CN111462145A - Active contour image segmentation method based on double-weight symbol pressure function - Google Patents

Active contour image segmentation method based on double-weight symbol pressure function Download PDF

Info

Publication number
CN111462145A
CN111462145A CN202010251518.8A CN202010251518A CN111462145A CN 111462145 A CN111462145 A CN 111462145A CN 202010251518 A CN202010251518 A CN 202010251518A CN 111462145 A CN111462145 A CN 111462145A
Authority
CN
China
Prior art keywords
function
image segmentation
weight
double
pressure 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
Application number
CN202010251518.8A
Other languages
Chinese (zh)
Other versions
CN111462145B (en
Inventor
房斌
符星宇
周明亮
李佳俊
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Sichuan Alcohol Research Institute
Original Assignee
Sichuan Alcohol Research Institute
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Sichuan Alcohol Research Institute filed Critical Sichuan Alcohol Research Institute
Priority to CN202010251518.8A priority Critical patent/CN111462145B/en
Publication of CN111462145A publication Critical patent/CN111462145A/en
Application granted granted Critical
Publication of CN111462145B publication Critical patent/CN111462145B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

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
    • G06T7/12—Edge-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0012—Biomedical image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/149—Segmentation; Edge detection involving deformable models, e.g. active contour models
    • 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
    • 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
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30101—Blood vessel; Artery; Vein; Vascular

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Medical Informatics (AREA)
  • Quality & Reliability (AREA)
  • Radiology & Medical Imaging (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Software Systems (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)

Abstract

The invention discloses a moving contour image segmentation method based on a double-weight symbol pressure function, which comprises the following steps of S1, inputting an original image to be segmented, initializing parameters of a moving contour model and setting an initial contour, S2, calculating a global gray term and a Legendre polynomial to obtain a symbol pressure function based on double weights, calculating an edge stop function g (| ▽ I |), S3, solving an energy function of the moving contour model by using a gradient descent algorithm to obtain a corresponding gradient flow equation to perform iteration of segmentation curves to obtain a level set function of a new moment, S4, performing binarization punishment and regular transformation processing of Gaussian filtering on the level set function of the new moment, S5, judging whether the curves continue to iterate, stopping the iteration if convergence conditions are met, completing the image segmentation, and executing S2 if the convergence conditions are not met, and performing next iteration.

Description

Active contour image segmentation method based on double-weight symbol pressure function
Technical Field
The invention relates to the field of image segmentation, in particular to a moving contour image segmentation method based on a double-weight symbol pressure function.
Background
Image segmentation plays an important role in image processing, machine learning, computer vision and other fields. An image segmentation technology based on an Active Contour Model (ACM) occupies a very important position at present, is widely applied to related fields such as medical images, radar images, composite image segmentation and the like at present, and has wide application prospect and application value. Medical images are available in a variety of image modalities such as MR, CT, PET, ultrasound imaging, etc., and various images can be obtained that reflect the physiological and physical properties of the human body in two-dimensional and three-dimensional regions. When the image segmentation technology based on the active contour model is applied to the field of medical images, the characteristic regions in the medical images can be segmented and extracted, so that reliable bases are provided for subsequent clinical diagnosis and pathology researches, and doctors are assisted to make more accurate diagnosis help, for example, in the fields of coronary artery CT (computed tomography) blood vessel segmentation shown in fig. 1, cerebral hemorrhage CT image segmentation shown in fig. 2 and the like.
Existing ACMs can be classified into two types: an edge-based active contour model and a region-based active contour model. The edge-based active contour model stops the evolution curve on the image edge to be segmented using an edge stopping function defined by the image gradient model. The region-based active contour model utilizes statistical information inside and outside the contour lines to control the evolution of the curve. In the image segmentation method based on the active contour model, complex conditions such as uneven gray scale, noise, fuzzy edge, time-consuming calculation, sensitivity to initial contour and the like have certain challenges for accurately and efficiently obtaining image segmentation results. Therefore, it is necessary to provide a more accurate and efficient method to ensure that the segmentation accuracy can be kept high in various complicated situations and the time and computation amount can be effectively controlled.
Disclosure of Invention
The invention aims to overcome the problem that the image segmentation method based on the active contour model in the prior art is low in real-time accuracy, and provides the active contour image segmentation method based on the double-weight symbol pressure function, so that the segmentation efficiency and accuracy are improved when the image to be segmented has complex conditions of uneven gray scale, noise, fuzzy edge and the like.
In order to achieve the above purpose, the invention provides the following technical scheme:
a method for segmenting an active contour image based on a double-weight symbol pressure function comprises the following steps:
step S1, inputting the obtained original image to be segmented; initializing parameters of the movable contour model and setting an initial contour;
step S2, calculating a global gray level term and a Legendre polynomial to obtain a symbol pressure function based on double weights, and calculating an edge stop function g (| ▽ I |);
step S3, using a gradient descent algorithm to carry out numerical solution on the energy function of the active contour model to obtain a corresponding gradient flow equation so as to carry out iteration of a segmentation curve and obtain a new moment level set function;
step S4, carrying out binarization punishment and Gaussian filter regularization processing on the new time level set function obtained in the step S3;
step S5, judging whether the curve continues iteration, if meeting the convergence condition, stopping iteration and finishing image segmentation; if the convergence condition is not satisfied, step S2 is executed to perform the next iteration.
Preferably, the detailed step of step S1 is as follows:
inputting an original image, calculating the gray value of each pixel point, and recording as I (x), wherein x is the pixel point in an image domain omega; initializing the parameters of the active contour model and setting the initial contour.
Preferably, in the step S2,
the dual weight based symbol pressure function is:
Figure BDA0002435657600000031
where I (x) is the gray value of a pixel in the image domain,
Figure BDA0002435657600000032
in the case of the legendre term,
Figure BDA0002435657600000033
is a global gray term, w1Weight occupied by Legendre terms in the symbol pressure function, w2The weight that the global gray term occupies in the symbol pressure function.
Preferably, the formula for calculating the legendre term and the global gray term is as follows:
Figure BDA0002435657600000034
Figure BDA0002435657600000035
where P (x) is Legendre polynomial vector η is used to adjust the weights occupied by the inner and outer portions;
Figure BDA0002435657600000036
is a closed-form solution of Legendre polynomial coefficient vector of the inner region of the contour line,
Figure BDA0002435657600000037
closed-form solution of Legendre polynomial coefficient vector for contour line outer region, c1Is the mean gray value inside the contour, c2Is the average gray value outside the contour; the 4 variables c1，c2And
Figure BDA0002435657600000038
the calculation formula of (a) is as follows:
Figure BDA0002435657600000039
Figure BDA00024356576000000310
Figure BDA00024356576000000311
Figure BDA00024356576000000312
wherein n is1(x)＝H(φ(x))，n2(x)＝1-H(φ(x))；
Figure BDA0002435657600000041
Where M denotes an identity matrix of K × K, K and L denote a Grymm matrix of K × K, λ1And λ2For two constants, default values are used; is a preset parameter.
Preferably, the edge-stopping function is a monotonically decreasing non-negative function.
Preferably, the edge stop function is composed of fractional and exponential forms, and the calculation formula is as follows:
Figure BDA0002435657600000042
Figure BDA0002435657600000043
where x and y represent the pixel points,
Figure BDA0002435657600000044
gaussian kernel function G representing a standard deviation ofA convolution operation with the image I to be segmented,
Figure BDA0002435657600000045
is a gradient operator.
Preferably, the numerical solution process of the energy function in step S3 is to obtain a gradient flow equation by calculation and then perform numerical iteration on the gradient flow equation, where the gradient flow equation is as follows:
Figure BDA0002435657600000046
wherein α, mu is a preset constant term, t is time,
Figure BDA0002435657600000047
representing the curvature.
Preferably, step S4 includes the steps of:
step S41, when the level set function phi at the new momenti+1>0 time phii+1＝1；φi+1When the value is less than or equal to 0, phii+1＝-1；
Step S42, regularizing the level set function phi by Gaussian filteringi+1，φi+1＝φi+1*G。
Preferably, the convergence condition of the numerical iteration operation of step S5 is:
Figure BDA0002435657600000048
wherein, for presetting the threshold value of the number of the pixel points, length (-) is used for calculating
Figure BDA0002435657600000051
I +1 is the number of iterations.
Compared with the prior art, the invention has the beneficial effects that:
(1) the Legendre term and the global gray level term are fused into the symbol pressure function, the improved symbol pressure function can adapt to more complex application environments, and the robustness on gray level unevenness, noise, edge blurring, multi-target images and different initial contours is good.
(2) One weight coefficient is used for controlling the influence degree of the Legendre term and the global gray term, and the other weight coefficient is introduced for adjusting the proportionality coefficient of the inside and outside fitting centers of the contour line, so that the curve can be better evolved and contracted into the inside of the region.
(3) And (4) providing a new edge stop function, incorporating the edge stop function into the gradient flow equation, and introducing edge information. The edge information is combined with the regional information provided by the symbol pressure function to drive the evolution of the contour line, so that the curve can better converge to the edge of the target, and the image segmentation is completed.
Description of the drawings:
FIG. 1 is an exemplary illustration of ACM for coronary CT vessel image segmentation;
FIG. 2 is an exemplary illustration of ACM for cerebral hemorrhage CT image segmentation;
FIG. 3 is a first flowchart of a method for segmenting an active contour image based on a dual-weight symbolic pressure function according to an exemplary embodiment 1 of the present invention;
fig. 4 is a flowchart of a method for segmenting an active contour image based on a dual-weight symbol pressure function according to an exemplary embodiment 1 of the present invention.
Detailed Description
The present invention will be described in further detail with reference to test examples and specific embodiments. It should be understood that the scope of the above-described subject matter is not limited to the following examples, and any techniques implemented based on the disclosure of the present invention are within the scope of the present invention.
Example 1
As shown in fig. 3 or fig. 4, the present embodiment provides an active contour image segmentation method based on a double-weight symbol pressure function, including the following steps:
step S1, inputting the obtained original image to be segmented; initializing parameters of the movable contour model and setting an initial contour;
step S2, calculating a global gray level term and a Legendre polynomial to obtain a symbol pressure function based on double weights, and calculating an edge stop function g (| ▽ I |);
step S3, using a gradient descent algorithm to carry out numerical solution on the energy function of the active contour model to obtain a corresponding gradient flow equation so as to carry out iteration of a segmentation curve and obtain a new moment level set function;
step S4, carrying out binarization punishment and Gaussian filter regularization processing on the new time level set function obtained in the step S3;
step S5, judging whether the curve continues iteration, if meeting the convergence condition, stopping iteration and finishing image segmentation; if the convergence condition is not satisfied, step S2 is executed to perform the next iteration.
The active contour image segmentation method is widely applied to the relevant fields of medical images, radar images, composite image segmentation and the like, so that the acquired original images can be medical images or radar images and the like. Taking medical images as an example, the raw images acquired by the present embodiment may be medical images of various image modalities (such as MR, CT, PET, ultrasound imaging, etc.), such as a coronary CT image shown in fig. 1 and a brain CT image shown in fig. 2. By extracting the feature region of the medical image by the active contour image segmentation method based on the double-weight symbol pressure function described in this embodiment, image segmentation is realized, for example, a blood vessel part of the coronary artery CT image shown in fig. 1 is extracted, and a cerebral hemorrhage position of the cerebral CT image shown in fig. 2 is extracted. The embodiment fuses the Legendre term and the global gray level term into the symbol pressure function, so that the improved symbol pressure function can adapt to more complex application environments, and has better robustness on uneven gray level, noise, fuzzy edge, multi-target images and different initial contours.
Step S1, inputting the obtained original image to be segmented; initializing parameters of the movable contour model and setting an initial contour;
the detailed steps of step S1 are as follows:
inputting an original image, calculating the gray value of each pixel point, and recording as I (x), wherein x is the pixel point in an image domain omega, and I is omega → R in the original image.
Initializing the parameters of the active contour model and setting the initial contour. Initializing parameters of the active contour model, specifically including preset parameter values in a symbol pressure function, an edge stopping function, a gradient flow equation and a convergence condition. The initial contour is an initialized contour curve, the energy of the initial contour is not necessarily minimum, and the target contour edge needs to be obtained through iterative evolution. The target contour edge can extract a characteristic region in the image to realize image segmentation. The initial contour, the curve generated by the intermediate iterative process, and the final target contour may all be referred to as the active contour.
The zero level set function is set to:
Φ0＝x:Φ(x)＝0
wherein phi is a level set function, and x is a pixel point in an image domain omega to be segmented; zero level set function phi0Divide the region omega into two non-adjacent regions omega1＝{x:Φ(x)>0} and Ω2＝{x:Φ(x)<0, representing foreground and background regions, respectively.
Step S2, calculating a global gray level term and a Legendre polynomial to obtain a symbol pressure function based on double weights, and calculating an edge stop function g (| ▽ I |);
wherein the dual weight based symbol pressure function is:
Figure BDA0002435657600000081
where I (x) is the gray value of a pixel in the image domain,
Figure BDA0002435657600000082
in the case of the legendre term,
Figure BDA0002435657600000083
is a global gray term, w1Weight occupied by Legendre terms in the symbol pressure function, w2The weight that the global gray term occupies in the symbol pressure function. And fusing the Legendre term and the global gray term into a symbol pressure function, respectively controlling the influence degree of the Legendre term and the global gray term by using one weight coefficient, and introducing another weight coefficient to adjust a proportional coefficient of an internal fitting center and an external fitting center of the contour line, so that the curve can be better evolved and contracted into the region. Legendre terms dominate to facilitate segmentation of non-uniform gray scale images, while global gray scale terms dominate to facilitate segmentation of noisy images. The weights of the Legendre term and the global gray term are adjusted according to specific image characteristics, and the influence degrees of the Legendre term and the global gray term are controlled, so that the improved symbol pressure function can adapt to more complex application environments and is suitable for uneven gray, noise and edgeThe fuzzy, multi-target images and different initial contours have better robustness.
The formula for calculating the Legendre term and the global gray term is as follows:
Figure BDA0002435657600000084
Figure BDA0002435657600000085
where P (x) is Legendre polynomial vector η is used to adjust the weights occupied by the inner and outer portions;
Figure BDA0002435657600000086
is a closed-form solution of Legendre polynomial coefficient vector of the inner region of the contour line (namely the foreground region),
Figure BDA0002435657600000087
is a closed-form solution of Legendre polynomial coefficient vector of the outer region of the contour line (i.e., the background region), c1Is the mean gray value inside the contour, c2The 4 variables c are the average gray value outside the contour1，c2And
Figure BDA0002435657600000091
the calculation formula of (a) is as follows:
Figure BDA0002435657600000092
Figure BDA0002435657600000093
Figure BDA0002435657600000094
Figure BDA0002435657600000095
wherein n is1(x)＝H(φ(x))，n2(x)＝1-H(φ(x))；
Where M denotes an identity matrix of K × K, K and L denote a Grymm matrix of K × K, λ1And λ2Two constants, a default value is typically used; hAnd (phi) is the Hei's function.
HThe formula for calculating (φ (x)) is as follows:
Figure BDA0002435657600000096
which is one of the initial parameters in step S1.
The edge stop function in step S2 may be any monotonically decreasing non-negative function. In the flat area of the image have
Figure BDA0002435657600000097
And
Figure BDA0002435657600000098
at the edge of the target have
Figure BDA0002435657600000099
Tends to be infinite and
Figure BDA00024356576000000910
that is, the difference between the background region and the foreground region is large, and if the region divided by the active contour meets the condition, the region to be segmented is considered to be identified, and the evolution of the active contour can be stopped. And introducing an edge stop function to constrain curvature in the energy function based on the region information, and combining the region information with the edge information to stop the evolution of the active contour so as to further obtain the target edge contour.
Preferably, in order to robustly capture the edge of the target and accelerate the segmentation speed of the multi-target image, two monotonically decreasing non-negative functions are introduced into the edge stopping function g (| ▽ I |), which respectively consist of fractions and exponential forms.
Figure BDA0002435657600000101
Figure BDA0002435657600000102
Wherein x and y represent pixel points, x being a central pixel point, and y being pixel points around the x neighborhood;
Figure BDA0002435657600000103
gaussian kernel function G representing a standard deviation ofA convolution operation with the image I to be segmented,
Figure BDA0002435657600000104
is a gradient operator. The standard deviation is set according to specific conditions, and in this embodiment, the value is one of the initial parameters in step S1.
The embodiment proposes a new edge stop function, which is incorporated into the gradient flow equation, and introduces edge information. The edge information is combined with the regional information provided by the symbol pressure function to drive the evolution of the contour line, so that the curve can better converge to the edge of the target, and the image segmentation is completed.
Step S3, the energy function of the active contour model is solved numerically by using a gradient descent algorithm to obtain a corresponding gradient flow equation for iteration of the segmentation curve, and a partial differential equation is calculated to obtain a new moment level set function phii+1(ii) a Wherein i +1 represents the iteration frequency, and if the iteration frequency is the first iteration, i is 0; phi is a0Level set function, phi, representing initialization1Representing the new time instant level set function obtained after the first iteration.
In another preferred embodiment of the present invention, the numerical solution process of the energy function is to obtain a gradient flow equation by calculation and then perform numerical iteration on the gradient flow equation, where the gradient flow equation is as follows:
Figure BDA0002435657600000105
wherein α, mu is a constant term, t is time,
Figure BDA0002435657600000106
constant terms α and μ, which may be set on a case-by-case basis, are one of the initial parameters in step S1 in this embodiment.
Step S4, for the new time level set function φ obtained in step S3i+1And carrying out binarization punishment and Gaussian filtering regularization processing.
Specifically, step S4 includes the following steps:
step S41, when phi isi+1>0 time phii+1＝1；φi+1When the value is less than or equal to 0, phii+1＝-1；
Step S42, regularizing the level set function phi by Gaussian filteringi+1E.g. phii+1＝φi+1*G。
In step S42, the standard deviation of the gaussian kernel function is a key parameter and should be selected appropriately according to the specific image. If the size is too small, the model is sensitive to noise, and the evolution is unstable; on the other hand, if too large, edge leakage may occur and the detected boundary may be inaccurate.
Step S5, judging whether the curve continues iteration, if meeting the convergence condition, stopping iteration and finishing image segmentation; if the convergence condition is not satisfied, step S2 is executed to perform the next iteration.
Specifically, the convergence condition of the numerical iteration operation in step S5 is:
Figure BDA0002435657600000111
wherein, the length (-) is used for calculating the threshold value of the number of the pixel points
Figure BDA0002435657600000112
I +1 is the number of iterations, and in the first iteration, i is 0.
The active contour model expresses a target contour by using a continuous curve, defines an energy functional, converts a segmentation process into a minimum value process for solving the energy functional, and obtains a gradient flow equation of the energy functional by solving an Euler (Euler-L) equation corresponding to the function when the numerical value is realized.
The foregoing is merely a detailed description of specific embodiments of the invention and is not intended to limit the invention. Various alterations, modifications and improvements will occur to those skilled in the art without departing from the spirit and scope of the invention.

Claims (9)

1. A movable contour image segmentation method based on a double-weight symbol pressure function is characterized by comprising the following steps:
step S1, inputting the obtained original image to be segmented; initializing parameters of the movable contour model and setting an initial contour;
step S2, calculating a global gray level term and a Legendre polynomial to obtain a symbol pressure function based on double weights, and calculating an edge stop function g (| ▽ I |);
step S3, using a gradient descent algorithm to carry out numerical solution on the energy function of the active contour model to obtain a corresponding gradient flow equation so as to carry out iteration of a segmentation curve and obtain a new moment level set function;
step S4, carrying out binarization punishment and Gaussian filter regularization processing on the new time level set function obtained in the step S3;
step S5, judging whether the curve continues iteration, if meeting the convergence condition, stopping iteration and finishing image segmentation; if the convergence condition is not satisfied, step S2 is executed to perform the next iteration.
2. The active contour image segmentation method based on the double-weight symbol pressure function as claimed in claim 1, wherein the detailed steps of the step S1 are as follows:
inputting an original image, calculating the gray value of each pixel point, and recording as I (x), wherein x is the pixel point in an image domain omega; initializing the parameters of the active contour model and setting the initial contour.
3. The active contour image segmentation method based on double weight symbol pressure function as claimed in claim 2, wherein in the step S2,
the dual weight based symbol pressure function is:
Figure FDA0002435657590000021
where I (x) is the gray value of a pixel in the image domain,
Figure FDA0002435657590000022
in the case of the legendre term,
Figure FDA0002435657590000023
is a global gray term, w1Weight occupied by Legendre terms in the symbol pressure function, w2The weight that the global gray term occupies in the symbol pressure function.
4. The active contour image segmentation method based on the double-weight symbol pressure function as claimed in claim 3, wherein the formula of the Legendre term and the global gray term is as follows:
Figure FDA0002435657590000024
Figure FDA0002435657590000025
where P (x) is Legendre polynomial vector η is used to adjust the weights occupied by the inner and outer portions;
Figure FDA0002435657590000026
is a closed-form solution of Legendre polynomial coefficient vector of the inner region of the contour line,
Figure FDA0002435657590000027
closed-form solution of Legendre polynomial coefficient vector for contour line outer region, c1Is the mean gray value inside the contour, c2Is the average gray value outside the contour; the 4 variables c1，c2And
Figure FDA0002435657590000028
Figure FDA0002435657590000029
the calculation formula of (a) is as follows:
Figure FDA00024356575900000210
Figure FDA00024356575900000211
Figure FDA00024356575900000212
Figure FDA00024356575900000213
wherein n is1(x)＝H(φ(x))，n2(x)＝1-H(φ(x))；
Figure FDA00024356575900000214
Where M denotes an identity matrix of K × K, K and L denote a Grymm matrix of K × K, λ1And λ2Is two timesNumber, using default values; is a preset parameter.
5. The active contour image segmentation method based on the dual weight sign pressure function as claimed in claim 3 wherein the edge stop function is a non-negative function that decreases monotonically.
6. The active contour image segmentation method based on the double-weight symbol pressure function as claimed in claim 5 is characterized in that the edge stopping function is composed of fractional and exponential forms, and the calculation formula is as follows:
Figure FDA0002435657590000031
Figure FDA0002435657590000032
where x and y represent the pixel points,
Figure FDA0002435657590000033
gaussian kernel function G representing a standard deviation ofA convolution operation with the image I to be segmented,
Figure FDA0002435657590000034
is a gradient operator.
7. The active contour image segmentation method based on the double-weight symbolic pressure function of claim 6, wherein the numerical solution process of the energy function in step S3 is to obtain a gradient flow equation by calculation and then perform numerical iteration operation on the gradient flow equation, and the gradient flow equation is as follows:
Figure FDA0002435657590000035
wherein α, mu is a preset constant term, t is time,
Figure FDA0002435657590000036
representing the curvature.
8. The active contour image segmentation method based on the dual-weight symbolic pressure function as claimed in claim 1, wherein the step S4 comprises the steps of:
step S41, when the level set function phi at the new momenti+1>0 time phii+1＝1；φi+1When the value is less than or equal to 0, phii+1＝-1；
Step S42, regularizing the level set function phi by Gaussian filteringi+1，φi+1＝φi+1*G。
9. The active contour image segmentation method based on the double-weight symbolic pressure function as claimed in claim 8, wherein the convergence condition of the numerical iteration operation of step S5 is as follows:
Figure FDA0002435657590000041
wherein, for presetting the threshold value of the number of the pixel points, length (-) is used for calculating
Figure FDA0002435657590000042
I +1 is the number of iterations.
CN202010251518.8A 2020-04-01 2020-04-01 Active contour image segmentation method based on double-weight symbol pressure function Active CN111462145B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010251518.8A CN111462145B (en) 2020-04-01 2020-04-01 Active contour image segmentation method based on double-weight symbol pressure function

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010251518.8A CN111462145B (en) 2020-04-01 2020-04-01 Active contour image segmentation method based on double-weight symbol pressure function

Publications (2)

Publication Number Publication Date
CN111462145A true CN111462145A (en) 2020-07-28
CN111462145B CN111462145B (en) 2023-04-07

Family

ID=71684353

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010251518.8A Active CN111462145B (en) 2020-04-01 2020-04-01 Active contour image segmentation method based on double-weight symbol pressure function

Country Status (1)

Country Link
CN (1) CN111462145B (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112102243A (en) * 2020-08-13 2020-12-18 哈尔滨工业大学（深圳） Active contour segmentation method and system combining general energy function and prior information item
CN112884765A (en) * 2021-03-25 2021-06-01 上海交通大学 2D image and 3D image registration method based on contour features
CN115984312A (en) * 2023-03-15 2023-04-18 苏州大学 Image segmentation method, electronic device and computer-readable storage medium
CN116777935A (en) * 2023-08-16 2023-09-19 天津市肿瘤医院（天津医科大学肿瘤医院） Deep learning-based method and system for automatically segmenting prostate whole gland
CN117593323A (en) * 2024-01-19 2024-02-23 苏州大学 Image segmentation method, system, medium and device based on non-local features

Citations (20)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO1999005640A1 (en) * 1997-07-25 1999-02-04 Arch Development Corporation Method and system for the segmentation of lung regions in lateral chest radiographs
US6282307B1 (en) * 1998-02-23 2001-08-28 Arch Development Corporation Method and system for the automated delineation of lung regions and costophrenic angles in chest radiographs
US20050232506A1 (en) * 2004-04-19 2005-10-20 Smith R T Enhancing images superimposed on uneven or partially obscured background
US20070047838A1 (en) * 2005-08-30 2007-03-01 Peyman Milanfar Kernel regression for image processing and reconstruction
US20100162197A1 (en) * 2008-12-18 2010-06-24 Brion Technologies Inc. Method and system for lithography process-window-maximixing optical proximity correction
CN102135606A (en) * 2010-12-13 2011-07-27 电子科技大学 KNN (K-Nearest Neighbor) sorting algorithm based method for correcting and segmenting grayscale nonuniformity of MR (Magnetic Resonance) image
CN102411794A (en) * 2011-07-29 2012-04-11 南京大学 Output method of two-dimensional (2D) projection of three-dimensional (3D) model based on spherical harmonic transform
US20140161352A1 (en) * 2012-12-07 2014-06-12 Commissariat A L'energie Atomique Et Aux Energies Alternatives Iterative method for determining a two-dimensional or three-dimensional image on the basis of signals arising from x-ray tomography
CN106204592A (en) * 2016-07-12 2016-12-07 东北大学 A kind of image level collection dividing method based on local gray level cluster feature
US20170032518A1 (en) * 2015-07-29 2017-02-02 Perkinelmer Health Sciences, Inc. Systems and methods for automated segmentation of individual skeletal bones in 3d anatomical images
US20170039704A1 (en) * 2015-06-17 2017-02-09 Stoecker & Associates, LLC Detection of Borders of Benign and Malignant Lesions Including Melanoma and Basal Cell Carcinoma Using a Geodesic Active Contour (GAC) Technique
CN106845512A (en) * 2016-11-30 2017-06-13 湖南文理学院 Beasts shape recognition method and system based on fractal parameter
CN107274414A (en) * 2017-05-27 2017-10-20 西安电子科技大学 Image partition method based on the CV models for improving local message
CN107730513A (en) * 2017-09-29 2018-02-23 华中科技大学 A kind of particle recognition and method for tracing based on spheric harmonic function invariant
JP2018077112A (en) * 2016-11-09 2018-05-17 学校法人同志社 Particle diameter analysis method and particle diameter analysis program
CN108090909A (en) * 2017-12-15 2018-05-29 中国人民解放军陆军军医大学第附属医院 A kind of ultrasonic contrast image partition method based on statistics Partial Differential Equation Model
CN109472792A (en) * 2018-10-29 2019-03-15 石家庄学院 In conjunction with the local energy functional of local entropy and the image partition method of non-convex regular terms
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
CN110211140A (en) * 2019-06-14 2019-09-06 重庆大学 Abdominal vascular dividing method based on 3D residual error U-Net and Weighted Loss Function
WO2019209179A1 (en) * 2018-04-27 2019-10-31 Agency For Science, Technology And Research System and method for intensity inhomogeneous image segmentation

Patent Citations (20)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO1999005640A1 (en) * 1997-07-25 1999-02-04 Arch Development Corporation Method and system for the segmentation of lung regions in lateral chest radiographs
US6282307B1 (en) * 1998-02-23 2001-08-28 Arch Development Corporation Method and system for the automated delineation of lung regions and costophrenic angles in chest radiographs
US20050232506A1 (en) * 2004-04-19 2005-10-20 Smith R T Enhancing images superimposed on uneven or partially obscured background
US20070047838A1 (en) * 2005-08-30 2007-03-01 Peyman Milanfar Kernel regression for image processing and reconstruction
US20100162197A1 (en) * 2008-12-18 2010-06-24 Brion Technologies Inc. Method and system for lithography process-window-maximixing optical proximity correction
CN102135606A (en) * 2010-12-13 2011-07-27 电子科技大学 KNN (K-Nearest Neighbor) sorting algorithm based method for correcting and segmenting grayscale nonuniformity of MR (Magnetic Resonance) image
CN102411794A (en) * 2011-07-29 2012-04-11 南京大学 Output method of two-dimensional (2D) projection of three-dimensional (3D) model based on spherical harmonic transform
US20140161352A1 (en) * 2012-12-07 2014-06-12 Commissariat A L'energie Atomique Et Aux Energies Alternatives Iterative method for determining a two-dimensional or three-dimensional image on the basis of signals arising from x-ray tomography
US20170039704A1 (en) * 2015-06-17 2017-02-09 Stoecker & Associates, LLC Detection of Borders of Benign and Malignant Lesions Including Melanoma and Basal Cell Carcinoma Using a Geodesic Active Contour (GAC) Technique
US20170032518A1 (en) * 2015-07-29 2017-02-02 Perkinelmer Health Sciences, Inc. Systems and methods for automated segmentation of individual skeletal bones in 3d anatomical images
CN106204592A (en) * 2016-07-12 2016-12-07 东北大学 A kind of image level collection dividing method based on local gray level cluster feature
JP2018077112A (en) * 2016-11-09 2018-05-17 学校法人同志社 Particle diameter analysis method and particle diameter analysis program
CN106845512A (en) * 2016-11-30 2017-06-13 湖南文理学院 Beasts shape recognition method and system based on fractal parameter
CN107274414A (en) * 2017-05-27 2017-10-20 西安电子科技大学 Image partition method based on the CV models for improving local message
CN107730513A (en) * 2017-09-29 2018-02-23 华中科技大学 A kind of particle recognition and method for tracing based on spheric harmonic function invariant
CN108090909A (en) * 2017-12-15 2018-05-29 中国人民解放军陆军军医大学第附属医院 A kind of ultrasonic contrast image partition method based on statistics Partial Differential Equation Model
WO2019209179A1 (en) * 2018-04-27 2019-10-31 Agency For Science, Technology And Research System and method for intensity inhomogeneous image segmentation
CN109472792A (en) * 2018-10-29 2019-03-15 石家庄学院 In conjunction with the local energy functional of local entropy and the image partition method of non-convex regular terms
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
CN110211140A (en) * 2019-06-14 2019-09-06 重庆大学 Abdominal vascular dividing method based on 3D residual error U-Net and Weighted Loss Function

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
JIM BRAUX-ZIN等: "A General Dense Image Matching Framework Combining Direct and Feature-Based Costs" *
JIM BRAUX-ZIN等: "A General Dense Image Matching Framework Combining Direct and Feature-Based Costs", 《PROCEEDINGS OF THE IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV)》 *
何宁等: "基于几何活动轮廓模型的图像分割方法研究综述" *
何宁等: "基于几何活动轮廓模型的图像分割方法研究综述", 《2008年全国射线数字成像与CT新技术研讨会论文集》 *
高婧婧: "脑部MR图像分割理论研究", 《中国知网博士电子期刊》 *

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112102243A (en) * 2020-08-13 2020-12-18 哈尔滨工业大学（深圳） Active contour segmentation method and system combining general energy function and prior information item
CN112102243B (en) * 2020-08-13 2023-06-09 哈尔滨工业大学（深圳） Active contour segmentation method and system combining general energy function and priori information item
CN112884765A (en) * 2021-03-25 2021-06-01 上海交通大学 2D image and 3D image registration method based on contour features
CN115984312A (en) * 2023-03-15 2023-04-18 苏州大学 Image segmentation method, electronic device and computer-readable storage medium
CN116777935A (en) * 2023-08-16 2023-09-19 天津市肿瘤医院（天津医科大学肿瘤医院） Deep learning-based method and system for automatically segmenting prostate whole gland
CN116777935B (en) * 2023-08-16 2023-11-10 天津市肿瘤医院（天津医科大学肿瘤医院） Deep learning-based method and system for automatically segmenting prostate whole gland
CN117593323A (en) * 2024-01-19 2024-02-23 苏州大学 Image segmentation method, system, medium and device based on non-local features
CN117593323B (en) * 2024-01-19 2024-03-29 苏州大学 Image segmentation method, system, medium and device based on non-local features

Also Published As

Publication number Publication date
CN111462145B (en) 2023-04-07

Similar Documents

Publication Publication Date Title
CN111462145B (en) Active contour image segmentation method based on double-weight symbol pressure function
Chen et al. A review of thyroid gland segmentation and thyroid nodule segmentation methods for medical ultrasound images
US9968257B1 (en) Volumetric quantification of cardiovascular structures from medical imaging
Chakravarty et al. RACE-net: a recurrent neural network for biomedical image segmentation
Lassen et al. Automatic segmentation of the pulmonary lobes from chest CT scans based on fissures, vessels, and bronchi
Lynch et al. Automatic segmentation of the left ventricle cavity and myocardium in MRI data
US9875581B2 (en) Automated 3D reconstruction of the cardiac chambers from MRI or ultrasound
US9111174B2 (en) Machine learnng techniques for pectoral muscle equalization and segmentation in digital mammograms
Arias-Lorza et al. Carotid artery wall segmentation in multispectral MRI by coupled optimal surface graph cuts
Chen et al. Image segmentation and bias correction via an improved level set method
Saritha et al. A comprehensive review: Segmentation of MRI images—brain tumor
Ertas et al. A computerized volumetric segmentation method applicable to multi-centre MRI data to support computer-aided breast tissue analysis, density assessment and lesion localization
Huynh et al. Fully automated MR liver volumetry using watershed segmentation coupled with active contouring
Larrey-Ruiz et al. Automatic image-based segmentation of the heart from CT scans
US20090060332A1 (en) Object segmentation using dynamic programming
Lee et al. Unsupervised segmentation of lung fields in chest radiographs using multiresolution fractal feature vector and deformable models
Bernier et al. Graph cut-based method for segmenting the left ventricle from MRI or echocardiographic images
Padmasini et al. State-of-the-art of level-set methods in segmentation and registration of spectral domain optical coherence tomographic retinal images
Yang et al. Split Bregman method based level set formulations for segmentation and correction with application to MR images and color images
Wang An efficient and robust method for automatically identifying the left ventricular boundary in cine magnetic resonance images
Khaniabadi et al. Comparative review on traditional and deep learning methods for medical image segmentation
CN111091578B (en) Segmentation method for vascular medical image
CN108416792B (en) Medical computed tomography image segmentation method based on active contour model
CN115393402A (en) Training method of image registration network model, image registration method and equipment
Kurugol et al. Centerline extraction with principal curve tracing to improve 3D level set esophagus segmentation in CT images

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