CN104123719A - Method for carrying out infrared image segmentation by virtue of active outline - Google Patents
Method for carrying out infrared image segmentation by virtue of active outline Download PDFInfo
- Publication number
- CN104123719A CN104123719A CN201410243635.4A CN201410243635A CN104123719A CN 104123719 A CN104123719 A CN 104123719A CN 201410243635 A CN201410243635 A CN 201410243635A CN 104123719 A CN104123719 A CN 104123719A
- Authority
- CN
- China
- Prior art keywords
- profile
- pixel
- infrared image
- local
- formula
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Abstract
The invention discloses a method for carrying out infrared image segmentation by virtue of an active outline. According to the method disclosed by the invention, the grey level distribution information of an image is adequately utilized by constructing an active outline model based on the combined drive of a local entropy and a local standard deviation, meanwhile, the infrared image is segmented by introducing a Gaussian kernel function to much accurately driving outline evolution in a local area. The method disclosed by the invention is capable of effectively realizing segmentation for the infrared image, so as to obtain a complete and correct target outline.
Description
Technical field
The invention belongs to technical field of image processing, be specifically related to a kind of method of carrying out infrared Image Segmentation with active profile.
Background technology
Infrared Image Segmentation has important effect to the recognition and tracking of infrared target, because infrared image has the features such as edge fog, gray scale be inhomogeneous conventionally, so conventional image partition method is difficult to the effect that reaches desirable for cutting apart of infrared image.In recent years, active contour model (ACM) is widely applied to during image cuts apart, and it is for having obtained good effect cutting apart of medical image, but the further research that needs in infrared Image Segmentation field.
Active contour model mainly can be divided into the model based on border and two classes of the model based on region, and the active contour model based on boundary information utilizes the border of the gradient objective definition of image, and wherein representative is GAC model; Active contour model based on area information utilizes regional statistical information (as gray scale) guiding curve to develop, and what be wherein widely used is CV model.
Document one (Zhang K, Zhang L, Song H, et al.Active contours with selective local or global segmentation:a new formulation and level set method[J] .Image and Vision Computing, 2010, 28 (4): 668-676.) SLGS model has been proposed, GAC and CV model are organically combined, it utilizes the half-tone information structure Signed Domination force function (SPF) defining in CV model, substitute the Edge-stopping function in GAC, thereby can control better the direction of curve evolvement, but its based on be this hypothesis of uniform gray level in region to be split, thereby this model is not very desirable to the segmentation effect of the inhomogeneous infrared image of gray scale.
Document two (Li C, Kao C Y, Gore J C, et al.Minimization of region-scalable fitting energy for image segmentation[J] .Image Processing, IEEE Transactions on, 2008, 17 (10): 1940-1949.) LBF model has been proposed, mainly utilize the half-tone information driving curve of image local area to develop, can cut apart the image that gray scale is inhomogeneous, but because the amount of image information that model utilizes is less, directly use it for cutting apart of infrared image, be difficult to equally obtain result accurately.
Summary of the invention
The present invention proposes a kind of method of carrying out infrared Image Segmentation with active profile, can effectively realize cutting apart of infrared image, obtains complete and correct objective contour.
In order to solve the problems of the technologies described above, the invention provides a kind of method of carrying out infrared Image Segmentation with active profile, comprise the following steps:
Step 1: calculate local entropy and Local standard deviation that in infrared image to be split, each pixel is corresponding, and use local entropy and the Local standard deviation that each pixel is corresponding to construct each pixel proper vector; Set initial profile by initialization level set function to infrared image; Calculate the mean vector of all pixel characteristic of correspondence vectors in the regional area of current evolution profile inside and outside regional area;
Step 2: the cosine similarity of each pixel characteristic of correspondence vector on the mean vector that calculation procedure one obtains and current evolution profile, determine the evolution direction of each pixel and the size of driving force on current evolution profile according to cosine similarity, thus the Signed Domination force function that tectonic forcing profile develops;
Step 3: by the Signed Domination force function substitution level set movements equation of step 2 structure, realize cutting apart of infrared image by the evolution of level set.
The present invention compared with prior art, its remarkable advantage is: (1) the inventive method is utilized local entropy and the Local standard deviation structural attitude vector of image, utilize local entropy and Local standard deviation to combine driving active contour model, utilize fully the intensity profile information of image, the direction that control wheel profile develops more accurately; (2) for the inhomogeneous feature of infrared image gray scale, the present invention introduces gaussian kernel function and image convolution to be split, the power that driving curve is developed only depends on when the effect of character pair vector in front profile local, can effectively overcome the inhomogeneous impact on segmentation result of gray scale; (3) the present invention can realize and better cutting apart for the inhomogeneous infrared image of gray scale.
Brief description of the drawings
Fig. 1 is process flow diagram of the present invention.
Fig. 2 is infrared image to be split used in emulation experiment of the present invention.
Fig. 3 is the local entropy image of Fig. 2 of emulation experiment acquisition of the present invention.
Fig. 4 is the standard deviation image of Fig. 2 of emulation experiment acquisition of the present invention.
Fig. 5 is the initial profile figure that emulation experiment of the present invention obtains.
Fig. 6 is the schematic diagram that is positioned at the pixel characteristic of correspondence vector f (x, y) of point (x, y) in emulation experiment infrared image of the present invention.
Fig. 7 is the mean vector schematic diagram of all pixel character pair vectors in the current evolution profile of emulation experiment of the present invention.
Fig. 8 is the segmentation result figure that uses SLGS model to obtain in emulation experiment of the present invention.
Fig. 9 is the segmentation result figure that uses LBF model to obtain in emulation experiment of the present invention.
Figure 10 is the segmentation result figure that uses the inventive method to obtain in emulation experiment of the present invention.
Embodiment
Infrared image has the features such as edge fog, gray scale be inhomogeneous, conventional active contour model is generally difficult to obtain desirable segmentation effect, the inventive method is by constructing a kind of active contour model of combining driving based on local entropy and Local standard deviation, make full use of the intensity profile information of image, simultaneously by introducing gaussian kernel function, in regional area, drive more accurately profile to develop, so that infrared image is cut apart.As shown in Figure 1, concrete steps of the present invention are as follows:
Step 1: calculate local entropy and Local standard deviation that in infrared image to be split, each pixel is corresponding, obtain corresponding local entropy image and the local standard difference image of infrared image, and use local entropy and the Local standard deviation that each pixel is corresponding to construct each pixel proper vector; Set initial profile by initialization level set function to infrared image again, introduce gaussian kernel function simultaneously, calculate the mean vector of (being called for short in inside and outside local) all pixel characteristic of correspondence vectors in the regional area and outside regional area of current evolution profile inside.
In described infrared image, the computing method of the local entropy of each pixel are as shown in formula (1),
In formula (1), entro (x, y) is the local entropy that pixel (x, y) is corresponding, the gray level sum that L is infrared image, n
ifor the number of the pixel that is i of gray level in infrared image, frequency p
ican be similar to that to regard gray level as be the probability that the pixel of i occurs in infrared image, M × N is the local window size centered by pixel (x, y).
Calculate local entropy corresponding to each pixel in infrared image to be split by formula (1), obtain local entropy image Entro corresponding to infrared image.
In described infrared image, the computing method of the Local standard deviation of each pixel are as shown in formula (2),
In formula (2), std (x, y) is the Local standard deviation that pixel (x, y) is corresponding, M × N is with pixel (x, y) the local window size centered by, (m, n) represents any one pixel in local window, s (m, n) be the grey scale pixel value that pixel (m, n) is corresponding
for the average gray of all pixels in local window, it defines as shown in formula (3):
In formula (3), m≤M-1, n≤N-1.
The local window of choosing while calculating local entropy and Local standard deviation can be the same window.
Calculate Local standard deviation corresponding to each pixel in infrared image to be split by formula (2), obtain local standard difference image Std corresponding to infrared image.
The local entropy that the each pixel of described use is corresponding and Local standard deviation are constructed each pixel proper vector as shown in formula (4),
f(x,y)=(entro(x,y),std(x,y)) (4)
In formula (4), f (x, y) is pixel (x, y) characteristic of correspondence vector;
The present invention f
ithe average that (x, y) represents all pixel character pair vectors in the inside and outside local of current evolution profile is to, mean vector f
i(x, y) can represent with formula (5),
In formula (5), f
1(x, y) is the mean vector of the interior all pixel characteristic of correspondence vectors of inside regional area of current evolution profile, entro
1(x, y) is the average of the local entropy that the interior all pixels of the inside regional area of current evolution profile are corresponding, std
1(x, y) is the average of the Local standard deviation that the interior all pixels of the inside regional area of current evolution profile are corresponding; f
2(x, y) is the mean vector of all pixel characteristic of correspondence vectors in current evolution profile outer partial region, entro
2(x, y) is the average of the local entropy that the interior all pixels of the inside regional area of current evolution profile are corresponding, std
2(x, y) is the average of the Local standard deviation that the interior all pixels of the inside regional area of current evolution profile are corresponding; W (x, y) is window function, and the present invention selects gaussian kernel function as W (x, y); Ω is infrared image to be split region; φ (x, y) is zero level set function corresponding to evolution profile, its initialization as shown in formula (6),
In formula (6), Ω
0a subset of image-region Ω,
Ω
0border.In the present invention, get ρ=1.
Mean vector f
ientro in (x, y)
1(x, y), std
1(x, y), entro
2(x, y) and std
2the calculating of (x, y) is as shown in formula (6):
In formula (6), K
σthat standard deviation is the gaussian kernel function of σ, H
ε(φ) be Heaviside function, H
ε(φ) structure as shown in public (7),
Step 2: the cosine similarity of utilizing each point characteristic of correspondence vector on mean vector in the inside and outside local of profile and current evolution profile, determine the evolution direction of each pixel and the size of driving force on profile, the Signed Domination force function that tectonic forcing profile develops.
Step 21: in two-dimensional space, vectorial A (x
1, y
1) and vectorial B (x
2, y
2) included angle cosine define as shown in formula (8):
Included angle cosine span is [1,1], and included angle cosine value is larger, and the angle of two vectors is less, shows that two vectorial similaritys are larger.
In the inventive method, first calculate the mean vector f in the inside and outside local of profile
ithe cosine similarity of each pixel characteristic of correspondence vector f (x, y) on (x, y) and current evolution profile, as shown in formula (9):
In formula (9), cos θ
1f (x, y) and f
1cosine similarity between (x, y), cos θ
2f (x, y) and f
2cosine similarity between (x, y), by comparing cos θ
1with cos θ
2size determine the evolution direction of each point on profile, if cos θ
1>=cos θ
2, profile should be to external expansion, on the contrary profile will inwardly shrink.
Step 21: by the mean vector f in local inside and outside profile
ion (x, y) and current evolution profile, each pixel characteristic of correspondence vector f (x, y) is made poor delivery, and determines the big or small F (x, y) of driving force by formula (10):
In formula (10), β is constant;
Step 23: use the evolution direction of each pixel and the size of corresponding driving force on profile, just formed the Signed Domination force function in the present invention, be designated as spf, as shown in formula (11):
Step 3: the level set movements equation by the Signed Domination force function substitution of structure as shown in formula (12), finally realize cutting apart of infrared image by the evolution of level set.
In formula (12), α is a constant, | ▽ φ | represent the directional derivative of zero level collection along gradient direction.
Carry out the evolution of level set according to formula (12), in the time meeting the condition of convergence, iteration stopping, obtains final segmentation result.
Beneficial effect of the present invention can be described further by following emulation experiment:
Selection infrared image as shown in Figure 2, as image to be split, uses the inventive method to test, and as shown in Figure 3, the local standard difference image of acquisition as shown in Figure 4 for the local entropy image obtaining in step 1.In the time calculating local entropy and Local standard deviation, all choose size and be 9 × 9 local window.In step 1, give the initial profile of image setting as shown in Figure 5.Fig. 6 is the schematic diagram that is positioned at the pixel characteristic of correspondence vector f (x, y) of point (x, y) in infrared image, and Fig. 7 is the mean vector schematic diagram of all pixel character pair vectors in current evolution profile, wherein, and f
1(x, y) is the mean vector of all pixel character pair vectors in the inner local of current evolution profile, f
2(x, y) is the mean vector of all pixel character pair vectors in the outside local of current evolution profile.
Contrast the inventive method and SLGS model and the segmentation result of LBF model to infrared image, wherein, be the segmentation result of the inventive method to infrared image shown in Fig. 8, is the segmentation result of SLGS model shown in Fig. 9, is the segmentation result of LBF model shown in Figure 10.From segmentation result, can find out, SLGS model and LBF model all could not accurately be partitioned into target, occurred much to cut apart, and the final profile obtaining are imperfect by mistake; And the inventive method can obtain complete and objective contour accurately.
Claims (9)
1. a method of carrying out infrared Image Segmentation with active profile, is characterized in that, comprises the following steps:
Step 1: calculate local entropy and Local standard deviation that in infrared image to be split, each pixel is corresponding, and use local entropy and the Local standard deviation that each pixel is corresponding to construct each pixel proper vector; Set initial profile by initialization level set function to infrared image; Calculate the mean vector of all pixel characteristic of correspondence vectors in the regional area of current evolution profile inside and outside regional area;
Step 2: the cosine similarity of each pixel characteristic of correspondence vector on the mean vector that calculation procedure one obtains and current evolution profile, determine the evolution direction of each pixel and the size of driving force on current evolution profile according to cosine similarity, thus the Signed Domination force function that tectonic forcing profile develops;
Step 3: by the Signed Domination force function substitution level set movements equation of step 2 structure, realize cutting apart of infrared image by the evolution of level set.
2. method of carrying out infrared Image Segmentation with profile initiatively as claimed in claim 1, is characterized in that, in step 1, in described infrared image the computing method of the local entropy of each pixel as shown in formula (1),
In formula (1), entro (x, y) is the local entropy that pixel (x, y) is corresponding, the gray level sum that L is infrared image, n
ifor the number of the pixel that is i of gray level in infrared image, frequency p
ican be similar to that to regard gray level as be the probability that the pixel of i occurs in infrared image, M × N is the local window size centered by pixel (x, y).
3. method of carrying out infrared Image Segmentation with profile initiatively as claimed in claim 1, is characterized in that, in step 1, in described infrared image the computing method of the Local standard deviation of each pixel as shown in formula (2),
In formula (2), std (x, y) is the Local standard deviation that pixel (x, y) is corresponding, M × N is with pixel (x, y) the local window size centered by, (m, n) represents any one pixel in local window, s (m, n) be the grey scale pixel value that pixel (m, n) is corresponding
for the average gray of all pixels in local window,
computing method as shown in formula (3):
In formula (3), m≤M-1, n≤N-1.
4. method of carrying out infrared Image Segmentation with active profile as claimed in claim 1, is characterized in that, in step 1, the local entropy that the each pixel of described use is corresponding and Local standard deviation are constructed each pixel proper vector as shown in formula (4),
f(x,y)=(entro(x,y),std(x,y)) (4)
In formula (4), f (x, y) is pixel (x, y) characteristic of correspondence vector.
5. method of carrying out infrared Image Segmentation with active profile as claimed in claim 1, is characterized in that, in step 1, uses f
ithe average that (x, y) represents all pixel character pair vectors in the inside and outside local of current evolution profile is to, mean vector f
i(x, y) as formula (5) so,
In formula (5), f
1(x, y) is the mean vector of the interior all pixel characteristic of correspondence vectors of inside regional area of current evolution profile, entro
1(x, y) is the average of the local entropy that the interior all pixels of the inside regional area of current evolution profile are corresponding, std
1(x, y) is the average of the Local standard deviation that the interior all pixels of the inside regional area of current evolution profile are corresponding; f
2(x, y) is the mean vector of all pixel characteristic of correspondence vectors in current evolution profile outer partial region, entro
2(x, y) is the average of the local entropy that the interior all pixels of the inside regional area of current evolution profile are corresponding, std
2(x, y) is the average of the Local standard deviation that the interior all pixels of the inside regional area of current evolution profile are corresponding; W (x, y) is window function, and Ω is infrared image to be split region; φ (x, y) is zero level set function corresponding to evolution profile, its initialization as shown in formula (6),
In formula (6), Ω
0a subset of image-region Ω,
Ω
0border.
6. method of carrying out infrared Image Segmentation with active profile as claimed in claim 5, is characterized in that, selects gaussian kernel function as window function W (x, y).
7. method of carrying out infrared Image Segmentation with active profile as claimed in claim 5, is characterized in that mean vector f
ientro in (x, y)
1(x, y), std
1(x, y), entro
2(x, y) and std
2the account form of (x, y) is as follows:
Wherein, K
σthat standard deviation is the gaussian kernel function of σ, H
ε(φ) be Heaviside function.
8. method of carrying out infrared Image Segmentation with active profile as claimed in claim 1, is characterized in that, the process of step 2 is specially:
Step 21: calculate the mean vector f in the inside and outside local of profile
ithe cosine similarity of each pixel characteristic of correspondence vector f (x, y) on (x, y) and current evolution profile, account form is as shown in formula (7):
In formula (7), cos θ
1f (x, y) and f
1cosine similarity between (x, y), cos θ
2f (x, y) and f
2cosine similarity between (x, y), by comparing cos θ
1with cos θ
2size determine the evolution direction of each point on profile, if cos θ
1>=cos θ
2, profile should be to external expansion, on the contrary profile will inwardly shrink;
Step 22: by the mean vector f in local inside and outside profile
ion (x, y) and current evolution profile, each pixel characteristic of correspondence vector f (x, y) is made poor delivery, and determines the big or small F (x, y) of driving force by formula (8):
In formula (8), β is constant;
Step 23: use the evolution direction of each pixel and the size of corresponding driving force on profile, structure Signed Domination force function spf, as shown in formula (9):
9. method of carrying out infrared Image Segmentation with profile initiatively as claimed in claim 1, is characterized in that, described in step 3 level set movements equation as shown in formula (10),
In formula (10), α is a constant,
represent the directional derivative of zero level collection along gradient direction.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410243635.4A CN104123719B (en) | 2014-06-03 | 2014-06-03 | Method for carrying out infrared image segmentation by virtue of active outline |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410243635.4A CN104123719B (en) | 2014-06-03 | 2014-06-03 | Method for carrying out infrared image segmentation by virtue of active outline |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104123719A true CN104123719A (en) | 2014-10-29 |
CN104123719B CN104123719B (en) | 2017-01-25 |
Family
ID=51769116
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410243635.4A Active CN104123719B (en) | 2014-06-03 | 2014-06-03 | Method for carrying out infrared image segmentation by virtue of active outline |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104123719B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109242877A (en) * | 2018-09-21 | 2019-01-18 | 新疆大学 | Image partition method and device |
CN111325727A (en) * | 2020-02-19 | 2020-06-23 | 重庆邮电大学 | Intracranial hemorrhage area three-dimensional segmentation method based on local entropy and level set algorithm |
CN112700459A (en) * | 2020-12-31 | 2021-04-23 | 南京理工大学 | Level set infrared image segmentation method based on multi-feature information fusion |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102609903A (en) * | 2011-12-21 | 2012-07-25 | 辽宁师范大学 | Method for segmenting moveable outline model image based on edge flow |
CN103093473A (en) * | 2013-01-25 | 2013-05-08 | 北京理工大学 | Multi-target picture segmentation based on level set |
CN103413299A (en) * | 2013-07-25 | 2013-11-27 | 北京交通大学 | Initiative outline model for image segmentation |
CN103413332A (en) * | 2013-08-23 | 2013-11-27 | 华北电力大学 | Image segmentation method based on two-channel texture segmentation active contour model |
-
2014
- 2014-06-03 CN CN201410243635.4A patent/CN104123719B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102609903A (en) * | 2011-12-21 | 2012-07-25 | 辽宁师范大学 | Method for segmenting moveable outline model image based on edge flow |
CN103093473A (en) * | 2013-01-25 | 2013-05-08 | 北京理工大学 | Multi-target picture segmentation based on level set |
CN103413299A (en) * | 2013-07-25 | 2013-11-27 | 北京交通大学 | Initiative outline model for image segmentation |
CN103413332A (en) * | 2013-08-23 | 2013-11-27 | 华北电力大学 | Image segmentation method based on two-channel texture segmentation active contour model |
Non-Patent Citations (4)
Title |
---|
XUE MEI ET AL.: "infrared image segmentation algorithm based on improved variation level set", 《PROCEEDINGS OF THE 2007 IEEE INTERNATIONAL CONFERENCE ON MECHATRONICS AND AUTOMATION》 * |
卢振泰等: "基于熵和局部邻域信息的高斯约束CV模型", 《计算机学报》 * |
张海波: "基于几何活动轮廓模型的图像分割方法研究", 《中国优秀硕士学位论文》 * |
苏日娜等: "改进符号压力函数的区域活动轮廓模型", 《中国图象图形学报》 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109242877A (en) * | 2018-09-21 | 2019-01-18 | 新疆大学 | Image partition method and device |
CN109242877B (en) * | 2018-09-21 | 2021-09-21 | 新疆大学 | Image segmentation method and device |
CN111325727A (en) * | 2020-02-19 | 2020-06-23 | 重庆邮电大学 | Intracranial hemorrhage area three-dimensional segmentation method based on local entropy and level set algorithm |
CN111325727B (en) * | 2020-02-19 | 2023-06-16 | 重庆邮电大学 | Intracranial hemorrhage area three-dimensional segmentation method based on local entropy and level set algorithm |
CN112700459A (en) * | 2020-12-31 | 2021-04-23 | 南京理工大学 | Level set infrared image segmentation method based on multi-feature information fusion |
CN112700459B (en) * | 2020-12-31 | 2023-10-17 | 南京理工大学 | Level set infrared image segmentation method based on multi-feature information fusion |
Also Published As
Publication number | Publication date |
---|---|
CN104123719B (en) | 2017-01-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102354397B (en) | Method for reconstructing human facial image super-resolution based on similarity of facial characteristic organs | |
CN102708370B (en) | Method and device for extracting multi-view angle image foreground target | |
CN101520894B (en) | Method for extracting significant object based on region significance | |
CN108053417A (en) | A kind of lung segmenting device of the 3DU-Net networks based on mixing coarse segmentation feature | |
US20160314338A1 (en) | Method and apparatus for fingerprint identification | |
CN101639935B (en) | Digital human serial section image segmentation method based on geometric active contour target tracking | |
WO2019007004A1 (en) | Image feature extraction method for person re-identification | |
CN111079685A (en) | 3D target detection method | |
CN102129573A (en) | SAR (Synthetic Aperture Radar) image segmentation method based on dictionary learning and sparse representation | |
CN101650777B (en) | Corresponding three-dimensional face recognition method based on dense point | |
CN101493935B (en) | Synthetic aperture radar image segmentation method based on shear wave hidden Markov model | |
CN107610223A (en) | Power tower three-dimensional rebuilding method based on LiDAR point cloud | |
CN102903102A (en) | Non-local-based triple Markov random field synthetic aperture radar (SAR) image segmentation method | |
CN105261047A (en) | Docking ring circle center extraction method based on close-range short-arc image | |
CN102622753B (en) | Semi-supervised spectral clustering synthetic aperture radar (SAR) image segmentation method based on density reachable measure | |
CN103839269A (en) | Image segmentation method based on quaternion and fuzzy C-means clustering | |
CN101996401A (en) | Target analysis method and device based on intensity image and range image | |
CN104036480A (en) | Surf algorithm based quick mismatching point eliminating method | |
CN105469408A (en) | Building group segmentation method for SAR image | |
CN104091339A (en) | Rapid image three-dimensional matching method and device | |
CN104036229A (en) | Regression-based active appearance model initialization method | |
CN103679734A (en) | Method for eyed typhoon two-dimensional surface wind field inversion on basis of SVM and PDE | |
CN102426700A (en) | Level set SAR image segmentation method based on local and global area information | |
CN104732546A (en) | Non-rigid SAR image registration method based on region similarity and local spatial constraint | |
CN105069796A (en) | Wavelet scatternet-based SAR image segmentation method |
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 |