CN101826209B - Canny model-based method for segmenting three-dimensional medical image - Google Patents
Canny model-based method for segmenting three-dimensional medical image Download PDFInfo
- Publication number
- CN101826209B CN101826209B CN2010101591299A CN201010159129A CN101826209B CN 101826209 B CN101826209 B CN 101826209B CN 2010101591299 A CN2010101591299 A CN 2010101591299A CN 201010159129 A CN201010159129 A CN 201010159129A CN 101826209 B CN101826209 B CN 101826209B
- Authority
- CN
- China
- Prior art keywords
- image
- mnk
- edge
- gradient
- pixel
- 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.)
- Expired - Fee Related
Links
Images
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention relates to a Canny model-based method for segmenting a three-dimensional medical image and belongs to the technical field of image processing. The method comprises the following steps of: capturing a three-dimensional section image Imnk comprising a user-interested target in an original three-dimensional medical image IMNK through user interaction; performing median filtering on thethree-dimensional section image Imnk to remove image noise; acquiring a Canny edge information image Cmnk of the three-dimensional section image Imnk which is subjected to the median filtering by a Canny method; searching a target edge and extracting a complete enclosed target edge according to interaction point coordinates (i0, j0) near the target edge of a certain frame of image of the user; and extracting interested target edges Emnk of all frames in the three-dimensional section image Imnk, wherein the extracted interested target edges Emnk serve as a segmentation result of the three-dimensional medical image. The Canny model-based method for segmenting the three-dimensional medical image uses a few user interaction processes, has a small calculated amount and can rapidly and effectively extract edge information of interested targets in the three-dimensional medical image so as to complete the segmentation of the three-dimensional medical image.
Description
Technical field
The invention belongs to technical field of image processing, relate to the interactive fast partition method of 3 d medical images.
Background technology
In order to differentiate normal structure structure and the abnormality in the medical image accurately, need cut apart medical image.Traditional image partition method mainly comprises:
(1) based on the dividing method at edge: utilize interregional heterogeneity (as gray scale uncontinuity in the zone) to mark off the separatrix between each zone usually, these class methods comprise parallel method of differential operator (as operators such as Roberts, Sobel, Laplacian, Marr), serial border search method, based on surface fitting method etc.;
(2) based on the dividing method in zone: utilize the zones of different in the homogeneity recognition image in the same area usually, comprise threshold method, region growing and division merging, sorter and cluster, based on the method for random field etc.
But because the image-forming principle of medical image and the property difference of tissue itself, and the formation of image is subjected to such as noise, field offset effect, local bulk effect and histokinesis, between tissue and the tissue, between tissue and the organ, influence between between organ and the organ or the like realizes fast and accurately that 3 d medical images is cut apart to be very important.Therefore, be necessary that studying a kind of method can cut apart the target area according to people's voluntary the going of the subjectivity meaning, and can extract object edge information fast and accurately again at this field of medical application.Like this, help fast interesting target being carried out whole observation and accurately analyzing.
Summary of the invention
The invention provides a kind of three-dimensional medical image segmentation method based on the Canny model, this method not only has comparatively fast image segmentation speed more accurately, and has human-computer interaction function, can cut apart according to operator's wish to obtain interested object edge; Simultaneously, the present invention is convenient to observe pathological tissues (or organ) shape etc., helps making medical analysis and further judgement accurately.
In order to describe content of the present invention easily, at first some terms are defined.
Definition 1: interesting target.Target object to be split, as tumour, liver etc.
Definition 2: picture size size.Two dimension medical section size of images size is M * N, and M represents the pixel number of two-dimentional medical section image length direction, and N represents the pixel number of two-dimentional medical section picture traverse direction.The size of 3 d medical images is M * N * K, and expression has K to open the two-dimentional medical section image that size is M * N.
Definition 3: gradation of image.Refer to the monochrome information in the image, without any colouring information.
Definition 4:3 * 3 median filters.With current pixel point and on every side 8 gray values of pixel points get and be arranged in the gray-scale value of that middle numerical value by arranging from big to small as current pixel point.
Definition 5: image gradient.Refer to the vector field that the gradation of image value changes, comprise gradient magnitude and gradient direction.Gradient magnitude refers to the maximum rate of change of gray-scale value, and gradient direction refers to the fastest-rising direction of gray-scale value.
Definition 6: growth algorithm.At first select an initial pixel point, then all pixels that satisfy certain rule, belong to same classification and be interconnected with the initial pixel point to round-looking scan.These pixels are preserved, can be obtained desired result.
Definition 7: edge end points.The breakaway poing that refers to edge line.
Definition 8: edge rift direction.Refer to the marginal position that tangential direction and the sensing of edge line at the end points place disconnects.The gradient direction at this direction and edge end points place is perpendicular.
Definition 9: burr.The little short-term that finger comes out from the edge line upper bifurcation is made of the line of end points and bifurcation.Generally since noise and thinning process cause.
Definition 10: mathematical morphology.Go to measure and extract in the image correspondingly-shaped to reach purpose with structural element to graphical analysis and identification with certain form.Expansion is one of them fundamental operation, and formula is:
Wherein A is an image collection, and B is a structural element, and ^ represents to do the mapping about initial point, ()
xExpression translation x, ⌒ represents to occur simultaneously, and φ represents empty set,
Be the dilation operation symbol.
Detailed technology scheme of the present invention is as follows:
A kind of three-dimensional medical image segmentation method based on the Canny model as shown in Figure 1, may further comprise the steps:
Step 1: intercepting comprises the 3D region image of interesting target.Be of a size of the original 3 d medical images I of M * N * K
MNKIn, intercept one and comprise 3D region image I interesting target, that be of a size of m * n * k
MnkM≤M wherein, n≤N, k≤K.
This step has just narrowed down to m * n * k from original M * N * K with the size of pending original 3 d medical images, can accelerate image processing speed, also helps improving the accuracy rate that target is cut apart.
Step 2: to the 3D region image I of step 1 gained
MnkCarry out medium filtering.The purpose of medium filtering is in order to reduce noise to the 3D region image I
MnkInfluence; During medium filtering, can adopt the median filter of 3 * 3 sizes or 5 * 5 sizes.
Step 3: the 3D region image I behind the calculating medium filtering
MnkThe gradient of all pixels obtains the 3D region image I
MnkGradient image D
Mnk
With the 3D region image I behind the medium filtering
MnkIn the pixel of any frame sectioning image be expressed as (i, j), pixel (i, gray-scale value j) be expressed as I (i, j), 1≤i≤m wherein, 1≤j≤n; Calculating pixel point (i, transverse gradients d j) at first
x(i is j) with vertical gradient d
y(i, j), d wherein
x(i, j)=I (i+1, j)-I (i, j), d
y(i, j)=I (i, j+1)-I (i, j); (i, (i, j) (i j), obtains the 3D region image I to Grad size M j) to the calculating pixel point with gradient direction θ then
MnkGradient image D
Mnk, wherein:
Step 4: extract gradient image D
MnkCanny marginal information image C
MnkTo gradient image D
MnkIn each frame gradient image D
Mn, carry out following operation:
Step 4-1: determined pixel point (i, j) gradient line, be about to gradient direction belong to (0, π/8], (7 π/8,9 π/8] or (15 π/8,2 π] pixel (i, j) be included into horizontal line gradient pixel, gradient direction is belonged to (π/8,3 π/8] or (9 π/8,11 π/8] pixel (i, j) be included into the first diagonal line gradient pixel, gradient direction is belonged to (3 π/8,5 π/8] or (11 π/8,13 π/8] pixel (i, j) be included into perpendicular line gradient pixel, gradient direction is belonged to (5 π/8,7 π/8] or (13 π/8,15 π/8] pixel (i j) is included into the second diagonal line gradient pixel.
Step 4-2: adopt non-maximum value inhibition method to extract the Grad ridge.The capture vegetarian refreshments (i, j) and be positioned at pixel (i, adjacent two pixels on gradient line j), the Grad size that compares three pixels, if pixel (i, Grad j) is less than adjacent two pixel Grad on its gradient line, then with pixel (i, j) Grad is changed to " 0 ", otherwise keeps pixel (i, Grad j), obtain the image N (i after non-maximum value suppresses, j), its wide ridge band has been refined into has only a pixel wide, and has kept the height value of ridge.
Step 4-3: adopt the double threshold threshold method to realize the extraction of Canny marginal information.Choose two fixing threshold tau
1And τ
2, and τ
2=2* τ
1, 0<τ
2<M
Max(i, j), M wherein
Max(i j) is this frame gradient image D
MnIn the greatest gradient value; Image N after non-maximum value suppressed (i, j) in the Grad and the threshold tau of all pixels
1Or τ
2Compare respectively, with Grad greater than threshold tau
1Or τ
2The Grad of pixel be changed to " 1 ", otherwise be changed to " 0 ", obtain two threshold value edge image T respectively
1And T
2Then with high threshold edge image T
2Be the basis, add low threshold value edge image T
1In with high threshold edge image T
2All marginal points that middle edge is connected obtain this frame gradient image D
MnCanny marginal information image C
Mn
Step 4-4: to gradient image D
MnkIn each frame gradient image D
MnCarry out the operation of step 4-1, obtain gradient image D to step 4-3
MnkCanny marginal information image C
Mnk
Step 5: user interactions.The user is in the 3D region image I
MnkMiddle t opens image I
MntDetermine a pixel position (i near the edge of middle interesting target
0, j
0), 0≤t≤k wherein.
Step 6: determine the 3D region image I
MnkMiddle t opens image I
MntInteresting target edge E
Mnt, specifically may further comprise the steps:
Step 6-1: searching near edge point position.Open Canny marginal information image C at corresponding t
MntIn, the pixel position (i definite according to step 5
0, j
0), from the close-by examples to those far off to the round-looking scan marginal point.
Step 6-2: extract edge line.The marginal point that searches with step 6-1 by growth algorithm, obtains coupled all logical marginal points as starting point, thereby extracts the whole piece edge line.
Step 6-3: the edge line that extracts sealing.If what step 6-2 obtained is the edge line of sealing, then change step 6-4 over to; Otherwise, find an edge line, search other edge line respectively along the direction of the fracture of two end points of current edge line then, the end points end points adjacent with current edge line of the edge line that finds coupled together as the interesting target edge line with straight line; Repeat this step, until the extraction of finishing the closed edge line.
Step 6-4: remove burr.In the closed edge line that step 6-3 obtains, delete all burr edges, obtain the 3D region image I
MnkMiddle t two field picture I
MntInteresting target edge E
Mnt
Step 7: determine interesting target fringe region in the consecutive frame Canny marginal information image.With t two field picture I
MntObject edge E
MntExpansion becomes the fringe region of 15~25 pixel wide, as t+1 two field picture I
Mn (t+1)With t-1 two field picture I
Mn (t-1)Region of search, interesting target edge.
Step 8: search interesting target edge line.Open with t-1 at t+1 and to open Canny marginal information image C
Mn (t+1)And C
Mn (t-1)Search for the longest edge line respectively in the middle corresponding determined interesting target fringe region of step 7.
Step 9: extract the sealing object edge and remove burr.The t+1 frame Canny marginal information image C that searches for step 8
Nm (t+1)With t-1 frame Canny edge image C
Mn (t-1)In longest edge edge line, according to the processing procedure of step 6-3 and step 6-4, can obtain the 3D region image I respectively
MnkIn t+1 two field picture I
Mn (t+1)Interesting target edge E
Mn (t+1)With t-1 two field picture I
Mn (t-1)Interesting target edge E
Mn (t-1)
Step 10: repeating step 7 is to step 9, up to obtaining whole 3D region image I
MnkInteresting target edge E
Mnk, just can be from whole original 3 d medical images I
MNKIn extract interesting target edge E
MnkThereby, finish cutting apart of 3 d medical images.
Need to prove:
1, the 3D region image I that comprises user's interesting target of step 1 acquisition
MnkSize is determined by user interactions.
2, the median filter window size in the step 2 is selected by the picture noise decision, and window is big more, and it is good more to remove noise effects, but the corresponding calculated amount also can increase.
3, step 3 has realized the 3D region image I to step 4
MnkThe extraction at Canny edge.Because the Canny edge can be described the accurate edge of destination object accurately, so just provides accurate guarantee for follow-up processing.Double threshold threshold tau described in the step 4-3 wherein
1And τ
2Satisfy 0<τ
1<τ
2<M
Max(i, relation j), wherein M
Max(i j) is this frame gradient image D
MnIn the greatest gradient value.Generally, if the double threshold threshold tau
1And τ
2Value is big more (to approach M
Max(i, j)), the Canny edge that is extracted is just few more, may miss real marginal information; If double threshold threshold tau
1And τ
2Value more little (approaching 0), the Canny edge that is then extracted is just many more, some noise mistakes may be extracted as marginal information.Therefore, double threshold threshold tau
1And τ
2Value should be moderate, preferably τ
2=2 τ
1=2/3M
Max(i, j).
The present invention adopts the method based on the Canny edge model, at first comprises the 3D region image I of user's interesting target by the user interactions intercepting
MnkThen to the 3D region image I
MnkCarry out medium filtering to remove picture noise; Then adopt the Canny method to obtain 3D region image I behind the medium filtering again
MnkCanny marginal information image C
MnkAfterwards according near the interaction point coordinate (i of user a certain two field picture object edge
0, j
0), the ferret out edge, extract the object edge of complete sealing; Extract the 3D region image I at last
MnkIn the interesting target edge E of all frames
MnkSegmentation result as 3 d medical images.The three-dimensional medical image segmentation method that adopts the present invention to propose based on the Canny model, used less user interaction process, calculated amount is less, can fast and effeciently extract the marginal information of interesting target in the 3 d medical images, thereby finish cutting apart of 3 d medical images.
Description of drawings
Fig. 1 is a schematic flow sheet of the present invention.
Embodiment
Technical solution of the present invention is at first used Matlab language compilation program when realizing; Use three-dimensional MRI or CT medical image to carry out parameter setting and program optimization processing then; Use C Plus Plus rewriting program code and interactive interface framework at last, to improve program feature.
Claims (3)
1. three-dimensional medical image segmentation method based on the Canny model may further comprise the steps:
Step 1: intercepting comprises the 3D region image of interesting target; Be of a size of the original 3 d medical images I of M * N * K
MNKIn, intercept one and comprise 3D region image I interesting target, that be of a size of m * n * k
MnkM≤M wherein, n≤N, k≤K;
Step 2: to the 3D region image I of step 1 gained
MnkCarry out medium filtering;
Step 3: the 3D region image I behind the calculating medium filtering
MnkThe gradient of all pixels obtains the 3D region image I
MnkGradient image D
Mnk
With the 3D region image I behind the medium filtering
MnkIn the pixel of any frame sectioning image be expressed as (i, j), pixel (i, gray-scale value j) be expressed as I (i, j), 1≤i≤m wherein, 1≤j≤n; Calculating pixel point (i, transverse gradients d j) at first
x(i is j) with vertical gradient d
y(i, j), d wherein
x(i, j)=I (i+1, j)-I (i, j), d
y(i, j)=I (i, j+1)-I (i, j); (i, (i, j) (i j), obtains the 3D region image I to Grad size M j) to the calculating pixel point with gradient direction θ then
MnkGradient image D
Mnk, wherein:
Step 4: extract gradient image D
MnkCanny marginal information image C
MnkTo gradient image D
MnkIn each frame gradient image D
Mn, carry out following operation:
Step 4-1: determined pixel point (i, j) gradient line, be about to gradient direction belong to (0, π/8], (7 π/8,9 π/8] or (15 π/8,2 π] pixel (i, j) be included into horizontal line gradient pixel, gradient direction is belonged to (π/8,3 π/8] or (9 π/8,11 π/8] pixel (i, j) be included into the first diagonal line gradient pixel, gradient direction is belonged to (3 π/8,5 π/8] or (11 π/8,13 π/8] pixel (i, j) be included into perpendicular line gradient pixel, gradient direction is belonged to (5 π/8,7 π/8] or (13 π/8,15 π/8] pixel (i j) is included into the second diagonal line gradient pixel;
Step 4-2: adopt non-maximum value inhibition method to extract the Grad ridge; Capture vegetarian refreshments (i, j) and be positioned at pixel (i, adjacent two pixels on gradient line j), relatively the Grad size of three pixels, if pixel (i, j) Grad is less than adjacent two pixel Grad on its gradient line, then with pixel (i, Grad j) is changed to " 0 ", otherwise keep pixel (i, j) Grad, obtain after non-maximum value suppresses image N (i, j);
Step 4-3: adopt the double threshold threshold method to realize the extraction of Canny marginal information; Choose two fixing threshold tau
1And τ
2, and 0<τ
1<τ
2<M
Max(i, j), M wherein
Max(i j) is this frame gradient image D
MnIn the greatest gradient value; Image N after non-maximum value suppressed (i, j) in the Grad and the threshold tau of all pixels
1Or τ
2Compare respectively, with Grad greater than threshold tau
1Or τ
2The Grad of pixel be changed to " 1 ", otherwise be changed to " 0 ", obtain two threshold value edge image T respectively
1And T
2Then with high threshold edge image T
2Be the basis, add low threshold value edge image T
1In with high threshold edge image T
2All marginal points that middle edge is connected obtain this frame gradient image D
MnCanny marginal information image C
Mn
Step 4-4: to gradient image D
MnkIn each frame gradient image D
MnCarry out the operation of step 4-1, obtain gradient image D to step 4-3
MnkCanny marginal information image C
Mnk
Step 5: user interactions; The user is in the 3D region image I
MnkMiddle t opens image I
MntDetermine a pixel position (i near the edge of middle interesting target
0, j
0), 0≤t≤k wherein;
Step 6: determine the 3D region image I
MnkMiddle t opens image I
MntInteresting target edge E
Mnt, specifically may further comprise the steps:
Step 6-1: searching near edge point position; Open Canny marginal information image C at corresponding t
MntIn, the pixel position (i definite according to step 5
0, j
0), from the close-by examples to those far off to the round-looking scan marginal point;
Step 6-2: extract edge line; The marginal point that searches with step 6-1 by growth algorithm, obtains coupled all logical marginal points as starting point, thereby extracts the whole piece edge line;
Step 6-3: the edge line that extracts sealing; If what step 6-2 obtained is the edge line of sealing, then change step 6-4 over to; Otherwise, find an edge line, search other edge line respectively along the direction of the fracture of two end points of current edge line then, the end points end points adjacent with current edge line of the edge line that finds coupled together as the interesting target edge line with straight line; Repeat this step, until the extraction of finishing the closed edge line;
Step 6-4: remove burr; In the closed edge line that step 6-3 obtains, delete all burr edges, obtain the 2 dimensional region image I
MnkMiddle t two field picture I
MntInteresting target edge E
Mnt
Step 7: determine interesting target fringe region in the consecutive frame Canny marginal information image; With t two field picture I
MntObject edge E
MntExpansion becomes the fringe region of 15~25 pixel wide, as t+1 two field picture I
Mn (t+1)With t-1 two field picture I
Mn (t-1)The interesting target fringe region;
Step 8: search interesting target edge line; Open with t-1 at t+1 and to open Canny marginal information image C
Mn (t+1)And C
Mn (t-1)Search for the longest edge line respectively in the middle corresponding determined interesting target fringe region of step 7;
Step 9: extract the sealing object edge and remove burr; The t+1 frame Canny marginal information image C that searches for step 8
Mn (t+1)With t-1 frame Canny edge image C
Mn (t-1)In longest edge edge line, according to the processing procedure of step 6-3 and step 6-4, can obtain the 3D region image I respectively
MnkIn t+1 two field picture I
Mn (t+1)Interesting target edge E
Mn (t+1)With t-1 two field picture I
Mn (t-1)Interesting target edge E
Mn (t-1)
Step 10: repeating step 7 is to step 9, up to obtaining whole 3D region image I
MnkInteresting target edge E
Mnk, just can be from whole original 3 d medical images I
MNKIn extract interesting target edge E
MnkThereby, finish cutting apart of 3 d medical images.
2. the three-dimensional medical image segmentation method based on the Canny model according to claim 1 is characterized in that, the 3D region image I of step 2 pair step 1 gained
MnkWhen carrying out medium filtering, adopt the median filter of 3 * 3 sizes or 5 * 5 sizes.
3. the three-dimensional medical image segmentation method based on the Canny model according to claim 1 is characterized in that, threshold tau described in the step 4-3
2=2 τ
1=2/3M
Max(i, j).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2010101591299A CN101826209B (en) | 2010-04-29 | 2010-04-29 | Canny model-based method for segmenting three-dimensional medical image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2010101591299A CN101826209B (en) | 2010-04-29 | 2010-04-29 | Canny model-based method for segmenting three-dimensional medical image |
Publications (2)
Publication Number | Publication Date |
---|---|
CN101826209A CN101826209A (en) | 2010-09-08 |
CN101826209B true CN101826209B (en) | 2011-12-21 |
Family
ID=42690115
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN2010101591299A Expired - Fee Related CN101826209B (en) | 2010-04-29 | 2010-04-29 | Canny model-based method for segmenting three-dimensional medical image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN101826209B (en) |
Families Citing this family (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102129692A (en) * | 2011-03-31 | 2011-07-20 | 中国民用航空总局第二研究所 | Method and system for detecting motion target in double threshold scene |
CN102831424B (en) * | 2012-07-31 | 2015-01-14 | 长春迪瑞医疗科技股份有限公司 | Method for extracting visible component by microscope system |
CN102968822A (en) * | 2012-08-23 | 2013-03-13 | 华南理工大学 | Three-dimensional medical image segmentation method based on graph theory |
CN103544695B (en) * | 2013-09-28 | 2015-12-23 | 大连理工大学 | A kind of efficiently based on the medical image cutting method of game framework |
CN104680506A (en) * | 2013-11-28 | 2015-06-03 | 方正国际软件(北京)有限公司 | Method and system for detecting boundary line along different directions |
CN105631869B (en) * | 2015-12-25 | 2019-03-26 | 东软集团股份有限公司 | A kind of tube dividing method, device and equipment |
CN109410181B (en) * | 2018-09-30 | 2020-08-28 | 神州数码医疗科技股份有限公司 | Heart image segmentation method and device |
CN109846513B (en) * | 2018-12-18 | 2022-11-25 | 深圳迈瑞生物医疗电子股份有限公司 | Ultrasonic imaging method, ultrasonic imaging system, image measuring method, image processing system, and medium |
CN110766694B (en) * | 2019-09-24 | 2021-03-26 | 清华大学 | Interactive segmentation method of three-dimensional medical image |
CN110672628B (en) * | 2019-09-27 | 2020-06-30 | 中国科学院自动化研究所 | Method, system and device for positioning edge-covering joint of plate |
CN111179289B (en) * | 2019-12-31 | 2023-05-19 | 重庆邮电大学 | Image segmentation method suitable for webpage length graph and width graph |
CN114066923A (en) * | 2021-11-29 | 2022-02-18 | 长春工业大学 | 3D Otsu threshold segmentation method based on iteration and dimension decomposition |
CN114842038B (en) * | 2022-04-14 | 2023-05-16 | 深圳市医未医疗科技有限公司 | Image data characteristic automatic generation method and device based on image histology |
CN117853507B (en) * | 2024-03-06 | 2024-06-18 | 阿里巴巴(中国)有限公司 | Interactive image segmentation method, device, storage medium and program product |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1301494C (en) * | 2004-06-07 | 2007-02-21 | 东软飞利浦医疗设备***有限责任公司 | Three-dimensional dividing method for medical images |
CN101699511A (en) * | 2009-10-30 | 2010-04-28 | 深圳创维数字技术股份有限公司 | Color image segmentation method and system |
-
2010
- 2010-04-29 CN CN2010101591299A patent/CN101826209B/en not_active Expired - Fee Related
Also Published As
Publication number | Publication date |
---|---|
CN101826209A (en) | 2010-09-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN101826209B (en) | Canny model-based method for segmenting three-dimensional medical image | |
CN105261017B (en) | The method that image segmentation based on road surface constraint extracts pedestrian's area-of-interest | |
Song et al. | A deep learning based framework for accurate segmentation of cervical cytoplasm and nuclei | |
Wan et al. | Robust nuclei segmentation in histopathology using ASPPU-Net and boundary refinement | |
CN102999886B (en) | Image Edge Detector and scale grating grid precision detection system | |
CN108564567A (en) | A kind of ultrahigh resolution pathological image cancerous region method for visualizing | |
CN110736747B (en) | Method and system for positioning under cell liquid-based smear mirror | |
CN105389586A (en) | Method for automatically detecting integrity of shrimp body based on computer vision | |
Zadeh et al. | Recognition and classification of the cancer cells by using image processing and labview | |
ATE453904T1 (en) | AUTOMATIC DETECTION OF PRENEOPLASTIC ANOMALIES IN ANATOMIC STRUCTURES BASED ON IMPROVED AREA GROWTH SEGMENTATION AND COMPUTER PROGRAM THEREOF | |
CN106157279A (en) | Eye fundus image lesion detection method based on morphological segment | |
CN108961230A (en) | The identification and extracting method of body structure surface FRACTURE CHARACTERISTICS | |
Oger et al. | A general framework for the segmentation of follicular lymphoma virtual slides | |
Cao et al. | An automatic breast cancer grading method in histopathological images based on pixel-, object-, and semantic-level features | |
CN110648312A (en) | Method for identifying wool and cashmere fibers based on scale morphological characteristic analysis | |
CN101430789B (en) | Image edge detection method based on Fast Slant Stack transformation | |
Bhadoria et al. | Image segmentation techniques for remote sensing satellite images | |
CN115206495A (en) | Renal cancer pathological image analysis method and system based on CoAtNet deep learning and intelligent microscopic device | |
CN104933723A (en) | Tongue image segmentation method based on sparse representation | |
CN117315210B (en) | Image blurring method based on stereoscopic imaging and related device | |
CN108242060A (en) | A kind of method for detecting image edge based on Sobel operators | |
Lu et al. | Breast cancer mitotic cell detection using cascade convolutional neural network with U-Net | |
CN116188786B (en) | Image segmentation system for hepatic duct and biliary tract calculus | |
Wang et al. | Binary and multiclass classification of histopathological images using machine learning techniques | |
Abdelsamea | An enhancement neighborhood connected segmentation for 2D-Cellular Image |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20111221 Termination date: 20190429 |
|
CF01 | Termination of patent right due to non-payment of annual fee |