CN106530314A - Multi-scale local statistic active contour model (LSACM) level set image segmentation method - Google Patents
Multi-scale local statistic active contour model (LSACM) level set image segmentation method Download PDFInfo
- Publication number
- CN106530314A CN106530314A CN201611190635.8A CN201611190635A CN106530314A CN 106530314 A CN106530314 A CN 106530314A CN 201611190635 A CN201611190635 A CN 201611190635A CN 106530314 A CN106530314 A CN 106530314A
- Authority
- CN
- China
- Prior art keywords
- level set
- lsacm
- evolution
- scale
- image segmentation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
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
- G06T2207/10072—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
- G06T2207/10—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
- G06T2207/20112—Image segmentation details
- G06T2207/20124—Active shape model [ASM]
-
- 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
-
- 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
- G06T2207/30056—Liver; Hepatic
-
- 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
- G06T2207/30096—Tumor; Lesion
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a multi-scale-based local statistic active contour model (LSACM) level set image segmentation method. The offset field B epsilon, a variance sigma i epsilon and a level set function Phi(x) in an LSACM level set method are initialized. The quantity L(x) for describing a local area characteristic in a multi-scale LSACM method is calculated. A differential characteristic d(X) which describes a multi-scale local area is calculated. The maximum response M of a high pass filter in the multi-scale LSACM method is calculated. A local area simulation gray Ci epsilon is updated. The offset field B epsilon is updated. The variance sigma i epsilon is updated. The purpose of curve evolution is achieved through solving the partial differential equation minimum value corresponding to a multi-scale LSACM level set energy function. If a set number of iterations is achieved, the iteration operation is stopped, the curve evolution is ended, and if the number of iterations is not achieved, the iteration is continued. The invention provides the multi-scale LSACM level set method, a gray uneven image can be effectively segmented, and the phenomena of excessive segmentation and insufficient segmentation in an image segmentation method are improved.
Description
Technical field
The present invention relates to the field such as medical science, computer vision, image procossing, specially a kind of to be based on multiple dimensioned partial statistics
Active contour model (LSACM) level set image segmentation method.
Background technology
The focus that image segmentation is studied in being always computer vision field, including medical image segmentation, such as liver
It is dirty that as parenchymatous organ maximum in human abdominal cavity, kinds of Diseases are more and sickness rate is high, liver neoplasm be segmented in liver diagnosis,
The effect of key is played in lesion segmentation, liver transplantation clinical practice, in various medical image methods, CT imagings and MR are imaged
Can reflect that pathomorphism is showed, but these image data amounts are big, contrast is low, and gradation of image is connect with surrounding tissue
Closely, obscurity boundary, conventional method are difficult segmentation.
In recent years, Level Set Method is widely used in medical image segmentation.As the evolution curve of level set is closure,
Object edge can be preferably converged to, so segmentation effect is relatively good.But existing Level Set Method, including institute of the present invention
The LSACM level sets being related to generally assume that the gradation of image in the regional area of very little is approaches uniformity, and in order to
Process convenient, typically all make a reservation for a yardstick in regional area unification, this hypothesis is for the common uneven image of gray scale
Segmentation can reach good effect, but be not suitable for splitting the serious uneven image of gray scale.
The content of the invention
For problem above, the invention provides a kind of be based on multiple dimensioned partial statistics active contour model (LSACM) water
Flat set image segmentation method, multi-scale information is combined with existing Level Set Method LSACM so that curve was developing
Do not limited by unified yardstick in journey, for the different gray scales of zones of different, adaptively changed scale size, so as to relatively be defined
True segmentation figure picture, can be with the problem in effectively solving background technology.
For achieving the above object, the present invention provides following technical scheme:It is a kind of to be divided based on multiple dimensioned LSACM level sets image
Segmentation method, multi-scale information is combined with partial statistics active contour model (LSACM), is constructed a kind of new multiple dimensioned
LSACM level set image segmentation methods, step are as follows:
Step 1:Initialized in multiple dimensioned partial statistics active contour model (LSACM) level set image segmentation method first
Several numerical value:Biased field Bε=1, ε=1...m, ε are selected yardstick number, same as below;Variances sigmaiε=i, i=1...2, ε
=1...m;Level set functionX is level set function variable, and χ ∈ inside represent level set
Inside, otherwise are represented outside level set;
Step 2:Calculate for describing multiple dimensioned partial statistics active contour model (LSACM) level set image segmentation method
In local features amount L (x), it is assumed that the uneven major part of gradation of image occurs in low frequency region, using multiple dimensioned low
Logical Gaussian filterTo construct local features description,
Pixel centered on wherein x, the neighborhood of pixel centered on y, the yardstick of neighborhood is by σεThe decisions of=2 ε+1, describe local features
Measure as L (x)=Bε(x)ciε, wherein BεX () is biased field, ciεFor regional area approximate gray-scale;
Step 3:Amount L (x) of the description local features obtained according to step 2 calculates the multiple dimensioned partial statistics master of description
Differential Characteristics d (x) of the regional area in dynamic skeleton pattern (LSACM) level set image segmentation method, wherein d (x)=(I
(x)-L(x))2=(I (x)-Bε(x)ciε)2, I (x) represent original image, d (x) is divided into into two according to the characteristics of evolution curve
Point, curvilinear inner din(χ) with curved exterior dout(χ), wherein din(χ)=(I (χ)-Lin(χ))2, Lin(χ) represent that L (x) is being drilled
Change the component of curvilinear inner, dout(χ)=(I (χ)-Lout(χ))2, Lout(χ) components of the L (x) in evolution curved exterior is represented, it is poor
Dtex levies the similarity degree that d (x) represents original image I (χ) and local feature L (χ), and d (x) values are less, represents that similarity degree is got over
Height, as L (x) is constructed based on low pass filter, so d (x) is apparent from similar to high pass filter, so Differential Characteristics phase
When in carrying out multi-scale filtering to image;
Step 4:Regional area Differential Characteristics d (x) obtained according to step 3 calculates multiple dimensioned partial statistics active profile die
The peak response M of the multiple dimensioned high pass filter in type (LSACM) level set image segmentation method:M=max (d (x)), i.e. M=
max((I(x)-L(x))2)=max ((I (x)-Bε(x)ciε)2), wherein M is divided into two parts, evolution curvilinear inner MinAnd evolution
Curved exterior Mout, Min=max ((I (x)-Lin(x))2), Mout=max ((I (x)-Lout(x))2), as the details of image is believed
, all in high-frequency region, the effect of M is the high-frequency information and the process in segmentation curve evolvement for retaining image for breath and marginal information
Middle searching optimal scale;
Step 5:The regional area approximate gray-scale c mentioned in calculation procedure 2iε, ciεBy formulaIterative calculation, wherein M1(φ)=H (φ), represents
Evolution curvilinear inner, M2(φ)=1-H (φ), represents evolution curved exterior, and H (φ) is jump function;
Step 6:Biased field B mentioned in calculation procedure 2ε(x), BεX () is by formulaIterative calculation, wherein M1(φ)=H (φ),
Represent evolution curvilinear inner, M2(φ)=1-H (φ), represents evolution curved exterior, and H (φ) is jump function;
Step 7:The variances sigma being previously mentioned in calculation procedure 1 and step 6iε, σiεRepresent the grey scale change of respective regions, σiεBy
FormulaIterative calculation,
Wherein M1(φ)=H (φ), represents evolution curvilinear inner, M2(φ)=1-H (φ), represents evolution curved exterior, and H (φ) is rank
Jump function;
Step 8:Summary step, multiple dimensioned partial statistics active contour model (LSACM) level set image segmentation method
Energy function represented with E,,
Curve evolvement equation is
By the minima of differential equation energy function E such that it is able to reach curve evolvement, the purpose of image segmentation, whereinFor the regular item of level set function, μ and υ is constant, and the effect of μ is to drive
Evolution curve is made to move to target, υ=o*255*255, o ∈ [0,1], υ is bigger, and the target that can be detected is bigger, and υ is less, then
The target of detection is less;
Step 9:Computing is iterated using the formula in above-mentioned steps, multiple dimensioned partial statistics active contour model is solved
(LSACM) energy-minimum of level set image segmentation method, if having arrived at the iterationses Ite of setting, interative computation
Stop, curve evolvement terminates, so as to image segmentation is completed, if being also not reaching to iterationses, return to step 2 continues iteration.
Preferably, parameter m in the step chooses 8.
Preferably, the constant v in the step 8 chooses 0.001*255*25 or 0.00001*255*255.
Preferably, iterationses parameter Ite in the step 9 chooses 40 or 100.
Compared with prior art, the invention has the beneficial effects as follows:
As existing Level Set Method LSACM is not suitable for splitting similar to this border of liver neoplasm unobvious, gray scale
Uneven image, a kind of feature extraction multi-scale level set side based on LSACM of the present invention for the uneven image of gray scale
Method.Multiple dimensioned main thought is that curve is developed under the driving of partial differential equation, and incorporating for multi-scale information makes song
Line adaptively changes scale size according to the gray scale feature of zones of different in evolutionary process, rather than is bound to LSACM side
Single yardstick predetermined in advance in method.For the larger image of grey scale change in adjacent or proximate region, this method ratio
More feasible, segmentation effect is more accurate, test result indicate that, this multiple dimensioned LSACM Level Set Method effectively can be split
The uneven image of gray scale.
Description of the drawings
Flow charts of the Fig. 1 for the inventive method;
Fig. 2 is liver neoplasm image to be split;(a)slice1;(b)slice2;(c)slice3;(d)slice4;
Fig. 3 is the inventive method segmentation result;(a) slice1 correspondence segmentation results;(b) slice2 correspondence segmentation results;
(c) slice3 correspondence segmentation results;(d) slice4 correspondence segmentation results;
Fig. 4 is the uneven image of gray scale to be split;(a) cerebral white matter and grey matter figure;The uneven petal figure of (b) gray scale;
Fig. 5 is the inventive method segmentation result;(a) brain image segmentation result;(b) petal image segmentation result.
Specific embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Site preparation is described.Obviously, described embodiment is only the part in the present invention, rather than the embodiment of whole.Based on this
Embodiment in invention, the every other reality obtained under the premise of creative work is not made by those of ordinary skill in the art
Example is applied, the scope of protection of the invention is belonged to.
Embodiment 1:
Fig. 1-Fig. 3 is referred to, flow charts of the wherein Fig. 1 for the inventive method, MRs of the Fig. 2 in medical image are imaged,
(a), (b), (c), (d) in slice1-slice4 reflect respectively be different modalities liver image section, Tumors display
For the white portion on liver organization, present invention offer is a kind of to split liver neoplasm using multiple dimensioned LSACM Level Set Method
Technical scheme, in the program, preferred constant v chooses 0.001*255*255, and iterationses Ite chooses (a) of 40, Fig. 3, (b),
C (), (d) corresponds to (a) of Fig. 2, (b), (c), (d) lesion segmentation result respectively.
The energy function of multiple dimensioned partial statistics active contour model (LSACM) level set image segmentation method is defined such as
Under:
Wherein ∫ΩMinH(φ)dx+∫ΩMout(1-H (φ)) dx is data item, is divided into evolution curvilinear inner energy ∫ΩMinH
(φ) dx and evolution curved exterior energy ∫ΩMout(1-H (φ)) dx, MinAnd MoutAll it is the maximum sound of multiple dimensioned high pass filter
Should.For the regular item of level set, in order to smooth evolution profile and
Avoid reinitializing.In order to solve the minima of energy function, EVOLUTION EQUATION is obtained using gradient descent method as follows:Should by solving
The minima of the differential equation can be developed with driving curve, and so as to reach the purpose of image segmentation, method for solving can use iterative method,
Formula is utilizedTo iterate to calculate, ΦnRepresent the evolution after nth iteration is calculated
Curve,For Laplace operator, Δ t is transmission intensity.
In order to obtain the M in the above-mentioned differential equationinAnd Mout, particularly may be divided into following step:
Step 1:To the partial zones in multiple dimensioned partial statistics active contour model (LSACM) level set image segmentation method
Domain is described, as gray scale is uneven and soft edge information is general all in low frequency region, so using multiple dimensioned
Low-pass Gaussian filterRetouch to construct local features
State, using formula L (x)=Bε(x)ciεCalculate regional area gray feature;
Step 2:Amount L (x) of the description local features obtained according to step 1 calculates the multiple dimensioned regional area of description
Differential Characteristics d (x), d (x)=(I (x)-L (x))2=(I (x)-Bε(x)ciε)2, I (x) represents original image, according to evolution curve
The characteristics of d (x) is divided into into two parts, curvilinear inner din(χ) with curved exterior dout(χ), wherein din(χ)=(I (χ)-Lin(χ)
)2, Lin(χ) components of the L (x) in evolution curvilinear inner, d are representedout(χ)=(I (χ)-Lout(χ))2, Lout(χ) represent that L (x) exists
The component of evolution curved exterior, Differential Characteristics d (x) represent the similarity degree of original image I (χ) and local feature L (χ), d (x)
Value is less, represents that similarity degree is higher, as L (x) is constructed based on low pass filter, so being apparent from d (x) similar to high pass
Wave filter, so Differential Characteristics carry out multi-scale filtering equivalent to image;
Step 3:Regional area Differential Characteristics d (x) obtained according to above-mentioned steps 2 calculates multiple dimensioned high pass filter most
Big response M:M=max (d (x)), i.e. M=max ((I (x)-L (x))2)=max ((I (x)-Bε(x)ciε)2), wherein M is divided into two
Part, evolution curvilinear inner MinWith evolution curved exterior Mout, Min=max ((I (x)-Lin(x))2), Mout=max ((I (x)-
Lout(x))2), as the detailed information and marginal information of image is all in high-frequency region, the effect of M is the high frequency letter for retaining image
Cease and optimal scale is found during segmentation curve evolvement;
Step 4:The regional area approximate gray-scale c mentioned in calculating above-mentioned steps 1iε, ciεBy formulaIterative calculation, wherein M1(φ)=H (φ), represents
Evolution curvilinear inner, M2(φ)=1-H (φ), represents evolution curved exterior, and H (φ) is jump function, below in the same manner;
Step 5:Biased field B mentioned in calculating above-mentioned steps 1ε(x), BεX () is by formulaIterative calculation;
Step 6:The variances sigma being previously mentioned in calculating above-mentioned steps 5iε, σiεRepresent the grey scale change of respective regions, σiεBy public affairs
FormulaIterative calculation;
Comprehensive above formula, it is possible to obtain the variable M in EVOLUTION EQUATIONinAnd Mout, fortune is iterated using above-mentioned formula
Calculate, solve the energy function minima of multiple dimensioned partial statistics active contour model (LSACM) level set image segmentation method, if
The iterationses Ite of setting is had arrived at, then interative computation stops, curve evolvement terminates, and image segmentation is completed, if also not reaching
To iterationses, then continue iteration.
Embodiment 2:
Fig. 1, Fig. 4 and Fig. 5 are referred to, wherein Fig. 1 is the flow chart of the inventive method, and (a) in Fig. 4 is from medical image
In MR imagings, reflection is white matter and grey matter image in brain, and (b) in Fig. 4 is the uneven petal image of gray scale, this
Invention provides a kind of technical scheme for splitting the uneven image of above-mentioned gray scale using multiple dimensioned LSACM Level Set Method, the program
In preferred constant v choose 0.00001*255*255, iterationses Ite chooses (a) and (b) respectively program of 100, Fig. 5
Segmentation result.
The energy function of multiple dimensioned partial statistics active contour model (LSACM) level set image segmentation method is defined such as
Under:
Wherein ∫ΩMinH(φ)dx+∫ΩMout(1-H (φ)) dx is data item, is divided into evolution curvilinear inner energy ∫ΩMinH
(φ) dx and evolution curved exterior energy ∫ΩMout(1-H (φ)) dx, MinAnd MoutAll it is the maximum sound of multiple dimensioned high pass filter
Should.For the regular item of level set, in order to smooth evolution profile and
Avoid reinitializing.In order to solve the minima of energy function, EVOLUTION EQUATION is obtained using gradient descent method as follows:It is micro- by solving this
Divide the minima of equation develop with driving curve, so as to reach the purpose of image segmentation, method for solving can use iterative method, i.e.,
Using formulaTo iterate to calculate, ΦnThe evolution represented after nth iteration is calculated is bent
Line,For Laplace operator, Δ t is transmission intensity.
In order to obtain the M in the above-mentioned differential equationinAnd Mout, particularly may be divided into following step:
Step 1:To the partial zones in multiple dimensioned partial statistics active contour model (LSACM) level set image segmentation method
Domain is described, as gray scale is uneven and soft edge information is general all in low frequency region, so using multiple dimensioned
Low-pass Gaussian filterRetouch to construct local features
State, using formula L (x)=Bε(x)ciεCalculate regional area gray feature;
Step 2:Amount L (x) of the description local features obtained according to step 1 calculates the multiple dimensioned regional area of description
Differential Characteristics d (x), d (x)=(I (x)-L (x))2=(I (x)-Bε(x)ciε)2, I (x) represents original image, according to evolution curve
The characteristics of d (x) is divided into into two parts, curvilinear inner din(χ) with curved exterior dout(χ), wherein din(χ)=(I (χ)-Lin(χ)
)2, Lin(χ) components of the L (x) in evolution curvilinear inner, d are representedout(χ)=(I (χ)-Lout(χ))2, Lout(χ) represent that L (x) exists
The component of evolution curved exterior, Differential Characteristics d (x) represent the similarity degree of original image I (χ) and local feature L (χ), d (x)
Value is less, represents that similarity degree is higher, as L (x) is constructed based on low pass filter, so being apparent from d (x) similar to high pass
Wave filter, so Differential Characteristics carry out multi-scale filtering equivalent to image;
Step 3:Regional area Differential Characteristics d (x) obtained according to above-mentioned steps 2 calculates multiple dimensioned high pass filter most
Big response M:M=max (d (x)), i.e. M=max ((I (x)-L (x))2)=max ((I (x)-Bε(x)ciε)2), wherein M is divided into two
Part, evolution curvilinear inner MinWith evolution curved exterior Mout, Min=max ((I (x)-Lin(x))2), Mout=max ((I (x)-
Lout(x))2), as the detailed information and marginal information of image is all in high-frequency region, the effect of M is the high frequency letter for retaining image
Cease and optimal scale is found during segmentation curve evolvement;
Step 4:The regional area approximate gray-scale c mentioned in calculating above-mentioned steps 1iε, ciεBy formulaIterative calculation, wherein M1(φ)=H (φ), represents
Evolution curvilinear inner, M2(φ)=1-H (φ), represents evolution curved exterior, and H (φ) is jump function, below in the same manner;
Step 5:Biased field B mentioned in calculating above-mentioned steps 1ε(x), BεX () is by formulaIterative calculation;
Step 6:The variances sigma being previously mentioned in calculating above-mentioned steps 5iε, σiεRepresent the grey scale change of respective regions, σiεBy public affairs
FormulaIterative calculation;
Comprehensive above formula, it is possible to obtain the variable M in EVOLUTION EQUATIONinAnd Mout, fortune is iterated using above-mentioned formula
Calculate, solve the energy function minima of multiple dimensioned partial statistics active contour model (LSACM) level set image segmentation method, if
The iterationses Ite of setting is had arrived at, then interative computation stops, curve evolvement terminates, and image segmentation is completed, if also not reaching
To iterationses, then continue iteration.
Presently preferred embodiments of the present invention is the foregoing is only, not to limit the present invention, all essences in the present invention
Any modification, equivalent and improvement made within god and principle etc., should be included within the scope of the present invention.
Claims (4)
1. a kind of multiple dimensioned partial statistics active contour model (LSACM) level set image segmentation method, it is characterised in that:Step
It is as follows:
Step 1:Initialize first several in multiple dimensioned partial statistics active contour model (LSACM) level set image segmentation method
Individual numerical value:Biased field Bε=1, ε=1...m, ε are selected yardstick number;Variances sigmaiε=i, i=1...2, ε=1...m;Level
Set functionX is level set function variable, and x ∈ inside are represented inside level set,
Otherwise is represented outside level set;
Step 2:Calculate for describing in multiple dimensioned partial statistics active contour model (LSACM) level set image segmentation method
Amount L (x) of local features, it is assumed that the uneven major part of gradation of image occurs in low frequency region is high using multiple dimensioned low pass
This wave filterTo construct local features description, wherein
Pixel centered on x, the neighborhood of pixel centered on y, the yardstick of neighborhood is by σε=2 ε+1 determines that the amount for describing local features is
L (x)=Bε(x)ciε, wherein ciεFor regional area approximate gray-scale;
Step 3:Amount L (x) of the description local features obtained according to step 2 calculates the multiple dimensioned partial statistics drivewheel of description
Differential Characteristics d (x) of the regional area in wide model (LSACM) level set image segmentation method, wherein d (x)=(I (x)-L
(x))2=(I (x)-Bε(x)ciε)2, I (x) represents original image, d (x) is divided into two parts according to the characteristics of evolution curve, bent
D inside linein(x) and curved exterior dout(x), wherein din(x)=(I (x)-Lin(x))2, LinX () represents L (x) in evolution curve
Internal component, dout(x)=(I (x)-Lout(x))2, LoutX () represents components of the L (x) in evolution curved exterior, Differential Characteristics
D (x) represents the similarity degree of original image I (x) and local feature L (x);
Step 4:Regional area Differential Characteristics d (x) obtained according to step 3 calculates multiple dimensioned partial statistics active contour model
(LSACM) the peak response M of the multiple dimensioned high pass filter of level set image segmentation method:M=max (d (x)), i.e. M=max
((I(x)-L(x))2)=max ((I (x)-Bε(x)ciε)2), wherein M is divided into two parts, evolution curvilinear inner MinWith evolution curve
Outside Mout, Min=max ((I (x)-Lin(x))2), Mout=max ((I (x)-Lout(x))2), the effect of M is the height for retaining image
Frequency information and segmentation curve evolvement during find optimal scale;
Step 5:The regional area approximate gray-scale c mentioned in calculation procedure 2iε, ciεBy formulaIterative calculation, wherein M1(φ)=H (φ), represents
Evolution curvilinear inner, M2(φ)=1-H (φ), represents evolution curved exterior, and H (φ) is jump function;
Step 6:Biased field B mentioned in calculation procedure 2ε(x), BεX () is by formulaIterative calculation, wherein M1(φ)=H (φ),
Represent evolution curvilinear inner, M2(φ)=1-H (φ), represents evolution curved exterior, and H (φ) is jump function;
Step 7:The variances sigma being previously mentioned in calculation procedure 1 and step 6iε, σiεRepresent the grey scale change of respective regions, σiεBy formulaIterative calculation, wherein M1
(φ)=H (φ), represents evolution curvilinear inner, M2(φ)=1-H (φ), represents evolution curved exterior, and H (φ) is step letter
Number;
Step 8:Summary step, multiple dimensioned partial statistics active contour model (LSACM) level set image segmentation method
Energy function represented with E,,
Curve evolvement equation isPass through
The minima of differential equation energy function E, so as to reach curve evolvement, the purpose of image segmentation, whereinFor the regular item of level set function, μ and υ is constant, and the effect of μ is to drive
Evolution curve is made to move to target, υ=o*255*255, o ∈ [0,1], υ are relevant with the size of detection target;
Step 9:Computing is iterated using the formula in above-mentioned steps, multiple dimensioned partial statistics active contour model is solved
(LSACM) energy-minimum of level set image segmentation method, if having arrived at the iterationses Ite of setting, interative computation
Stop, curve evolvement terminates, so as to image segmentation is completed, if being also not reaching to iterationses, return to step 2 continues iteration.
2. method according to claim 1, it is characterised in that:The m chooses 8.
3. method according to claim 1, it is characterised in that:Constant v in the step 8 chooses 0.001*255*255
Or 0.00001*255*255.
4. method according to claim 1, it is characterised in that:Iterationses parameter Ite in the step 9 choose 40 or
100。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611190635.8A CN106530314A (en) | 2016-12-21 | 2016-12-21 | Multi-scale local statistic active contour model (LSACM) level set image segmentation method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611190635.8A CN106530314A (en) | 2016-12-21 | 2016-12-21 | Multi-scale local statistic active contour model (LSACM) level set image segmentation method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106530314A true CN106530314A (en) | 2017-03-22 |
Family
ID=58340304
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611190635.8A Pending CN106530314A (en) | 2016-12-21 | 2016-12-21 | Multi-scale local statistic active contour model (LSACM) level set image segmentation method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106530314A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109886977A (en) * | 2019-02-19 | 2019-06-14 | 闽南师范大学 | A kind of image partition method, terminal device and storage medium with neighborhood constraint |
CN111145179A (en) * | 2019-11-20 | 2020-05-12 | 昆明理工大学 | Gray uneven image segmentation method based on level set |
CN114241186A (en) * | 2021-11-24 | 2022-03-25 | 长春工业大学 | Finger vein image region-of-interest segmentation method based on active contour method |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101571951A (en) * | 2009-06-11 | 2009-11-04 | 西安电子科技大学 | Method for dividing level set image based on characteristics of neighborhood probability density function |
US20120230572A1 (en) * | 2011-03-10 | 2012-09-13 | Siemens Molecular Imaging Limited | Method and System for Multi-Organ Segmentation Using Learning-Based Segmentation and Level Set Optimization |
CN105160660A (en) * | 2015-08-17 | 2015-12-16 | 中国科学院苏州生物医学工程技术研究所 | Active contour blood vessel extraction method and system based on multi-feature Gaussian fitting |
-
2016
- 2016-12-21 CN CN201611190635.8A patent/CN106530314A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101571951A (en) * | 2009-06-11 | 2009-11-04 | 西安电子科技大学 | Method for dividing level set image based on characteristics of neighborhood probability density function |
US20120230572A1 (en) * | 2011-03-10 | 2012-09-13 | Siemens Molecular Imaging Limited | Method and System for Multi-Organ Segmentation Using Learning-Based Segmentation and Level Set Optimization |
CN105160660A (en) * | 2015-08-17 | 2015-12-16 | 中国科学院苏州生物医学工程技术研究所 | Active contour blood vessel extraction method and system based on multi-feature Gaussian fitting |
Non-Patent Citations (3)
Title |
---|
CHUNMING LI 等: "A Level Set Method for Image Segmentation in the Presence of Intensity Inhomogeneities With Application to MRI", 《IEEE TRANSACTIONS ON IMAGE PROCESSING》 * |
CHUNMING LI 等: "Implicit Active Contours Driven by Local Binary Fitting Energy", 《2007 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION》 * |
KAIHUA ZHANG 等: "A Level Set Approach to Image Segmentation With Intensity Inhomogeneity", 《IEEE TRANSACTIONS ON CYBERNETICS》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109886977A (en) * | 2019-02-19 | 2019-06-14 | 闽南师范大学 | A kind of image partition method, terminal device and storage medium with neighborhood constraint |
CN111145179A (en) * | 2019-11-20 | 2020-05-12 | 昆明理工大学 | Gray uneven image segmentation method based on level set |
CN111145179B (en) * | 2019-11-20 | 2023-07-25 | 昆明理工大学 | Gray-scale uneven image segmentation method based on level set |
CN114241186A (en) * | 2021-11-24 | 2022-03-25 | 长春工业大学 | Finger vein image region-of-interest segmentation method based on active contour method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Udupa et al. | Disclaimer:" Relative fuzzy connectedness and object definition: theory, algorithms, and applications in image segmentation" | |
Balocco et al. | SRBF: Speckle reducing bilateral filtering | |
Rusko et al. | Fully automatic liver segmentation for contrast-enhanced CT images | |
Tosun et al. | Cortical surface segmentation and mapping | |
CN110648338B (en) | Image segmentation method, readable storage medium, and image processing apparatus | |
Hardisty et al. | Quantitative characterization of metastatic disease in the spine. Part I. Semiautomated segmentation using atlas‐based deformable registration and the level set method | |
JP2011514822A (en) | Method and system for segmenting CT scan data | |
Ajam et al. | A review on segmentation and modeling of cerebral vasculature for surgical planning | |
López-Mir et al. | Liver segmentation in MRI: A fully automatic method based on stochastic partitions | |
Milenković et al. | Automated breast-region segmentation in the axial breast MR images | |
Hille et al. | Vertebral body segmentation in wide range clinical routine spine MRI data | |
Tunçay et al. | Realistic microwave breast models through T1-weighted 3-D MRI data | |
Foruzan et al. | A knowledge-based technique for liver segmentation in CT data | |
CN110910405A (en) | Brain tumor segmentation method and system based on multi-scale cavity convolutional neural network | |
CN109685814A (en) | Cholecystolithiasis ultrasound image full-automatic partition method based on MSPCNN | |
CN106573150A (en) | Suppression of vascular structures in images | |
CN106530314A (en) | Multi-scale local statistic active contour model (LSACM) level set image segmentation method | |
CN107507189A (en) | Mouse CT image kidney dividing methods based on random forest and statistical model | |
Forsberg | Atlas-based registration for accurate segmentation of thoracic and lumbar vertebrae in CT data | |
Alirr et al. | An automated liver tumour segmentation from abdominal CT scans for hepatic surgical planning | |
Powell et al. | Atlas-based segmentation of temporal bone surface structures | |
CN103700068B (en) | A kind of method that in CTA image, liver and blood vessel are split simultaneously | |
CN102419864A (en) | Method and device for extracting skeletons of brain CT (computerized tomography) image | |
Wong et al. | A comparison of peripheral imaging technologies for bone and muscle quantification: a review of segmentation techniques | |
O'Donnell | Semi-automatic medical image segmentation |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170322 |
|
RJ01 | Rejection of invention patent application after publication |