CN108986108A - A kind of SAR image sample block selection method based on sketch line segment aggregation properties - Google Patents
A kind of SAR image sample block selection method based on sketch line segment aggregation properties Download PDFInfo
- Publication number
- CN108986108A CN108986108A CN201810670378.0A CN201810670378A CN108986108A CN 108986108 A CN108986108 A CN 108986108A CN 201810670378 A CN201810670378 A CN 201810670378A CN 108986108 A CN108986108 A CN 108986108A
- Authority
- CN
- China
- Prior art keywords
- sketch
- sar image
- synthetic aperture
- aperture radar
- pixel
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- 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
- G06T2207/10044—Radar 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
- G06T2207/20016—Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention discloses a kind of SAR image sample block selection method based on sketch line segment aggregation properties, mainly solve the problems, such as that prior art SAR image mixing aggregated structure atural object pixel subspace segmentation 1 sliding window sample set redundant samples of interval are excessive.Its implementation is: according to the sketch model extraction sketch map of SAR image;Sketch map compartmentalization is obtained into administrative division map, the extremely not homogeneous region of SAR image is obtained according to administrative division map;All sketch line segments in each aggregation zone with bilateral aggregation properties are extracted, the rectangle set of blocks of the sketch line segment composition of these sketch line segments and k nearest neighbor is solved;It is sampled and is expanded in corresponding extremely not homogeneous region according to the coordinate of rectangular block in rectangle set of blocks, obtain sample set of blocks.The present invention can construct the sample set in the extremely not homogeneous region of SAR image, and the sample set constructed can more comprehensively represent the structure feature in extremely not homogeneous region.
Description
Technical field
The invention belongs to technical field of image processing, in particular to a kind of SAR image based on sketch line segment aggregation properties
Sample block selection method is applied in synthetic aperture radar SAR image mixed pixel subspace mutually disconnected extremely not homogeneous area
The determination of domain training sample set is further used for the feature learning in the extremely not homogeneous region in SAR image.
Background technique
Synthetic aperture radar SAR is the impressive progress in remote sensing technology field, for obtaining the full resolution pricture of earth surface.
Compared with other kinds of imaging technique, SAR is particularly suitable for the terrestrial reference imaging of large area, can penetrate cloud and mist, sleet, have
All weather operations ability.With the fast development of SAR image technology, more and more SAR image data are obtained, pass through calculating
Machine automatic interpretation SAR image becomes a urgent problem needed to be solved.And SAR image segmentation is the premise of SAR image interpretation,
Segmentation result influences further detection, identification very big.
The extremely not homogeneous region ground object structure of SAR image is complicated, can be learnt by machine learning model training sample set
To feature characterize, then these regions are clustered using the feature in each extremely not homogeneous region, to reach segmentation
Purpose.The scale Yu quality for being input to sample set in machine learning model influence whether the quality of feature learning, and extract
Feature quality key effect is played for SAR image segmentation result.Therefore for each extremely not homogeneous region, building
The sample set of one high quality is used for feature learning, it appears particularly important.
Patent " SAR image based on deconvolution network Yu mapping inference network of the Xian Electronics Science and Technology University in its application
A kind of deconvolution net is disclosed in dividing method " (number of patent application CN201510679181.X, publication number CN105389798 A)
The SAR image segmentation method of network and mapping inference network.This method is drawn according to the sketch map of synthetic aperture radar SAR image
Point administrative division map, by administrative division map be mapped to original image assembled, homogeneous and structural region.Respectively to each in aggregation and homogenous region
A mutually disconnected region carries out unsupervised training, obtains the filter set for characterizing each mutual not connected region structure feature.
Reasoning is compared between the structure feature the mutual not connected region in two class regions respectively, obtains point of aggregation and homogenous region
Cut result.Structural region is split under the guidance of sketch line segment using super-pixel merging.It is complete to merge each region segmentation result
Divide at SAR image.Shortcoming existing for this method is, to mutually connected region does not carry out structure feature in aggregation zone
When habit, the sample block for being sent into deconvolution learning network is not selected, it is big that this allows for sample size, leads to network training
Time is long, and space complexity is high, so that the cost of e-learning feature is larger.
Xian Electronics Science and Technology University " is based on G in the patent of its application0Stochastic gradient variation Bayes's SAR image of distribution
It is disclosed in dividing method " (number of patent application 201710702367.1) a kind of based on G0The stochastic gradient variation Bayes of distribution
SAR image segmentation method.The administrative division map that this method is divided according to the sketch map of synthetic aperture radar SAR image, by region
Figure be mapped to original image assembled, homogeneous and structural region.Then it uses based on stochastic gradient variation Bayesian network model
Method carries out pinpoint target to the segmentation of mixing aggregated structure atural object pixel subspace, using the method for gathering feature based on sketch line
Segmentation carries out line target segmentation using view-based access control model semantic rules, even using being carried out based on multinomial logistic regression prior model
The segmentation of matter area pixel subspace, last combination and segmentation is as a result, and obtain the segmentation result of SAR image.Existing for this method not
Foot place is that mutually disconnected extremely not homogeneous region carries out feature learning in mixing aggregated structure atural object pixel subspace
When, this method is to carry out sample taken at random every the collected sample block of 1 sliding window to each extremely not homogeneous region, this to pass through machine
The feature that learning model is acquired has randomness, it is possible to and it needs through multiple sample taken at random, and passes through multiple feature learning,
It is likely to practise in certain middle school to the extremely not homogeneous good character representation in region, with affecting SAR image mixing aggregated structure
The accuracy of image sub-prime space segmentation, and increase trained cost.
Summary of the invention
It is an object of the invention to propose a kind of based on sketch line segment aggregation properties for the deficiency in above-mentioned prior art
SAR image sample block selection method, it is richer and more representational to select structure from each extremely not homogeneous region
Sample block.
To achieve the above object, including technical solution it is as follows: a kind of SAR image sample based on sketch line segment aggregation properties
This block selection method, comprising the following steps:
1. a kind of SAR image sample block selection method based on sketch line segment aggregation properties, which is characterized in that including following
Step:
Step 1): input synthetic aperture radar SAR image establishes the sketch model of synthetic aperture radar SAR image, from element
Retouch the sketch map B1 that synthetic aperture radar SAR image is extracted in model;
Step 2): compartmentalization processing is carried out to the sketch map B1 for the SAR image that step 1) obtains, obtains synthetic aperture radar
The administrative division map of SAR image, administrative division map are formed by aggregation zone, structural region and without sketch line segment region;
Step 3): aggregation zone is that spatially have unilateral aggregation and bilateral aggregation properties by those in sketch map B1
Sketch line segment composition, there are corresponding extremely not homogeneous regions in SAR image pixel space for each aggregation zone;
Step 4): all sketch line segments in each aggregation zone with bilateral aggregation properties are extracted, these sketches are solved
The rectangle set of blocks D1 that the sketch line segment of line segment and k nearest neighbor is constituted;
Step 5): for the rectangle set of blocks D1 of aggregation zone, according to the coordinate of each rectangular block in rectangle set of blocks D1,
The collecting sample block in extremely not homogeneous region corresponding with aggregation zone, collected sample block is put together, is constituted each
The sample set of blocks D2 in extremely not homogeneous region.
The step 4) specifically includes the following steps:
Rectangular coordinate system 4a) is established using the column of SAR image sketch map as y-axis with the behavior x-axis of SAR image sketch map,
SAR image sketch map is placed in coordinate system, define the first sketch point in the SAR image sketch map upper left corner coordinate be (1,
1);
4b) find the coordinate of current bilateral aggregation two endpoints of sketch line segment, respectively starting point coordinate (start_x,
Start_y) with terminal point coordinate (end_x, end_y);
4c) find the starting point coordinate and terminal that other sketch line segments of k nearest neighbor are constituted with current bilateral aggregation sketch line segment
Coordinate;
The value of all x coordinates in starting point coordinate in step 4b) and step 4c) and terminal point coordinate 4d) is stored in vector X
In, the value of all y-coordinates is stored in vector Y, and acquire in vector X the smallest value x_min and maximum value x_max and
The smallest value y_min and maximum value y_max in vector Y;
It is 4e) coordinate in the rectangular block upper left corner with (x_min, y_min), and with max { (x_max-x_min), (y_max-
Y_min it) } as the width of rectangular block and height, is sampled on the aggregation zone of SAR image, collected rectangular block is placed on
Together, rectangle set of blocks D1 is obtained.
K value is 5 in the step 4).
The step 5) specifically includes the following steps:
The size for being image block with 32 × 32, to all rectangular blocks in rectangle set of blocks D1, respectively with rectangular block upper left
Angular coordinate is the coordinate in the upper left corner for the image block that will be acquired, using rectangular block bottom right angular coordinate as the seat in the image block lower right corner
Mark, using the coordinate in the rectangular block lower left corner as the coordinate in the image block lower left corner, the upper right angular coordinate of rectangular block is as image block
Upper right angular coordinate is sampled on the extremely not homogeneous region of SAR image using the center of rectangular block as the center of image block,
Acquired image block is put together, the sample set of blocks D2 in each extremely not homogeneous region is obtained.
The step 1) embodies the following steps are included: 1a) input synthetic aperture radar SAR image, establish synthetic aperture
The sketch model of radar SAR image:
1a1) in [100,150] range, a number, the sum as template are arbitrarily chosen;
A template for 1a2) constructing 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, the weighting of each pixel in calculation template
Coefficient, the weighting coefficient of all pixels point in statistical mask, wherein scale number value is 3~5, and direction number value is 18;
1a3) according to the following formula, pixel in synthetic aperture radar SAR image corresponding with template area coordinate is calculated
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;
1a4) according to the following formula, pixel in synthetic aperture radar SAR image corresponding with template area coordinate is calculated
Variance yields:
Wherein, ν indicates the variance of all pixels point in synthetic aperture radar SAR image corresponding with template area coordinate
Value;
1a5) according to the following formula, the response that each pixel in synthetic aperture radar SAR image is directed to ratio operator is calculated:
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;
1a6) according to the following formula, the response that each pixel in synthetic aperture radar SAR image is directed to correlation operator is calculated:
Wherein, C indicates that for the response of correlation operator, a and b divide each pixel in synthetic aperture radar SAR image
It Biao Shi not two different zones, ν in templateaIndicate the variance yields of all pixels point in a of template area, νbIt indicates in the b of template area
The variance yields of all pixels point, μaIndicate the mean value of all pixels point in a of template area, μbIndicate all pixels in the b of template area
The mean value of point;
1a7) according to the following formula, the response that each pixel in synthetic aperture radar SAR image is directed to each template is calculated:
Wherein, F indicates that for the response of each template, R and C divide each pixel in synthetic aperture radar SAR image
Not Biao Shi in synthetic aperture radar SAR image pixel for pixel needle in ratio operator and synthetic aperture radar SAR image
To the response of correlation operator;
1a8) judge whether constructed template is equal to the sum of selected template, if it is not, then returning to 1a2), otherwise,
Execute 1a9);
1a9) template of the selection with maximum response, the mould as synthetic aperture radar SAR image from each template
Plate, and using the maximum response of the template as the intensity of pixel in synthetic aperture radar SAR image, by the direction of the template
As the direction of pixel in synthetic aperture radar SAR image, the sideline response diagram and ladder of synthetic aperture radar SAR image are obtained
Degree figure;
1a10) according to the following formula, the intensity value for calculating synthetic aperture radar SAR image intensity map, obtains intensity map:
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;
Non-maxima suppression method 1a11) is used, intensity map is detected, suggestion sketch is obtained;
The pixel suggested in sketch with maximum intensity 1a12) is chosen, will suggest the picture in sketch with the maximum intensity
The pixel of vegetarian refreshments connection connects to form suggestion line segment, obtains suggestion sketch map;
1a13) according to the following formula, the code length gain for suggesting sketch line in sketch map is calculated:
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;
1a14) in [5,50] range, a number is arbitrarily chosen, as threshold value T;
The suggestion sketch line for 1a15) selecting CLG > T in all suggestion sketch lines, is combined into synthetic aperture radar SAR
The sketch map of image;
The sketch map of synthetic aperture radar SAR image 1b) is extracted from sketch model: scheming synthetic aperture radar SAR
As being input in sketch model, the sketch map B1 of synthetic aperture radar SAR image is obtained.
The step 2) specifically includes the following steps:
Sketch line fields method 2a) is used, compartmentalization processing is carried out to the sketch map of synthetic aperture radar SAR image,
Obtain include aggregation zone administrative division map, its step are as follows:
2a1) according to the concentration class of sketch line segment in the sketch map of synthetic aperture radar SAR image, sketch line is divided into
Indicate aggregation atural object aggregation sketch line and expression boundary, line target, the boundary sketch line of isolated target, line target sketch line,
Isolated target sketch line;
2a2) according to the statistics with histogram of sketch line segment concentration class, the sketch line segment that concentration class is equal to optimal concentration class is chosen
As seed line-segment sets { Ek, k=1,2 ..., m }, wherein EkIndicate that any bar sketch line segment in seed line-segment sets, k indicate
The label of any bar sketch line segment in seed line-segment sets, m indicate the total number of seed line segment, and { } indicates set operation;
2a3) 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;
The round primitive that a radius is the optimal concentration class section upper bound 2a4) is constructed, with the circle primitive to line segment aggregate
In line segment expanded, the line segment aggregate ecto-entad after expansion is corroded, obtained in sketch map be with sketch point
The aggregation zone of unit;
2a5) to the sketch line for indicating boundary, line target and isolated target, it is with each sketch point of each sketch line
The geometry window that central configuration size is 5 × 5, obtains structural region;
The part other than aggregation zone and structural region will 2a6) be removed in sketch map as can not sketch region;
2a7) by sketch map aggregation zone, structural region and can not sketch region merging technique, obtain include aggregation zone,
The administrative division map of structural region and the synthetic aperture radar SAR image without sketch line region.
Compared with prior art, the present invention at least has the advantages that due to the present invention to mixing aggregated structure
In image sub-prime space A1 when each extremely not homogeneous region collecting sample block, by extract bilateral aggregation properties sketch line segment with
The rectangular block that its k nearest neighbor is constituted is as ground object structure, so that the block that the present invention selects more can comprehensively represent extremely not homogeneous area
The sample block in domain;Moreover, the present invention acquires sample to extremely not homogeneous region each in mixing aggregated structure atural object pixel subspace A1
When this block, rectangular block is expanded, so that the ground object structure that rectangular block represents in the present invention is more complete, sample set can be made
In sample block include more comprehensively and more representative structure;The present invention by positioning bilateral aggregation properties sketch line segment with
The rectangular block that its sketch line segment for constituting k nearest neighbor collectively forms, can effectively extract more representative in extremely not homogeneous region
Ground object structure (i.e. rectangular block), is expanded by the structure for being included to rectangular block, so that the ground that the image block obtained includes
Object structure is more perfect.
Sample size of the invention has the number of sketch line segment of bilateral aggregation properties related with sketch map, sketch line
Section is a kind of rarefaction representation to SAR image, and number is it is believed that in sketch line segment there is bilateral to assemble sketch characteristic
The number of sketch line segment be also it is believed that in the case, the number of the rectangular block comprising ground object structure then can be true
It is fixed, finally, the number of obtained sample block is also it is believed that therefore can be preferable by expanding rectangular block
Control the cost of e-learning or training.
Detailed description of the invention
Fig. 1 is workflow schematic diagram of the invention;
Fig. 2 is the sketch map extracted in the present invention, administrative division map and mixing aggregated structure atural object pixel subspace schematic diagram;Its
In, (a) is the original image of SAR image Pyramid figure;(b) sketch map to be extracted to (a);(c) for sketch map shown in (b)
Carry out the administrative division map of mixing aggregated structure atural object pixel subspace obtained after compartmentalization;(d) being will be in administrative division map shown in (c)
After gray area is mapped to (a), obtained Pyramid mixing aggregated structure atural object pixel subspace figure.
Fig. 3 is respectively to obtained mixing aggregated structure after the sample set that selects of the present invention and random sample collection training
Image sub-prime space segmentation result figure, wherein (a) is that the sample set selected using the present invention carries out the mixing that feature learning obtains
The segmentation result figure of aggregated structure atural object pixel subspace;It (b) is that the mixing that feature learning obtains is carried out using random sample collection
The segmentation result figure of aggregated structure atural object pixel subspace.
Specific embodiment
The present invention is described in further detail with reference to the accompanying drawing.
Referring to Fig.1, steps are as follows for realization of the invention.
Step 1, to SAR image sketch, the specific steps are as follows:
Synthetic aperture radar SAR image 1a) is inputted, the sketch model of synthetic aperture radar SAR image is established:
1a1) in [100,150] range, a number, the sum as template are arbitrarily chosen;
A template for 1a2) constructing 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, the weighting of each pixel in calculation template
Coefficient, the weighting coefficient of all pixels point in statistical mask, wherein scale number value is 3~5, and direction number value is 18;
1a3) according to the following formula, pixel in synthetic aperture radar SAR image corresponding with template area coordinate is calculated
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;
1a4) according to the following formula, pixel in synthetic aperture radar SAR image corresponding with template area coordinate is calculated
Variance yields:
Wherein, ν indicates the variance of all pixels point in synthetic aperture radar SAR image corresponding with template area coordinate
Value;
1a5) according to the following formula, the response that each pixel in synthetic aperture radar SAR image is directed to ratio operator is calculated:
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;
1a6) according to the following formula, the response that each pixel in synthetic aperture radar SAR image is directed to correlation operator is calculated:
Wherein, C indicates that for the response of correlation operator, a and b divide each pixel in synthetic aperture radar SAR image
It Biao Shi not two different zones, ν in templateaIndicate the variance yields of all pixels point in a of template area, νbIt indicates in the b of template area
The variance yields of all pixels point, μaIndicate the mean value of all pixels point in a of template area, μbIndicate all pixels in the b of template area
The mean value of point;
1a7) according to the following formula, the response that each pixel in synthetic aperture radar SAR image is directed to each template is calculated:
Wherein, F indicates that for the response of each template, R and C divide each pixel in synthetic aperture radar SAR image
Not Biao Shi in synthetic aperture radar SAR image pixel for pixel needle in ratio operator and synthetic aperture radar SAR image
To the response of correlation operator;
1a8) judge whether constructed template is equal to the sum of selected template, if it is not, then returning to 1a2), otherwise,
Execute 1a9);
1a9) template of the selection with maximum response, the mould as synthetic aperture radar SAR image from each template
Plate, and using the maximum response of the template as the intensity of pixel in synthetic aperture radar SAR image, by the direction of the template
As the direction of pixel in synthetic aperture radar SAR image, the sideline response diagram and ladder of synthetic aperture radar SAR image are obtained
Degree figure;
1a10) according to the following formula, the intensity value for calculating synthetic aperture radar SAR image intensity map, obtains intensity map:
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;
Non-maxima suppression method 1a11) is used, intensity map is detected, suggestion sketch is obtained;
The pixel suggested in sketch with maximum intensity 1a12) is chosen, will suggest the picture in sketch with the maximum intensity
The pixel of vegetarian refreshments connection connects to form suggestion line segment, obtains suggestion sketch map;
1a13) according to the following formula, the code length gain for suggesting sketch line in sketch map is calculated:
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;
1a14) in [5,50] range, a number is arbitrarily chosen, as threshold value T;
The suggestion sketch line for 1a15) selecting CLG > T in all suggestion sketch lines, is combined into synthetic aperture radar SAR
The sketch map of image;
The sketch map of synthetic aperture radar SAR image 1b) is extracted from sketch model: scheming synthetic aperture radar SAR
As being input in sketch model, the sketch map B1 of synthetic aperture radar SAR image is obtained.
Step 2, pixel space is divided.
Sketch line fields method 2a) is used, compartmentalization processing is carried out 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, step is such as
Under:
2a1) according to the concentration class of sketch line segment in the sketch map of synthetic aperture radar SAR image, sketch line is divided into
Indicate aggregation atural object aggregation sketch line and expression boundary, line target, the boundary sketch line of isolated target, line target sketch line,
Isolated target sketch line;
2a2) according to the statistics with histogram of sketch line segment concentration class, the sketch line segment that concentration class is equal to optimal concentration class is chosen
As seed line-segment sets { Ek, k=1,2 ..., m }, wherein EkIndicate that any bar sketch line segment in seed line-segment sets, k indicate
The label of any bar sketch line segment in seed line-segment sets, m indicate the total number of seed line segment, and { } indicates set operation;
2a3) 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;
The round primitive that a radius is the optimal concentration class section upper bound 2a4) is constructed, with the circle primitive to line segment aggregate
In line segment expanded, the line segment aggregate ecto-entad after expansion is corroded, obtained in sketch map be with sketch point
The aggregation zone of unit;
2a5) to the sketch line for indicating boundary, line target and isolated target, it is with each sketch point of each sketch line
The geometry window that central configuration size is 5 × 5, obtains structural region;
The part other than aggregation zone and structural region will 2a6) be removed in sketch map as can not sketch region;
2a7) by sketch map aggregation zone, structural region and can not sketch region merging technique, obtain include aggregation zone,
The administrative division map of structural region and the synthetic aperture radar SAR image without sketch line region;
Step 3, aggregation zone is that those in sketch map B1 spatially have the unilateral element gathered with bilateral aggregation properties
Line segment composition is retouched, some aggregation zone is phase on spatial position with some extremely not homogeneous region in SAR image pixel space
It is corresponding.
The specific implementation of this step " is based on Primal Sketch Map and semantic information in Master's thesis referring to Yuan Jia woods
The SAR image of classification is divided " proposed in model.
Step 4, it is assumed that some aggregation zone is indicated with C1, extremely not homogeneous region corresponding with C1 is indicated with S1.In C1
In, all sketch line segments with bilateral aggregation properties are extracted, the sketch line segment composition of these sketch line segments and k nearest neighbor is solved
Rectangle set of blocks D1, in an embodiment of the present invention, K value takes 5.
Rectangular coordinate system 4a) is established using the column of SAR image sketch map as y-axis with the behavior x-axis of SAR image sketch map,
SAR image sketch map is placed in coordinate system, define the first sketch point in the SAR image sketch map upper left corner coordinate be (1,
1);
4b) look for the coordinate for seeing current bilateral aggregation two endpoints of sketch line segment, respectively starting point coordinate (start_x,
Start_y) with terminal point coordinate (end_x, end_y);
It 4c) looks for and sees that the starting point for the other 5 sketch line segments for constituting k nearest neighbor (K=5) with current bilateral aggregation sketch line segment is sat
Mark and terminal point coordinate are respectively: (start_x1, start_y1) and (end_x1, end_y1), (start_x2, start_y2)
With (end_x2, end_y2), (start_x3, start_y3) and (end_x3, end_y3), (start_x4, start_y4) with
(end_x4, end_y4), (start_x5, start_y5) and (end_x5, end_y5);
4d) by 4b) and 4c) in 6 starting point coordinates and 6 terminal point coordinates the value of all x coordinates be stored in vector X,
The value of all y-coordinates is stored in vector Y, and acquires the smallest value x_min and maximum value x_max, vector Y in vector X
In the smallest value y_min and maximum value y_max;
It is 4e) coordinate in the rectangular block upper left corner with (x_min, y_min), and with max { (x_max-x_min), (y_max-
Y_min it) } as the width of rectangular block and height, is sampled on the aggregation zone of SAR image, collected rectangular block is placed on
Together, rectangle set of blocks D1 is obtained.
Step 5, rectangle set of blocks D1 corresponding to each extremely not homogeneous region expands, and obtained image block is placed on
Together, the image block sample set D2 in each extremely not homogeneous region is constituted.
To the size for 32 × 32 being image block, to all rectangular blocks in rectangle set of blocks D1, respectively with a rectangular block left side
Upper angular coordinate is the coordinate in the upper left corner for the image block that will be acquired, i.e., (x_min, y_min), with rectangular block bottom right angular coordinate work
For the coordinate in the image block lower right corner, i.e. (x_max-31, y_max-31), using the coordinate in the rectangular block lower left corner as image block lower-left
The coordinate at angle, i.e. (x_max-31, y_min), using the upper right angular coordinate of rectangular block as the upper right angular coordinate of image block, i.e. (x_
Min, y_max-31), using the center of rectangular block as the center of image block, i.e. (x_min/2+x_max-15, y_min/2+y_
Max-15), sampled on the extremely not homogeneous region of SAR image, acquired image block is put together, obtains each pole not
The sample set of blocks D2 of homogenous region.
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 Pyramid that X-band resolution ratio is 1 meter schemes.
2. emulation content:
Emulation 1 is schemed mixing aggregated structure atural object pixel subspace to SAR image Pyramid and is emulated, as a result such as Fig. 2,
Wherein:
Fig. 2 (a) is the original image of SAR image Pyramid figure;
Fig. 2 (b) is the sketch map extracted to Fig. 2 (a);
Fig. 2 (c) is mixing aggregated structure atural object pixel for carrying out obtaining after compartmentalization to sketch map shown in Fig. 2 (b)
The administrative division map in space;
Fig. 2 (d) is after the gray area in administrative division map shown in Fig. 2 (c) is mapped to Fig. 2 (a), and obtained Pyramid is mixed
Close aggregated structure atural object pixel subspace figure.
Emulation 2 constructs sample set with method of the invention, and correspondingly to the extremely not homogeneous region each of Fig. 2 (d)
Each extremely not homogeneous region is constituted every the sample block for randomly choosing same number in the sample set that the sampling of 1 sliding window is constituted with press proof
This collection;
Using in number of patent application 201710702367.1 based on G0The stochastic gradient variation Bayesian model of distribution point
The other sample set constructed to the present invention and random sample collection carry out feature learning, obtain respective structural eigenvector, then sharp
Respective structural eigenvector is clustered with hierarchical clustering, obtains the segmentation result of mixed pixel subspace, as a result as schemed
3, in which:
Fig. 3 (a) is that the sample set selected using the present invention carries out the mixing aggregated structure atural object pixel that feature learning obtains
The segmentation result figure of subspace;
Fig. 3 (b) is to carry out the mixing aggregated structure atural object pixel subspace that feature learning obtains using random sample collection
Segmentation result figure.
3. simulated effect is analyzed:
By the comparison segmentation result figure of Fig. 3 (a) and Fig. 3 (b) it may be concluded that the method for the present invention can obtain mixing aggregation
Structurally more comprehensively and more representative sample block, therefore the feature learnt is more preferable, is conducive to improve in image sub-prime space
Mix the accuracy of aggregated structure atural object pixel subspace segmentation.
Claims (6)
1. a kind of SAR image sample block selection method based on sketch line segment aggregation properties, which is characterized in that including following step
It is rapid:
Step 1): input synthetic aperture radar SAR image establishes the sketch model of synthetic aperture radar SAR image, from sketch mould
The sketch map B1 of synthetic aperture radar SAR image is extracted in type;
Step 2): compartmentalization processing is carried out to the sketch map B1 for the SAR image that step 1) obtains, obtains synthetic aperture radar SAR
The administrative division map of image, administrative division map are formed by aggregation zone, structural region and without sketch line segment region;
Step 3): aggregation zone is the sketch spatially by those in sketch map B1 with unilateral aggregation and bilateral aggregation properties
Line segment composition, there are corresponding extremely not homogeneous regions in SAR image pixel space for each aggregation zone;
Step 4): all sketch line segments in each aggregation zone with bilateral aggregation properties are extracted, these sketch line segments are solved
The rectangle set of blocks D1 constituted with the sketch line segment of k nearest neighbor;
Step 5): for the rectangle set of blocks D1 of aggregation zone, according to the coordinate of each rectangular block in rectangle set of blocks D1, with
Collecting sample block in the corresponding extremely not homogeneous region of aggregation zone, collected sample block is put together, constitutes each pole not
The sample set of blocks D2 of homogenous region.
2. a kind of SAR image sample block selection method based on sketch line segment aggregation properties according to claim 1, special
Sign is, the step 4) specifically includes the following steps:
Rectangular coordinate system 4a) is established using the column of SAR image sketch map as y-axis with the behavior x-axis of SAR image sketch map, by SAR
Image sketch map is placed in coordinate system, and the coordinate for defining the first sketch point in the SAR image sketch map upper left corner is (1,1);
4b) find the coordinate of current bilateral aggregation two endpoints of sketch line segment, respectively starting point coordinate (start_x, start_y)
With terminal point coordinate (end_x, end_y);
4c) find the starting point coordinate and terminal point coordinate that other sketch line segments of k nearest neighbor are constituted with current bilateral aggregation sketch line segment;
4d) value of all x coordinates in starting point coordinate in step 4b) and step 4c) and terminal point coordinate is stored in vector X, institute
There is the value of y-coordinate to be stored in vector Y, and acquires the smallest value x_min and maximum value x_max and vector Y in vector X
In the smallest value y_min and maximum value y_max;
It is 4e) coordinate in the rectangular block upper left corner with (x_min, y_min), and with max { (x_max-x_min), (y_max-y_
Min it) } as the width of rectangular block and height, is sampled on the aggregation zone of SAR image, collected rectangular block is placed on one
It rises, obtains rectangle set of blocks D1.
3. a kind of SAR image sample block selection method based on sketch line segment aggregation properties according to claim 1, special
Sign is that K value is 5 in the step 4).
4. a kind of SAR image sample block selection method based on sketch line segment aggregation properties according to claim 1, special
Sign is, the step 5) specifically includes the following steps:
The size for being image block with 32 × 32, to all rectangular blocks in rectangle set of blocks D1, respectively with rectangular block upper left corner seat
It is designated as the coordinate in the upper left corner for the image block to be acquired, using rectangular block bottom right angular coordinate as the coordinate in the image block lower right corner,
Using the coordinate in the rectangular block lower left corner as the coordinate in the image block lower left corner, upper right of the upper right angular coordinate of rectangular block as image block
Angular coordinate is sampled on the extremely not homogeneous region of SAR image, will be adopted using the center of rectangular block as the center of image block
The image block collected is put together, and the sample set of blocks D2 in each extremely not homogeneous region is obtained.
5. a kind of SAR image sample block selection method based on sketch line segment aggregation properties according to claim 1, special
Sign is that the step 1) embodies the following steps are included: 1a) input synthetic aperture radar SAR image, establish synthetic aperture thunder
Up to the sketch model of SAR image:
1a1) in [100,150] range, a number, the sum as template are arbitrarily chosen;
A template for 1a2) constructing the side being made of pixel with different directions and scale, line, utilizes the direction of template
With dimensional information structural anisotropy's Gaussian function, by the Gaussian function, the weighting coefficient of each pixel in calculation template,
The weighting coefficient of all pixels point in statistical mask, wherein scale number value is 3~5, and direction number value is 18;
1a3) according to the following formula, the mean value of pixel in synthetic aperture radar SAR image corresponding with template area coordinate is calculated:
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;
1a4) according to the following formula, the variance of pixel in synthetic aperture radar SAR image corresponding with template area coordinate is calculated
Value:
Wherein, ν indicates the variance yields of all pixels point in synthetic aperture radar SAR image corresponding with template area coordinate;
1a5) according to the following formula, the response that each pixel in synthetic aperture radar SAR image is directed to ratio operator is calculated:
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;
1a6) according to the following formula, the response that each pixel in synthetic aperture radar SAR image is directed to correlation operator is calculated:
Wherein, C indicates that for each pixel for the response of correlation operator, a and b distinguish table in synthetic aperture radar SAR image
Show two different zones, ν in templateaIndicate the variance yields of all pixels point in a of template area, νbIndicate own in the b of template area
The variance yields of pixel, μaIndicate the mean value of all pixels point in a of template area, μbIndicate all pixels point in the b of template area
Mean value;
1a7) according to the following formula, the response that each pixel in synthetic aperture radar SAR image is directed to each template is calculated:
Wherein, F indicates that for each pixel for the response of each template, R and C distinguish table in synthetic aperture radar SAR image
Show in synthetic aperture radar SAR image pixel for pixel in ratio operator and synthetic aperture radar SAR image for phase
The response of closing property operator;
1a8) judge whether constructed template is equal to the sum of selected template, if it is not, then returning to 1a2), otherwise, execute
1a9);
1a9) template of the selection with 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, using the direction of the template as
The direction of pixel in synthetic aperture radar SAR image obtains the sideline response diagram and gradient of synthetic aperture radar SAR image
Figure;
1a10) according to the following formula, the intensity value for calculating synthetic aperture radar SAR image intensity map, obtains intensity map:
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;
Non-maxima suppression method 1a11) is used, intensity map is detected, suggestion sketch is obtained;
The pixel suggested in sketch with maximum intensity 1a12) is chosen, will suggest the pixel in sketch with the maximum intensity
The pixel of connection connects to form suggestion line segment, obtains suggestion sketch map;
1a13) according to the following formula, the code length gain for suggesting sketch line in sketch map is calculated:
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;
1a14) in [5,50] range, a number is arbitrarily chosen, as threshold value T;
The suggestion sketch line for 1a15) selecting CLG > T in all suggestion sketch lines, is combined into synthetic aperture radar SAR image
Sketch map;
The sketch map of synthetic aperture radar SAR image 1b) is extracted from sketch model: i.e. that synthetic aperture radar SAR image is defeated
Enter into sketch model, obtains the sketch map B1 of synthetic aperture radar SAR image.
6. a kind of SAR image sample block selection method based on sketch line segment aggregation properties according to claim 1, special
Sign is, the step 2) specifically includes the following steps:
Sketch line fields method 2a) is used, compartmentalization processing is carried out to the sketch map of synthetic aperture radar SAR image, is obtained
Administrative division map including aggregation zone, its step are as follows:
2a1) 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;
2a2) 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;
2a3) 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 2a4) 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;
2a5) 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 2a6) be removed in sketch map as can not sketch region;
2a7) by sketch map aggregation zone, structural region and can not sketch region merging technique, obtain including aggregation zone, structure
The administrative division map in region and the synthetic aperture radar SAR image without sketch line region.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810670378.0A CN108986108B (en) | 2018-06-26 | 2018-06-26 | SAR image sample block selection method based on sketch line segment aggregation characteristics |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810670378.0A CN108986108B (en) | 2018-06-26 | 2018-06-26 | SAR image sample block selection method based on sketch line segment aggregation characteristics |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108986108A true CN108986108A (en) | 2018-12-11 |
CN108986108B CN108986108B (en) | 2022-04-19 |
Family
ID=64538364
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810670378.0A Active CN108986108B (en) | 2018-06-26 | 2018-06-26 | SAR image sample block selection method based on sketch line segment aggregation characteristics |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108986108B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110188682A (en) * | 2019-05-30 | 2019-08-30 | 西安电子科技大学 | Remote sensing image object detection method based on geometry two-way convolutional network |
CN111402232A (en) * | 2020-03-16 | 2020-07-10 | 深圳市瑞图生物技术有限公司 | Method for detecting sperm aggregation in semen |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101297189A (en) * | 2005-10-26 | 2008-10-29 | 通用电气公司 | Methods and systems for delivery of fluidic samples to sensor arrays |
US20120141021A1 (en) * | 2006-03-23 | 2012-06-07 | Intelliscience Corporation | Methods and systems for data analysis and feature recognition |
CN103198479A (en) * | 2013-03-27 | 2013-07-10 | 西安电子科技大学 | SAR image segmentation method based on semantic information classification |
CN105787450A (en) * | 2016-02-26 | 2016-07-20 | 中国空间技术研究院 | Urban area building detection method based on high-resolution SAR image |
CN106611420A (en) * | 2016-12-30 | 2017-05-03 | 西安电子科技大学 | SAR image segmentation method based on deconvolution network and sketch direction constraint |
CN107239797A (en) * | 2017-05-23 | 2017-10-10 | 西安电子科技大学 | Polarization SAR terrain classification method based on full convolutional neural networks |
CN107403434A (en) * | 2017-07-28 | 2017-11-28 | 西安电子科技大学 | SAR image semantic segmentation method based on two-phase analyzing method |
CN107464247A (en) * | 2017-08-16 | 2017-12-12 | 西安电子科技大学 | One kind is based on G0Stochastic gradient variation Bayes's SAR image segmentation method of distribution |
-
2018
- 2018-06-26 CN CN201810670378.0A patent/CN108986108B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101297189A (en) * | 2005-10-26 | 2008-10-29 | 通用电气公司 | Methods and systems for delivery of fluidic samples to sensor arrays |
US20120141021A1 (en) * | 2006-03-23 | 2012-06-07 | Intelliscience Corporation | Methods and systems for data analysis and feature recognition |
CN103198479A (en) * | 2013-03-27 | 2013-07-10 | 西安电子科技大学 | SAR image segmentation method based on semantic information classification |
CN105787450A (en) * | 2016-02-26 | 2016-07-20 | 中国空间技术研究院 | Urban area building detection method based on high-resolution SAR image |
CN106611420A (en) * | 2016-12-30 | 2017-05-03 | 西安电子科技大学 | SAR image segmentation method based on deconvolution network and sketch direction constraint |
CN107239797A (en) * | 2017-05-23 | 2017-10-10 | 西安电子科技大学 | Polarization SAR terrain classification method based on full convolutional neural networks |
CN107403434A (en) * | 2017-07-28 | 2017-11-28 | 西安电子科技大学 | SAR image semantic segmentation method based on two-phase analyzing method |
CN107464247A (en) * | 2017-08-16 | 2017-12-12 | 西安电子科技大学 | One kind is based on G0Stochastic gradient variation Bayes's SAR image segmentation method of distribution |
Non-Patent Citations (5)
Title |
---|
BYONGJUN (PHIL) HWANG等: "A practical algorithm for the retrieval of floe size distribution of Arctic sea ice from high-resolution satellite Synthetic Aperture Radar imagery", 《ELEM SCI ANTH》 * |
JIE WU ET AL: "Local Maximal Homogeneous Region Search for SAR Speckle Reduction With Sketch-Based Geometrical Kernel Function", 《IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING》 * |
王阳阳: "基于反卷积映射推理网络的SAR图像分割", 《中国优秀硕士学位论文全文数据库 (信息科技辑)》 * |
赵昊等: "相似度自适应样本块的高分辨SAR图像阴影修复", 《中国图象图形学报》 * |
门龙生: "基于素描线补全策略和区域特征学习的SAR图像分割", 《万方数据知识服务平台》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110188682A (en) * | 2019-05-30 | 2019-08-30 | 西安电子科技大学 | Remote sensing image object detection method based on geometry two-way convolutional network |
CN111402232A (en) * | 2020-03-16 | 2020-07-10 | 深圳市瑞图生物技术有限公司 | Method for detecting sperm aggregation in semen |
CN111402232B (en) * | 2020-03-16 | 2023-05-16 | 深圳市瑞图生物技术有限公司 | Sperm aggregation detection method in semen |
Also Published As
Publication number | Publication date |
---|---|
CN108986108B (en) | 2022-04-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108596055B (en) | Airport target detection method of high-resolution remote sensing image under complex background | |
Huang et al. | A new building extraction postprocessing framework for high-spatial-resolution remote-sensing imagery | |
CN105374033B (en) | SAR image segmentation method based on ridge ripple deconvolution network and sparse classification | |
CN104408482B (en) | A kind of High Resolution SAR Images object detection method | |
CN106611420B (en) | The SAR image segmentation method constrained based on deconvolution network and sketch map direction | |
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 | |
CN106683102B (en) | SAR image segmentation method based on ridge ripple filter and convolutional coding structure learning model | |
CN106611423B (en) | SAR image segmentation method based on ridge ripple filter and deconvolution structural model | |
CN106846322A (en) | Based on the SAR image segmentation method that curve wave filter and convolutional coding structure learn | |
CN106408030A (en) | SAR image classification method based on middle lamella semantic attribute and convolution neural network | |
CN106611421A (en) | SAR image segmentation method based on feature learning and sketch line constraint | |
CN105335975B (en) | Polarization SAR image segmentation method based on low-rank decomposition and statistics with histogram | |
CN105389799B (en) | SAR image object detection method based on sketch map and low-rank decomposition | |
CN106651884B (en) | Mean field variation Bayes's SAR image segmentation method based on sketch structure | |
JP2018072938A (en) | Number-of-targets estimation device, number-of-targets estimation method, and program | |
CN113223042B (en) | Intelligent acquisition method and equipment for remote sensing image deep learning sample | |
CN106611422A (en) | Stochastic gradient Bayesian SAR image segmentation method based on sketch structure | |
CN109034258A (en) | Weakly supervised object detection method based on certain objects pixel gradient figure | |
CN108734200A (en) | Human body target visible detection method and device based on BING features | |
Yang et al. | Toward country scale building detection with convolutional neural network using aerial images | |
CN108986108A (en) | A kind of SAR image sample block selection method based on sketch line segment aggregation properties | |
CN111950345A (en) | Camera identification method and device, electronic equipment and storage medium | |
CN105389798B (en) | SAR image segmentation method based on deconvolution network Yu mapping inference network | |
Lv et al. | Improved object-based convolutional neural network (IOCNN) to classify very high-resolution remote sensing images |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |