CN104794730B - SAR image segmentation method based on super-pixel - Google Patents
SAR image segmentation method based on super-pixel Download PDFInfo
- Publication number
- CN104794730B CN104794730B CN201510230209.1A CN201510230209A CN104794730B CN 104794730 B CN104794730 B CN 104794730B CN 201510230209 A CN201510230209 A CN 201510230209A CN 104794730 B CN104794730 B CN 104794730B
- Authority
- CN
- China
- Prior art keywords
- super
- mrow
- pixel
- msub
- sar image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Radar Systems Or Details Thereof (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of SAR image segmentation method based on super-pixel, it is high mainly to solve prior art, it is impossible to the problem of differentiating thin objects.Implementation step is：1.SAR images input, and complete the input of SAR image to be split and obtain image information；2. the SAR image of pair input produces super-pixel, to super-pixel image；3. extract the textural characteristics and space characteristics of super-pixel image；4. being merged by being clustered to textural characteristics, and with reference to space characteristics to super-pixel, the final segmentation result of SAR image is exported.The present invention can effectively reduce the computation complexity of traditional algorithm, shorten the processing time of SAR image segmentation, can tell thin objects, improve the degree of accuracy of segmentation, available for airfield runway, the image procossing of farmland distribution and geological prospecting.
Description
Technical field
The invention belongs to technical field of image processing, more particularly to image partition method, can be used for airfield runway, agriculture
Field is distributed and the image procossing of geological prospecting.
Background technology
Synthetic aperture radar SAR is a kind of Coherent Imaging RADAR for being operated in microwave band.It is with its high-resolution and entirely
Weather, round-the-clock, large area data retrieval capabilities and as current remote sensing observations important means, in resource, environment, archaeology
And military affairs etc. are widely used.SAR image includes abundant target classification, but is only possible in image understanding pair
Subregion therein is interested.Area-of-interest is different and different with application purpose.For example, in detection flood
When, region interested is waters, and in military activity, bridge, road near military base are all likely to become interested
Target, therefore be partitioned into these target regions, the amount of calculation of computer can not only be efficiently reduced, improve algorithm
Real-time, and it is significant for correct identification target.SAR image segmentation is always that SAR image understands and interpretation
Where key problem and difficult point.
SAR image comes from the back scattering of electromagnetic wave, therefore substantial amounts of coherent speckle noise in SAR image be present, and this causes
Each pixel often differs greatly with its actual value, therefore when the partitioning algorithm of routine is applied to SAR image, general effect is not
It is highly desirable.It can be said that coherent speckle noise is a key factor for influenceing SAR image segmentation quality.At present, three kinds are primarily present
Method reduces the influence that coherent speckle noise is split to image：1st, the statistical model of coherent speckle noise is established；2nd, carry out multiple dimensioned
SAR image segmentation method；3rd, the pretreatment of noise reduction is carried out to SAR image.
When splitting to natural image, often using probabilistic model of the Gauss model of additivity as image, and it is right
For SAR image, due to the difference of imaging mechanism and the presence of coherent speckle noise, it is impossible to use the additivity in natural image
Gauss model represents the statistical distribution characteristic of SAR image.Therefore, it is widely used in some statistical probabilities of natural image segmentation
Model, such as Markov random field model, Bayesian model, when splitting to SAR image, it is assumed that probability distribution meets
Rayleigh is distributed, Gamma distributions, K distributions etc., referring to Song Jianshe, Zheng Yongan, and Yuan Lihai,《Diameter radar image is managed
Solution and application》, Beijing:Science Press, 2008.These algorithms achieve relatively good segmentation effect, but such method by
Single pixel is operated in it, so computation complexity is often larger, still not can solve coherent speckle noise
Influence to segmentation result.
SAR image partitioning algorithm of the representative of second method as multiscale analysis model, the core of such algorithm are thought
Want the texture that SAR image is regarded as to different scale, different texture this property is presented according to different classes of target, will be right
The segmentation of SAR image, it is converted into the identification to different texture in SAR image.It is many using the algorithm quantity of this method, such as
U.Kandaswamy etc. is split using co-occurrence matrix to SAR image, referring to U.Kandaswamy, D.A.Adjeroh,
and M.C.Lee,《Efficient texture analysis of SAR imagery》,IEEE Transactions on
Geoscience and Remote Sensing, vol.43, no.9, pp.2075-2083,2005.X.Zhang use symbiosis square
The energy feature of battle array and wavelet decomposition is split to SAR image, referring to X.Zhang, L.Jiao, F.Liu et al.,
《Spectral clustering ensemble applied to SAR image segmentation》,IEEE
Transactions on Geoscience and Remote Sensing,vol.46,no.7,pp.2126-2136,2008.
Hou Biao etc. is split using second generation Bandelet domain Hidden markov tree models to SAR image, referring to Hou Biao, Xu Jing,
Liu Feng etc.,《Image segmentation based on second generation Bandelet domains Hidden markov tree model》, automation journal, vol.35,
No.5,2009.But up to the present, still none of these methods can effectively be modeled to whole textures, and should
Class algorithm is still to be handled for single pixel, for the SAR image containing a large amount of pixels, the calculating of this kind of algorithm
Efficiency comparison is low, and calculating speed is slow.
The third method is than relatively straightforward, i.e., before splitting to SAR image, carries out the pre- place of noise reduction to it first
Reason, to reduce influence of the coherent speckle noise to partitioning algorithm.Simplest method is that multiple pixels are averaged in such,
This is a very effective preprocess method, but it the shortcomings that be that while noise reduction the edge of image can not be kept to believe well
Breath.
The content of the invention
It is an object of the invention to overcome the shortcomings of above-mentioned prior art, a kind of SAR image based on super-pixel point is proposed
Segmentation method, to effectively reduce computation complexity, shorten the time of segmentation, keep the edge of image well while noise reduction
Information, improve segmentation accuracy.
To achieve the above object, the present invention provides the SAR image segmentation method based on super-pixel, comprises the following steps：
1. a kind of SAR image segmentation method based on super-pixel, comprises the following steps：
(1) to the SAR image I of input, its length L, width W and resolution ratio R are extractedI, and own according in SAR image storehouse
Image estimation go out the resolution ratio section R of minimum targets；
(2) super-pixel number N is estimated according to image parameters, generate the super-pixel set S={ s of SAR image to be spliti,
Wherein siFor i-th of super-pixel i=1,2 ..., NS；
(3) super-pixel set S textural characteristics F is calculatedgAnd space characteristics F (i)n(i, j), wherein i, j=1,2 ..., Ns
And i ≠ j；
(4) number K, the textural characteristics F using K-means algorithms to super-pixel set S are split according to SAR imageg(i) enter
Row cluster；
(5) according to super-pixel set S textural characteristics Fg(i) cluster result, any two in super-pixel set S is surpassed
Pixel siAnd sj, using Nearest neighbor rule, judge whether super-pixel corresponds to same cluster, if siAnd sjCorresponding same cluster, and
And their space characteristics Fn(i, j)=1, then merged, and to super-pixel set S textural characteristics FgAnd space characteristics (i)
Fn(i, j) is updated；Conversely, without merging；
(6) the super-pixel number N of the super-pixel set S after statistical updatings(t), and by the super-pixel number Ns(t) with it is upper
Statistical result Ns(t-1) it is compared：If Ns(t)==Ns(t-1) stop condition, output SAR image point, are then met
Cut result；If Ns(t) ＜ Ns(t-1) stop condition, return to step (4), are then unsatisfactory for.
The present invention has advantages below compared with prior art：
1. the present invention takes into full account that SAR image contains substantial amounts of coherent speckle noise, by the yardstick bigger than single pixel
The elementary cell that super-pixel is split as image, effectively reduce influence of the SAR image coherent speckle noise to segmentation result；
2. the present invention fully takes into account SAR image enormous size, containing a large amount of pixels, segmentation work is set to face a large amount of calculating
The problem of, the elementary cell using super-pixel as segmentation, and the textural characteristics and space characteristics extracted have simple to operate, fortune
The advantages of calculation amount is small, the computation complexity of partitioning algorithm is effectively reduced, shorten the processing time of image segmentation；
3. the present invention uses adaptive super-pixel merging method, the edge of image is kept to believe well while noise reduction
Breath, improve segmentation accuracy.
Brief description of the drawings
Fig. 1 is the implementation process figure of the present invention；
Fig. 2 is the present invention to super-pixel image caused by a SAR image of the width containing organic field；
Fig. 3 is that the present invention contains super-pixel image caused by the SAR image of river and different vegetation to a width；
Fig. 4 is the present invention to super-pixel image caused by SAR image of the width containing organic field and building；
Fig. 5 is the original SAR image containing organic field, building and vegetation that emulation experiment of the present invention uses；
Fig. 6 is the result figure split with emulation of the present invention to Fig. 5；
Fig. 7 is the original SAR image containing river, the woods and farmland that emulation experiment of the present invention uses；
Fig. 8 is the result figure split with emulation of the present invention to Fig. 7.
Embodiment
Reference picture 1, the SAR image segmentation method of the invention based on super-pixel, comprises the following steps：
Step 1：According to the SAR image of input, image information is obtained.
Input image I all in SAR image storehouse, and to the SAR image I of input, extract length L per piece image,
Width W and resolution ratio RI, the resolution ratio section R of minimum target is gone out according to these parameter Estimationss。
Step 2：Estimate super-pixel number, obtain super-pixel set.
2a) according to the length L of piece image, width W, resolution ratio RIWith the resolution ratio section R of minimum targets, by following two
The super-pixel number N of kind situation estimation diagram pictures：
The first situation is：According to length L, width W and the resolution ratio R for inputting SAR image to be splitI, estimate excess of export picture
Plain number NsInterval：
Wherein, Num (s) is the interval of the number of pixels contained by each super-pixel, and interval is [50,200], is surpassed
Number of pixels NsTo appoint the integer value taken in interval；
Second of situation be：As the resolution ratio section R for being unable to estimate out minimum targetsWhen, scheme according to SAR to be split is inputted
Length L, the width W of picture, estimate super-pixel number NsInterval be：
2b) generate the super-pixel set S={ s of SAR image to be spliti, wherein siFor i-th of super-pixel i=1,2 ...,
Ns：
2b1) obtain the super-pixel number N of SAR imagesAfterwards, inside SAR image to be split, uniform design NsIndividual pixel
Seed, and centered on these seeds, the level set that the method for use level collection allows these seeds to be expanded according to following seed is drilled
Change equation to be expanded：
Φn+1=Φn+F|▽Φn|Δt
Wherein ΦnThe level set function after n-th evolution is represented, | ▽ Φn| represent the level set function after n-th evolution
Grad, F represent velocity function, Δ t represent curve evolvement time interval；
2b2) when the seed-bearing edge of institute overlaps, just stop the expansion of seed, each seed is a super-pixel
si, so as to complete to inputting SAR image super-pixel set S={ siGeneration work.
Step 3：Calculate super-pixel set S textural characteristics FgAnd space characteristics F (i)n(i,j)。
3a) for any one super-pixel si, pass through equation below calculating super-pixel set S textural characteristics Fg(i)：
Wherein, H (i, 1:M) it is super-pixel siGrey level histogram, m represent image number of greyscale levels, Num (si) represent super
Pixel siContained number of pixels；
3b) for any two super-pixel siAnd sj, pass through equation below calculating super-pixel set S space characteristics Fn(i,
j)：
Step 4：Textural characteristics are clustered, and super-pixel merged with reference to space characteristics.
The existing clustering method to textural characteristics has a variety of, such as K-means, fuzzy C-mean algorithm method, Nearest neighbor rule etc.,
This example uses but is not limited to K-means.
This step is implemented as follows：
4a) for any two super-pixel siAnd sj, it is calculated respectively into all clusters according to variable volume distance VBSD
Heart cnDistance, dist (si,cn)、dist(sj,cn), wherein n=1,2 ..., K；
4b) by super-pixel siI-th of super-pixel s is distributed to apart from cluster centre corresponding to minimum valuei, as super-pixel si
Corresponding cluster ci：
4c) by super-pixel sjJ-th of super-pixel s is distributed to apart from cluster centre corresponding to minimum valuej, as super-pixel sj
Corresponding cluster cj：
4d) judge the two super-pixel siAnd sjWhether identical cluster centre is corresponded to, if the two super-pixel siAnd sj
Corresponding cluster centre is identical, then super-pixel siAnd sjBelong to same cluster, conversely, being not belonging to same cluster；
4e) judge the super-pixel s of same clusteriAnd sjSpace characteristics, if their space characteristics Fn(i, j)=1,
Then by super-pixel siAnd sjMerge, obtain new super-pixel set S*, conversely, without merging.
Step 5：To new super-pixel set S*Space characteristics and textural characteristics be updated.
5a) in two super-pixel siAnd sjAfter merging, the sequence number i of one of two super-pixel is removed, with another super-pixel
Sequence number j come represent merge after super-pixel, obtain new textural characteristics Fg′(j)：
Wherein Fg(i) super-pixel s before merging is representediTextural characteristics, Fg(j) super-pixel s before merging is representedjTexture it is special
Sign, Num (si) represent super-pixel s before mergingiComprising number of pixels, Num (sj) represent super-pixel s before mergingjComprising pixel
Number；
5b) delete s before mergingiTextural characteristics Fg(i), the textural characteristics F caused by newg' (j) substitutes sjAfter merging
Textural characteristics.
Step 6：Super-pixel amalgamation result is differentiated.
Statistics super-pixel set S new after merging every time*Super-pixel number Ns(t), and by the super-pixel number Ns(t)
With last statistical result Ns(t-1) it is compared：If Ns(t)==Ns(t-1) stop condition, output SAR figures, are then met
As segmentation result；If Ns(t) ＜ Ns(t-1) stop condition, is then unsatisfactory for, return to step (4), continues to close super-pixel
And.
The effect of the present invention can be by further illustrating to the emulation experiment of true SAR image below：
1st, emulation experiment condition
The emulation of the present invention is in windows XP, SP2, CPU Pentium (R) 4, fundamental frequency 2.4GHZ, software platform
To be realized on Matlab7.0.1.The true SAR image that emulation experiment is selected is the SAR image of two width Ku wave bands, such as Fig. 5 and Fig. 7
Shown, wherein Fig. 5 is the SAR image on California, United States airport, and size is 522 × 446, and resolution ratio is 3 meters, containing organic
The ground targets such as field, building and vegetation；Fig. 7 is the width SAR image near the A Erbukeerke cities of New Mexico,
Size is 600 × 432, and resolution ratio is 1 meter, contains the ground targets such as river, the woods and farmland.
2nd, emulation content and result
Emulation one, with the present invention in a SAR image of the width containing organic field caused super-pixel, formed as shown in Figure 2
Super-pixel image.
Emulation two, super-pixel is produced in the SAR image that a width contains river and different vegetation with the present invention, is formed as schemed
Super-pixel image shown in 3.
Emulation three, with the present invention to caused super-pixel in SAR image of the width containing organic field and building, formed as schemed
Super-pixel image shown in 4.
Emulation four, emulation experiment segmentation is carried out with the present invention to Fig. 5, as a result such as Fig. 6.
From fig. 6, it can be seen that the present invention can accurately position the edge of airfield runway, and accurately by runway with it is other
Ground target separates.For the building of near airports, the present invention can be had found, even if building isolated individually
The ground target of itself and surrounding can be distinguished.For ground vegetation, the present invention effectively can will have different visions
The vegetation region of feature separates.
Emulation five, emulation experiment segmentation is carried out with the present invention to Fig. 7, as a result such as Fig. 8.
From figure 8, it is seen that the present invention can be accurately positioned the edge in river course and the edge of road, and by ground not
Same vegetation is separated, and the present invention can also effectively tell special target, in the trees on meadow, river
Island etc..For same class ground target, the segmentation result of present invention gained has good region consistency, and mistake point is very
It is few.
Can be seen that the present invention from above the simulation experiment result can not only produce super-pixel image, and by based on super
The mode of pixel effectively can be split to SAR image, and the edge of Different Ground target can be accurately positioned in segmentation,
The region consistency of similar target can be kept, and accurately can position and find the special mesh isolated in SAR image
Mark.
Claims (4)
1. a kind of SAR image segmentation method based on super-pixel, comprises the following steps：
(1) to the SAR image I of input, its length L, width W and resolution ratio R are extractedI, and according to figure all in SAR image storehouse
Resolution ratio section R as estimating minimum targets；
(2) super-pixel number N is estimated according to image parameters, generate the super-pixel set S={ s of SAR image to be spliti, wherein
siFor i-th of super-pixel i=1,2 ..., NS；
(3) super-pixel set S textural characteristics F is calculatedgAnd space characteristics F (i)n(i, j), wherein i, j=1,2 ..., NsAnd i ≠
j；
(4) number K, the textural characteristics F using K-means algorithms to super-pixel set S are split according to SAR imageg(i) gathered
Class；
(5) according to super-pixel set S textural characteristics Fg(i) cluster result, to any two super-pixel in super-pixel set S
siAnd sj, using Nearest neighbor rule, judge whether super-pixel corresponds to same cluster, if siAnd sjCorresponding same cluster, and it
Space characteristics Fn(i, j)=1, then merged, and to super-pixel set S textural characteristics FgAnd space characteristics F (i)n
(i, j) is updated；Conversely, without merging；
It is described to use Nearest neighbor rule, judge whether super-pixel corresponds to same cluster, carry out as follows：
5a) for any two super-pixel siAnd sj, it is calculated respectively according to variable volume distance VBSD arrives all cluster centre cn
Distance dist (si,cn)、dist(sj,cn), wherein n=1,2 ..., K；
5b) by super-pixel siI-th of super-pixel s is distributed to apart from cluster centre corresponding to minimum valuei, as super-pixel siIt is corresponding
Cluster centre ci：
<mrow>
<msup>
<mi>c</mi>
<mi>i</mi>
</msup>
<mo>=</mo>
<munder>
<mi>arg</mi>
<mi>n</mi>
</munder>
<mi>m</mi>
<mi>i</mi>
<mi>n</mi>
<mrow>
<mo>(</mo>
<mi>d</mi>
<mi>i</mi>
<mi>s</mi>
<mi>t</mi>
<mo>(</mo>
<mrow>
<msub>
<mi>s</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<msub>
<mi>c</mi>
<mi>n</mi>
</msub>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
</mrow>
5c) by super-pixel sjJ-th of super-pixel s is distributed to apart from cluster centre corresponding to minimum valuej, as super-pixel sjIt is corresponding
Cluster centre cj：
<mrow>
<msup>
<mi>c</mi>
<mi>j</mi>
</msup>
<mo>=</mo>
<munder>
<mi>arg</mi>
<mi>n</mi>
</munder>
<mi>m</mi>
<mi>i</mi>
<mi>n</mi>
<mrow>
<mo>(</mo>
<mi>d</mi>
<mi>i</mi>
<mi>s</mi>
<mi>t</mi>
<mo>(</mo>
<mrow>
<msub>
<mi>s</mi>
<mi>j</mi>
</msub>
<mo>,</mo>
<msub>
<mi>c</mi>
<mi>n</mi>
</msub>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
</mrow>
Judge super-pixel siAnd sjWhether identical cluster centre is corresponded to, if super-pixel siAnd sjCorresponding cluster centre is identical,
Then super-pixel siAnd sjBelong to same cluster；
The textural characteristics F to super-pixel set SgAnd space characteristics F (i)n(i, j) is updated, and is carried out as follows：
5d) in two super-pixel siAnd sjAfter merging, the sequence number i of one of two super-pixel is removed, with the sequence of another super-pixel
Number j represents the super-pixel after merging, and obtains new textural characteristics Fg′(j)：
<mrow>
<msubsup>
<mi>F</mi>
<mi>g</mi>
<mo>&prime;</mo>
</msubsup>
<mrow>
<mo>(</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<msub>
<mi>F</mi>
<mi>g</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
</mrow>
<mo>&times;</mo>
<mi>N</mi>
<mi>u</mi>
<mi>m</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>s</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>+</mo>
<msub>
<mi>F</mi>
<mi>g</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>&times;</mo>
<mi>N</mi>
<mi>u</mi>
<mi>m</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>s</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mi>N</mi>
<mi>u</mi>
<mi>m</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>s</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>+</mo>
<mi>N</mi>
<mi>u</mi>
<mi>m</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>s</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
</mrow>
Wherein Fg(i) super-pixel s before merging is representediTextural characteristics, Fg(j) super-pixel s before merging is representedjTextural characteristics, Num
(si) represent super-pixel s before mergingiComprising number of pixels, Num (sj) represent super-pixel s before mergingjComprising number of pixels；
5e) delete s before mergingiTextural characteristics Fg(i), the textural characteristics F caused by newg' (j) substitutes sjTexture after merging is special
Sign；
(6) the super-pixel number N of the super-pixel set S after statistical updatings(t), and by the super-pixel number Ns(t) it is and last
Statistical result Ns(t-1) it is compared：If Ns(t)==Ns(t-1) stop condition, output SAR image segmentation knot, are then met
Fruit；If Ns(t) ＜ Ns(t-1) stop condition, return to step (4), are then unsatisfactory for.
2. the SAR image segmentation method according to claim 1 based on super-pixel, wherein the step (2) according to figure
As parameter Estimation super-pixel number Ns, carried out by the following two kinds situation：
The first situation is：According to length L, width W and the resolution ratio R for inputting SAR image to be splitI, estimate super-pixel number
NsInterval：
<mrow>
<msub>
<mi>N</mi>
<mi>s</mi>
</msub>
<mo>&Element;</mo>
<mfrac>
<mrow>
<mi>L</mi>
<mo>&times;</mo>
<mi>W</mi>
<mo>&times;</mo>
<msubsup>
<mi>R</mi>
<mi>I</mi>
<mn>2</mn>
</msubsup>
</mrow>
<msubsup>
<mi>R</mi>
<mi>s</mi>
<mn>2</mn>
</msubsup>
</mfrac>
<mo>&cap;</mo>
<mfrac>
<mrow>
<mi>L</mi>
<mo>&times;</mo>
<mi>W</mi>
</mrow>
<mrow>
<mi>N</mi>
<mi>u</mi>
<mi>m</mi>
<mrow>
<mo>(</mo>
<mi>s</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
</mrow>
Wherein, Num (s) is the interval of the number of pixels contained by each super-pixel, and interval is [50,200], super-pixel
Number NsTo appoint the integer value taken in interval；
Second of situation be：As the resolution ratio section R for being unable to estimate out minimum targetsWhen, according to input SAR image to be split
Length L, width W, estimate super-pixel number NsInterval be：
<mrow>
<msub>
<mi>N</mi>
<mi>s</mi>
</msub>
<mo>&Element;</mo>
<mfrac>
<mrow>
<mi>L</mi>
<mo>&times;</mo>
<mi>W</mi>
</mrow>
<mrow>
<mi>N</mi>
<mi>u</mi>
<mi>m</mi>
<mrow>
<mo>(</mo>
<mi>s</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>.</mo>
</mrow>
3. the SAR image segmentation method according to claim 1 based on super-pixel, wherein being calculated in the step (3) super
Pixel set S textural characteristics Fg(i), for any one super-pixel si, calculated by equation below：
<mrow>
<msub>
<mi>F</mi>
<mi>g</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<mi>H</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mn>1</mn>
<mo>:</mo>
<mi>m</mi>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mi>N</mi>
<mi>u</mi>
<mi>m</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>s</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
</mrow>
Wherein, H (i, 1:M) it is super-pixel siGrey level histogram, m represent image number of greyscale levels, Num (si) represent siIt is contained
Number of pixels.
4. the SAR image segmentation method according to claim 1 based on super-pixel, wherein being calculated in the step (3) super
Pixel set S space characteristics Fn(i, j), for any two super-pixel siAnd sj, calculated by equation below：
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510230209.1A CN104794730B (en) | 2015-05-07 | 2015-05-07 | SAR image segmentation method based on super-pixel |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510230209.1A CN104794730B (en) | 2015-05-07 | 2015-05-07 | SAR image segmentation method based on super-pixel |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104794730A CN104794730A (en) | 2015-07-22 |
| CN104794730B true CN104794730B (en) | 2018-03-06 |
Family
ID=53559510
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510230209.1A Active CN104794730B (en) | 2015-05-07 | 2015-05-07 | SAR image segmentation method based on super-pixel |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104794730B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106203399B (en) * | 2016-07-27 | 2019-06-04 | 厦门美图之家科技有限公司 | A kind of image processing method, device and calculate equipment |
| CN106558029A (en) * | 2016-10-28 | 2017-04-05 | 成都西纬科技有限公司 | A kind of image filtering method and device |
| CN106780485B (en) * | 2017-01-12 | 2020-01-03 | 西安电子科技大学 | SAR image change detection method based on super-pixel segmentation and feature learning |
| CN107895139B (en) * | 2017-10-19 | 2021-09-21 | 金陵科技学院 | SAR image target identification method based on multi-feature fusion |
| CN108828583B (en) * | 2018-06-15 | 2022-06-28 | 西安电子科技大学 | Point trace clustering method based on fuzzy C mean value |
| CN110533669B (en) * | 2019-08-06 | 2023-01-06 | 西安电子科技大学 | SAR image super-pixel segmentation method based on variational level set |
| CN115131373B (en) * | 2022-07-14 | 2023-10-27 | 西安电子科技大学 | SAR image segmentation method based on texture features and SLIC |
| CN116030037A (en) * | 2023-02-23 | 2023-04-28 | 西南石油大学 | Ultrasonic image simulation method based on self-supervision learning |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103824300A (en) * | 2014-03-12 | 2014-05-28 | 西安电子科技大学 | SAR (synthetic aperture radar) image segmentation method based on spatial correlation feature ultra-pixel block |
| CN104463210B (en) * | 2014-12-08 | 2017-10-24 | 西安电子科技大学 | Classification of Polarimetric SAR Image method based on object-oriented and spectral clustering |
-
2015
- 2015-05-07 CN CN201510230209.1A patent/CN104794730B/en active Active
Non-Patent Citations (2)
| Title |
|---|
| A Novel Iris Segmentation Approach Based on Superpixel Method;Liuyang Cao et al.;《2014 Fourth International Conference on Instrumentation and Measurement, Computer, Communication and Control》;20141231;第826-831页 * |
| Histogram similarity measure using variable bin size distance;Yu Ma et al.;《Computer Vision and Images Understanding》;20101231;第981-989页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104794730A (en) | 2015-07-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104794730B (en) | SAR image segmentation method based on super-pixel | |
| CN102096825B (en) | Graph-based semi-supervised high-spectral remote sensing image classification method | |
| CN108830870B (en) | Satellite image high-precision farmland boundary extraction method based on multi-scale structure learning | |
| CN104077599B (en) | Polarization SAR image classification method based on deep neural network | |
| CN102013017B (en) | Method for roughly sorting high-resolution remote sensing image scene | |
| CN102402685B (en) | Method for segmenting three Markov field SAR image based on Gabor characteristic | |
| CN110852245B (en) | Precipitation particle classification method of double-polarization weather radar based on discrete attribute BNT | |
| CN102982338B (en) | Classification of Polarimetric SAR Image method based on spectral clustering | |
| CN103839073B (en) | Polarization SAR image classification method based on polarization features and affinity propagation clustering | |
| CN107491734B (en) | Semi-supervised polarimetric SAR image classification method based on multi-core fusion and space Wishart LapSVM | |
| CN104408483B (en) | SAR texture image classification methods based on deep neural network | |
| Fathizad et al. | Evaluating desertification using remote sensing technique and object-oriented classification algorithm in the Iranian central desert | |
| CN103971115A (en) | Automatic extraction method for newly-increased construction land image spots in high-resolution remote sensing images based on NDVI and PanTex index | |
| CN103745472B (en) | SAR image segmentation method based on condition triple Markov field | |
| Tyralis et al. | Estimation of intensity–duration–frequency curves using max-stable processes | |
| CN106611422A (en) | Stochastic gradient Bayesian SAR image segmentation method based on sketch structure | |
| CN104346814B (en) | Based on the SAR image segmentation method that level vision is semantic | |
| CN109300115B (en) | Object-oriented multispectral high-resolution remote sensing image change detection method | |
| CN102496142B (en) | SAR (synthetic aperture radar) image segmentation method based on fuzzy triple markov fields | |
| CN106096622B (en) | Semi-supervised Classification of hyperspectral remote sensing image mask method | |
| CN103886335A (en) | Polarized SAR image classifying method based on fuzzy particle swarms and scattering entropy | |
| CN105184297A (en) | Polarized SAR image classification method based on tensor and sparse self-coder | |
| CN104751183A (en) | Polarimetric SAR image classification method based on tensor MPCA | |
| Xia et al. | Land resource use classification using deep learning in ecological remote sensing images | |
| CN102609721B (en) | Remote sensing image clustering method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |