CN106846322B - The SAR image segmentation method learnt based on curve wave filter and convolutional coding structure - Google Patents
The SAR image segmentation method learnt based on curve wave filter and convolutional coding structure Download PDFInfo
- Publication number
- CN106846322B CN106846322B CN201611260213.3A CN201611260213A CN106846322B CN 106846322 B CN106846322 B CN 106846322B CN 201611260213 A CN201611260213 A CN 201611260213A CN 106846322 B CN106846322 B CN 106846322B
- Authority
- CN
- China
- Prior art keywords
- sketch
- line
- pixel
- sar image
- sketch line
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
- G06T2207/10041—Panchromatic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
Landscapes
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention discloses a kind of SAR image segmentation methods learnt based on curve wave filter and convolutional coding structure, mainly solve the problems, such as prior art segmentation SAR image inaccuracy.Implementation step is: (1) SAR image sketch obtains sketch map;(2) according to the administrative division map of SAR image, the pixel subspace of SAR image is divided;(3) curve ripple filter set is constructed;(4) convolutional coding structure learning model is constructed;(5) SAR image segmentation method based on curve wave filter and convolutional coding structure learning model, segmentation mixing aggregated structure atural object pixel subspace are used;(6) gather the pinpoint target segmentation of feature based on sketch line;(7) the line target segmentation of view-based access control model semantic rules;(8) homogenous region pixel subspace is split using based on multinomial logistic regression prior model;(9) combination and segmentation is as a result, obtain SAR image segmentation result.Present invention obtains the good segmentation effects of SAR image, can be used for the semantic segmentation of SAR image.
Description
Technical field
The invention belongs to technical field of image processing, further relate to one of technical field of image segmentation and are based on song
Image segmentation side synthetic aperture radar SAR (Synthetic Aperture Radar) of line wave filter and convolutional coding structure study
Method.The present invention can accurately be split the region with different characteristic of synthetic aperture radar SAR image, and can be used
In the object detection and recognition of subsequent synthetic aperture radar SAR image.
Background technique
Synthetic aperture radar SAR obtains answering extensively in military and civilian multiple fields due to its special image-forming mechanism
With.SAR has many advantages, such as round-the-clock, round-the-clock, multiband, multipolarization, variable side view angle and high-resolution, not only can be detailed
Carefully, landform, landforms are accurately surveyed and drawn, the information of earth surface is obtained, underground letter can also be collected through earth's surface and natural vegetation
Breath, or even under rugged environment also detailed ground surveying and mapping data and image can be provided with higher resolution ratio.
As one of technology basic and crucial in image understanding and interpretation, image segmentation is connection image procossing and image
The tie of analysis and understanding, occupies consequence in Image Engineering, and the superiority and inferiority of segmentation result will affect subsequent picture analysis, reason
The quality and efficiency of solution.But SAR image but has coherent speckle noise obvious, similar target differs greatly, and heterogeneous destinations connect very much
Closely, exist while multiscale target, data scale is huge, the problems such as lacking effective semantic expressiveness, and these problems all increase
The difficulty of SAR image segmentation.
Simultaneously SAR image segmentation needs to introduce high-level semantic knowledge, when people face piece image when, concern and
Pixel not instead of not one by one, target made of multiple pixel aggregations, these significant targets are only it is appreciated that scheming
The basic unit of picture, and the numeric string of pixel only image discretization, therefore the semantic primitive of image and single
Pixel is differentiated, this namely famous " semantic gap " problem.The SAR image of previous work is understood as SAR image
Segmentation also includes semantic requirement.However, associated research is considerably less, existing semantic representation method is built
It stands on natural image, how for the SAR image effective semantic rules of design and a urgent problem to be solved.
Paper that the Zhong Weiyu and Shen Ting of Center for Earth Observation and Digital Earth Chinese Academy of Sciences are delivered at it " in conjunction with
One kind is proposed in the SAR image of multiple features and SVM segmentation " (" computer application research ", 2013,30 (9): 2846-2851.)
SAR image segmentation is carried out in conjunction with the texture characteristic extracting method of non-down sampling profile transformation (NSCT) and GLCM.This method is
It realizes that gray level co-occurrence matrixes (GLCM) multiple dimensioned, multidirectional texture feature extraction, proposes a kind of combination non-down sampling profile
Convert the texture characteristic extracting method of (NSCT) and GLCM.First with NSCT to synthetic aperture radar (SAR) SAR) image carry out it is multiple dimensioned,
Multi-direction decomposition;Gray scale symbiosis amount is extracted using GLCM to obtained sub-band images again;Then to the gray scale symbiosis amount of extraction into
Row correlation analysis removes redundancy feature amount, and it is constituted multiple features vector with gray feature;Finally, making full use of support
Advantage of the vector machine (SVM) in terms of Small Sample Database library and generalization ability is completed the division of multiple features vector by SVM, is realized
SAR image segmentation.But the shortcoming that this method still has is, does not introduce the high-level semantic knowledge of SAR image, only
SAR image is divided in pixel scale, results in SAR image segmentation result inaccuracy.
The paper " a kind of effective MSTAR SAR image segmentation method " that Linda of Wuhan University et al. delivers at it is (" military
Chinese college journal ", 2015,40 (10): 1377-1380.) in propose a kind of MSTAR SAR image segmentation method.This method is first
Over-segmentation operation first is carried out to image to be processed, segmented image region is obtained.Secondly image is carried out to the image after over-segmentation
The feature extraction of region class and Pixel-level obtains the feature vector for indicating image, hidden to MSTAR SAR image use space
The distribution model of Cray containing Di Li (sLDA) and markov random file (MRF) establish proposed model, and it is general to obtain energy
Letter.Finally energy functional is optimized with Graph-Cut algorithm and Branch-and-Bound algorithm, obtains final point
Cut result.Shortcoming existing for this method is, in the feature vector for acquiring SAR image, do not learn in SAR image due to
Correlation and distinctive structure feature between pixel.
Liu Fang, Duan Yiping, Li Lingling, burnt Lee " are based on the semantic and adaptive neighbour of level vision in its paper delivered at equal
The SAR image of the hidden model of domain multinomial is divided " (IEEE Trancactions on Geoscience and Remote
Sensing, 2016,54 (7): 4287-4301.) in propose it is a kind of based on level vision semanteme and adaptive neighborhood multinomial
The SAR image segmentation method of hidden model, this method propose the level vision of SAR image on the basis of SAR image sketch map
It is semantic.SAR image is divided into aggregation zone, structural region and homogenous region by the level vision semanteme.Based on the division, to not
Region with characteristic uses different dividing methods.For aggregation zone, gray level co-occurrence matrixes feature is extracted, and using part
The method of linear restriction coding obtains the expression of each aggregation zone, and then is split using the method for hierarchical clustering.To knot
Structure region devises vision semantic rules positioning boundary and line target by analysis side mode type and line model.In addition, boundary and
Line target contains strong directional information, therefore devises the hidden model of multinomial based on geometry window and be split.It is right
Homogenous region is gone to indicate center pixel, devises the hidden mould of multinomial based on self-adapting window in order to find appropriate neighborhood
Type is split.The segmentation result in these three regions be integrated into together segmentation result to the end.The deficiency of this method
Place is, inaccurate to the boundary alignment of aggregation zone, not reasonable to the determination of homogenous region classification number, the area of segmentation result
Domain consistency is poor, and does not handle pinpoint target in the segmentation of structural region.
Summary of the invention
It is an object of the invention to be directed to the deficiency of above-mentioned prior art, propose a kind of based on curve wave filter and convolution
The SAR image segmentation method of structural model, with the significantly more efficient segmentation for completing SAR image.
To achieve the above object, technical scheme is as follows:
(1) SAR image sketch:
(1a) obtains its sketch model according to the characteristic distributions of SAR image to the synthetic aperture radar SAR image of input;
(1b) utilizes SAR image sketch model, carries out sketch processing to the synthetic aperture radar SAR image of input, obtains
The corresponding sketch map of synthetic aperture radar SAR image that must be inputted;
(2) pixel subspace is divided:
(2a) uses sketch line fields method, carries out compartmentalization processing to the sketch map of synthetic aperture radar SAR image,
Obtain include aggregation zone, synthetic aperture radar SAR image without sketch line region and structural region administrative division map;
(2b) will include aggregation zone, the administrative division map without sketch line region and structural region, be mapped to the synthesis hole of input
In diameter radar SAR image, mixing aggregated structure atural object pixel subspace, the homogeneous area in synthetic aperture radar SAR image are obtained
Domain pixel subspace and structure-pixel subspace;
(3) curve ripple filter set is constructed:
(3a) extracts pair of mixing aggregated structure atural object pixel subspace from the administrative division map of synthetic aperture radar SAR image
The aggregation zone answered is divided into 10 ° in [0 °, 180 °] section with and is divided into 18 sections i.e. 18 directions, and statistics is every respectively
In a section in the aggregation zone sketch line segment line segment item number;
Sketch line segment item number (3b) all to the aggregation zone, according to how much progress of the line segment item number in each interval
Sequence, obtains the collating sequence in direction, using the degree in 6 directions preceding in the collating sequence in direction as curve wave filter
In directioin parameter;
The scale parameter of curve wave filter in [- 4,0] section, is divided into 0.2 carry out discretization with, obtained by (3c)
Scale parameter after discretization;It is discrete that the displacement parameter of curve wave filter is divided into 0.5 carry out in [0,9] section
Change, the displacement parameter after obtaining discretization;
(3d) according to the following formula, the curve wave function of calculated curve wave filter:
Wherein, t indicates the curve wave function of curve wave filter, and b indicates the scale parameter in curve wave filter, DbTable
Show using scale parameter b as the scale operator in the curve wave filter of parameter, θ indicates the direction ginseng in curve wave filter
Number, RθIndicate that, using directioin parameter θ as the rotation operator in the curve wave filter of parameter, (m, n) indicates curve wave filter
The coordinate position of middle pixel, m0Indicate the displacement parameter of horizontal direction, n0Indicate the displacement parameter of vertical direction,It indicates
By the displacement parameter m of horizontal direction0With the displacement parameter n of vertical direction0As the displacement ginseng in the curve wave filter of parameter
Number;
(3e) according to the following formula, calculates each curve wave filter:
Wherein, c (t) indicates that the curve wave filter using curve wave function t as parameter, cos indicate cosine operation, exp
It indicates using natural constant e as the index operation at bottom;
Each the curve ripple filter bank being calculated is become curve ripple filter set by (3f);
(4) convolutional coding structure learning model is constructed:
(4a) is not connected to each of the mixing aggregated structure atural object pixel subspace in synthetic aperture radar SAR image mutually
Region, by 31 × 31 window carry out every a sliding window sample, the corresponding multiple images block in each region is obtained, by multiple images
Block is sequentially inputted in convolutional coding structure learning model, obtains the input layer of convolutional coding structure learning model;
(4b) uses curve wave filter, carries out convolution behaviour to the image block in the input layer of convolutional coding structure learning model
Make, obtains the convolutional layer of convolutional coding structure learning model;
(4c) according to the following formula, calculates data fidelity term:
Wherein, E (c) indicates data fidelity term, and c indicates that the filter in convolutional coding structure learning model convolutional layer, N indicate volume
Each of to be learnt the total number for the image block that mutually disconnected region includes in product Structure learning model, | | | |FExpression is done
The operation of frobenius norm,Indicate the square operation of frobenius norm, xiIt indicates to input in convolutional coding structure learning model
I-th of image block,It indicates to extract xiIntermediate size is the characteristic image block of n × n, MiIndicate convolutional coding structure learning model
The sum of the corresponding curve wave filter of i-th of image block of middle input, ∑ indicate sum operation, and * indicates convolution operation,Table
Show corresponding j-th of curve wave filter of i-th of the image block inputted in convolutional coding structure learning model;
(4d) according to the following formula, calculates structure fidelity term:
Wherein, G (c) indicates structure fidelity term, and R () indicates to ask the operation of all sketch line total lengths in sketch map, SM
() indicates to extract the operation with the one-to-one sketch segment of input picture block;
(4e) according to the following formula, calculating target function:
Wherein, L (c) indicates objective function,It indicates in objective function L (c) value minimum, seeks curve ripple filter
The operation of wave device c;
(4f) exports the curve wave filter obtained by objective function guidance learning, obtains the output of convolutional coding structure model
Layer;
(5) training convolutional Structure learning model:
(5a) sets 0.1 for structural failure threshold value;
Step (4a) sampling is obtained image block and is sequentially inputted in convolutional coding structure learning model by (5b);
(5c) randomly selects six filters, directioin parameter is by system in step (3b) from curve ripple filter set
6 directions of meter obtain, displacement parameter and scale parameter random initializtion, the filtering that these initial six filters are formed
Device set is as selected curve ripple filter set;
(5d) by each curve filter in image block currently entered and selected curve ripple filter set into
Row convolution operation obtains characteristic pattern corresponding with each curve filter;
(5e) utilizes the structure fidelity term formula in step (4d), the structure fidelity term of calculating input image block;
(5f) judges whether the structure fidelity term of current input image block is less than structural failure threshold value, if so, executing step
Suddenly (5i) is otherwise executed step (5g);
(5g) utilizes scale parameter more new formula and displacement parameter more new formula, updates step (4c) data fidelity term respectively
The scale parameter and displacement parameter of curve wave filter in formula, obtain updated curve wave filter;
(5h) is using updated curve ripple filter set as selected curve ripple filter set, return step
(5d) re-starts training to input picture block;
The curve wave filter that study obtains is saved the curve wave filter collection succeeded in school to the input picture block by (5i)
In conjunction, the study to the feature of the input picture block is completed, and export the curve wave filter collection that the input picture block succeeds in school
It closes;
(5j) judges whether all image blocks pass through the study that convolutional coding structure learning model completes feature, if so, terminating
Otherwise program inputs next image block and executes step (5c);
(6) segmentation SAR image mixes aggregated structure atural object pixel subspace:
(6a) splices the characteristic set of all mutual connected regions, using spliced characteristic set as code book;
(6b) calculates separately the inner product with each feature in code book, obtains to all features of each mutual not connected region
To projection vector of all features in each region on code book;
(6c) it is corresponding to obtain each region to each mutually all projection vectors progress maximum value convergence of connected region
One structural eigenvector;
(6d) propagates AP clustering algorithm using neighbour, does not cluster to the structural eigenvector of all mutual connected regions,
Obtain the segmentation result of mixing aggregated structure atural object pixel subspace;
(7) segmenting structure pixel subspace:
(7a) uses vision semantic rules, divides line target;
The feature of gathering of (7b) based on sketch line divides pinpoint target;
The result that (7c) divides line target and pinpoint target merges, and obtains the segmentation knot of structure-pixel subspace
Fruit;
(8) divide homogenous region pixel subspace:
Using based on multinomial logistic regression prior model without sketch line region segmentation method, to homogenous region pixel
Space is split, and obtains the segmentation result of homogenous region pixel subspace;
(9) combination and segmentation result:
The segmentation of aggregated structure atural object pixel subspace, homogenous region pixel subspace and structure-pixel subspace will be mixed
As a result it merges, obtains the final segmentation result of synthetic aperture radar SAR image.
Compared with the prior art, the present invention has the following advantages:
First, the present invention is closed using the sketch map of synthetic aperture radar SAR image using sketch line fields method
At the administrative division map of aperture radar SAR image, administrative division map is mapped to the synthetic aperture radar SAR image of input, obtains synthesis hole
Mixing aggregated structure atural object pixel subspace, homogenous region pixel subspace and structure-pixel in diameter radar SAR image is empty
Between, sampling and feature learning are carried out in the pixel subspace of mixing aggregated structure atural object, are overcome the prior art and are closed in segmentation
At the deficiency being only split in pixel scale to synthetic aperture radar SAR image when aperture radar SAR image, so that of the invention
With introducing the high-level semantic knowledge of synthetic aperture radar SAR image, and improve synthetic aperture radar SAR image segmentation knot
The accuracy of fruit.
Second, in the convolutional coding structure model that the present invention constructs, schemed using the data fidelity term and input of input picture block
The structural information of input picture block is extracted as the structure fidelity term of block, the prior art is overcome and acquires synthetic aperture radar
When the feature vector of SAR image, the Pixel-level Image Segmentation Methods Based on Features synthetic aperture radar SAR of synthetic aperture radar SAR image is only used
Image, and the deficiency of the structural information of synthetic aperture radar SAR image is had ignored, so that the present invention, which has, can learn to synthesize
Due to the correlation between pixel and the advantages of peculiar structure feature in aperture radar SAR image.
Third, the present invention is to each of the mixing aggregated structure atural object pixel subspace in synthetic aperture radar SAR image
Mutual disconnected region sample every a sliding window by 32 × 32 window, obtains multiple images block, and obtained image block is defeated
Enter into convolutional coding structure model, the input for overcoming the depth self-encoding encoder that characteristics of image is automatically extracted used in the prior art is
One-dimensional vector can destroy the deficiency of the spatial structure characteristic of image, so that the present invention has the substantive characteristics that can extract image,
The advantages of promoting the precision of synthetic aperture radar SAR image segmentation.
Detailed description of the invention
Fig. 1 is flow chart of the invention;
Fig. 2 is analogous diagram of the invention;
Fig. 3 is pinpoint target segmentation result figure of the present invention;
Fig. 4 is simulation result schematic diagram of the present invention.
Specific embodiment
The present invention will be further described with reference to the accompanying drawing.
Referring to attached drawing 1, the specific steps of the present invention are as follows.
Step 1, SAR image sketch.
To the synthetic aperture radar SAR image of input, its sketch model is obtained according to the characteristic distributions of SAR image.
According to the following steps, using SAR image sketch model, sketch is carried out to the synthetic aperture radar SAR image of input
Change processing, obtains the corresponding sketch map of synthetic aperture radar SAR image of input:
Step 1 arbitrarily chooses a number, the sum as template in [100,150] range;
Step 2 constructs a template on the side being made of pixel with different directions and scale, line, utilizes template
Direction and dimensional information structural anisotropy's Gaussian function, by the Gaussian function, in calculation template each pixel plus
Weight coefficient, the weighting coefficient of all pixels point in statistical mask, wherein scale number value is 3~5, and direction number value is
18;
Step 3 calculates pixel in synthetic aperture radar SAR image corresponding with template area coordinate according to the following formula
Mean value:
Wherein, μ indicates the equal of all pixels point in corresponding with template area coordinate synthetic aperture radar SAR image
Value, ∑ indicate sum operation, and g indicates the corresponding coordinate of any one pixel in the Ω region of template, and ∈ expression belongs to symbol
Number, wgIndicate weight coefficient of the pixel at coordinate g in the Ω region of template, wgValue range be wg∈ [0,1], Ag
Indicate the value of the pixel with pixel in the Ω region of template at the coordinate g in corresponding synthetic aperture radar SAR image;
Step 4 calculates pixel in synthetic aperture radar SAR image corresponding with template area coordinate according to the following formula
Variance yields:
Wherein, ν indicates the variance of all pixels point in synthetic aperture radar SAR image corresponding with template area coordinate
Value;
Step 5 calculates the response that each pixel in synthetic aperture radar SAR image is directed to ratio operator according to the following formula
Value:
Wherein, R indicates response of each pixel for ratio operator, min { } in synthetic aperture radar SAR image
Indicate minimum Value Operations, a and b respectively indicate two different regions in template, μaIndicate all pixels point in a of template area
Mean value, μbIndicate the mean value of all pixels point in the b of template area;
Step 6 calculates the response that each pixel in synthetic aperture radar SAR image is directed to correlation operator according to the following formula
Value:
Wherein, C indicate synthetic aperture radar SAR image in each pixel be directed to correlation operator response,It indicates
Square root functions, a and b respectively indicate two different zones, ν in templateaIndicate the variance of all pixels point in a of template area
Value, νbIndicate the variance yields of all pixels point in the b of template area, μaIndicate the mean value of all pixels point in a of template area, μbTable
Show the mean value of all pixels point in the b of template area;
Step 7 calculates the response that each pixel in synthetic aperture radar SAR image is directed to each template according to the following formula
Value:
Wherein, F indicate synthetic aperture radar SAR image in each pixel be directed to each template response,It indicates
Square root functions, R and C respectively indicate pixel in synthetic aperture radar SAR image and are directed to ratio operator and synthetic aperture radar
Pixel is directed to the response of correlation operator in SAR image;
Step 8, judges whether constructed template is equal to the sum of selected template, if so, step 9 is executed, otherwise,
Execute step 2;
Step 9, selection has the template of maximum response from each template, as synthetic aperture radar SAR image
Template, and using the maximum response of the template as the intensity of pixel in synthetic aperture radar SAR image, by the side of the template
The direction of pixel in as synthetic aperture radar SAR image, obtain synthetic aperture radar SAR image sideline response diagram and
Gradient map;
Step 10 calculates the intensity value of synthetic aperture radar SAR image intensity map, obtains intensity map according to the following formula:
Wherein, I indicates that the intensity value of synthetic aperture radar SAR image intensity map, r indicate synthetic aperture radar SAR image
Value in the response diagram of sideline, t indicate the value in synthetic aperture radar SAR image gradient map;
Step 11 detects intensity map using non-maxima suppression method, obtains suggestion sketch;
Step 12, choose suggest sketch in maximum intensity pixel, will suggest sketch in the maximum intensity
The pixel of pixel connection connects to form suggestion line segment, obtains suggestion sketch map;
Step 13 calculates the code length gain for suggesting sketch line in sketch map according to the following formula:
Wherein, CLG indicates to suggest the code length gain of sketch line in sketch map, and ∑ indicates sum operation, and J indicates current
The number of pixel, A in sketch line neighborhoodjIndicate the observation of j-th of pixel in current sketch line neighborhood, Aj,0It indicates
In the case that current sketch line cannot indicate structural information, the estimated value of j-th of pixel, ln () table in the sketch line neighborhood
Show the log operations using e the bottom of as, Aj,1Indicate the sketch line neighborhood in the case where current sketch line can indicate structural information
In j-th of pixel estimated value;
Step 14 arbitrarily chooses a number, as threshold value T in [5,50] range;
Step 15 selects the suggestion sketch line of CLG > T in all suggestion sketch lines, is combined into synthetic aperture radar
The sketch map of SAR image.
The synthetic aperture radar SAR image sketch model that the present invention uses is that Jie-Wu et al. in 2014 was published in IEEE
Article " Local maximal on Transactions on Geoscience and Remote Sensing magazine
homogenous region search for SAR speckle reduction with sketch-based
Geometrical kernel function " proposed in model.
Step 2, pixel subspace is divided.
Using sketch line fields method, compartmentalization processing is carried out to the sketch map of synthetic aperture radar SAR image, is obtained
The administrative division map of synthetic aperture radar SAR image including aggregation zone, without sketch line region and structural region.
According to the concentration class of sketch line segment in the sketch map of synthetic aperture radar SAR image, sketch line is divided into expression
Assemble the aggregation sketch line of atural object and indicates boundary, line target, the boundary sketch line of isolated target, line target sketch line, isolates
Target sketch line.
According to the statistics with histogram of sketch line segment concentration class, the sketch line segment conduct that concentration class is equal to optimal concentration class is chosen
Seed line-segment sets { Ek, k=1,2 ..., m }, wherein EkIndicate that any bar sketch line segment in seed line-segment sets, k indicate seed
The label of any bar sketch line segment in line-segment sets, m indicate the total number of seed line segment, and { } indicates set operation.
As basic point with the unselected line segment for being added to seed line-segment sets sum, with this basic point recursive resolve line segment aggregate.
The round primitive that a radius is the optimal concentration class section upper bound is constructed, with the circle primitive in line segment aggregate
Line segment is expanded, and is corroded to the line segment aggregate ecto-entad after expansion, is obtained as unit of sketch point in sketch map
Aggregation zone.
To the sketch line for indicating boundary, line target and isolated target, centered on each sketch point of each sketch line
The geometry window that size is 5 × 5 is constructed, structural region is obtained.
The part other than aggregation zone and structural region will be removed in sketch map as can not sketch region.
By in sketch map aggregation zone, structural region and can not sketch region merging technique, to obtain include aggregation zone, without element
Retouch the administrative division map of the synthetic aperture radar SAR image of line region and structural region.
It will include aggregation zone, the administrative division map without sketch line region and structural region, be mapped to the synthetic aperture thunder of input
Up in SAR image, mixing aggregated structure atural object pixel subspace in synthetic aperture radar SAR image, homogenous region picture are obtained
Sub-prime space and structure-pixel subspace.
Step 3, curve wave filter is constructed.
The corresponding of mixing aggregated structure atural object pixel subspace is extracted from the administrative division map of synthetic aperture radar SAR image
Aggregation zone is divided into 10 ° in [0 °, 180 °] section with and is divided into 18 sections i.e. 18 directions, counts each area respectively
The line segment item number of sketch line segment in the interior aggregation zone.
The sketch line segment item number all to the aggregation zone is arranged according to the number of the line segment item number in each interval
Sequence obtains the collating sequence in direction, using the degree in 6 directions preceding in the collating sequence in direction as curve wave filter in
Directioin parameter.
By the scale parameter of curve wave filter in [- 4,0] section, it is divided into 0.2 carry out discretization with, is obtained discrete
Scale parameter after change;The displacement parameter of curve wave filter is divided into 0.5 carry out discretization in [0,9] section with, is obtained
Displacement parameter after to discretization.
According to the following formula, the curve wave function of calculated curve wave filter:
Wherein, t indicates the curve wave function of curve wave filter, and b indicates the scale parameter in curve wave filter, DbTable
Show using scale parameter b as the scale operator in the curve wave filter of parameter, θ indicates the direction ginseng in curve wave filter
Number, RθIndicate that, using directioin parameter θ as the rotation operator in the curve wave filter of parameter, (m, n) indicates curve wave filter
The coordinate position of middle pixel, m0Indicate the displacement parameter of horizontal direction, n0Indicate the displacement parameter of vertical direction,It indicates
By the displacement parameter m of horizontal direction0With the displacement parameter n of vertical direction0As the displacement ginseng in the curve wave filter of parameter
Number.
According to the following formula, each curve wave filter is calculated:
Wherein, c (t) indicates that the curve wave filter using curve wave function t as parameter, cos indicate cosine operation, exp
It indicates using natural constant e as the index operation at bottom.
Each the curve ripple filter bank being calculated is become into curve ripple filter set.
Step 4, convolutional coding structure learning model is constructed.
To the mutual disconnected area in each of mixing aggregated structure atural object pixel subspace in synthetic aperture radar SAR image
Domain, by 31 × 31 window carry out every a sliding window sample, obtain the corresponding multiple images block in each region, by multiple images block according to
It is secondary to be input in convolutional coding structure learning model, obtain the input layer of convolutional coding structure learning model.
Using curve wave filter, convolution operation is carried out to the image block in the input layer of convolutional coding structure learning model, is obtained
To the convolutional layer of convolutional coding structure learning model.
According to the following formula, data fidelity term is calculated:
Wherein, E (c) indicates data fidelity term, and c indicates that the filter in convolutional coding structure learning model convolutional layer, N indicate volume
Each of to be learnt the total number for the image block that mutually disconnected region includes in product Structure learning model, | | | |FExpression is done
The operation of frobenius norm,Indicate the square operation of frobenius norm, xiIt indicates to input in convolutional coding structure learning model
I-th of image block,It indicates to extract xiIntermediate size is the characteristic image block of n × n, MiIndicate convolutional coding structure learning model
The sum of the corresponding curve wave filter of i-th of image block of middle input, ∑ indicate sum operation, and * indicates convolution operation,Table
Show corresponding j-th of curve wave filter of i-th of the image block inputted in convolutional coding structure learning model.
According to the following formula, structure fidelity term is calculated:
Wherein, G (c) indicates structure fidelity term, and R () indicates to ask the operation of all sketch line total lengths in sketch map, SM
() indicates to extract the operation with the one-to-one sketch segment of input picture block.
According to the following formula, calculating target function:
Wherein, L (c) indicates objective function,It indicates in objective function L (c) value minimum, seeks curve ripple filter
The operation of wave device c.
The curve wave filter obtained by objective function guidance learning is exported, the output layer of convolutional coding structure model is obtained.
Step 5, according to the following steps, training convolutional Structure learning model:
Step 1 sets 0.1 for structural failure threshold value;
Step 4 sampling is obtained image block and is sequentially inputted in convolutional coding structure learning model by step 2;
Step 3 randomly selects six filters, directioin parameter is by step (3b) from curve ripple filter set
6 directions of statistics obtain, displacement parameter and scale parameter random initializtion, the filter that these initial six filters are formed
Wave device set is as selected curve ripple filter set;
Step 4, by each curve filter in image block currently entered and selected curve ripple filter set
Convolution operation is carried out, characteristic pattern corresponding with each curve filter is obtained;
Step 5 utilizes the structure fidelity term formula in step 4, the structure fidelity term of calculating input image block;
Step 6, judges whether the structure fidelity term of current input image block is less than structural failure threshold value, if so, executing
Otherwise step 9 executes step 7;
Step 7 updates step 4 data fidelity term using scale parameter more new formula and displacement parameter more new formula respectively
The scale parameter and displacement parameter of curve wave filter in formula, obtain updated curve wave filter;
Scale parameter more new formula is as follows:
Wherein, btIndicate the scale parameter of the curve wave filter of the t times iteration, bt-1Indicate the scale of the t-1 times iteration
Parameter, η indicate that step-length, value range are [0,1], and c () indicates that curve wave filter, sin indicate cosine operation, and X indicates bent
The horizontal direction parameter of line wave filter, value is by formula X=DbRθ(m-m0) obtain, Y indicates the Vertical Square of curve wave filter
To parameter, value is by formula Y=DbRθ(n-n0) obtain,Indicate curve wave filter to the local derviation of independent variable X, value by
Following formula obtains:
Curve wave filter horizontal direction parameter X is indicated to the local derviation of scale parameter, value is obtained by the following formula:
Curve wave filter is indicated to the local derviation of independent variable Y, value is obtained by the following formula:
Curve wave filter vertical direction Y is indicated to the local derviation of scale parameter, value is obtained by the following formula:
Displacement parameter more new formula is as follows:
Wherein,Indicate the horizontal direction displacement parameter of the curve wave filter of the t times iteration,It indicates to change for the t-1 times
The horizontal direction displacement parameter in generation,
Wherein,Indicate the vertical direction displacement parameter of the curve wave filter of the t times iteration,It indicates to change for the t-1 times
The vertical direction displacement parameter in generation.
Step 8, using updated curve ripple filter set as selected curve ripple filter set, return step
Step 4 re-starts training to input picture block;
The curve wave filter that study obtains is saved the curve wave filter succeeded in school to the input picture block by step 9
In set, the study to the feature of the input picture block is completed, and export the curve wave filter that the input picture block succeeds in school
Set;
Step 10, judges whether all image blocks pass through the study that convolutional coding structure learning model completes feature, if so, knot
Otherwise Shu Chengxu inputs next image block and executes step 3.
Step 6, segmentation SAR image mixes aggregated structure atural object pixel subspace.
By all mutually not characteristic set splicings of connected region, using spliced characteristic set as code book.
To all features of each mutual not connected region, the inner product with each feature in code book is calculated separately, is obtained every
Projection vector of a all features in region on code book.
To each mutually all projection vectors progress maximum value convergence of connected region, it is one corresponding to obtain each region
Structural eigenvector.
AP clustering algorithm is propagated using neighbour, the structural eigenvector of all mutual connected regions is not clustered, is obtained
Mix the segmentation result of aggregated structure atural object pixel subspace.
Step 7, segmenting structure pixel subspace.
With vision semantic rules, divide line target.
If i-th sketch line liWith j-th strip sketch line ljThe distance between be Dij, liDirection be Oi, ljDirection be Oj,
I, j ∈ [1,2 ..., S], S are the total number of sketch line.
Width is greater than the line target of 3 pixels with two sketch line liAnd ljIt indicates, liAnd ljThe distance between DijIt is less than
T1And direction difference (Oi-Oj) less than 10 degree, wherein T1=5.
If the s articles sketch line lsGeometry window wsThe average gray of interior each column is AiIf the gray scale difference of adjacent column is
ADi=| Ai-Ai+1|, if zs=[zs1,zs2,...,zs9] be adjacent column gray scale difference ADiLabel vector.
By width less than the line target of 3 pixels with single sketch line lsIt indicates, lsGeometry window wsIt is interior, calculate phase
The gray scale difference AD of adjacent columniIf ADi> T2, then zsi=1.Otherwise zsi=0, zsIn there are two element value be 1, remaining is 0,
Middle T2=34.
If L1,L2It is the set for indicating the sketch line of line target, if Dij< T1And | Oi-Oj| < 10, then li,lj∈
L1;If sum (zs)=2, then ls∈L2, wherein sum () indicate to vector important summation operation.
In structure-pixel subspace, according to the set L of the sketch line of line target1, by i-th sketch line liWith j-th strip sketch
Line ljBetween region as line target.
In structure-pixel subspace, according to the set L of the sketch line of line target2, the s articles sketch line l will be coveredsRegion
As line target.
According to the following steps, based on the feature of gathering of sketch line, divide pinpoint target.
Step 1 would not indicate all sketch wire tags of line target in the structural region of administrative division map as candidate sketch line
Sketch line in set;
Step 2 randomly selects a sketch line from candidate sketch line set, with an endpoint of selected sketch line
Centered on, construct the geometry window that size is 5 × 5;
Step 3 judges the endpoint that whether there is other sketch lines in geometry window, and if it exists, execute step 4;Otherwise,
Execute step 6;
Step 4 judges whether to only exist an endpoint, if so, sketch line where the endpoint and current sketch line are carried out
Connection;Otherwise, step 5 is executed;
Step 5 connects the sketch line where selected sketch line and each endpoint, wherein angle is chosen from all connecting lines
The sketch line that maximum two sketch lines are completed as connection;
Step 6 judges the endpoint that whether there is other sketch lines in the geometry window of another endpoint of sketch line, if
In the presence of execution step 4;Otherwise, step 7 is executed;
Step 7 chooses the sketch line comprising two and two or more sketch line segments to the sketch line for completing attended operation,
The item number n comprising sketch line segment in selected sketch line is counted, wherein n >=2;
Step 8, judges whether the item number n of sketch line is equal to 2, if so, executing step 9;Otherwise, step 10 is executed;
Sketch line of the angle value on sketch line vertex in the range of [10 °, 140 °] is used as to have and gathers spy by step 9
The sketch line of sign;
Step 10 selects sketch line of the angle value on the corresponding n-1 vertex of sketch line all in [10 °, 140 °] range;
Step 11 is defined as follows two kinds of situations in selected sketch line:
Whether the first situation judges adjacent (i-1)-th, the two sketch line segments of i-th sketch line segment, i+1 item i-th
The same side of straight line, 2≤i≤n-1, if all sketch line segments and adjacent segments on sketch line are all same where sketch line segment
Side, then marking the sketch line is with the sketch line for gathering feature;
Whether second situation judges adjacent (i-1)-th, the two sketch line segments of i-th sketch line segment, i+1 item i-th
The same side of straight line, 2≤i≤n-1, if having n-1 sketch line segment and adjacent segments same on sketch line where sketch line segment
Side, and have a sketch line segment line segment adjacent thereto in non-the same side, also marking the sketch line is with the element for gathering feature
Retouch line;
Step 12, an optional sketch line in there is the sketch line for gathering feature, by two ends of selected sketch line
Point coordinate, determine the distance between two endpoints, if the end-point distances in [0,20] range, then using selected sketch line as table
Show the sketch line of pinpoint target;
Step 13, judge it is untreated have gather the sketch line of feature and whether all selected, if so, executing step 14;
Otherwise, step 12 is executed;
Step 14, with the method for super-pixel segmentation, to the sketch line for indicating pinpoint target in synthetic aperture radar SAR image
The pixel of surrounding carries out super-pixel segmentation, by super-pixel of the gray value of super-pixel after segmentation in [0,45] or [180,255]
As pinpoint target super-pixel;
Step 15 merges pinpoint target super-pixel, using the boundary of the pinpoint target super-pixel after merging as pinpoint target
Boundary, obtain the segmentation result of pinpoint target.
The result divided to line target and pinpoint target merges, and obtains the segmentation result of structure-pixel subspace.
Step 8, divide homogenous region pixel subspace.
According to the following steps, using the homogenous region dividing method based on multinomial logistic regression prior model, to homogeneous
Area pixel subspace is split, and obtains the segmentation result of homogenous region pixel subspace:
Step 1 arbitrarily chooses a pixel, centered on selected pixel from the pixel subspace of homogenous region
The square window for establishing 3 × 3 calculates the standard deviation sigma of the window1;
The side length of square window is increased by 2, obtains new square window, calculate the standard deviation of new square window by step 2
σ2;
Step 3, if standard deviation threshold method T3=3, if | σ1-σ2| < T3, then it is σ by standard deviation2Square window as most
Whole square window executes step 4;Otherwise, step 2 is executed;
Step 4 calculates the prior probability of center pixel in square window according to the following formula:
Wherein, p1' indicate square window in center pixel prior probability, exp () indicate exponential function operation, η ' table
Show that probabilistic model parameter, η ' value are 1, xk′' indicate to belong to the number of pixels of kth ' class in square window, k' ∈ [1 ...,
I' ..., K'], K' indicates the classification number of segmentation, and K' value is 5, xi′Belong to i-th ' in the square window that ' expression step 3 obtains
The number of pixels of class;
The probability density of pixel grey scale is multiplied with the probability density of texture, obtains likelihood probability p' by step 52, wherein
The probability density of gray scale is distributed to obtain by fading channel Nakagami, and the probability density of texture is distributed to obtain by t;
Step 6, by prior probability p1' and likelihood probability p2' be multiplied, obtain posterior probability p12';
Step 7 judges whether there are also untreated pixels in the pixel subspace of homogenous region, if so, executing step 1;
Otherwise, step 8 is executed;
Step 8 obtains the segmentation result of homogenous region pixel subspace according to maximum posteriori criterion.
Step 9, combination and segmentation result.
The segmentation of aggregated structure atural object pixel subspace, homogenous region pixel subspace and structure-pixel subspace will be mixed
As a result merge, obtain the final segmentation result of synthetic aperture radar SAR image.
Effect of the invention is further described below with reference to analogous diagram.
1. simulated conditions:
The hardware condition that the present invention emulates are as follows: Intellisense and image understanding laboratory graphics workstation;Present invention emulation
Used synthetic aperture radar SAR image are as follows: the Piperiver that Ku wave band resolution ratio is 1 meter schemes.
2. emulation content:
Emulation experiment of the invention is split to the Piperiver figure in SAR image, as shown in Fig. 2 (a)
Piperiver figure.The synthetic aperture radar SAR image that the figure is 1 meter from Ku wave band resolution ratio.
Using SAR image sketch step of the invention, to retouching of Piperiver pixel shown in Fig. 2 (a), obtain as
Sketch map shown in Fig. 2 (b).
Sketch map compartmentalization shown in Fig. 2 (b) is obtained such as Fig. 2 using division pixel subspace step of the invention
(c) administrative division map shown in.White space in Fig. 2 (c) indicates mixing aggregated structure semantic space, and others are homogeneous texture language
Adopted space and structure semantics space.Administrative division map shown in Fig. 2 (c) is mapped to original SAR image shown in Fig. 2 (a), is obtained such as Fig. 2 (d)
Shown in SAR image mix aggregated structure atural object pixel subspace figure.
Image Segmentation Methods Based on Features pinpoint target step is gathered based on sketch line using of the invention, extracts non-agglomerated area in Fig. 2 (b)
The sketch line in domain obtains Fig. 3 (a), extracts the sketch line of the expression line target in Fig. 3 (a), obtains result shown in Fig. 3 (b).
Black sketch line in Fig. 3 (b) is the sketch line for indicating line target.To the sketch line for not indicating line target in Fig. 3 (b), extract
With the sketch line for gathering feature, obtain shown in Fig. 3 (c) as a result, wherein black sketch line indicates pinpoint target.Pairing pore-forming
Diameter radar SAR image is sought indicating the super-pixel around doubtful pinpoint target sketch line, obtains result shown in Fig. 3 (d).It will
Super-pixel of the gray value of super-pixel in [0,45] or [180,255] is as pinpoint target super-pixel after segmentation, and merges
Pinpoint target super-pixel, using the boundary of the pinpoint target super-pixel after merging as the boundary of pinpoint target, obtained independent mesh
It marks shown in segmentation result such as Fig. 3 (e).
Aggregated structure atural object pixel subspace step is mixed using segmentation SAR image of the invention, to shown in Fig. 2 (d)
The mixing aggregated structure atural object pixel subspace figure of Piperiver figure is split, and obtains mixed land cover picture shown in Fig. 4 (a)
Sub-prime space segmentation result figure, grey area indicate untreated ground object space, and the region of remaining same color indicates same
A kind of atural object, the region of different colours indicate different atural object.
Using combination and segmentation result step of the invention, it is empty to merge mixing aggregated structure atural object pixel shown in Fig. 4 (a)
Between segmentation result and homogenous region pixel subspace segmentation result and structure-pixel subspace segmentation result, obtain Fig. 4 (b), Fig. 4
(b) be Fig. 2 (a) Piperiver image final segmentation result, Fig. 4 (c) be based on level vision semanteme and adaptive neighborhood it is more
Final segmentation result figure of the SAR image segmentation method of Xiang Shiyin model to Piperiver image.
3. simulated effect is analyzed:
Fig. 4 (b) is final segmentation result figure of the method for the present invention to Piperiver image, and Fig. 4 (c) is regarded based on level
Feel semantic and the hidden model of adaptive neighborhood multinomial SAR image segmentation method to the final segmentation result of Piperiver image
Figure, by the comparison of segmentation result figure, it could be assumed that, the method for the present invention is for mixing aggregated structure atural object pixel subspace
Boundary determination is more accurate, and the segmentation for homogenous region pixel subspace, region consistency is obviously preferable, classification number more adduction
Reason, and preferable dividing processing has been carried out to the pinpoint target in structure-pixel subspace.Use the method for the present invention pairing pore-forming
Diameter radar SAR image is split, and is effectively divided SAR image, and improves the accuracy of SAR image segmentation.
Claims (9)
1. a kind of SAR image segmentation method learnt based on curve wave filter and convolutional coding structure, is included the following steps:
(1) SAR image sketch:
(1a) obtains its sketch model according to the characteristic distributions of SAR image to the synthetic aperture radar SAR image of input;
(1b) utilizes SAR image sketch model, carries out sketch processing to the synthetic aperture radar SAR image of input, obtains defeated
The corresponding sketch map of synthetic aperture radar SAR image entered;
(2) pixel subspace is divided:
(2a) uses sketch line fields method, carries out compartmentalization processing to the sketch map of synthetic aperture radar SAR image, obtains
The administrative division map of synthetic aperture radar SAR image including aggregation zone, without sketch line region and structural region;
(2b) will include aggregation zone, the administrative division map without sketch line region and structural region, be mapped to the synthetic aperture thunder of input
Up in SAR image, mixing aggregated structure atural object pixel subspace in synthetic aperture radar SAR image, homogenous region picture are obtained
Sub-prime space and structure-pixel subspace;
(3) curve ripple filter set is constructed:
(3a) extracts the corresponding of mixing aggregated structure atural object pixel subspace from the administrative division map of synthetic aperture radar SAR image
Aggregation zone is divided into 10 ° in [0 °, 180 °] section with and is divided into 18 sections i.e. 18 directions, counts each area respectively
The line segment item number of sketch line segment in the interior aggregation zone;
Sketch line segment item numbers (3b) all to the aggregation zone is arranged according to the number of the line segment item number in each interval
Sequence obtains the collating sequence in direction, using the degree in 6 directions preceding in the collating sequence in direction as curve wave filter in
Directioin parameter;
The scale parameter of curve wave filter in [- 4,0] section, is divided into 0.2 carry out discretization with, obtained discrete by (3c)
Scale parameter after change;The displacement parameter of curve wave filter is divided into 0.5 carry out discretization in [0,9] section with, is obtained
Displacement parameter after to discretization;
(3d) according to the following formula, the curve wave function of calculated curve wave filter:
Wherein, t indicates the curve wave function of curve wave filter, and b indicates the scale parameter in curve wave filter, DbIndicating will
Scale parameter b indicates the directioin parameter in curve wave filter, R as the scale operator in the curve wave filter of parameter, θθ
Indicate that, using directioin parameter θ as the rotation operator in the curve wave filter of parameter, (m, n) indicates pixel in curve wave filter
The coordinate position of point, m0Indicate the displacement parameter of horizontal direction, n0Indicate the displacement parameter of vertical direction,Expression will be horizontal
The displacement parameter m in direction0With the displacement parameter n of vertical direction0As the displacement parameter in the curve wave filter of parameter;
(3e) according to the following formula, calculates each curve wave filter:
Wherein, c (t) indicates that the curve wave filter using curve wave function t as parameter, cos indicate cosine operation, and exp is indicated
Using natural constant e as the index operation at bottom;
Each the curve ripple filter bank being calculated is become curve ripple filter set by (3f);
(4) convolutional coding structure learning model is constructed:
(4a) is to the mutual disconnected area in each of mixing aggregated structure atural object pixel subspace in synthetic aperture radar SAR image
Domain, by 31 × 31 window carry out every a sliding window sample, obtain the corresponding multiple images block in each region, by multiple images block according to
It is secondary to be input in convolutional coding structure learning model, obtain the input layer of convolutional coding structure learning model;
(4b) uses curve wave filter, carries out convolution operation to the image block in the input layer of convolutional coding structure learning model, obtains
To the convolutional layer of convolutional coding structure learning model;
(4c) according to the following formula, calculates data fidelity term:
Wherein, E (c) indicates data fidelity term, and c indicates that the filter in convolutional coding structure learning model convolutional layer, N indicate convolution knot
Each of to be learnt the total number for the image block that mutually disconnected region includes in structure learning model, | | | |FExpression is done
The operation of frobenius norm,Indicate the square operation of frobenius norm, xiIt indicates to input in convolutional coding structure learning model
I-th of image block,It indicates to extract xiIntermediate size is the characteristic image block of n × n, MiIndicate convolutional coding structure learning model
The sum of the corresponding curve wave filter of i-th of image block of middle input, ∑ indicate sum operation, and * indicates convolution operation,Table
Show corresponding j-th of curve wave filter of i-th of the image block inputted in convolutional coding structure learning model;
(4d) according to the following formula, calculates structure fidelity term:
Wherein, G (c) indicates structure fidelity term, and R () indicates to ask the operation of all sketch line total lengths in sketch map, SM ()
It indicates to extract the operation with the one-to-one sketch segment of input picture block;
(4e) according to the following formula, calculating target function:
Wherein, L (c) indicates objective function,It indicates in objective function L (c) value minimum, seeks curve wave filter c
Operation;
(4f) exports the curve wave filter obtained by objective function guidance learning, obtains the output layer of convolutional coding structure model;
(5) training convolutional Structure learning model:
(5a) sets 0.1 for structural failure threshold value;
Step (4a) sampling is obtained image block and is sequentially inputted in convolutional coding structure learning model by (5b);
(5c) randomly selects six filters, directioin parameter is by 6 of statistics in step (3b) from curve ripple filter set
A direction obtains, displacement parameter and scale parameter random initializtion, the filter collection that these initial six filters are formed
Cooperation is selected curve ripple filter set;
(5d) rolls up image block currently entered and each curve filter in selected curve ripple filter set
Product operation, obtains characteristic pattern corresponding with each curve filter;
(5e) utilizes the structure fidelity term formula in step (4d), the structure fidelity term of calculating input image block;
(5f) judges whether the structure fidelity term of current input image block is less than structural failure threshold value, if so, thening follow the steps
(5i) is otherwise executed step (5g);
(5g) utilizes scale parameter more new formula and displacement parameter more new formula, updates step (4c) data fidelity term formula respectively
The scale parameter and displacement parameter of middle curve wave filter obtain updated curve wave filter;
(5h) using updated curve ripple filter set as selected curve ripple filter set, return step (5d),
Training is re-started to input picture block;
(5i) saves the curve wave filter that study obtains in the curve ripple filter set succeeded in school to the input picture block,
The study to the feature of the input picture block is completed, and exports the curve ripple filter set that the input picture block succeeds in school;
(5j) judges whether all image blocks pass through the study that convolutional coding structure learning model completes feature, if so, terminate program,
Otherwise, it inputs next image block and executes step (5c);
(6) segmentation SAR image mixes aggregated structure atural object pixel subspace:
(6a) splices the characteristic set of all mutual connected regions, using spliced characteristic set as code book;
(6b) calculates separately the inner product with each feature in code book to all features of each mutual not connected region, obtains every
Projection vector of a all features in region on code book;
(6c) it is one corresponding to obtain each region to each mutually all projection vectors progress maximum value convergence of connected region
Structural eigenvector;
(6d) propagates AP clustering algorithm using neighbour, does not cluster, obtains to the structural eigenvector of all mutual connected regions
Mix the segmentation result of aggregated structure atural object pixel subspace;
(7) segmenting structure pixel subspace:
(7a) uses vision semantic rules, divides line target;
The feature of gathering of (7b) based on sketch line divides pinpoint target;
The result that (7c) divides line target and pinpoint target merges, and obtains the segmentation result of structure-pixel subspace;
(8) divide homogenous region pixel subspace:
Using based on multinomial logistic regression prior model without sketch line region segmentation method, to homogenous region pixel subspace
It is split, obtains the segmentation result of homogenous region pixel subspace;
(9) combination and segmentation result:
The segmentation result of aggregated structure atural object pixel subspace, homogenous region pixel subspace and structure-pixel subspace will be mixed
It merges, obtains the final segmentation result of synthetic aperture radar SAR image.
2. the SAR image segmentation method according to claim 1 learnt based on curve wave filter and convolutional coding structure,
It is characterized in that, SAR image sketch model is utilized described in step (1b), element is carried out to the synthetic aperture radar SAR image of input
Specific step is as follows for retouching processing:
Step 1 arbitrarily chooses a number, the sum as template in [100,150] range;
Step 2 constructs a template on the side being made of pixel with different directions and scale, line, utilizes the side of template
To with dimensional information structural anisotropy's Gaussian function, by the Gaussian function, the weighting system of each pixel in calculation template
It counts, the weighting coefficient of all pixels point in statistical mask, wherein scale number value is 3~5, and direction number value is 18;
Step 3, according to the following formula, pixel is equal in calculating synthetic aperture radar SAR image corresponding with template area coordinate
Value:
Wherein, μ indicates the mean value of all pixels point in synthetic aperture radar SAR image corresponding with template area coordinate, ∑
Indicate sum operation, g indicates the corresponding coordinate of any one pixel in the Ω region of template, and ∈ expression belongs to symbol, wg
Indicate weight coefficient of the pixel at coordinate g in the Ω region of template, wgValue range be wg∈ [0,1], AgIndicate with
The value of pixel of the pixel at the coordinate g in corresponding synthetic aperture radar SAR image in the Ω region of template;
Step 4 calculates the side of pixel in synthetic aperture radar SAR image corresponding with template area coordinate according to the following formula
Difference:
Wherein, ν indicates the variance yields of all pixels point in synthetic aperture radar SAR image corresponding with template area coordinate;
Step 5 calculates the response that each pixel in synthetic aperture radar SAR image is directed to ratio operator according to the following formula:
Wherein, R indicates response of each pixel for ratio operator, min { } expression in synthetic aperture radar SAR image
Minimum Value Operations, a and b respectively indicate two different regions in template, μaAll pixels point is equal in expression template area a
Value, μbIndicate the mean value of all pixels point in the b of template area;
Step 6 calculates the response that each pixel in synthetic aperture radar SAR image is directed to correlation operator according to the following formula:
Wherein, C indicate synthetic aperture radar SAR image in each pixel be directed to correlation operator response,
Indicate square root functions, a and b respectively indicate two different zones, ν in templateaIndicate all pixels in a of template area
The variance yields of point, νbIndicate the variance yields of all pixels point in the b of template area, μaAll pixels point is equal in expression template area a
Value, μbIndicate the mean value of all pixels point in the b of template area;
Step 7 calculates the response that each pixel in synthetic aperture radar SAR image is directed to each template according to the following formula:
Wherein, F indicate synthetic aperture radar SAR image in each pixel be directed to each template response,Expression square
Root operation, R and C respectively indicate pixel in synthetic aperture radar SAR image to scheme for ratio operator and synthetic aperture radar SAR
Pixel is directed to the response of correlation operator as in;
Step 8, judges whether constructed template is equal to the sum of selected template, if so, executing step 9, otherwise, executes
Step 2;
Step 9, selection has the template of maximum response from each template, as the template of synthetic aperture radar SAR image,
And using the maximum response of the template as the intensity of pixel in synthetic aperture radar SAR image, the direction of the template is made
For the direction of pixel in synthetic aperture radar SAR image, the sideline response diagram and gradient of synthetic aperture radar SAR image are obtained
Figure;
Step 10 calculates the intensity value of synthetic aperture radar SAR image intensity map, obtains intensity map according to the following formula:
Wherein, I indicates that the intensity value of synthetic aperture radar SAR image intensity map, r indicate synthetic aperture radar SAR image sideline
Value in response diagram, t indicate the value in synthetic aperture radar SAR image gradient map;
Step 11 detects intensity map using non-maxima suppression method, obtains suggestion sketch;
Step 12 chooses the pixel suggested in sketch with maximum intensity, will suggest the pixel in sketch with the maximum intensity
The pixel of point connection connects to form suggestion line segment, obtains suggestion sketch map;
Step 13 calculates the code length gain for suggesting sketch line in sketch map according to the following formula:
Wherein, CLG indicates to suggest the code length gain of sketch line in sketch map, and ∑ indicates sum operation, and J indicates current sketch
The number of pixel, A in line neighborhoodjIndicate the observation of j-th of pixel in current sketch line neighborhood, Aj,0It indicates current
In the case that sketch line cannot indicate structural information, the estimated value of j-th of pixel in the sketch line neighborhood, ln () indicate with
E is the log operations at bottom, Aj,1Indicate the jth in the sketch line neighborhood in the case where current sketch line can indicate structural information
The estimated value of a pixel;
Step 14 arbitrarily chooses a number, as threshold value T in [5,50] range;
Step 15 selects the suggestion sketch line of CLG > T in all suggestion sketch lines, is combined into synthetic aperture radar SAR figure
The sketch map of picture.
3. the SAR image segmentation method according to claim 1 learnt based on curve wave filter and convolutional coding structure,
It is characterized in that, specific step is as follows for sketch line fields method described in step (2a):
Sketch line is divided into table according to the concentration class of sketch line segment in the sketch map of synthetic aperture radar SAR image by step 1
Show the aggregation sketch line of aggregation atural object and indicates boundary, line target, the boundary sketch line of isolated target, line target sketch line, orphan
Vertical target sketch line;
Step 2 chooses the sketch line segment work that concentration class is equal to optimal concentration class according to the statistics with histogram of sketch line segment concentration class
For seed line-segment sets { Ek, k=1,2 ..., m }, wherein EkIndicate that any bar sketch line segment in seed line-segment sets, k indicate kind
Sub-line section concentrates the label of any bar sketch line segment, and m indicates the total number of seed line segment, and { } indicates set operation;
Step 3, as basic point with the unselected line segment for being added to seed line-segment sets sum, with this basic point recursive resolve line-segment sets
It closes;
Step 4 constructs the round primitive that a radius is the optimal concentration class section upper bound, with the circle primitive in line segment aggregate
Line segment expanded, the line segment aggregate ecto-entad after expansion is corroded, is obtained in sketch map with sketch point being single
The aggregation zone of position;
Step 5, to the sketch line for indicating boundary, line target and isolated target, during each sketch point with each sketch line is
The heart constructs the geometry window that size is 5 × 5, obtains structural region;
Step 6 will remove the part other than aggregation zone and structural region as can not sketch region in sketch map;
Step 7, by sketch map aggregation zone, structural region and can not sketch region merging technique, obtain including aggregation zone, nothing
The administrative division map of the synthetic aperture radar SAR image of sketch line region and structural region.
4. the SAR image segmentation method according to claim 1 learnt based on curve wave filter and convolutional coding structure, special
Sign is that the more new formula of scale parameter described in step (5g) is as follows:
Wherein, btIndicate the scale parameter of the curve wave filter of the t times iteration, bt-1Indicate the scale parameter of the t-1 times iteration,
η indicates that step-length, value range are [0,1], and c () indicates that curve wave filter, sin indicate cosine operation, and X indicates curve ripple filter
The horizontal direction parameter of wave device, value is by formula X=DbRθ(m-m0) obtain, Y indicates the vertical direction ginseng of curve wave filter
Number, value is by formula Y=DbRθ(n-n0) obtain,Curve wave filter is indicated to the local derviation of independent variable X, value is by following
Formula obtains:
Curve wave filter horizontal direction parameter X is indicated to the local derviation of scale parameter, value is obtained by the following formula:
Curve wave filter is indicated to the local derviation of independent variable Y, value is obtained by the following formula:
Curve wave filter vertical direction Y is indicated to the local derviation of scale parameter, value is obtained by the following formula:
5. the SAR image segmentation method according to claim 1 learnt based on curve wave filter and convolutional coding structure, special
Sign is that the more new formula of displacement parameter described in step (5g) is as follows:
Wherein,Indicate the horizontal direction displacement parameter of the curve wave filter of the t times iteration,Indicate the t-1 times iteration
Horizontal direction displacement parameter,
Wherein,Indicate the vertical direction displacement parameter of the curve wave filter of the t times iteration,Indicate the t-1 times iteration
Vertical direction displacement parameter.
6. the SAR image segmentation method according to claim 1 learnt based on curve wave filter and convolutional coding structure, special
Sign is that vision semantic rules described in step (7a) are as follows:
If i-th sketch line liWith j-th strip sketch line ljThe distance between be Dij, liDirection be Oi, ljDirection be Oj, i, j
∈ [1,2 ..., S], S are the total number of sketch line;
Width is greater than the line target of 3 pixels with two sketch line liAnd ljIt indicates, liAnd ljThe distance between DijLess than T1And
Direction difference (Oi-Oj) less than 10 degree, wherein T1=5;
If the s articles sketch line lsGeometry window wsThe average gray of interior each column is AiIf the gray scale difference of adjacent column is ADi=
|Ai-Ai+1|, if zs=[zs1,zs2,...,zs9] be adjacent column gray scale difference ADiLabel vector;
By width less than the line target of 3 pixels with single sketch line lsIt indicates, lsGeometry window wsIt is interior, calculate adjacent column
Gray scale difference ADiIf ADi> T2, then zsi=1;Otherwise zsi=0, zsIn there are two element value be 1, remaining is 0, wherein T2
=34;
If L1,L2It is the set for indicating the sketch line of line target, if Dij< T1And | Oi-Oj| < 10, then li,lj∈L1;If
sum(zs)=2, then ls∈L2, wherein sum () indicate to vector important summation operation.
7. the SAR image segmentation method according to claim 1 learnt based on curve wave filter and convolutional coding structure, special
Sign is that specific step is as follows for segmentation line target described in step (7a):
Step 1, in structure-pixel subspace, according to the set L of the sketch line of line target1, by i-th sketch line liWith j-th strip element
Retouch line ljBetween region as line target;
Step 2, in structure-pixel subspace, according to the set L of the sketch line of line target2, the s articles sketch line l will be coveredsArea
Domain is as line target.
8. the SAR image segmentation method according to claim 1 learnt based on curve wave filter and convolutional coding structure, special
Sign is that specific step is as follows for segmentation pinpoint target described in step (7b):
Step 1 would not indicate all sketch wire tags of line target in the structural region of administrative division map as candidate sketch line set
In sketch line;
Step 2 randomly selects a sketch line from candidate sketch line set, during an endpoint with selected sketch line is
The heart, the geometry window that construction size is 5 × 5;
Step 3 judges the endpoint that whether there is other sketch lines in geometry window, and if it exists, execute step 4;Otherwise, it executes
Step 6;
Step 4 judges whether to only exist an endpoint, if so, sketch line where the endpoint and current sketch line are attached;
Otherwise, step 5 is executed;
Step 5 connects the sketch line where selected sketch line and each endpoint, and it is maximum that wherein angle is chosen from all connecting lines
Two sketch lines as connection complete sketch line;
Step 6 judges the endpoint that whether there is other sketch lines in the geometry window of another endpoint of sketch line, if depositing
Executing step 4;Otherwise, step 7 is executed;
Step 7 chooses the sketch line comprising two and two or more sketch line segments, statistics to the sketch line for completing attended operation
It include the item number n of sketch line segment, wherein n >=2 in selected sketch line;
Step 8, judges whether the item number n of sketch line is equal to 2, if so, executing step 9;Otherwise, step 10 is executed;
Step 9, by sketch line of the angle value on sketch line vertex in the range of [10 °, 140 °] as have gather feature
Sketch line;
Step 10 selects sketch line of the angle value on the corresponding n-1 vertex of sketch line all in [10 °, 140 °] range;
Step 11 is defined as follows two kinds of situations in selected sketch line:
Whether the first situation judges adjacent (i-1)-th, the two sketch line segments of i-th sketch line segment, i+1 item in i-th element
The same side of straight line, 2≤i≤n-1, if all sketch line segments and adjacent segments on sketch line are all same where retouching line segment
Side, then marking the sketch line is with the sketch line for gathering feature;
Whether second situation judges adjacent (i-1)-th, the two sketch line segments of i-th sketch line segment, i+1 item in i-th element
The same side of straight line where line segment, 2≤i≤n-1 are retouched, if having n-1 sketch line segment and adjacent segments on sketch line in the same side,
And have a sketch line segment line segment adjacent thereto in non-the same side, also marking the sketch line is with the sketch line for gathering feature;
Step 12, an optional sketch line in having the sketch line for gathering feature are sat by two endpoints of selected sketch line
Mark, determines the distance between two endpoints, if the end-point distances are in [0,20] range, then only using selected sketch line as expression
The sketch line of vertical target;
Step 13, judge it is untreated have gather the sketch line of feature and whether all selected, if so, executing step 14;Otherwise,
Execute step 12;
Step 14, with the method for super-pixel segmentation, around the sketch line for indicating pinpoint target in synthetic aperture radar SAR image
Pixel carry out super-pixel segmentation, by super-pixel conduct of the gray value of super-pixel after segmentation in [0,45] or [180,255]
Pinpoint target super-pixel;
Step 15 merges pinpoint target super-pixel, using the boundary of the pinpoint target super-pixel after merging as the side of pinpoint target
Boundary obtains the segmentation result of pinpoint target.
9. the SAR image segmentation method according to claim 1 learnt based on curve wave filter and convolutional coding structure, special
Sign is, based on the specific step without sketch line region segmentation method of multinomial logistic regression prior model described in step (8)
It is rapid as follows:
Step 1 is arbitrarily chosen a pixel from the pixel subspace of homogenous region, is established centered on selected pixel
3 × 3 square window calculates the standard deviation sigma of the window1;
The side length of square window is increased by 2, obtains new square window, calculate the standard deviation sigma of new square window by step 22;
Step 3, if standard deviation threshold method T3=3, if | σ1-σ2| < T3, then it is σ by standard deviation2Square window as final
Square window executes step 4;Otherwise, step 2 is executed;
Step 4 calculates the prior probability of center pixel in square window according to the following formula:
Wherein, p '1Indicate that the prior probability of center pixel in square window, η ' indicate that probabilistic model parameter, η ' value are 1, xk′′
Indicating the number of pixels for belonging to kth ' class in square window, k' ∈ [1 ..., i' ..., K'], K' indicate the classification number of segmentation,
K' value is 5, xi′Belong to the number of pixels of the i-th ' class in the square window that ' expression step 3 obtains;
The probability density of pixel grey scale is multiplied with the probability density of texture, obtains likelihood probability p' by step 52, wherein gray scale
Probability density is distributed to obtain by fading channel Nakagami, and the probability density of texture is distributed to obtain by t;
Step 6, by prior probability p1' and likelihood probability p2' be multiplied, obtain posterior probability p12';
Step 7 judges whether there are also untreated pixels in the pixel subspace of homogenous region, if so, executing step 1;Otherwise,
Execute step 8;
Step 8 obtains the segmentation result of homogenous region pixel subspace according to maximum posteriori criterion.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611260213.3A CN106846322B (en) | 2016-12-30 | 2016-12-30 | The SAR image segmentation method learnt based on curve wave filter and convolutional coding structure |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611260213.3A CN106846322B (en) | 2016-12-30 | 2016-12-30 | The SAR image segmentation method learnt based on curve wave filter and convolutional coding structure |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106846322A CN106846322A (en) | 2017-06-13 |
CN106846322B true CN106846322B (en) | 2019-06-21 |
Family
ID=59115010
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611260213.3A Active CN106846322B (en) | 2016-12-30 | 2016-12-30 | The SAR image segmentation method learnt based on curve wave filter and convolutional coding structure |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106846322B (en) |
Families Citing this family (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107341813B (en) * | 2017-06-15 | 2019-10-25 | 西安电子科技大学 | SAR image segmentation method based on Structure learning and sketch characteristic inference network |
CN107403434B (en) * | 2017-07-28 | 2019-08-06 | 西安电子科技大学 | SAR image semantic segmentation method based on two-phase analyzing method |
CN109409375B (en) * | 2018-10-11 | 2021-12-14 | 西安电子科技大学 | SAR image semantic segmentation method based on contour structure learning model |
CN109523077B (en) * | 2018-11-15 | 2022-10-11 | 云南电网有限责任公司 | Wind power prediction method |
CN110458203B (en) * | 2019-07-19 | 2021-11-30 | 北京科技大学 | Advertisement image material detection method |
CN111080666B (en) * | 2019-12-27 | 2022-07-15 | 浙江大学 | Object segmentation method and device based on cyclic convolution |
CN113077486B (en) * | 2021-04-30 | 2021-10-08 | 深圳世源工程技术有限公司 | Method and system for monitoring vegetation coverage rate in mountainous area |
CN115720254B (en) * | 2023-01-09 | 2023-04-18 | 北京无线电测量研究所 | Stable video stream generation method and system for video synthetic aperture radar |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105374033B (en) * | 2015-10-19 | 2017-11-21 | 西安电子科技大学 | SAR image segmentation method based on ridge ripple deconvolution network and sparse classification |
CN105389798B (en) * | 2015-10-19 | 2018-08-24 | 西安电子科技大学 | SAR image segmentation method based on deconvolution network Yu mapping inference network |
CN105427313B (en) * | 2015-11-23 | 2018-03-06 | 西安电子科技大学 | SAR image segmentation method based on deconvolution network and adaptive inference network |
CN105608692B (en) * | 2015-12-17 | 2018-05-04 | 西安电子科技大学 | Polarization SAR image segmentation method based on deconvolution network and sparse classification |
-
2016
- 2016-12-30 CN CN201611260213.3A patent/CN106846322B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN106846322A (en) | 2017-06-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106846322B (en) | The SAR image segmentation method learnt based on curve wave filter and convolutional coding structure | |
CN106683102B (en) | SAR image segmentation method based on ridge ripple filter and convolutional coding structure learning model | |
CN106611420B (en) | The SAR image segmentation method constrained based on deconvolution network and sketch map direction | |
CN106611423B (en) | SAR image segmentation method based on ridge ripple filter and deconvolution structural model | |
CN106778821B (en) | Classification of Polarimetric SAR Image method based on SLIC and improved CNN | |
Zhang et al. | Boundary-constrained multi-scale segmentation method for remote sensing images | |
CN105608692B (en) | Polarization SAR image segmentation method based on deconvolution network and sparse classification | |
CN106529508B (en) | Based on local and non local multiple features semanteme hyperspectral image classification method | |
CN106909902B (en) | Remote sensing target detection method based on improved hierarchical significant model | |
CN103049763B (en) | Context-constraint-based target identification method | |
CN106611422B (en) | Stochastic gradient Bayes's SAR image segmentation method based on sketch structure | |
CN106651884B (en) | Mean field variation Bayes's SAR image segmentation method based on sketch structure | |
CN106611421B (en) | The SAR image segmentation method constrained based on feature learning and sketch line segment | |
CN107341813B (en) | SAR image segmentation method based on Structure learning and sketch characteristic inference network | |
CN107909015A (en) | Hyperspectral image classification method based on convolutional neural networks and empty spectrum information fusion | |
CN107403434B (en) | SAR image semantic segmentation method based on two-phase analyzing method | |
CN105335975B (en) | Polarization SAR image segmentation method based on low-rank decomposition and statistics with histogram | |
Buddhiraju et al. | Comparison of CBF, ANN and SVM classifiers for object based classification of high resolution satellite images | |
CN106203444B (en) | Classification of Polarimetric SAR Image method based on band wave and convolutional neural networks | |
CN113223042B (en) | Intelligent acquisition method and equipment for remote sensing image deep learning sample | |
Xiao et al. | Segmentation of multispectral high-resolution satellite imagery using log Gabor filters | |
CN109584284B (en) | Hierarchical decision-making coastal wetland ground object sample extraction method | |
CN108229551A (en) | A kind of Classification of hyperspectral remote sensing image method based on compact dictionary rarefaction representation | |
CN109034213B (en) | Hyperspectral image classification method and system based on correlation entropy principle | |
CN107292268A (en) | The SAR image semantic segmentation method of quick ridge ripple deconvolution Structure learning model |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |