》", 《计算机辅助设计与图形学学报》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107516314A (en) * | 2017-08-25 | 2017-12-26 | 山东大学 | The super voxel dividing method of medical image and device |
CN107516314B (en) * | 2017-08-25 | 2020-03-27 | 山东大学 | Medical image hyper-voxel segmentation method and device |
CN108596236A (en) * | 2018-04-18 | 2018-09-28 | 东南大学 | It is a kind of to roll into a ball partition method based on the thalamic nuclei of global connection features and geodesic distance |
CN112132842A (en) * | 2020-09-28 | 2020-12-25 | 华东师范大学 | Brain image segmentation method based on SEEDS algorithm and GRU network |
CN113052859A (en) * | 2021-04-20 | 2021-06-29 | 哈尔滨理工大学 | Super-pixel segmentation method based on self-adaptive seed point density clustering |
Also Published As
Publication number | Publication date |
---|---|
CN106919950B (en) | 2019-10-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Huang et al. | Brain image segmentation based on FCM clustering algorithm and rough set | |
Ahmmed et al. | Classification of tumors and it stages in brain MRI using support vector machine and artificial neural network | |
Al-Dmour et al. | A clustering fusion technique for MR brain tissue segmentation | |
CN109767440A (en) | A kind of imaged image data extending method towards deep learning model training and study | |
CN106919950A (en) | Probability density weights the brain MR image segmentation of geodesic distance | |
CN106295124A (en) | Utilize the method that multiple image detecting technique comprehensively analyzes gene polyadenylation signal figure likelihood probability amount | |
DE102008060789A1 (en) | System and method for unmonitored detection and Gleason grading for a prostate cancer preparation (whole-mount) using NIR fluorescence | |
CN109447998A (en) | Based on the automatic division method under PCANet deep learning model | |
CN104217213B (en) | A kind of medical image multistage sorting technique based on symmetric theory | |
CN112418337B (en) | Multi-feature fusion data classification method based on brain function hyper-network model | |
CN109871875A (en) | A kind of building change detecting method based on deep learning | |
CN106650768A (en) | Gaussian image model-based brain network modeling and mode classification method | |
CN111524140B (en) | Medical image semantic segmentation method based on CNN and random forest method | |
CN109871869A (en) | A kind of Lung neoplasm classification method and its device | |
Remya et al. | Enhanced DWT Filtering technique for brain tumor detection | |
CN111460966B (en) | Hyperspectral remote sensing image classification method based on metric learning and neighbor enhancement | |
CN110021019A (en) | A kind of thickness distributional analysis method of the AI auxiliary hair of AGA clinical image | |
CN104573745B (en) | Fruit-fly classified method based on magnetic resonance imaging | |
Kshirsagar et al. | Classification and detection of brain tumor by using GLCM texture feature and ANFIS | |
Bansal et al. | An improved hybrid classification of brain tumor MRI images based on conglomeration feature extraction techniques | |
Borman et al. | Classification of Medicinal Wild Plants Using Radial Basis Function Neural Network with Least Mean Square | |
CN113066054B (en) | Cervical OCT image feature visualization method for computer-aided diagnosis | |
Wisaeng et al. | Brain tumor segmentation using fuzzy otsu threshold morphological algorithm | |
Chao et al. | Brain image segmentation based on the hybrid of back propagation neural network and AdaBoost system | |
CN104050486A (en) | Polarimetric SAR image classification method based on maps and Wishart distance |
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 |